<%BANNER%>

Measuring Post-Stroke Arm Motor Ability: Measurement Properties of the Fugl-Meyer Upper Extremity Assessment Examined wi...

xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID E20101219_AAAADU INGEST_TIME 2010-12-19T20:54:09Z PACKAGE UFE0015615_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES
FILE SIZE 46649 DFID F20101219_AACBKS ORIGIN DEPOSITOR PATH woodbury_m_Page_168.pro GLOBAL false PRESERVATION BIT MESSAGE_DIGEST ALGORITHM MD5
28abe0fd24171367e28a47d4b8c1fee4
SHA-1
4248694d12159893767e36113ac68d82f4c845ae
29194 F20101219_AACCHL woodbury_m_Page_173.QC.jpg
87236f96aeff9a9d79d56a732986f8cf
e4f51a523241b1042a22d910f3a0d1b42d0ea020
120184 F20101219_AACANZ woodbury_m_Page_219.jpg
c6fb3bae09728c2ca717c0a1c3d6c254
08fad574641b340b802740de6bfc452eddbe6a39
49892 F20101219_AACBKT woodbury_m_Page_169.pro
bf582fd042623516480c3193e14afb0f
e9f2b393f729c2906d07de0dfb959d8ad1e8156b
7402 F20101219_AACCHM woodbury_m_Page_173thm.jpg
835401d31fb356149fb9f43eb6080cd1
f23745d8a63ed8fa297a117cbc668e2cc0e396ff
40950 F20101219_AACBKU woodbury_m_Page_170.pro
8306f74d4a98b7300d4d6256badb3ca8
c23b5fba4515e6d18836ad920da1a2f39aae69d3
8246 F20101219_AACCHN woodbury_m_Page_174thm.jpg
a4cf80139290a05a713a92003f7b5883
1c4115576b5f0269a538f5c712ba56d3622dc2b2
44932 F20101219_AACBKV woodbury_m_Page_171.pro
09ea622018ed3915d9a19b4f3fcfcce2
9de9e237cba3157ce4b5f35c535fa9406a8eb062
26754 F20101219_AACCHO woodbury_m_Page_175.QC.jpg
dec825580221ee6b33845068092fce7b
b53c1d43baecaf0c90f843c302d9f1b7d9382226
44980 F20101219_AACBKW woodbury_m_Page_172.pro
d645619279d8eb8880925d3dbe76d404
7e1e5ec2d893a8e21ce074e01a3bab3c168a10d6
6964 F20101219_AACCHP woodbury_m_Page_175thm.jpg
512ab6ac9902dae4a084ec9401f1dfc1
6af5b906c7f067b6ae164c7ca063de25b412c29a
44385 F20101219_AACBKX woodbury_m_Page_173.pro
c228f7b20da82c615151646697ddefdf
9ed130999b112a0148ac99174b24138f13080cb4
7795 F20101219_AACCHQ woodbury_m_Page_176thm.jpg
f17ca9f6a9521169bc8728a2b9111882
2436113993d9890b0947a5090882bd79fae965a6
8071 F20101219_AACCHR woodbury_m_Page_177thm.jpg
c8a35e915c5e4cb98d82e437fba7c184
33945354601ca75d6b41b32eee120d979d1e7e29
49885 F20101219_AACBKY woodbury_m_Page_174.pro
0f2cde5669180cdfb4a69060aa288ab5
35e0a6d5802a005bede474999c4e311d1f7a8640
34073 F20101219_AACCHS woodbury_m_Page_178.QC.jpg
53c296e7933b8fcd6ffcc13247d83a13
ebc5fc0cbfc8583ebec4a6295f489e1d732f6319
41601 F20101219_AACBKZ woodbury_m_Page_175.pro
3ef66e93b723cdaa418572ba32a1163f
2f7e1ed2186d74c9aebc53cc8c8dfd18a1860026
8138 F20101219_AACCHT woodbury_m_Page_178thm.jpg
f39f836cc94b7adf971502acbbaa6cbf
3676abce13da78a9363344c5511b7f15845a6b86
107495 F20101219_AACATA woodbury_m_Page_129.jp2
812c4a158d0200bd15acac85e5bf1b7a
a22d9a62e581420be54da9591a59154f486e6f50
32763 F20101219_AACCHU woodbury_m_Page_179.QC.jpg
7caabd5971ae289eb193ca132b3577f5
f239eadf46fd27789dde7dc716a91a26ef2a70a4
109405 F20101219_AACATB woodbury_m_Page_130.jp2
1cee4a5cc3f8a4a5a24842ec4929b103
2eebbb90024352ebccef6983fe7f6e417d551810
8433 F20101219_AACCHV woodbury_m_Page_180thm.jpg
6114c7aa6c0b95e9fdf081172d98d441
8c05ac17326c4e86a29a47322523e5b21f29adb2
22957 F20101219_AACCHW woodbury_m_Page_181.QC.jpg
3152dde9c399c41b3e95cc525d6d268b
9dc058d597daf7ca843892b16cfa24da620424b4
105355 F20101219_AACATC woodbury_m_Page_131.jp2
bdc4453971aa35eeb6b4327c73a60038
530e44bc8ec1f517346dac898133df931048eeaf
5794 F20101219_AACCHX woodbury_m_Page_181thm.jpg
2ceecdb161206a61e207257d6d3d6471
ee2fceedde56ffa54d9bd3e8bf66a2295d93377b
104203 F20101219_AACATD woodbury_m_Page_132.jp2
edc56cc5be5f321a97db17c2a86f8338
3d36d663487ef4f48a6474b31b3d2ca0e496c478
23130 F20101219_AACCHY woodbury_m_Page_182.QC.jpg
e218d7cb868792de0f8c0e460aa78b51
94ce0f374ded36c1eba7d8493cb2c891c2e44e5b
105577 F20101219_AACATE woodbury_m_Page_133.jp2
1adf223257636bff7feaf55dce2102fc
29f9621c001bfdf19f04d3511798b58afb9728a5
5319 F20101219_AACCHZ woodbury_m_Page_182thm.jpg
5ae29cff614e8c97076f5407b749d558
cd0c0949f28496413c7ca34a8b0eb3b660bb787c
107149 F20101219_AACATF woodbury_m_Page_134.jp2
64454ed652bb2aa9f0aa0d1267af3340
dc522c87181a9073c7365cb389a3f09896240bb6
111971 F20101219_AACATG woodbury_m_Page_135.jp2
d6a08b287d89f383a979b6fa69387410
f76ad557d3099e982cd2445ca9e111bd9d6cfd2f
2060 F20101219_AACBQA woodbury_m_Page_089.txt
3d5f4fc416a43ef5c4cbd6a4d48f44c6
152abc38e51285b95d82085af46aa497502da546
113337 F20101219_AACATH woodbury_m_Page_136.jp2
d9500762ef825da852c7abf07155a37d
b56b7052a1ef70810618850eeddf3d82eb333bbe
436 F20101219_AACBQB woodbury_m_Page_090.txt
ed48a8fb03c193c106e2b94e58ddd2a2
3b5d75c5af5a39ef159a9cb8046666146b2823f6
110245 F20101219_AACATI woodbury_m_Page_137.jp2
0e7046a90273428125910662384b12cd
6022e626c1f8fac3a738a1639005221a6fa19214
1396 F20101219_AACBQC woodbury_m_Page_091.txt
ce68b20fb1ca8741e019148d1b1eaf79
5c855afb93136bc8a879bc818f9059272fb7bb2b
100593 F20101219_AACATJ woodbury_m_Page_138.jp2
6bb3bbcbf6b47b94c4f0b8f1f40969e9
ae732ac531b9ed43325e83319c66586330381145
1912 F20101219_AACBQD woodbury_m_Page_092.txt
0a449c094c4d9b19b4c4776aaa16aae4
8d0e661222e002b4dacc64f277fc99416a984a87
103831 F20101219_AACATK woodbury_m_Page_139.jp2
3be59f7ccd7ba0637590f646046780e9
3d8b4e99d2f0e0b5249529e16da318bae8f82f82
2227 F20101219_AACBQE woodbury_m_Page_093.txt
656cd5220bc54958ffbafa7faa0fa842
88c800a2150bf2a08c27b8a959cec8dabba7b644
105461 F20101219_AACAGA woodbury_m_Page_040.jp2
d41cf1c0b52708cafb52e1eb37ff659e
1197a5031ce8eb0e830795e38d5bbbc488a3d1fb
105387 F20101219_AACATL woodbury_m_Page_140.jp2
652425b88d52e038bf77c1d6de0461a5
3a6c2f047c6e322ff4e5593f74645da540fcb6a4
1463 F20101219_AACBQF woodbury_m_Page_094.txt
4456b5e5f78c43d888de09a1e91e3ab7
7fcc6da79895711340db1086452ffc51d9b64bfe
32504 F20101219_AACAGB woodbury_m_Page_169.QC.jpg
e1b3922503dc1b201349b5926f9433af
3e39bd1eb2aa7a339f5c5819864482480ec00aa0
115200 F20101219_AACATM woodbury_m_Page_142.jp2
3d047c6e734d44fd8ec5ef2011ea78c9
70a96680382ef68f743ad55e1177cf49b0eb26fc
4587 F20101219_AACBQG woodbury_m_Page_095.txt
b4d12997798c0efe494465557f64a230
e49f3379609963b2e01e82639e7cd11a6a063adb
342382 F20101219_AACAGC UFE0015615_00001.xml FULL
a54113c5567fe2159147461ded82542b
fb5f13c3845ec07610e7ec694aab7f3d78f038b5
BROKEN_LINK
woodbury_m_Page_001.tif
109535 F20101219_AACATN woodbury_m_Page_144.jp2
144cfbb818e678364407e58d2962c20a
794d3ce72537b4fe2ba7ec35874cc992618d58ec
105633 F20101219_AACATO woodbury_m_Page_145.jp2
71ef403e5d30e979705ec38169efd28c
8e768b7792014b92f2949dbbb136f01a498189fc
1823 F20101219_AACBQH woodbury_m_Page_096.txt
f147eeb563db6b293c15e3b9794e0f89
d7f3221675b8d1e265dfdae10cfab24b1821176a
99934 F20101219_AACATP woodbury_m_Page_146.jp2
df6f95b5178133f090dbe9a61282859f
b31cddecb2d9b2a451f8b2958c82818b05484d47
1875 F20101219_AACBQI woodbury_m_Page_097.txt
6c1819e3597c2b1bb3acf1d79bcbe2a3
779b3b489433327b0c82695bdccb190218811c4d
34889 F20101219_AACAGF woodbury_m_Page_001.jpg
512c2ee5ffa7874569b8f9a319dd5a52
a72d7fd9464dba3c48ff552d796b018d479933d2
102973 F20101219_AACATQ woodbury_m_Page_147.jp2
fb1d40bd9799a28f559b4631f76d9071
b53159428a455ccaa3947f7b1e4c81bded4e954c
2013 F20101219_AACBQJ woodbury_m_Page_098.txt
68a084c14895d863fa85bf5397c42253
e928cd8738e4020daefcb99a2826c86c364f20dc
5051 F20101219_AACAGG woodbury_m_Page_002.jpg
3ca62d65b6158d4c08f51e306aeb7251
7e4a86151451763db94887ed3d5105c5f73fd213
108738 F20101219_AACATR woodbury_m_Page_148.jp2
038975922bd4f498caf3378700b1fb21
9749b39c49a9dbbe5aaea6ecef3fd11d60c11263
1763 F20101219_AACBQK woodbury_m_Page_099.txt
41cac2397f9ff47dc4dc28f46e14a0ea
51cbf1f67d253fc9eb3b6baaf4095fe38383abf4
5486 F20101219_AACAGH woodbury_m_Page_003.jpg
70197b6e1b9db50e3df76c6453db1cc0
889f4c1839713076b45f1d093aa1d96b01b658e9
1053954 F20101219_AACBDA woodbury_m_Page_181.tif
aae44a3a786217201984c69cd60d8339
0f12dd8399f47f4320ca3f727d1873a9d64e885f
108731 F20101219_AACATS woodbury_m_Page_149.jp2
78ff2255366b58d97a206ff076a8f66f
939576f3a9bc1769d287d859de64c5dc33119aac
2008 F20101219_AACBQL woodbury_m_Page_100.txt
0e72fdf91b4822859b36e8da02c5d751
88889f9ce8bc49de63c72c60be2f572c4d3e451f
90977 F20101219_AACAGI woodbury_m_Page_004.jpg
e6295bccbd3943e4713d1064718028d0
9dfde40babfd7e7b865ba5bcea647bcb1399c3eb
8423998 F20101219_AACBDB woodbury_m_Page_182.tif
528939c234743ad57be5ff9dfd670401
defbe6fafc7851a6c4eb8e0482043dff188a9dfd
112568 F20101219_AACATT woodbury_m_Page_150.jp2
b4ce73173d3e579e858eef4223c8c020
f1b8a554add8c8e46d8a88aa6d3d561d1242619a
1740 F20101219_AACBQM woodbury_m_Page_102.txt
025c07870e94e20163ac640ac08a2b1b
93bacf8b5ef8ddc7abeb59382837831728c24164
102310 F20101219_AACAGJ woodbury_m_Page_005.jpg
3e4f0426faae2e2411380f4c4a053315
4029d179b19ad72bc495bec3fb57c0cde4e4d728
F20101219_AACBDC woodbury_m_Page_183.tif
5b6285c1ee46670755a4df049305372f
6bc62a629130b07b1ae77c856f1b6f842338da11
111116 F20101219_AACATU woodbury_m_Page_151.jp2
02dc2345a839d5a594feb585705a9a20
0eaccba31540557370f13b2410bf6627bfd97dd1
1753 F20101219_AACBQN woodbury_m_Page_103.txt
f6e5b75a9fa3cb3255b9829363aef443
647c64863244086174af38e7e70f4dde07fde424
108906 F20101219_AACAGK woodbury_m_Page_006.jpg
0b987cde6e934f340d414588320dc20c
45a7ee2ba46b371a83f347fde60af72e36e15983
F20101219_AACBDD woodbury_m_Page_184.tif
8c412539db272c29effc12e0fc16b2cf
8759e2ef65290a2346091bb19596f871eca445e6
109239 F20101219_AACATV woodbury_m_Page_152.jp2
b7190a9e5503515bc8a3ef06af97788b
69630c5af0d2c9552686c9085a283894708f0fa8
2019 F20101219_AACBQO woodbury_m_Page_104.txt
724742c8e8c15d6a724f1cb376c9b1ab
2b249ede99edd57ec23876470870592fa3f69e82
136154 F20101219_AACAGL woodbury_m_Page_007.jpg
47d1f8dc269cba32029dcba76ceaeb99
ce477ea3b2592e164b846354396afe97db37d216
25271604 F20101219_AACBDE woodbury_m_Page_185.tif
fe9aa3ed33949888baa1d39be7095245
41fef2cd67b6571683947a3716252ee1b863c66b
102521 F20101219_AACATW woodbury_m_Page_154.jp2
663620610103e18c214f1d65637d518c
67007cae7c09622fe0e50e1658e25cd002aae891
2101 F20101219_AACBQP woodbury_m_Page_105.txt
932e2c70bfa78a7fc55645f381b98664
0ffca6820179dfa0572619de5a9b1139bfcbcc88
157743 F20101219_AACAGM woodbury_m_Page_008.jpg
d1b62fdbcd7eb86e9321ee8173b09f4c
77e6118c9edd59e5d8ea681bfa8d561fded6d08d
F20101219_AACBDF woodbury_m_Page_186.tif
21864f5fd7e205a03ff5ca90856e39d8
86de77a6446d58aa759ba8f97360aa4fad7b0562
37477 F20101219_AACATX woodbury_m_Page_156.jp2
b634158a1f3e5db55584b14952cab93b
7e99254328540c1e6a4c0d8364429320b124b7b8
1956 F20101219_AACBQQ woodbury_m_Page_106.txt
cb97968619d988c67a9dbf5036121d9e
f8be4ea0acfbd6d11c0a838ade1cfff050f5a554
117288 F20101219_AACAGN woodbury_m_Page_009.jpg
47c4b3977fc88a65e8aa1c9af727b978
53babf4b23ecd264c58f2d84f7affe5e43e777c9
F20101219_AACBDG woodbury_m_Page_187.tif
ff9d54c9f87ce9be6ca9537edad59595
6b85d45414234d64e21e8b690d8adb7d7df40acd
305791 F20101219_AACATY woodbury_m_Page_157.jp2
95c481e7ee5325c117f791f779d757f6
147c622ead69c87abecd4e3d916d219863a22a54
2047 F20101219_AACBQR woodbury_m_Page_107.txt
eec0541de5ec8c284c31c619c43acb80
17a8105f6757619d278df6523479c8c90f472323
99896 F20101219_AACAGO woodbury_m_Page_010.jpg
0bc8c342b28086d25953617f91df3dcc
8672d544e13eb65ed935f95bbd38dbbc4e370de1
F20101219_AACBDH woodbury_m_Page_188.tif
0b8af69ed6d565cdadd7a35f7a97ecc6
59e371cf9e646e60557343ee20778db9f0d39604
470431 F20101219_AACATZ woodbury_m_Page_158.jp2
f02c19e6d3622c10ca2fe72b4183e56a
3f3aae10f8f0290a038636dbfa99175379ae4d73
29918 F20101219_AACCAA woodbury_m_Page_068.QC.jpg
06d5a953f3c5076d5cd7b202667b5f6f
8dfe0ea35dbc52fe9af44018e355afec9a41a163
2027 F20101219_AACBQS woodbury_m_Page_108.txt
d9c9d288b914fa41a716e522ef99d542
36e8e4611f3f9135d7f23b6c04e316045c3ef043
96876 F20101219_AACAGP woodbury_m_Page_012.jpg
ec90813d56c2eccfdd34ad1b17c7ba4e
45e9d8add0e4305fb59195b5570af8bef1b0f32c
F20101219_AACBDI woodbury_m_Page_189.tif
ccf5ec859f1c4423641bc16afc2ee4b5
0b7f25fe2b14290fd761ef38e1f7c10e4c0d6df0
7506 F20101219_AACCAB woodbury_m_Page_068thm.jpg
c88a7e07086b10ae07ed912cf3a6c5f4
0818657b35a0a20bd77bd952d7064cc9092f2823
1952 F20101219_AACBQT woodbury_m_Page_109.txt
3da26e076773784e7f37d18cce9bb203
e795bfb1ec5514702780a29ac344f03ad19544a8
87080 F20101219_AACAGQ woodbury_m_Page_013.jpg
80b95a888cc54329d057356b3a95525a
f6ca34a1d2aec2654f792498dc75dcda42c10c64
F20101219_AACBDJ woodbury_m_Page_190.tif
bcce1878ad1af58d41b4f353b3598104
8175c4597083e86e6cc8d3a6ffea78c07a481cdd
31299 F20101219_AACCAC woodbury_m_Page_069.QC.jpg
de84a7653ceece8c084dac46eb16c7d5
1da0996df5911e6fba51a2f47849c73f3c8ed454
1999 F20101219_AACBQU woodbury_m_Page_110.txt
2e27df60e08b15551d47338cd4eef6d2
134ee0c4dfd6e49ca51ae626bdf2c284c385d297
74475 F20101219_AACAGR woodbury_m_Page_014.jpg
48b94d256841ee029a01912a8d3ffa55
660a394fbb121e7252b59270761875dd8fbb8374
F20101219_AACBDK woodbury_m_Page_191.tif
56947f556ec21ead94f7ff017688323e
38e4c3774bdd2231faaf89a000d7b7b80f35c19b
7836 F20101219_AACCAD woodbury_m_Page_069thm.jpg
ba5ea0ec3b5175d904d93f37c4b1a8a0
905e19143709dac703d5a17e09d3bab0e8224fbb
2022 F20101219_AACBQV woodbury_m_Page_111.txt
93906216b6c4d6d67f70179a0d143b10
103a7499dccef58f8ff61891a1537c59260e543a
92866 F20101219_AACAGS woodbury_m_Page_015.jpg
2c85939533052101f2355f33b8eea309
fd921e9ac20a620cc5f3f20a75a7199787d75ec3
F20101219_AACBDL woodbury_m_Page_192.tif
e01a32abed7b3550255839f254169427
0cb38aedaca7228347bf2ffa9c73917454a14209
35197 F20101219_AACCAE woodbury_m_Page_070.QC.jpg
1be0eefdd24b4931b0e96821102a666f
9ad2bb446a89406acd8e12ffb1de85fbe89f8a6f
2006 F20101219_AACBQW woodbury_m_Page_112.txt
cc65da9e09c4401a4078c171eb6f665c
bce1e3e5f99e0b8f003d0a409573ef7ec99f5311
102041 F20101219_AACAGT woodbury_m_Page_016.jpg
b85a43a40ecf6a0f03d746ae448c8eeb
d4b05787b905afbd8df4f55d261c7686733d64df
F20101219_AACBDM woodbury_m_Page_193.tif
aa7c43a514eb411747c797324a020b2a
93a8d37d3d8587d8b63c88896b937fb0fc582de2
8277 F20101219_AACCAF woodbury_m_Page_070thm.jpg
85fed53c76438c6daa6acddf9a0e2823
e0c23308c06d77ff66e0d45285b927dbfffdfe32
1913 F20101219_AACBQX woodbury_m_Page_113.txt
bd3f0e68547261f960f2e675c5134329
b07cdc0fb7963719d57a0fdb6ea868dba098a4d8
99891 F20101219_AACAGU woodbury_m_Page_017.jpg
eaad738593bfaa3a650c74f962f69f41
d8d7c512584c561ee7e08f230e9b8795cb5dce80
F20101219_AACBDN woodbury_m_Page_194.tif
f39e8bfb1f67f7761ea6f5a300e9e847
856eefcf60cc1ee8ed401a10ac27de0a204d8740
31911 F20101219_AACCAG woodbury_m_Page_071.QC.jpg
7de00e16f22f86a8e17ef4946c599002
8e123e1907d028111a57463e8505de0b5ebd131f
1007 F20101219_AACBQY woodbury_m_Page_114.txt
4c343415d294018fa230f5ac69989aa1
da8b7d96d0de8ee38b72ecb2ee3b2212f81d595c
F20101219_AACBDO woodbury_m_Page_195.tif
aec887d066120a6af16283e1a420db6a
b5761dbaab00eefabb07ca1595197d15f0ee535f
7702 F20101219_AACCAH woodbury_m_Page_071thm.jpg
85f7725316bec2d3da338769d8317832
e543f9e2624516eaa4264dfac511aa8369daa3d3
561 F20101219_AACBQZ woodbury_m_Page_115.txt
81d376b517a549a6966dd9e7c12799fd
2d0138bd8f61902dd8cbc64726d8269ae3f9975c
103481 F20101219_AACAGV woodbury_m_Page_018.jpg
8fb479e06975bbcf38aae7de0efa1f6e
0150f4c8f118b062f11e2be509aafb47ff4ceeb3
F20101219_AACBDP woodbury_m_Page_196.tif
ef42865478818b31e5214f437ddac574
2ac9fd0550ec28e7d77b11906295e1a17569d564
33041 F20101219_AACCAI woodbury_m_Page_072.QC.jpg
772cc5ec17a8c8df8f8270a3ce9fee2d
29235585041269dde91ef86f72db1768553d6834
104258 F20101219_AACAGW woodbury_m_Page_019.jpg
e24dcca2d1e5234894db8b842e12f4a6
b67a0ad40013473b5e6bd085d837386f1ae6cc5b
F20101219_AACAZA woodbury_m_Page_074.tif
47b825d4022741de34593ca8dded2560
9738e6e105eba952c13f2d5fed6da1f8b2a57f86
F20101219_AACBDQ woodbury_m_Page_197.tif
297de376309640bd17631b769cfdd594
bebd8291b7fd3b689c46908aafa45c04401ab389
8336 F20101219_AACCAJ woodbury_m_Page_072thm.jpg
ed0be9e76c46e21b5704c91d06e0f905
5b9e7eb4711ec03d5efb6fe7e3e6324d08f2d342
96218 F20101219_AACAGX woodbury_m_Page_020.jpg
8c164c513c3175765589ca4db147f2a4
ebab44d681aff655899877ebbc1c8e46449fa419
F20101219_AACAZB woodbury_m_Page_075.tif
4b913dda0f77b4e3d80d3acbe381b0b8
2f057cf422fa71bb00d7d6d1c894791dce499c25
33243 F20101219_AACCAK woodbury_m_Page_073.QC.jpg
6e2375e94bace585319de787793701b8
1954b92ee550b8793bbc56d85d833fb9110da8e4
99453 F20101219_AACAGY woodbury_m_Page_021.jpg
f09fe8853759810686a2c2b7a3297a7e
4225e96f0e14b7aaa0ce6fb0bc7908bd61cca142
F20101219_AACAZC woodbury_m_Page_076.tif
b0202d5d73f2b141f9a5daaedc9ba9b6
3debcc826e4489eb1b655e4c8ac493394ef691cd
F20101219_AACBDR woodbury_m_Page_199.tif
1d86cd6a6ea9322774e242e80016e19d
5d7f88cab912d25eea7e0ec403aaf50dc4f3c555
8392 F20101219_AACCAL woodbury_m_Page_073thm.jpg
973bf4ff4f1085a3dde011440102e52e
5ac4a3889a77d6c10dd7575f11f0e101dbede32b
95121 F20101219_AACAGZ woodbury_m_Page_022.jpg
21107a8876ae49bb2d8b0053310192ea
adea33bdcc27ccd769a7231da0f763c08c504584
F20101219_AACAZD woodbury_m_Page_077.tif
7cf26272268d7f6109959f87994c8b05
9f3d573fa6570c4a6236af0413dccfb9ea7450f5
F20101219_AACBDS woodbury_m_Page_200.tif
5b7ade65daddccd0a954196de8c2c8b2
715a72a56fc1144f42d666d15bac1b547e6bd020
33124 F20101219_AACCAM woodbury_m_Page_074.QC.jpg
7f7a92052e03e3af9cc08dac7104a127
c5ece8483a62c9ce5d89600e762dcf393c5071b7
F20101219_AACAZE woodbury_m_Page_078.tif
da0a4ca09ae7316c101b33aa59eff450
8d198a128b4446e6bb56da9330a8c5b68d5a15c7
F20101219_AACBDT woodbury_m_Page_201.tif
0efd978a8ada6c26f92563bc79920060
d8bb2b0b4cb39cb4a7c5a8c7e525d13484029e52
F20101219_AACAZF woodbury_m_Page_079.tif
7e715efa5c2e04ad7d43c7c17451bf0f
0b1c4c7177a84de1dab328d2c8c1ef49203ebf7d
F20101219_AACBDU woodbury_m_Page_202.tif
887e61a63e99e96486b8400a73001564
3f8a47af954b451f562ac81b92ed997671efb2ef
8287 F20101219_AACCAN woodbury_m_Page_074thm.jpg
507c590367a908482aa9d5b3c42ac161
72c27f8415194e9fb867a786b6427aeeb86f09ee
F20101219_AACAZG woodbury_m_Page_080.tif
c023566164494814d3a1fd14bc8a7e21
85e2ab1428737dfc3c20b599bf87c0658e4c7769
F20101219_AACBDV woodbury_m_Page_203.tif
c1d6e0a65b0eedc21b304ee3e09a237f
ef9928664d331649c0ee619341d2c94128f043af
33532 F20101219_AACCAO woodbury_m_Page_075.QC.jpg
5ad6cf8403cb4b5e9c7f50d6bd66f410
d76dbe48ea7168cf4eba0180d98f45a2c4e401d5
F20101219_AACAZH woodbury_m_Page_081.tif
88d0deb5dc3680fae0a16db693835808
9fc97e828fce90fb9df6f1c8b786910be9a80797
F20101219_AACBDW woodbury_m_Page_204.tif
3b958a57e599bef541a5d1d41fabb892
ebe49fa9d2f8bac4f0fcfd8f75fef031beaa2f87
3876 F20101219_AACBWA woodbury_m_Page_011thm.jpg
d1186242cfc812733fe0a0377f164f72
7b3075927309f8805470ad1fe3bd1b7abadf5d08
8183 F20101219_AACCAP woodbury_m_Page_075thm.jpg
2c7410563330ec2cb3414b9713d8e8d7
eee9e0597ae335ec81c5a307bf576010736482d4
F20101219_AACBDX woodbury_m_Page_205.tif
ecb9b7888a2d3a5bb8778ddcb41d4d29
865976717f4ef22ccd910222561bb1f473ac22df
28212 F20101219_AACBWB woodbury_m_Page_012.QC.jpg
c2f6a39d469ec38578e033c03e5efbae
06758b8f5c15b1b4c08b9a52297c878659a42167
32498 F20101219_AACCAQ woodbury_m_Page_076.QC.jpg
026ad152904c4e402864f0d4589ad066
523f89a3e59df54ad45552c13c09e116761774ed
F20101219_AACAZI woodbury_m_Page_082.tif
5492c040a9d56b95a13229f570acaec0
6201e14d7c4f86c517996946575f7eb476c23687
F20101219_AACBDY woodbury_m_Page_206.tif
0575cbda080100c228c67762ea078d59
d6cdb4bf684da5ad6528f6885ab0d13c887d702d
6725 F20101219_AACBWC woodbury_m_Page_012thm.jpg
f48184387e93115d7d6862d39cc29744
c60b1882367a67f66d973f53d61b4adcdeb3e433
8274 F20101219_AACCAR woodbury_m_Page_076thm.jpg
64d94617e77355021d92b2846c7b01cb
b2d3db2a68837b244aee42e37c44da0d2da9132f
F20101219_AACAZJ woodbury_m_Page_083.tif
ff39cc02b1a514c62dc7625d731a7242
01cbf8a164f930b49e5693a505d89c4a2f02578a
F20101219_AACBDZ woodbury_m_Page_207.tif
aeaf3faccf2d781a8d8ada500414f21e
16fe2883ce7dadee117d7820f335165e9519eb91
26500 F20101219_AACBWD woodbury_m_Page_013.QC.jpg
20e090bd4c26463873ccea1dcb3e891b
09c5c0cf2f7772961c9bb511879713230e809fba
33959 F20101219_AACCAS woodbury_m_Page_077.QC.jpg
78aebb957be6a8e7be544d08b5b962f2
89eb863063e21cc4bb454f2d3f05c4ad266a3897
F20101219_AACAZK woodbury_m_Page_084.tif
3cae00f68b10d40b858549df40a09a7b
a51c591993ee74aa038304463ba0468f70f071ff
8417 F20101219_AACCAT woodbury_m_Page_077thm.jpg
92d101d364de4355146054c4f34333ab
d3b68b8953eb8a0f72fa3f97e4383a0a2675caf7
38262 F20101219_AACAMA woodbury_m_Page_164.jpg
9049da3e927de83f3f8ac7d92ba90751
288696bbaca07a096082717c7ae01bc4af928312
F20101219_AACAZL woodbury_m_Page_085.tif
bbb09b187097df4a44050cf5c122e38c
6920bf3e4ba4113f3f3bf5e170801ac149eb67b9
6438 F20101219_AACBWE woodbury_m_Page_013thm.jpg
7eef582be051bd931f5cd88ef6e6a7fd
49dfda5400986828b495d848d3ba942265d0b8ff
31672 F20101219_AACCAU woodbury_m_Page_078.QC.jpg
75807a342a8901bd96b89915dc00db48
8f245ba75c1539eb8ca9bbfbcb9216014d753e97
27801 F20101219_AACAMB woodbury_m_Page_165.jpg
7254c76b9c69e135ac73bba7d74d5985
ccbad0bd9388c20b628d0d8b3026164178a2ebdf
F20101219_AACAZM woodbury_m_Page_086.tif
632f54274ddcbad30353e6580d0cf1c0
7ecb016b0e12a23d22da1ed45041f15b48ac4300
24895 F20101219_AACBWF woodbury_m_Page_014.QC.jpg
fbf94e213ff6591821afea6ec716f94a
4be129fab7654c63c5a8aa576e62e2be5ec94a5c
7916 F20101219_AACCAV woodbury_m_Page_078thm.jpg
b38f23f9b6047af38f10de709c484e68
85d694bf3158fd288f1a6e41340e4f3f5f314f44
30588 F20101219_AACAMC woodbury_m_Page_166.jpg
81210f607c09b498d71f2df9ad76e50e
abbbcf2741ce527e354031250ced4710b4b5f644
F20101219_AACAZN woodbury_m_Page_087.tif
f550fc69d328b978a5267846799614f1
078ef31e675998da7399be6c951e456c5742124b
6105 F20101219_AACBWG woodbury_m_Page_014thm.jpg
31bb898790193568c58805d41052023a
9aa185e56c1fcbba174e516d13d90ae3b908b814
31136 F20101219_AACCAW woodbury_m_Page_079.QC.jpg
152ff40230b5d43f706c972374210a96
2a2398ce382febf6ce1f586ee841049a0aecd93c
35024 F20101219_AACAMD woodbury_m_Page_167.jpg
fb3b8359d11221015c4d1fc833a887a1
f1b9c4c43c340a46698628e0f18e3e431b3fb3d4
F20101219_AACAZO woodbury_m_Page_088.tif
f049ca456f217c54380b517aa350d801
f07f5988a5203b0f6a9a2a23c3b4fbe30eff9e3a
29746 F20101219_AACBWH woodbury_m_Page_015.QC.jpg
7a8e9e1088432129ee7170964336a18b
5c343de0f891a7fc366bf34d2d0dcf93e45aee82
7811 F20101219_AACCAX woodbury_m_Page_079thm.jpg
86d8e1c745cfac51ae96177f67f330fe
8e65f5590a3929d04dd4b5597c1089efedc9aa71
98678 F20101219_AACAME woodbury_m_Page_168.jpg
f68978e0a861ac036fd52c590c217974
bb88980bac05e80b031b53a19dc54f6829b86da0
1054428 F20101219_AACAZP woodbury_m_Page_089.tif
afdfd3b0df71b529b7efa0db07a4b3e5
9c2a666e4eb1391b5b80bf7c5a589ac75f654f5c
7415 F20101219_AACBWI woodbury_m_Page_015thm.jpg
e1665358e98481d46015deb4024bfa45
0901a08cbb2364058863ebe5573133158a8cceef
34388 F20101219_AACCAY woodbury_m_Page_080.QC.jpg
2ba5b94570b46c61fe344db60ac43672
7b268005381cfff572c56791b8428fa738b265aa
100736 F20101219_AACAMF woodbury_m_Page_169.jpg
cdf7d4023ed893c72f2cd8158be490cb
2d4f5751fa25bd019048d45241be7919b24dd7fc
8425398 F20101219_AACAZQ woodbury_m_Page_090.tif
866ebb2af4239e88bbd7c58351050ea6
061a949b7d7a6b674c3e23af362bbeea321f85c3
32907 F20101219_AACBWJ woodbury_m_Page_016.QC.jpg
b66acd60ca742181fb78932ecf478e0f
af3d0b9387d71d3854456676b8f2e334bc15593c
8496 F20101219_AACCAZ woodbury_m_Page_080thm.jpg
70c5a5f0c97425ae3037a3df3f25c342
cbb2ebcda48d582ccc2ad238394f0fb9e7a8c1ff
84463 F20101219_AACAMG woodbury_m_Page_170.jpg
af16cc48282abd904bbe26d98bfb4c34
0e7359f419eabe789fc73ba8f4f57b30e308e812
F20101219_AACAZR woodbury_m_Page_091.tif
fbab89e04b5fcf6e7a3773abb8b919f0
c093764228bf6e548a13ec139d64d2b3a1a9481a
8088 F20101219_AACBWK woodbury_m_Page_016thm.jpg
4fb9adfe0b5405a3ff3de778a28a8100
1d5c28c1f223ce01800b75048155c161ff8bd478
91197 F20101219_AACAMH woodbury_m_Page_171.jpg
6f612d8e2ca0008c0a267a9af37d2e49
5b7e2ad3df4897f9f925ae2e016822d4ac556ad2
51032 F20101219_AACBJA woodbury_m_Page_122.pro
89b46a3db09c07e8e3cff8aa132370a7
2f4f0e13691b424b4eed76784c3a5d29744e779d
F20101219_AACAZS woodbury_m_Page_093.tif
abb9c599e214e79c98a3b00677721e20
571a19817d2d71d378d3f991c22400469d1581dc
32303 F20101219_AACBWL woodbury_m_Page_017.QC.jpg
1b26f23148c8984413c5e308e1cfdfe8
103df03525113ff86df0e493714cbd0a14e6bcb0
91188 F20101219_AACAMI woodbury_m_Page_172.jpg
f81031ba35dc02d6d80bce8ea0dd642a
db36478fb63dc74afc8ebb3c5ea40a53c6ddfd90
42277 F20101219_AACBJB woodbury_m_Page_123.pro
9016c17bca2b00acf16a3d5f74617c26
3a70fe6b6c39db2a73f6aefa9aefc74680eff5dd
F20101219_AACAZT woodbury_m_Page_094.tif
6d9f15566a47b352fb7506be31165d67
9124cc4b0970b8a5d5e8446a6c1e11eefb021d86
7944 F20101219_AACBWM woodbury_m_Page_017thm.jpg
abb5b85899396fd7573a6885f80c76d5
b6b43d3a5107db0b310e3fa7e0e60889949d57a7
91891 F20101219_AACAMJ woodbury_m_Page_173.jpg
f611442c82f500dc426164172df54f05
a18b2ab5a76f76b2bc236725786a6cb758ccf42a
45778 F20101219_AACBJC woodbury_m_Page_125.pro
03cd02ae268dc2a6d1696f544602f643
362180106857ec8c6e5c523df948b4db5d5d95c3
25265604 F20101219_AACAZU woodbury_m_Page_095.tif
edc9cf4a11089b28a409a66fe86bdfc0
34780ab2e423b5e35d9790bf409e9db24c58a042
33776 F20101219_AACBWN woodbury_m_Page_018.QC.jpg
bfe45362f8a7149386649b3a78032271
5644799031989a0d828bb79c41e9c40edddb5259
99705 F20101219_AACAMK woodbury_m_Page_174.jpg
ca7448c7de6601acd1a97867e5339c22
ed42cddf7578a451dd1985724463b5b6b4afd8c5
58428 F20101219_AACBJD woodbury_m_Page_126.pro
84671c9f5a63ba59a07625a9aeca70fe
d21305d786f2f39ed4206a3f73e39b5dd7e5227a
F20101219_AACAZV woodbury_m_Page_096.tif
e934e6a15f76705f8a5d0f6ab1fc5b0a
73abd53f52cddaf4efbcfe04fea683e5a92e6e0d
8472 F20101219_AACBWO woodbury_m_Page_018thm.jpg
5156a0f2bb5ede658ca8d89803c807a4
bcd99cd72d1f88229adfdbcb5f9f5f5ab1a2c340
84897 F20101219_AACAML woodbury_m_Page_175.jpg
3aed072679a4e5ca569d55ebc8f3d1f9
8fd89cc90ef1d22e36d3d856e1d8f206071c76fa
51016 F20101219_AACBJE woodbury_m_Page_127.pro
1824db5b6a6c6177205b86bd8f2a62ef
09bd78a3c285db1281f9bc86d8e8fdcd9641d4a4
F20101219_AACAZW woodbury_m_Page_097.tif
4598c0168823f756f769c104f3633028
4a79cd42fe11e6b8244d83e2b069466e0bba8784
33755 F20101219_AACBWP woodbury_m_Page_019.QC.jpg
ea9bbe942b2f930f9697fbe3c65337f7
945218a45b184c7129cb07a3d84592ed50e0f292
92501 F20101219_AACAMM woodbury_m_Page_176.jpg
570748a0c060aa13ac9b976902b25449
168693e3013e1aead44a09ef54ebf032a449c3ed
52782 F20101219_AACBJF woodbury_m_Page_128.pro
2276806480fce5b37e4b2ab520147355
f62b47ecf320c222c4af078f645962f54a0923b9
F20101219_AACAZX woodbury_m_Page_098.tif
2213738d39226494f058b759c6c987b7
758eeda0ee7d27dfef5166b9bb50714ce5bb28a1
8193 F20101219_AACBWQ woodbury_m_Page_019thm.jpg
e514f92560755d75f384bafc9fed4a5d
7b353e9116769d23f9050ec36edde227428d18a2
103756 F20101219_AACAMN woodbury_m_Page_177.jpg
274308d0c80b8593f5359333dff1d222
31dd2b6ad564de59207c2437ceb26c0cfd1f991d
49602 F20101219_AACBJG woodbury_m_Page_129.pro
8eab25e0ec675c021f9112b3fc2c950a
904b5dfd9d93b705e38dd8bb6bdac8bf4cc15a94
F20101219_AACAZY woodbury_m_Page_099.tif
cd04ae38d5384abab929e4f1ae8bf2f5
7a9a58db503a668254dcbcb7a1cdc073540cd5fe
31104 F20101219_AACBWR woodbury_m_Page_020.QC.jpg
c630e99373403a7fc6224c49563ca6b1
90930268a7b1491d0206a4689cb8315bd9c92792
101677 F20101219_AACAMO woodbury_m_Page_178.jpg
54c88dfe9008f1f3784113ed6e2684c6
d1a629b5713d927c91d0b67d11e0a004bf9b7137
50740 F20101219_AACBJH woodbury_m_Page_130.pro
76e71826aa24408037b113de9e2a535b
becec2827f58f3feb2c60984377ad61cad9bae4a
F20101219_AACAZZ woodbury_m_Page_100.tif
39ade9ea38047b2570e9acb6b2e5bbc0
6a4d12ce783a58b82d1c73e43526255fe8645a80
30382 F20101219_AACCGA woodbury_m_Page_154.QC.jpg
c6e1b9f05849cb060528a6b2f1a00200
9117851d6e87856f99b01d973ce057aeedd72a90
7717 F20101219_AACBWS woodbury_m_Page_020thm.jpg
ea2bc715294da4fe79122ae388ca9bb8
cadc573af594dfe14e96a73238d90c47546e6008
100862 F20101219_AACAMP woodbury_m_Page_179.jpg
826b27033dce47bd2cf73b77f3c8c0c5
5acb621a931db93d121504554ed176b7ab3edb90
48589 F20101219_AACBJI woodbury_m_Page_131.pro
ae1d7fff8d1a794d4447da4686577fdb
2d669d47a56464de7e78d31af95cad7603d7e72e
7857 F20101219_AACCGB woodbury_m_Page_154thm.jpg
6a590213608d92de5f9e4bc8f5a196d2
b5e42fbbad2e67f9a820f7b5fbc9ace3f3d88288
33083 F20101219_AACBWT woodbury_m_Page_021.QC.jpg
22e8d01eb06c453e89f29518e7c3f1af
46586043504e2309d0fedc804059a6b5103723c1
104680 F20101219_AACAMQ woodbury_m_Page_180.jpg
83e1a098b0180b5b6f5fd25bac5361c7
c0a04ff73f4475dbb8fab70e560aae2ae86871bd
47807 F20101219_AACBJJ woodbury_m_Page_132.pro
90b8f239d73937cb8f68e9d57d078751
b9ccfc4749baf1c05d90b65d44fe057ee52df047
25091 F20101219_AACCGC woodbury_m_Page_155.QC.jpg
7083fcaa2b4ee38531f36f4108cdc662
3c22aded6a09f55f8c2cae0c604d8a9ad03c4b5d
8111 F20101219_AACBWU woodbury_m_Page_021thm.jpg
f68f18f964b9a61b9d52b3cfac7a2a01
b59e5e5a98d833628d5eb467c11c17657306b81b
49022 F20101219_AACBJK woodbury_m_Page_134.pro
8b08269594d8b8c9c298f1027dc26520
953997e86047c7759b404cd56f11fdc18d68ecd6
6240 F20101219_AACCGD woodbury_m_Page_155thm.jpg
c86995fafdf4d350f499c7b75446d60e
ab1e297c33632c573ae62fd52b2b0692ff7a902f
7872 F20101219_AACBWV woodbury_m_Page_022thm.jpg
aeca1fd48fdef0c21f2275b1516d4722
7cf17125c725a2816a48ce9dff31ccd2fe9ceafc
70065 F20101219_AACAMR woodbury_m_Page_181.jpg
e3b13687856de9369b27cbdd138d65cf
599fe5ee5da37b6d93bd7d7269ba074a72b046f3
52995 F20101219_AACBJL woodbury_m_Page_135.pro
5db6ec7ad1753cecf4774f6b99f0a2fa
2a07527a5674b859172e72f1a0ff5d841ebd038a
12987 F20101219_AACCGE woodbury_m_Page_156.QC.jpg
55da06099754c0a7fd360cac2ff012b2
5df2993e2887ffb332186fbeb9afb38f6d07baa0
33715 F20101219_AACBWW woodbury_m_Page_023.QC.jpg
baf5753370144f3a9f9efffa014cc3b0
28b190b1820da766e795d3e111f8d29e0f98927c
79968 F20101219_AACAMS woodbury_m_Page_182.jpg
543afb96b8acbe2a9c0156d3b031ed6d
1551d5e177913c95e88b0474e07feffc785ed7e1
53833 F20101219_AACBJM woodbury_m_Page_136.pro
1b644c1b60857c34c02caf3f3f440424
b425560fceb4adf421da3d527face7b431f1a5d9
3788 F20101219_AACCGF woodbury_m_Page_156thm.jpg
471bd32c32d9e1c2d922bb7bd2e9d5dd
2e7bec45035ee6f3f84c5fc0f921cd5f272d8a4c
8451 F20101219_AACBWX woodbury_m_Page_023thm.jpg
f85a2a1d031e774a9357e0cddf665fec
dd7fe073813b73da0327af1895a2ff941ff5c397
71670 F20101219_AACAMT woodbury_m_Page_183.jpg
2e4bc31cb4e8d8815d801afc846c5410
91ed4b321b7bfeeb299b1333d6871e0056e5830c
51705 F20101219_AACBJN woodbury_m_Page_137.pro
b15f2e89ba2ee86eadddd2dbb159f252
4bd02a9d806f32761281ec9131dbfc0f14d13f50
11402 F20101219_AACCGG woodbury_m_Page_157.QC.jpg
25538de07d28a2b88be5d17c9d23d344
7490bbe7d32bb9e5aaa41d624ec81199b80eb57e
34130 F20101219_AACBWY woodbury_m_Page_024.QC.jpg
2633e7ee804ff46abcb6d6794643f4f2
77ca348ef0c5c197cf23fba5de2e2b442c616f64
95198 F20101219_AACAMU woodbury_m_Page_184.jpg
8a7b507a09213a6dc4c6c05071fad195
58b5e9eaa71afe3015dea117d4a969d2e3602276
47242 F20101219_AACBJO woodbury_m_Page_138.pro
3dbd36e9480764016c4907b88c02767e
134759ca381ea6a783bcf2be8ea79ff63128dfb3
3254 F20101219_AACCGH woodbury_m_Page_157thm.jpg
5560d163b19100e42cf9ba2f46738b0c
af51e9af308139a93993361557f0b63d73ef562b
8147 F20101219_AACBWZ woodbury_m_Page_024thm.jpg
c242bbc647f83cf96812aaabc0405df0
39200d5751e60999338101730ec4f211423590af
65790 F20101219_AACAMV woodbury_m_Page_185.jpg
cb952bae65f090c33c2eb8a72f668d06
0fa5da69fc6d1f82ab2933f934269eccf8ce5b8b
47216 F20101219_AACBJP woodbury_m_Page_139.pro
dd809318e2d9cdcc4a47c7a1f311285b
73d43ede8ce3cc56ec3abbfbe804586d34879097
15318 F20101219_AACCGI woodbury_m_Page_158.QC.jpg
fdaaa6e228b1033930d03164717f47ce
c962adb75432dd93353f56938116853c8c7f5f8e
94384 F20101219_AACAMW woodbury_m_Page_186.jpg
b996e016c64178400deb2f0568e434f1
4ffb1282b03c0802287e7029872ceb9719847770
49660 F20101219_AACBJQ woodbury_m_Page_140.pro
d9c1d54dbc747873bec21b276fc58c22
e5a2d7871abf627b729b51d9f9376615543ef89c
3754 F20101219_AACCGJ woodbury_m_Page_158thm.jpg
91111b23d6cc7eca7123294dd85d8668
c01ad8a248ff7ad17c85ba878d1bf33aff0276be
64444 F20101219_AACAMX woodbury_m_Page_187.jpg
c1e62e051c52552d8d0f99c49b7c7ab9
70db37d3dbc5a745c1b8080a36fd69032340d97c
47287 F20101219_AACBJR woodbury_m_Page_141.pro
012a8ab544c9a225d9658644db6572d1
adba1b77e7d48aab442a82907b0c093b274f3a19
12796 F20101219_AACCGK woodbury_m_Page_159.QC.jpg
97d8a5ac5696341717cfc4a066dac8f7
0e153baf22cdb776a4e0294494011a7eeaa83265
43350 F20101219_AACAMY woodbury_m_Page_188.jpg
47671cc8150c2799739ce06031a54e36
dfc700e0a717924f7805615fc61a80cd767f8815
55196 F20101219_AACBJS woodbury_m_Page_142.pro
0cecf5bb9346c44cad96c98641cbd8cb
aba5fab5ca786938861ce8dab28981d5a39c0ba0
3527 F20101219_AACCGL woodbury_m_Page_159thm.jpg
9bd3c6057e740152d3e8641982cac7bf
1049ba42745e0e88d90c4af5c3fd7e284ec5af66
94476 F20101219_AACAMZ woodbury_m_Page_189.jpg
85fc904680bcc05c13e563ba11534c7e
5bd8df1104a817371986d89a445e356e0a560e66
50183 F20101219_AACBJT woodbury_m_Page_143.pro
ba69dc751bfbd5f632110e623abe56b8
2626178f7c676589f8a3837dee4aab7186149904
26908 F20101219_AACCGM woodbury_m_Page_160.QC.jpg
122b0630fa2b460db51ce7918fdc4b29
03b22dbcd8cd96722974eb7b29a14a878018dbb0
50638 F20101219_AACBJU woodbury_m_Page_144.pro
5b15241157c8a2130a24157d7a07ab56
c6a31286f60916d292480901991e206c96c0395c
6653 F20101219_AACCGN woodbury_m_Page_160thm.jpg
6a6184a33f13c339ab753250a617b2d4
318b474789f27d1af83cf210883105ce51aff7a5
50230 F20101219_AACBJV woodbury_m_Page_145.pro
ebfba8cb9aa0e547a3d50c093866786b
11adb402af21c731ee812d38e12fa1546492d8dc
14097 F20101219_AACCGO woodbury_m_Page_161.QC.jpg
b46289467f48337dbdeafaefe5fca47a
db63cb4e4e0373515c05713681ac47ad5bd3ab6a
46209 F20101219_AACBJW woodbury_m_Page_146.pro
5be7a25308394a6fdd0219a3e1691bfd
c721c3ebcfb3f92bc5292303d0152d61c881c56f
3970 F20101219_AACCGP woodbury_m_Page_161thm.jpg
5108c9b34c74db134c9efd34b1c36188
0396ce2385e4c5bb2375dec599885ef60ca7a749
27135 F20101219_AACCGQ woodbury_m_Page_162.QC.jpg
e91a03305c4934f844ce9b213111cbf7
b8f28ff9d1c1500538448a9a2b35c26103e40c55
47786 F20101219_AACBJX woodbury_m_Page_147.pro
101be8da97cd62b9a7b7f5c9e67ac6ae
c9ab541bd31fc4eacc38bd5c230da8e25b2065a9
6852 F20101219_AACCGR woodbury_m_Page_162thm.jpg
e0fcd544fbc1902d24954d1f9a89d90c
d938e710bb6de71dbda6663c76f148a78ca9e480
50225 F20101219_AACBJY woodbury_m_Page_148.pro
ed3a97748e4b21f85e40f2037c0bbe2f
b504bc4ab99c4dcc5d31433069c9709d877a0488
17465 F20101219_AACCGS woodbury_m_Page_163.QC.jpg
56f0c6280a25239c19a947614d30ce0d
8b5aca90d68e4875a14c051fbe95103d0a9bf38c
49909 F20101219_AACBJZ woodbury_m_Page_149.pro
a7c6a2f0f94500c4dca0076c74a88049
8ec153fead6bca00b676f5dbeb99cdbfa4d6adec
4163 F20101219_AACCGT woodbury_m_Page_163thm.jpg
826bba78234671bf75ea587ed400faf0
5267b3c4cc15b17ae4248c0648b2afa4674bf36a
106980 F20101219_AACASA woodbury_m_Page_100.jp2
3f4a31a0991c1f9b6bedbf54220f9911
cbe09df01b531b528b6d7529c62102f78f539b7a
10879 F20101219_AACCGU woodbury_m_Page_164.QC.jpg
93d2f337ef550adc8172b27c7c0c5418
3d81aad0c4797b6e8c1a60b7651c74e0d73f0d64
2601 F20101219_AACCGV woodbury_m_Page_164thm.jpg
d9b59bdebe2c0241db3749e6c129f0f3
ce27980ca97c91ac9686461299dc5355f593ce04
108922 F20101219_AACASB woodbury_m_Page_101.jp2
c94d52cc26952491797870f070871307
2f2f3c8d986193ae6f9b140bb01644d1dbb00fbc
10389 F20101219_AACCGW woodbury_m_Page_165.QC.jpg
86e5a10424c7bb981bc2b03e8d1f6d0f
49165d0bcbbb16dd430fcfa8c09037bcee24bfad
95470 F20101219_AACASC woodbury_m_Page_102.jp2
fb111b611b4e3778249191913cc87292
594422bf5d945a9fa993b26e0c8faf4b8267cb8c
3113 F20101219_AACCGX woodbury_m_Page_165thm.jpg
1849d1567e37eab43b4311ff00692d15
bdb84c2aa10308469ef000097b2538b11e93f035
92269 F20101219_AACASD woodbury_m_Page_103.jp2
519b4e0e9172da4eb69e7fc21992a133
681f0ee61b84aafce8fab2466f40ed6263e2628d
10528 F20101219_AACCGY woodbury_m_Page_166.QC.jpg
b8bbf44c16c2046685717ef8c3d3d70b
d2330f2a9659f5407f4281b092c050e7245873b3
110363 F20101219_AACASE woodbury_m_Page_104.jp2
441cbdba12bc9a96acfcfb0b4e167f24
310e766762846a4621f80a1a5995891a8e015978
3711 F20101219_AACCGZ woodbury_m_Page_166thm.jpg
8b3d9640a15a1d90f409e3a41b02fa54
32a3443b9eca8b185fe119e885e815ff1d459998
103598 F20101219_AACASF woodbury_m_Page_106.jp2
86d577d8b65d89915ffed28e4f1a66ee
452e94ad43528caa2124772761a0a3dffb086009
110668 F20101219_AACASG woodbury_m_Page_107.jp2
329be70f7a493320e94a8517f062d2ad
c50341a343dc57ff005360d51a50ae7270dc1ae9
1944 F20101219_AACBPA woodbury_m_Page_060.txt
7f0e11474fd230af682d128eef0210d4
9d5c09bb22bed08aeffd0854b82af4edc994e00e
110064 F20101219_AACASH woodbury_m_Page_108.jp2
30d60cdbb705e28ee45659ab90582fff
7a09a5284a97864c5222a0f05198fe72986c3f63
2117 F20101219_AACBPB woodbury_m_Page_062.txt
0c3dd26a75317562387c881948935700
e90d4a5e98b363c153c7f147a2ac7a22117c3a9c
108249 F20101219_AACASI woodbury_m_Page_109.jp2
0fccf20b0fbb016d86515932b3a81476
56c15b59b1231debcf45c5438a5f9a512914277d
1918 F20101219_AACBPC woodbury_m_Page_063.txt
ba35aca4028b3e0211886ed98c8861c4
eeb0fa6fa72462bcc884c2cfe33f41af05de63ff
110549 F20101219_AACASJ woodbury_m_Page_110.jp2
29f9c70ab10266cb3b4e85ff586d3fef
95a318cc24432a8b641bf539c6fd10dd705d8c6e
2045 F20101219_AACBPD woodbury_m_Page_064.txt
cb4f2542c0b9706ef9a0d56f97a82874
928470e5cd1b652c3b7396a0265c351cd7701bd7
109271 F20101219_AACASK woodbury_m_Page_111.jp2
a5621be6022595a6ad198098fc2e3223
b81d15ff023611f29198f5fcd4a81d7e2a3f0fb9
1895 F20101219_AACBPE woodbury_m_Page_065.txt
c5351526db4e369c870df7e3c94bd1eb
215a7044d0a0a4a6d566b5dca9f1e4a27b68db31
99428 F20101219_AACAFA woodbury_m_Page_089.jp2
e7cc1d654b7e9f87b875cf87e179016f
10ec70aefb5eec40c584cdd23260c3f3cbb96384
110229 F20101219_AACASL woodbury_m_Page_112.jp2
d437f41b5b3c0a70d6964028dd5e1b5f
8e59c73502e52bb0bbe07803306f773405ec527a
1703 F20101219_AACBPF woodbury_m_Page_066.txt
15fb13d3b1c23fa6a166e40d138dd3ba
c31ddf8fd8fb14906a84ae87dee3ebdd82c9bf43
8286 F20101219_AACAFB woodbury_m_Page_179thm.jpg
5aa47506ab8ef716dcd8392535e9f683
3733e4b78cf7776847858fc5d4966bbad79f01fc
101230 F20101219_AACASM woodbury_m_Page_113.jp2
5fab7814fdef2f74b996755dd5caa843
d09f8dddc19721e871136fbd2cd7c3ccaad29f51
391 F20101219_AACBPG woodbury_m_Page_067.txt
c02a2bde30dbd0413b4110e11cac412e
d96b0d437e48ee689d543e6797dfc640022e2cfe
35587 F20101219_AACAFC woodbury_m_Page_202.pro
51a9dcb2fa34541f34d7ba66e033bd3c
82646e39462d2c74a31d975bbd71b5d46c519513
56319 F20101219_AACASN woodbury_m_Page_114.jp2
4307521b9109ad8eae47b990e1fa0058
a25d7cf3d6707ef079aecd2ce385ad948b14d271
1831 F20101219_AACBPH woodbury_m_Page_068.txt
fd37f86adb736c002b3ebf65342d69ea
03223cf219ff3e8745e824d47ed0fae146e1e66b
48191 F20101219_AACAFD woodbury_m_Page_113.pro
7814e219fb69c262951f9293ae4d5ab2
598e0277df20b461ca587b41e720c22a63c5c605
902597 F20101219_AACASO woodbury_m_Page_116.jp2
cb181aa9837bd3a87fc524de417b0320
be994fc356f74a52b8c2ece478658ab406fe5adb
2095 F20101219_AACBPI woodbury_m_Page_069.txt
781881e61addf23b182775f5689482c0
4cd38469df6946bdeed55ba5b388bb55cbbb5758
102241 F20101219_AACAFE woodbury_m_Page_145.jpg
9352e096661fdb168a0d5a7ffa65550e
67cbc367ad2e4f8e42546e757a4b7f9fd60864c7
953657 F20101219_AACASP woodbury_m_Page_117.jp2
da26da2dea8074418daa023ba6cab8b2
9c5bbb9505e3e2b246ad464006aea80a7536050f
2225 F20101219_AACBPJ woodbury_m_Page_070.txt
a029fd50de2e0f066cf2405922b86d44
b5bc8e975b8daa9eea1e37eae9f27f2386a0bcad
F20101219_AACAFF woodbury_m_Page_112.tif
dda2515d59ee9bb5265fefcec8039b77
0801c2377b269ae5f772b45249b428278a353062
590148 F20101219_AACASQ woodbury_m_Page_118.jp2
674542d0d4e2abda19463a874a95a5e8
2ac0bb5e23bad8ffadc878cc9e9c92a7cde30834
1939 F20101219_AACBPK woodbury_m_Page_071.txt
b1678bf59fc7aa15f6e51cd3d1adecf8
6bcc72c872c48cb41bc35e075a2de1864b932c8c
29996 F20101219_AACAFG woodbury_m_Page_102.QC.jpg
3af99aa4d742600d39cb749675bdcd51
e2040ba2044095e967de520dfb4068f8aafe94f1
98623 F20101219_AACASR woodbury_m_Page_119.jp2
6d6f79b4a2444ce95b5f4b9d9bead4f7
ba79138a3aebcc3e6a7f30856ef0d39118cf4e7d
2018 F20101219_AACBPL woodbury_m_Page_072.txt
9cc6f537e177fd595864640c1cb0a4e1
cef5b220d786c436a42ed468b56a4bb319e74f16
91125 F20101219_AACAFH woodbury_m_Page_204.jpg
19d2adbf18e17a94132f5baf09c3ec8a
e5541fcfed980a4f943c1e17d700c3768e882b04
F20101219_AACBCA woodbury_m_Page_154.tif
059eed8977f272b2a0cac89f26c53c38
43799adb51f040a71afc96e6e9e691b7e8149306
107822 F20101219_AACASS woodbury_m_Page_120.jp2
4006953228690b5e8c789e7eb5f6b7f9
bb72f7689d35d1b0f19dc1afb68c4a9e99d0b5a7
1988 F20101219_AACBPM woodbury_m_Page_073.txt
3e5681e80b4a6bcf50743d1237ae33b7
60f315a30186d718cd194d5c50345742e0ea3a7a
34472 F20101219_AACAFI woodbury_m_Page_136.QC.jpg
41c7a363080d7586ff1db6fc2acf20ce
f0368600eb63f747621a0e5f8fd2c291208892a5
F20101219_AACBCB woodbury_m_Page_155.tif
b36e460092df9e5a389b371c906d03fb
311f1bb3f32c765699569cce3e66fde0fa7f8741
102259 F20101219_AACAST woodbury_m_Page_121.jp2
a36e5d7eb29a9b435946edbc48535b28
0d98cd60d160d70f2ff641b32eae12a79703ad68
F20101219_AACBPN woodbury_m_Page_074.txt
90a5a5117f119adb92dc6d49b95dded0
e098befecaa74b0ef0a3b600c9be69ef5d3078a2
113786 F20101219_AACAFJ woodbury_m_Page_105.jp2
d620e23467b3745e9ac46474ad9ec0e9
148bb3c10b31786991479263500db0b712755778
F20101219_AACBCC woodbury_m_Page_157.tif
acdabf124d72cfe376a10c4505a51c77
c381c7416b1624fcc86960f69a51fe179b271ee5
110053 F20101219_AACASU woodbury_m_Page_122.jp2
ea8317ef6b2c3b19f74d6010c9f87022
2793653947f108e0229ace863ce92f88687001d4
2023 F20101219_AACBPO woodbury_m_Page_075.txt
3d33baa567ef2756323c43341b163969
650eb3fa529a227be4c11a45267c2e22605f1ecf
101431 F20101219_AACAFK woodbury_m_Page_086.jpg
989b13ea6a07edeacdcd286b60bb41ff
2285f84411bb45bef370b8552b8c4dd205c15a21
F20101219_AACBCD woodbury_m_Page_158.tif
4d5480dd021a5d8e5dc0040780b23926
1b4ab4257064d8a6912e3a62ae46d63241c55327
101891 F20101219_AACASV woodbury_m_Page_124.jp2
43ed065d49b7572fb5089da07da3012e
0094cb38ac174805a95f38a5fd565a64f4184589
1986 F20101219_AACBPP woodbury_m_Page_076.txt
69c2652cf16abebd210d16bc1da6ade0
0c63aea5356096a9a6c6dabbbac91b58853f94a1
32912 F20101219_AACAFL woodbury_m_Page_104.QC.jpg
f6035399baad361a8a3ce0b320f3d70b
4d48dd6114fb44735f7cc7be13c36c6d99b04ba0
F20101219_AACBCE woodbury_m_Page_159.tif
c6564b91f323d38ac153eba07b5b9502
abfba99ee371816790e6b62433d54349ff8fd570
99538 F20101219_AACASW woodbury_m_Page_125.jp2
85a588fde5669de6db93f7177176ac95
30a5e7d07cf9c071653fd1b00683bb2ef111bd94
2074 F20101219_AACBPQ woodbury_m_Page_077.txt
344f9af6342ecdde73d6cc5325ecd9b6
e5ec6e106d36ddeb6105b8473266d9b42bfb54b4
33803 F20101219_AACAFM woodbury_m_Page_180.QC.jpg
ed59f79e36a2140a227183437ce4ef65
3b13bec12144cb657cc4aeeb15145b96a288dde2
F20101219_AACBCF woodbury_m_Page_160.tif
226b01af51f72889e2b2d29072cbabdc
722abeb85883eea65e19a244fcbd114e116ba21b
123564 F20101219_AACASX woodbury_m_Page_126.jp2
6c738cdb47291dc9b976cc0b421cd36f
02e731d1b731bc00327aebc689327137bca8d473
1861 F20101219_AACBPR woodbury_m_Page_078.txt
319c7053a7c548b8458f9f0464e754db
7d9facae09d51e2b8605aaaaf55c656680b78fde
F20101219_AACAFN woodbury_m_Page_144.txt
460c081aaf3a520942c656bef37acc67
39c80acf6baae5e08c2561a50076a16bc9e85cc5
F20101219_AACBCG woodbury_m_Page_161.tif
516c1331dbaa9d48f8bc0646d9eb7bb7
33e08db78d0334554eff7ed6b576e774a53ca271
110310 F20101219_AACASY woodbury_m_Page_127.jp2
4b11e69aaa2db9dad2459e77b0a836ea
f56073bfea70de9c9507500b4d3b77e0f36c0221
F20101219_AACBPS woodbury_m_Page_079.txt
9c17237a52d165831ee3c9476bf8b5fe
c1b3f4dfecb2cb585d0ca0029727dc1168cc76df
F20101219_AACAFO woodbury_m_Page_198.tif
075afbf991bd423f9cd78b54dc4e1d76
54232725ec6a1f6bae7040fadb023e5ef84791c1
F20101219_AACBCH woodbury_m_Page_162.tif
bbe42b5a9b7100de72fe12f77c901841
10a5c9fbfb6b187b53c03acea86403abf21d4f13
111949 F20101219_AACASZ woodbury_m_Page_128.jp2
36f701e2b456f79ce391c5dc3b923640
7a283cf6f17bfebb9463529b717913d69c1c459e
1942 F20101219_AACAFP woodbury_m_Page_037.txt
99d5f3209e911cbaa3865c7cd026d85e
abbe3496b64a36141d5ea0c80aa5a19456316871
F20101219_AACBCI woodbury_m_Page_163.tif
55f2ad212e98b998bdaf19986a146854
82dc2b0412a9f14f83ea482d71e9b0ed06b2e243
2115 F20101219_AACBPT woodbury_m_Page_080.txt
9200d17ddced6acd2cca4ba4ec3ec2d9
4309adef1516216f6d37cfe2a5262662a3360ff0
33728 F20101219_AACAFQ woodbury_m_Page_101.QC.jpg
a86d264d8e2f8111324a8083c7ebd081
919c9d62faa2a708866ec1ce30be497989be2273
F20101219_AACBCJ woodbury_m_Page_164.tif
15005d5ec796664ba05ee1ea4751b2ca
6501a8c3889cd42e98eb4e3e22349654c9eaaa57
1865 F20101219_AACBPU woodbury_m_Page_081.txt
6aaf3fc3ac51766c187257eba345baee
6b38f72f363f81d844edb565a698ab9ab4d0fab9
8609 F20101219_AACAFR woodbury_m_Page_149thm.jpg
9beb07a56b4cf7725497062ad35fdc1a
5fa5fb4af0a946adf05455ec21a91d8da44e4a03
F20101219_AACBCK woodbury_m_Page_165.tif
bf3779af833532fdfe754d8b4e0e60fb
bd36b84b5ffc0ae3630474a232ad0a406897cfc3
1995 F20101219_AACBPV woodbury_m_Page_084.txt
6926690a6d4de6432b4745d6ab2edb1b
9afc3d4153d5bb8ed2e71e913a1ef7cfd6b7b776
29875 F20101219_AACAFS woodbury_m_Page_207.QC.jpg
52f2e2d9216644324e9099f9c26caa6a
76860897f7413d34006a50102aeb7e4e9efa92a1
F20101219_AACBCL woodbury_m_Page_166.tif
e5de92b476686679f9eb5f50a6fe0460
ecd23fd78f3d746088600e80c4c3c0effd4057d0
F20101219_AACBPW woodbury_m_Page_085.txt
e0a4516997d88d8b5104a62ea28ffa29
b7b1f6523147ad8849e262ef12e28d109e1a4ae3
1051983 F20101219_AACAFT woodbury_m_Page_218.jp2
9a70b436a3ba401a5a410808fc822fa6
efcfa61286df6246d8d01f84268e3a2a02b62f41
F20101219_AACBCM woodbury_m_Page_167.tif
52fc9c9452bf796cef33f3894ed2ecf9
c7eb8eae2adb6348658a0af89089bbf6d10d3dfd
F20101219_AACBPX woodbury_m_Page_086.txt
9662cbf6e839405251702938c1377149
e883e6911c9e9dca0d8b903eb92825af7c9b0ac3
F20101219_AACBCN woodbury_m_Page_168.tif
bf77245b7bf157e7b30ed1db5bd346c5
85a5bb283e15369fd606027578c75c9b1b9ad19c
1911 F20101219_AACBPY woodbury_m_Page_087.txt
aacd833c5bcd4c0d87dce92fc5f42991
713758acddd109a9a81c22dc0999b4df5f0b4744
F20101219_AACAFU woodbury_m_Page_069.tif
f80d57e0574175b6ef122565b0e189d7
4baff9f93fc1027d1181224041d4234b6dc040c2
F20101219_AACBCO woodbury_m_Page_169.tif
161a55caf59513879256cdb92efd78fd
1436506be8418a642149439f5617e9f379dc313c
1591 F20101219_AACBPZ woodbury_m_Page_088.txt
af9f1b0713471676e303fd4ad8fd2052
3ac5346ead3d69dbd81dab6569b12ede7c27e3b7
1863 F20101219_AACAFV woodbury_m_Page_059.txt
a850914c24b5e932b90b38030c1c2129
b5a9ce306a7202b79ee176f318a7e84db820b797
F20101219_AACBCP woodbury_m_Page_170.tif
c4a3f512c81ff346f6e5a999f0008f96
215d3876d3286cb1ab2851b5f4a083287b6f32f9
122824 F20101219_AACAFW woodbury_m_Page_219.jp2
fd82994b76038901b410e204ef79db1f
90581b7b6480862e9a9b7c87f4c2ecec684702ae
F20101219_AACAYA woodbury_m_Page_045.tif
bd5ed18792d1926db3403ca67785b442
947f0c971e41a302926f0df9e47e271652131112
65309 F20101219_AACAFX woodbury_m_Page_220.pro
92d27aa49180b74f2ff732bc5500120b
cfdbbe4104ddd3daf95faa9be5288647aef58cc5
F20101219_AACAYB woodbury_m_Page_047.tif
732555d003c661cfe27728259a46cdc5
3b2a1e99c493aba5ec944084ac29f28596f3a707
F20101219_AACBCQ woodbury_m_Page_171.tif
85704a1f086741dfb8fd66eef704b33c
44d75a118cf06f08890ca0239f49f51945503339
F20101219_AACAFY woodbury_m_Page_147.QC.jpg
ec4ce658cdf13039a61668fac6da1966
d197342d06332d95c2b7660a93a9acc24f7b6e6f
F20101219_AACAYC woodbury_m_Page_048.tif
301e94917254e75dbb4cdbca17ebbfab
0a1ba347843467c16ac5b4440c2e8a671276c667
F20101219_AACBCR woodbury_m_Page_172.tif
d0ea835301c44e0cef99ed40398b0424
313d9229ff64b1f8d82f4580b8e6fad043b49bdd
F20101219_AACAFZ woodbury_m_Page_082.txt
09218ccfae5eda432b7b961f3d10d9dd
3a677b334004065dea17fbf4f6bf4ba83bc3deeb
F20101219_AACAYD woodbury_m_Page_049.tif
a7ba9a27b2f34cd8f0400a8aa7473d17
4b8653aaffa422ec21fef9405359c09b4c5309f5
F20101219_AACBCS woodbury_m_Page_173.tif
efca4fd3e20a2548a993ea289ccf8366
10ec70cfdd0c83c30b3f444d32a20213b8443ae5
F20101219_AACAYE woodbury_m_Page_050.tif
64ad11031aa0731d9e0186ec66fbcf2a
3a804574d4a1214b1fcef67414d5f1e85db712ee
F20101219_AACBCT woodbury_m_Page_174.tif
04a13f06fd189f3c338666efafc4a182
3d793c0ea58e98eb50c13b94caf30d660ab7c4d6
F20101219_AACAYF woodbury_m_Page_051.tif
1206369a28634b5c051e3ba65ae3a557
0f2ba60968ba0f52778eff5c739e747ba6b6ebf4
F20101219_AACBCU woodbury_m_Page_175.tif
0d7033841301beeaabacd7c4a51a6aa7
1479c645ef766427af4aeed24dc9dd08a6af45d7
F20101219_AACAYG woodbury_m_Page_052.tif
4a467e1b20e51c78c9ec59049625b6da
200a9597b3af26de56ec6117576600362c927d11
F20101219_AACBCV woodbury_m_Page_176.tif
36953016b050b9b0f0f4eddcc8e1117e
4d88faada5892564642b925abd5e437792f40219
F20101219_AACBCW woodbury_m_Page_177.tif
17ee3d42b1e120bb13afcde9347f878a
6fe195cde900a2986a7f3952a67c17fe2c92370c
2448 F20101219_AACBVA woodbury_m_Page_226.txt
e6c92e41d7e2ab1160317a32554a0679
136bd9d4165125de1300f7b185d401e05c190412
F20101219_AACAYH woodbury_m_Page_053.tif
d818a1133a5eb564bdb4fc4e8aa8061c
dc2546c87c582f9703caea8bd5af5f10444196de
F20101219_AACBCX woodbury_m_Page_178.tif
3148f19f6aa106716bd4ede87b9e70d5
be0d88cdff7cd9ef7a7f5bbb25b5bcdb4a8aeed0
2319 F20101219_AACBVB woodbury_m_Page_227.txt
44254f8f7364b4a58f7fd659a1c114d7
54a52dd2931772e5e011f13a7f93fb610cdfef52
F20101219_AACAYI woodbury_m_Page_054.tif
01cdaff5adb9115df001db7c91fe0798
ed6b5736986c1372424d5cb5e4757f8bdb557195
F20101219_AACBCY woodbury_m_Page_179.tif
32832602537caa61058abab10387f93b
14dd8c9dbb09539e8d18a3a11f377240793485a5
994 F20101219_AACBVC woodbury_m_Page_228.txt
11d40c989f43cf374713265e4ceba76b
b1a0b7c44cb697d1299395de36c7c1e9af9df604
F20101219_AACAYJ woodbury_m_Page_055.tif
efbbd437e45f5c77ae061ba3e580b3eb
76c88462854712ab6a854dbfaedff9282adace5f
F20101219_AACBCZ woodbury_m_Page_180.tif
ffef9677c90f05dc2e7b4ce77da9dcfd
cf1f700dbd73ed03f7332d907c52e6ed9d492c62
F20101219_AACAYK woodbury_m_Page_056.tif
8ffd693460f94b35858c57b82aba7d7a
e31128c948e63ed62c021a3764c67e43557f8d75
2579 F20101219_AACBVD woodbury_m_Page_001thm.jpg
13c9d9e4f6a4f45775c2ad614d0a15b8
249580b788fc05f38c29f1b0da1ad4f4c9e2c45f
106907 F20101219_AACALA woodbury_m_Page_136.jpg
7264cbc961c95d94a5f7b1508088b26e
01cd641b26c54b7a364698ef14e3ebac7357b2fd
F20101219_AACAYL woodbury_m_Page_057.tif
a34f8420c3201c298e4fba2fccd69361
a0deabfb5b377b4ef9432a6990560bb8af750a5f
805026 F20101219_AACBVE woodbury_m.pdf
5e2279e4924ccf2ec4ba890fcd786c64
bd82128ecc1096b39c9490581878aa02dcdf246a
104344 F20101219_AACALB woodbury_m_Page_137.jpg
1ec2cb96575122aa38f9999839e1742c
fd434c3a322436e76fb305f4c86b1b5f16f2ed6c
F20101219_AACAYM woodbury_m_Page_058.tif
10c7f936dc568b96d502ae6e09cf3c2c
b243b35a68b9f094e178ae6b39df77d429635d59
264110 F20101219_AACBVF UFE0015615_00001.mets
3b6f2d88e4d69a9ef5c6302abe1ad267
a93ed47b2959691f0bce2841421db4cda17c64d3
woodbury_m_Page_001.tif
96036 F20101219_AACALC woodbury_m_Page_138.jpg
c18f70039a5902f3eb674053ffb2ec2e
9b81e19e04af790a3618111852045efa37409a71
F20101219_AACAYN woodbury_m_Page_059.tif
a3e020185b8d6077851fa2d76a7aa5b5
d7f0318f97ef413c9b949294c04c61406d94ab06
10362 F20101219_AACBVG woodbury_m_Page_001.QC.jpg
1b850bcf73b4b9ba5deef91a1a2a5511
ec9ded8b6f935983ee0d1524af84631e25551563
99406 F20101219_AACALD woodbury_m_Page_139.jpg
6408a6ca579b16d91c26b33e5d266689
cab93968aa4f7d83565256d1606641f5d6ec8776
F20101219_AACAYO woodbury_m_Page_060.tif
2e8bdea7b9351e043f3acd55b72374d0
efcc0a82c4a8ace0897525033b4610f1e5b0363e
1821 F20101219_AACBVH woodbury_m_Page_002.QC.jpg
50bbb221fbfa27d65828be9cbb3fa772
693969dc39b1872f5efdcb74571206b533444fb7
100343 F20101219_AACALE woodbury_m_Page_140.jpg
5c85dba3d73fd4dc1f82fc633e9929cf
9ca1435c3085e62f5946933cbe896f8aaca30b1b
F20101219_AACAYP woodbury_m_Page_061.tif
27b48704d6e9bda8a1005f35ec9d9500
922f4ec0783df0348d543bfe6f8ab49d1b16528a
669 F20101219_AACBVI woodbury_m_Page_002thm.jpg
5dc42a1e81bac97df0d589a8654c0a22
847d7371454a1db200f9bf8c43aa108fa0b7da87
96975 F20101219_AACALF woodbury_m_Page_141.jpg
518ef2f7dc5a5471bc49cf139efa4f79
179f1ed77c0e5ae7345316e1d7b06bcc67c8f205
F20101219_AACAYQ woodbury_m_Page_062.tif
7f2ce9a27ca2992e79143d28152338d8
686be4071131ad2d43923e9cb9ad504318ec1454
1461 F20101219_AACBVJ woodbury_m_Page_003.QC.jpg
25f5fa58bd5e068e3bbfdbd7a28fd3a0
e4e7014e88874fa88644c5b0ce7c4a7437de4e78
111701 F20101219_AACALG woodbury_m_Page_142.jpg
fd78b979c21eefe5ba0d53190270e790
d714df467bfac5b51fd8ac3659b441633a2502f6
F20101219_AACAYR woodbury_m_Page_063.tif
0831ec3fa4ab9a6ff5ff3444f9a3daae
08cf772407b44c9fc98d6645cfa1a6b5b95187ab
589 F20101219_AACBVK woodbury_m_Page_003thm.jpg
189a9fd9dfc88b7eb6a4adc979f5e8c5
e2f74144ad61a7e6c9105f076d46339b8fbc375b
103358 F20101219_AACALH woodbury_m_Page_143.jpg
8961192dc37b6e0d13b2a87707f0d17e
d3d2c91cfc1e84ef2f23646bcd5e664dd0c76d6d
44554 F20101219_AACBIA woodbury_m_Page_092.pro
d831fa3d5d1bdbe19e8f35ea55f20448
7fb4e43fa19e1f669e1d58c0058522ef7a4bf1ca
F20101219_AACAYS woodbury_m_Page_064.tif
8850a72e041003cdf27bfaf57933d3ee
4a6c231bbcc21dc96e59a33a8402b78d04602419
29838 F20101219_AACBVL woodbury_m_Page_004.QC.jpg
6385192453df1b7416bb953bb4a1560e
0a9239189a160aa8122b1d863c0177447a665cf8
104287 F20101219_AACALI woodbury_m_Page_144.jpg
348a714704ffaa9fb0c163a2d630b31f
3068d60209137cde1778530524f6732f3415afd9
49523 F20101219_AACBIB woodbury_m_Page_093.pro
24dc1dcb0f318fd7fc99b31722139461
806826a198795945bf4194d4a1e3b799668cf266
F20101219_AACAYT woodbury_m_Page_065.tif
23b80fc0a1de84e2f8f3c43af4b49097
38f624adea16502b1e95a323c345477214f6b9dc
7411 F20101219_AACBVM woodbury_m_Page_004thm.jpg
050ea95657c2ea05b7f80332480a159e
9d8c9082528f1194b1c13d3e17df271aea9e2972
95956 F20101219_AACALJ woodbury_m_Page_146.jpg
e10677dfb76c7f1065bbce4530ec7065
872a4ac923c5f6e965000297bb0f6e3601ffde60
33473 F20101219_AACBIC woodbury_m_Page_094.pro
b486aac0e5c7b05f4901c130dfd333fb
b5899ce375bdfa25a40e4c3568f00318b1349e8f
F20101219_AACAYU woodbury_m_Page_066.tif
0dfad3b9641ad181e504fd9600db9b35
97a52000d9f278af5e1b5a33b038e0fe5a647c15
32813 F20101219_AACBVN woodbury_m_Page_005.QC.jpg
4b09bfa0000b14a04da2e7091739398a
c240d827772fd3c164ed120b69f57a3d1277f494
98821 F20101219_AACALK woodbury_m_Page_147.jpg
0d566ef971af8264342c3ec31a46ed3d
203070af86ccbf978c0e7e2388a1c9ac28d2caa2
72206 F20101219_AACBID woodbury_m_Page_095.pro
6f6a18f9b4df8633dd515d6d674ac395
612eaa377bffb56da869ead29d7a1f5225e49014
F20101219_AACAYV woodbury_m_Page_068.tif
04448e11e49c58b2e43e80c4276bc70e
4c9feb062538e0ce1d9059e58924eca8544d79e6
8279 F20101219_AACBVO woodbury_m_Page_005thm.jpg
4f5035653db42efbfc6be7f8a3adb621
c9f92fbffa3160a0b95a651c6f503a4df4f1b92d
102014 F20101219_AACALL woodbury_m_Page_148.jpg
a368ae774ebdc5a45475733eae6d22df
3cd781333a0886dc34928474c71af048524a5a27
44742 F20101219_AACBIE woodbury_m_Page_096.pro
6de27f783aa7bb6afd7cbd0733c23362
6df47fbe9b022f7086d662ed836abbf2eb98a667
F20101219_AACAYW woodbury_m_Page_070.tif
769f8f7ea23db43b5fc7732b52fa006b
d85b4fc20d2e7815de0e8ebbc179f1f47301b87a
25364 F20101219_AACBVP woodbury_m_Page_006.QC.jpg
62612dd574b34cb0472f0fdd366eb6f0
b80cdf876b2278523a1a0449947778943dcaf7ab
103457 F20101219_AACALM woodbury_m_Page_149.jpg
cf36bbf9d86a881932d1b3d7479fa545
081fef2835606a303a3c0cc03a7dc96e9bda7b84
47024 F20101219_AACBIF woodbury_m_Page_097.pro
897e93acb0ef85e6e1f232ef26f5a71f
be6fa084a6244899dd48e8d913155e8ea1b5943b
F20101219_AACAYX woodbury_m_Page_071.tif
5510212804ceea5e654b94139ed5c6a0
8a8dcff1b5458dcf412f0a4dc7930bd3ac293e3d
6348 F20101219_AACBVQ woodbury_m_Page_006thm.jpg
dcdf6c1d89cc24ab19bfb458ceff1994
f2881d8eef263632e9ed7bd2bea6a026329dd582
106381 F20101219_AACALN woodbury_m_Page_150.jpg
17fa248ba1012680e8d1caa54d712cd1
e4019a02ef97968c08fb1a97a215ccec8e6403fa
50234 F20101219_AACBIG woodbury_m_Page_098.pro
5ef5f21465967f63960ab1e05eed4c32
1ca3eed2a91e0c006fdf740a9284017183c0ecc0
F20101219_AACAYY woodbury_m_Page_072.tif
69b4d08fdf66e41a2fab09671a1de835
be74f92132f7f4181628ebe407a9136979859b02
30150 F20101219_AACBVR woodbury_m_Page_007.QC.jpg
82126f7636c11f765ed40326f0c0638a
9d678a3360c20491a1dd6074763079f0b9f9b9de
102216 F20101219_AACALO woodbury_m_Page_152.jpg
1a4916f63b15a360f38784c92f776bcd
4f76fa062537d8374f83084697e0c2c4ece5c73e
50849 F20101219_AACBIH woodbury_m_Page_100.pro
8de4b9972cf29522a80824895a98d305
d0e7f95b389995e8b9dc959c4ddab28fdf051b49
F20101219_AACAYZ woodbury_m_Page_073.tif
63d03abc4e9276986b0ec65d8bb0e072
7257aa03f2e4e99437aa5a18130503c8892ba052
7898 F20101219_AACCFA woodbury_m_Page_138thm.jpg
7ee5e39c879597bc1068443cbeb3f959
03804d454b09281f647451fbaf8ff62e62333b83
7188 F20101219_AACBVS woodbury_m_Page_007thm.jpg
dd7dffb6e7e99c19227ee9641720957f
712d995ee99289a418260e780635c4d86ecc2685
107248 F20101219_AACALP woodbury_m_Page_153.jpg
6cc722b0c28e0b01c87a51df124c4710
b3fd822e2f3c2850e67d3368d863189183d512ef
50902 F20101219_AACBII woodbury_m_Page_101.pro
e935cf5163ed2958d8efc64182d7a8c3
17b1f052509e371febd4230b44476fe2676bdb44
32567 F20101219_AACCFB woodbury_m_Page_139.QC.jpg
be7d1f7a0fe438571101770400a33a67
863cfdd3881337439b1e5d0305046625856827bc
37070 F20101219_AACBVT woodbury_m_Page_008.QC.jpg
7d1df8fd2c51f14f3b93aed7e10196a6
9db6e38a457e3b90cea6d0a9d3b68ce86c9c23db
96727 F20101219_AACALQ woodbury_m_Page_154.jpg
57e63981e79189378fbbcb06dfb38be2
f39d6df7ed8925b353572075889883b3a433b4cc
43647 F20101219_AACBIJ woodbury_m_Page_102.pro
43a3ec97b4d1d7fb9a6a7b5fa5e14a23
fec94ec2521d6660b8053142fe56154faf6aa66d
8361 F20101219_AACCFC woodbury_m_Page_139thm.jpg
ab34cb7285240847b7009cea7abc15e6
5dd6b62eb1b8ad1547963c26eb17b8f77d38c3b5
8480 F20101219_AACBVU woodbury_m_Page_008thm.jpg
ad78b575ea360035bd95796a0f611903
1e72cc8a1d38c18e29c60d5f42e0cfbb5753c3b4
80003 F20101219_AACALR woodbury_m_Page_155.jpg
203b66d80e5507ba84637e9b45177862
195e9843def02cc5967d55d0ece3ef9c6e40184e
43103 F20101219_AACBIK woodbury_m_Page_103.pro
03ba75a6d22a634f4f3d33041f57fe12
7450e145e38538611270bab88ebc72a865fe92c5
32577 F20101219_AACCFD woodbury_m_Page_140.QC.jpg
c16064835aad13279a74e3ef34bb6167
9450992d82b0fbf59f3d44690ed415c11621cfd9
28050 F20101219_AACBVV woodbury_m_Page_009.QC.jpg
534175586b4c85613917cb34e913f355
10928ddc382c2e5bed1be0ba52627da5c3410467
38657 F20101219_AACALS woodbury_m_Page_156.jpg
a7056be40b4a8f141e04c0909c89f4b2
208339864e1d0eacbf1551f411b4417cd0c3c607
50826 F20101219_AACBIL woodbury_m_Page_104.pro
0a1bd2b9939330a87c90c7191d114a6a
700115e6368830afb1b86160faac768a52d84a6b
8288 F20101219_AACCFE woodbury_m_Page_140thm.jpg
f2335321e24a935a02faed64c76e1dc5
20d6be8d5f914b0ddfcbb3c93199fa7013e2e95d
6760 F20101219_AACBVW woodbury_m_Page_009thm.jpg
cba25c0701b6c6a76716138de765f573
4211cf8933ae0fc4e2f592b201440303f4d9b9fa
33883 F20101219_AACALT woodbury_m_Page_157.jpg
ee8c1265a0a27a1bc3d5b6de8a224a35
b9c7956bd64ad68f1d0a8168eb5f4658a2e067d6
53405 F20101219_AACBIM woodbury_m_Page_105.pro
bb2b9b59de567d4513fcb0929bac6d60
c888b57ef8518ac06d7836bc034e0a6a5285085c
30547 F20101219_AACCFF woodbury_m_Page_141.QC.jpg
0fb7c0e1d4506d1f8c9d115020a173b8
3a4765f2dd8158b7beea37e3cbc320536bb457f0
27701 F20101219_AACBVX woodbury_m_Page_010.QC.jpg
df4a29bceda59c662d264aa517889ac0
f2d21d9e2fbf91334c3ad0312d066c9e7f9e4a4d
49965 F20101219_AACALU woodbury_m_Page_158.jpg
3ac4cbeb46e99ec69df25456e29017f0
744cd5bbd1d8c6380ea8428222209faf77d011c5
48467 F20101219_AACBIN woodbury_m_Page_106.pro
239f484a1505516489d696551279c2d2
166e4c5082e41e17f25d2eaf59cbbb61ec2cfec5
7921 F20101219_AACCFG woodbury_m_Page_141thm.jpg
1c3a77fcacfb77fe39b8534abd13f1f5
0e6c1eb15ccec1e66a5f1c417ffeae78c44262a0
6973 F20101219_AACBVY woodbury_m_Page_010thm.jpg
00835aff0eec5ea7842be7ec4955eff9
e89ff8aedb02a8b5952c9f393e933d1c4a2a303c
44984 F20101219_AACALV woodbury_m_Page_159.jpg
1cf9be4d5c598d2c904b5a1a202c0412
e7ec3dbd303df762e45313ec7f6e54c662c50f8a
51973 F20101219_AACBIO woodbury_m_Page_107.pro
2804baf6fd5bd48183e15ce0696ca0f4
94fc7cd7287cbbbe73d36edf4c7e4343ddd1dfe3
37162 F20101219_AACCFH woodbury_m_Page_142.QC.jpg
10d3fc5299e8a0c2d77930067388da30
e4a34529f555c838acaee575029547f06c921665
16194 F20101219_AACBVZ woodbury_m_Page_011.QC.jpg
baa819da08226647a66d24da2182cb49
aaf5071669c3851ecf10bf096fcd4125c0fea7ee
88685 F20101219_AACALW woodbury_m_Page_160.jpg
fc49146491e01ce9b9261480001f284b
2a0419dc5b650f35156f9521c4232405b82177b2
51444 F20101219_AACBIP woodbury_m_Page_108.pro
e14c2e1cdc5599a30b582772471c1e2b
be38ea85c0b0745a7a55390d3005f63ac4a1c120
8831 F20101219_AACCFI woodbury_m_Page_142thm.jpg
751d6f524b1454311d024a10cb4a82b6
82769d3329b69d16182f69501a42452c9ca5bf61
45166 F20101219_AACALX woodbury_m_Page_161.jpg
ffce8380a637557f77b65492b7ff85a6
e095504d734c146ffaf4b04555746c2a68c30ec9
49471 F20101219_AACBIQ woodbury_m_Page_109.pro
d5c36dff2b12e9fa484426ddf471aea8
4b7b433050a1da50d0bf8c025a0bc1ae2dd54fc2
34019 F20101219_AACCFJ woodbury_m_Page_143.QC.jpg
a7ed0cf74a924b744b1320b87cca63a8
7b01f033e238caace93e294cdf030204dead65d6
99730 F20101219_AACALY woodbury_m_Page_162.jpg
4c96d7d74e9b6fbf35d060263ac0ecf6
057812a68d7aa7880e12074661017045b7395db4
50928 F20101219_AACBIR woodbury_m_Page_110.pro
85e1461c9d5efc88f61d0b99812c0fe7
17a3c5c51c8b33e46b7e9ecafb2a723589fdb1a3
8495 F20101219_AACCFK woodbury_m_Page_143thm.jpg
88fb81150b30119d9b8361c1fcc44887
59ae99348d695fe3985aa9f443fd781366145d56
60086 F20101219_AACALZ woodbury_m_Page_163.jpg
02b87acb7319015efd98dff37af6050d
e47579d8c88dbf37044c54a83e5884330046784e
51392 F20101219_AACBIS woodbury_m_Page_111.pro
625a43d947e4d158d0b47c5e828378f6
c3caf4eb0dd704597817df3240d3df7898cf2ed1
34249 F20101219_AACCFL woodbury_m_Page_144.QC.jpg
e1177380986c8fc93210b3933b07fcdf
0215bab7a37a405b925d79ebf3999510c9309970
51031 F20101219_AACBIT woodbury_m_Page_112.pro
ccc8bf7da594f1b60461f5aa17e273fb
62ca30ff835fb74575e3421cb960cf86f3198b5b
8693 F20101219_AACCFM woodbury_m_Page_144thm.jpg
f5b161e6ec25c7d4f4a0259899e51776
b00b4874212c6f9736f5d5c37540a53a76ed1f42
12983 F20101219_AACBIU woodbury_m_Page_115.pro
609dc1cf369aeed7c0d681395480cde0
fe1672e6c649d47f49dd83b7c27b7acf36f96825
32949 F20101219_AACCFN woodbury_m_Page_145.QC.jpg
875d37b722011f6d03222a830339eaba
c416b26865839576a13a3a469f90dc7a1c9e5e52
49597 F20101219_AACBIV woodbury_m_Page_116.pro
8b107d8bcbd148dacb771403cdc7fb86
291c8e469cec9bab24845ca48d24287197e37de3
8398 F20101219_AACCFO woodbury_m_Page_145thm.jpg
b58fa480f68360b07a1ab47c5e124725
787b3ac978d6f6144cce3267ffba02a22cd15bbf
31916 F20101219_AACCFP woodbury_m_Page_146.QC.jpg
9cf38ef0b3dc5786368cc54856b3caae
14f13d855d8ef3c0b016154ee9c5fd8a7e75d709
35276 F20101219_AACBIW woodbury_m_Page_117.pro
976aa9bb5a0f720681227e816882aac0
3767c3e0b91b6c8c028948038c223be9bc1911b1
8095 F20101219_AACCFQ woodbury_m_Page_146thm.jpg
d3b248cb7e96c4573ef11d504f93c3a0
215ef3e9b49154467bb995f50b8be71b2cf53a78
20006 F20101219_AACBIX woodbury_m_Page_118.pro
60d3e6393c9f78fa9230e5efa8c16eff
01580265f6a9017fd06b55f92e7179efbd39f1eb
8141 F20101219_AACCFR woodbury_m_Page_147thm.jpg
ab6ea6455e58db7ba4896f0f512f50e1
7e50668913e4b09c453844759e3d0b764cc68a86
45831 F20101219_AACBIY woodbury_m_Page_119.pro
61e1d2d4c0e2b3d0acd6566564d36cea
1bd66ee78457a19002f7b8973ebe95313ad55978
49569 F20101219_AACBIZ woodbury_m_Page_120.pro
cef3cf1e89f741f238d97a8ed14d8ed1
8ffa5ebcd32fadca0f9914369561de05e17ec5f1
34978 F20101219_AACCFS woodbury_m_Page_148.QC.jpg
ea0b8bab4e5e7676e9614aba073cd9e6
e2ca65c90ace459d7970765b3b8b307b073c51e1
8415 F20101219_AACCFT woodbury_m_Page_150thm.jpg
0e7bce663843ad1100c4cfba7dff021d
d09ef041865fd00764ac98ce596b8b10f1b2e4ac
34749 F20101219_AACCFU woodbury_m_Page_151.QC.jpg
10c9120a84177759392873fd0742cf54
de989fcef062cfbc57223d267c90f92a421f87f9
109729 F20101219_AACARA woodbury_m_Page_073.jp2
6673263b2839eaaa5d44eafc5ac8cb32
30194d56e8e87c280db80df913dc5e0e832606b8
8412 F20101219_AACCFV woodbury_m_Page_151thm.jpg
636f494cb74dadaac2664c4de415e5c1
86d160a95ea2a8e6d1a6ed31fa5b616ae4bdb109
109809 F20101219_AACARB woodbury_m_Page_074.jp2
732bc29ce18fd07b8f4dae11049f5c33
c79f02a772357d1de966486e45d3fe413948c2b2
33781 F20101219_AACCFW woodbury_m_Page_152.QC.jpg
9a0c2ca9c1a7b983ddd90c6874163e0f
cc9cd025bb610fd238ae6cedab5cc28c646f27c9
109035 F20101219_AACARC woodbury_m_Page_075.jp2
1614f20eecf81d8ae28a507277455eca
54a28f2115094da211ec3db2b5aafea648b47bff
8016 F20101219_AACCFX woodbury_m_Page_152thm.jpg
8948c5c69a1765c8f9355f39eafe36bd
4bd054ded959c3ad196be2e20be0c55667f60c08
108642 F20101219_AACARD woodbury_m_Page_076.jp2
ad761a7f644856a9a72f48a6846e9814
3f43fe8aabf684c717ef9de8616bb3603677f13e
34727 F20101219_AACCFY woodbury_m_Page_153.QC.jpg
684163569349347fd67a2f0d45a49c82
e19730ba076a883c096547882eeb91c2c8d1a361
1051976 F20101219_AACARE woodbury_m_Page_077.jp2
e616f4e696691f88dd121a67673f427a
c8452122aff3dcdc3297ae64df03231635262b76
8331 F20101219_AACCFZ woodbury_m_Page_153thm.jpg
e099d140ac5fc91945503a2bef259ce1
d8d3913f965c4433c0492a13c911e58f56436e21
101978 F20101219_AACARF woodbury_m_Page_078.jp2
61fcb3e489bcf8a08c4ef0496db00be3
1d12fb8c6d57e482212bea6213c5b04fc7be42c8
104513 F20101219_AACARG woodbury_m_Page_079.jp2
0b2741b7b8633db421fa0fd7c4eadce1
b803580c17c3cbc65280abf8e76cdba14fd1d989
1949 F20101219_AACBOA woodbury_m_Page_032.txt
f6d362ed4c7366f662d0f715c1d79660
4113940b135d22137889c739c6eeb0aa624ba9ba
114202 F20101219_AACARH woodbury_m_Page_080.jp2
4afc3c225e49224a92ca76cd204d26f4
f18915bea548b3e341bba4c3369eaa7f7bbfc4c5
1932 F20101219_AACBOB woodbury_m_Page_033.txt
8a8967d2a48157b6cde2ce4c8c8c1a70
21cce0817fe7882b9d5acb8d83ba3933bf468075
99690 F20101219_AACARI woodbury_m_Page_081.jp2
7a45f19a5863c91e41e2fda430ba889c
f931f42460997e6510df1da71497cd55c49daa23
1985 F20101219_AACBOC woodbury_m_Page_034.txt
bb3a8ff07700a2be19b62db1253dda37
769ae9e5061268c7da0fc1ed3c73984b69023e49
111961 F20101219_AACARJ woodbury_m_Page_082.jp2
0890f2d68b3b01cae27b436366875c9e
6dbeadf5b5b6c7fc76b36adcf945f3f275640fe4
1767 F20101219_AACBOD woodbury_m_Page_035.txt
60b47d7702e762b1ad8bc4a36f5e7b7c
f19966000a3959e0df32fde3a5c95947aa6bf837
110865 F20101219_AACARK woodbury_m_Page_083.jp2
bcc6db3448a857585fc012bded76ad9a
49018b2b105596178eebb5cf617f69d7a4596623
1976 F20101219_AACBOE woodbury_m_Page_036.txt
e77ba93a1cb408bf46be42f05bbeef7e
842a4c18347e26954cca990ee44f88472fdb8ea6
8237 F20101219_AACAEA woodbury_m_Page_148thm.jpg
50d1d7d49b7fa999bfbaed7a6ea8bbd5
c3c0f8b391462229af8a35ea513323694104907c
110190 F20101219_AACARL woodbury_m_Page_084.jp2
7c8bfd172d57c0737c046812d828c500
4be2cab3d1d846e599cc83a9d76d6b2286a62bdf
2053 F20101219_AACBOF woodbury_m_Page_038.txt
6605b4cdaf96f625ade0724901cc3f1e
5311a333220a7b4c61111f6e8c70026327382810
107644 F20101219_AACAEB woodbury_m_Page_143.jp2
bcfab4e51de38de5fa55dc695c803be6
c02b796d06bd922ceaaa271f915713598ffbf8c6
112220 F20101219_AACARM woodbury_m_Page_085.jp2
a23cf22350349e35f7896e9721fc5b6b
a0ae9b3166b3fcdde82c85ecfbe0b4da57d32cc3
2000 F20101219_AACBOG woodbury_m_Page_039.txt
63cb4beada62068a2cc54ed7569a70e8
9fc06b00ad0d2cb8ffbfd05edfc21a170477bb4d
44218 F20101219_AACAEC woodbury_m_Page_099.pro
a153bce0cc211f94131876c43d8ae723
6cfcaf27d0903738fe8a325e21517825d7358235
110921 F20101219_AACARN woodbury_m_Page_086.jp2
d08700f4bb22ed73272b1202dc97718d
ad85679a82d636bd2d65656af86d422a0a5380d1
1916 F20101219_AACBOH woodbury_m_Page_040.txt
4f8a49e8902591aa0aa391b0adecd8e1
74695f8d13763cb3ae71ffb8f0ff1559806ddd15
34985 F20101219_AACCLA woodbury_m_Page_224.QC.jpg
137326258eb31bc682f01f80ad10fe1b
633fbc88497f135d179e95dbc404a1435a021939
31108 F20101219_AACAED woodbury_m_Page_048.QC.jpg
0806f30121fc63d8f3c7040e4d03fc0f
692203f3944e6f6478654b32d339fb2deab27af6
104657 F20101219_AACARO woodbury_m_Page_087.jp2
3241ef5c815161a699cf80a92765d80b
628563134465fa460e882afc9d450697d04ace99
2198 F20101219_AACBOI woodbury_m_Page_041.txt
df6de98b53ad4da8bbbeed0a535c7027
1a140227ff150295d0e31a4ad72805c63db1152f
8557 F20101219_AACCLB woodbury_m_Page_224thm.jpg
35d2e97cec1e8aa29d127a873362b435
1aadfb169bbb709d0ae4b3005fbaf08393e707ba
97477 F20101219_AACAEE woodbury_m_Page_035.jp2
e775f3a7cc6b06ec5d99e46b78ae3732
137255bf912da5e9be8a27afd1135053ee2a67ef
75214 F20101219_AACARP woodbury_m_Page_088.jp2
67891ad1f8cbc8d1e1b2fca78892c625
883474b4d50d847679c2fe8e0fbaadd64a6e5078
2010 F20101219_AACBOJ woodbury_m_Page_042.txt
8be4de29e4b6a245ebadb51cdab10f0a
9b213aa7abdd54ca89b33463b896525c83f9883d
35454 F20101219_AACCLC woodbury_m_Page_225.QC.jpg
959278984baae35abca61b1d0ef882e8
81b5a1d2f41a3980d3e27dcc761701b23a03f17e
54321 F20101219_AACAEF woodbury_m_Page_011.jpg
04f141021e44d087c8618a2b8fffd085
f2fc3cefeea25ea4ffed1ea6a364bf809cb512ec
232465 F20101219_AACARQ woodbury_m_Page_090.jp2
99206eef553f6293539e22b309dcd552
c4c6c2730f0a3f12919502b54569c80d32600c40
1750 F20101219_AACBOK woodbury_m_Page_043.txt
cd221cce710e90c6ce3163aee7761842
e9a36473ccbe24804d786b7d4168db2cc866b4af
8476 F20101219_AACCLD woodbury_m_Page_225thm.jpg
2757f4acabbe6684e4a45a0101d72e34
9a21417d25c4cc41dce75c7b8c25994444fad884
2026 F20101219_AACAEG woodbury_m_Page_083.txt
cd18edde397f9ec3fec776f99c020463
bdd27ebd8bc43f26f83bbff89c235474eb468e92
825059 F20101219_AACARR woodbury_m_Page_091.jp2
d9d1106c98b17d3ed56a640f01f20f42
7de4d8fa2641bb0a64fa33d55026d1cd2b0ecad4
1827 F20101219_AACBOL woodbury_m_Page_044.txt
81949b9bc54252bd730d9977f8aa6f3f
3beb5253ffcb1e02d8555e699d02faf8c9558da2
36160 F20101219_AACCLE woodbury_m_Page_226.QC.jpg
1639cbd7b66c478c4a96dfb461d9c1c5
174ecd595941e94b221f63c4e985e7aaaa3dbd61
416954 F20101219_AACAEH woodbury_m_Page_188.jp2
8f17052f09f43e0b9d009e4e9811f705
a763c67428aa450d10eb79bf49d64e15a04bd7d3
F20101219_AACBBA woodbury_m_Page_128.tif
ef8ea1a6e614dc0a0f62249d7057ad87
826b4473223c48a9132c0526a1e1d5ae3b7286c6
1019333 F20101219_AACARS woodbury_m_Page_092.jp2
94d94bf2e22349ae5a2de137755f7a10
06d8bb1ee03080949445c8c9e3c8c53985775e18
1890 F20101219_AACBOM woodbury_m_Page_045.txt
dbba16df0faf3805f535cb4869bc829a
b16ba7e61d3acb1b3019e829e2a6e72ed49e2a8c
8870 F20101219_AACCLF woodbury_m_Page_226thm.jpg
4f510394b7af21c64b5570989f56627d
08d6c9917180ccdcc3a366a44768f263afaf071e
83778 F20101219_AACAEI woodbury_m_Page_155.jp2
3333f9ff9828d9e8db2abe44bc6dbde5
33b584a5bd30823dc36796447fcc9e065632bc72
F20101219_AACBBB woodbury_m_Page_129.tif
729ebdb4a6a9d3e2c19b4bf73b36cd75
6ba02036d234a1b933354cc11710ae0f344d77ec
48110 F20101219_AACART woodbury_m_Page_093.jp2
5c2b3396500f468c9d9bc01eabbe1abf
cfc6c3e1131f4fe92daac23a7028f835c7d75654
2098 F20101219_AACBON woodbury_m_Page_046.txt
19596173d3adb042a238331a152a010a
255d76194375fca6efed882075e894f21d6bea72
33739 F20101219_AACCLG woodbury_m_Page_227.QC.jpg
c90085bdca15df18975d75a8cb6463cd
bc1282a0bbca1c50edb90a81ccfaf1e8d0259bed
47298 F20101219_AACAEJ woodbury_m_Page_124.pro
3a419bce04c8db70cdd7bf92beee7d2a
88eca3a17d8150ee07796f2d8a8f491736e18f99
F20101219_AACBBC woodbury_m_Page_130.tif
7e31d301cd90cc46069f3fb66f824e8e
20c70e438dcdca63bb0273a73a7e416ca5d5a575
76627 F20101219_AACARU woodbury_m_Page_094.jp2
d3e4dea9bc80b383e6a1b52a946c8e6b
9faea6366a2ae8c5ab9ce47474082482849abb9b
1933 F20101219_AACBOO woodbury_m_Page_047.txt
6d0327df84c432cf9c0d67c759c44a40
41a97e28371adc60709fed9759f5c3b40fcd218d
8146 F20101219_AACCLH woodbury_m_Page_227thm.jpg
5a82dcac49f2a518faf8c14fb44f14cc
92783725446a19552c87ce4e425b872648b4adf8
8153 F20101219_AACAEK woodbury_m_Page_190thm.jpg
64faf7ca2517aa08004d6078744babcf
933e9b93aeca1b9286ac195e765578f7fba60c67
F20101219_AACBBD woodbury_m_Page_131.tif
70c64fd60ffcd9923e022ce3d779edfd
69f077e97f12f9b02fbb552b4053f4daca13fa2b
824380 F20101219_AACARV woodbury_m_Page_095.jp2
d6fa04893c2c4f80e30ccbc58c195423
b50062aefd3d7375d3d4a0a2bb0c513e91989a24
1925 F20101219_AACBOP woodbury_m_Page_048.txt
ded75aacf5057c1e5117ca0e1a8f9be0
bbcc695b337916740608a1349d3fc234916e1fa2
17337 F20101219_AACCLI woodbury_m_Page_228.QC.jpg
ee9ec387b66ab2b7b8f9a4a94f23f2c9
9156753a13841821857dd3003596683132665dcb
1162 F20101219_AACAEL woodbury_m_Page_203.txt
4e90df7f2a7bc5d7dab2d9d1ca47cf0d
001aaccf1bde8e721fdb4b9bf44aeb733462bd0a
F20101219_AACBBE woodbury_m_Page_132.tif
8d04cf51b732b2c10f566c0c4ee5e3d9
ab774c3f71a1499134b40b960bd092124a2da82f
98174 F20101219_AACARW woodbury_m_Page_096.jp2
bedf801394d99908534e2e2984ac2c8a
6e05c5222b5f36f50fb85d00b19faf31785ceae4
F20101219_AACBOQ woodbury_m_Page_049.txt
c4b2b91ac08c60795f153f1cf4810cf6
794b9b888e10516205aadb8e98294bf73392e649
4238 F20101219_AACCLJ woodbury_m_Page_228thm.jpg
169387786bbf19fbc74f04eede054a17
ab43cb615f425064ac86ed3fb61eed1c1ea7538d
32224 F20101219_AACAEM woodbury_m_Page_201.jp2
9572150c34860674d2c0379451093469
0481ed98d65854b39b15a754ab263ccecda97942
F20101219_AACBBF woodbury_m_Page_133.tif
60380f1b743d07c665f1bdf278cd5218
65ba6194d82039c9371f7a20d7417defeece5f05
102543 F20101219_AACARX woodbury_m_Page_097.jp2
fcf9d179d110e2fa4e034170fa90f636
452906c136c7bf0a70a0b36c59f5f01172890ff8
2096 F20101219_AACBOR woodbury_m_Page_050.txt
e0b22983cd8c80c2000c06e3ae2d577b
9281098ba95121da1a1b6c36fb8f71f80a8ffa05
32273 F20101219_AACAEN woodbury_m_Page_100.QC.jpg
bfd7275ea6013c0beda037051f129b20
429bec313f97bb4d9d912db829e821ef65f36eff
F20101219_AACBBG woodbury_m_Page_134.tif
7e99caab2b3f4b6b0b0f831db506506b
6b318401eef3decd2df405bb233f697b291b23b0
107947 F20101219_AACARY woodbury_m_Page_098.jp2
b35dd90400b09e26612f588132a97490
3b9855db0b2f7fc10e8caf7d1bed9b966b12e704
1945 F20101219_AACBOS woodbury_m_Page_051.txt
88e8bcaf1fbdd2ae0f95a67749870013
0eea54d30d0058f284c4e2b1c3dca2740f7100e9
2005 F20101219_AACAEO woodbury_m_Page_101.txt
a95cd294e9080e82f7e58cba04b90152
400f0be5a5a2d52f8b0cea853c9b81b5201c41b7
F20101219_AACBBH woodbury_m_Page_135.tif
0fb802829a2a69c59b12e95a4c2092d7
d2ca3c9091e08fb2f30376c2bffd6ff56c137e02
96502 F20101219_AACARZ woodbury_m_Page_099.jp2
97d18ae3e3004f819c070384c86b438d
0ac90679d069e012156514dc4fee1304662092d9
F20101219_AACBOT woodbury_m_Page_052.txt
4499dc724f00559b829eb3c281516afe
eadd036ce6116a2063b94a84d2f17b6f6ca4d376
88551 F20101219_AACAEP woodbury_m_Page_103.jpg
f1e3225439a8754cc06e4979227a34eb
58bf0d63d00eb1a6f33acab941aa681bb2ae9334
F20101219_AACBBI woodbury_m_Page_136.tif
7c66e9000768079b91e1dfc79eefe065
6c3fe26ba8b906420836d9b4dee33ee084e95804
2020 F20101219_AACBOU woodbury_m_Page_053.txt
b5db7add6b7774dff9a8a38a4e7ff420
a430b4cea5dd0034ff28d47ca780239b76151f2f
130143 F20101219_AACAEQ woodbury_m_Page_224.jp2
c915e0ab778e567698a80cb1d5f5c3d3
d5c4f293c077b9a259d94cabf1709c30e6d2c946
F20101219_AACBBJ woodbury_m_Page_137.tif
167f58466f331f037bddf173007906d2
849df43a26ff202fc29980fe2f080c518796c00e
2001 F20101219_AACBOV woodbury_m_Page_054.txt
d5c5e52e91b678095a2fcbca18753169
fa113ca97ff76691454d403685938988cf1d12a5
96504 F20101219_AACAER woodbury_m_Page_097.jpg
9fe91f00e065690ade679adcebd226cb
a67ee6ade6ab93c059ad0c27ef033bc8ee2c1051
F20101219_AACBBK woodbury_m_Page_138.tif
3218365667886cd6f640c55c55c17dc8
d8cef9f3906ed7e15ec2246439815fb08641d951
2012 F20101219_AACBOW woodbury_m_Page_055.txt
413333cedba7511abb2d5339347abfa1
18906ced32b7a848369327c7b1e9a87be1bceedd
1959 F20101219_AACAES woodbury_m_Page_190.txt
0cd49a27c20a320115078cf2d75d19ed
632a272fd13861cc707a11f94202f2a81c2dc5d9
F20101219_AACBBL woodbury_m_Page_139.tif
582e13eb1dd11bc36419dcde9fff66c8
6e12aa5783703e66bb43852644cf27c9ddb6ae85
F20101219_AACBOX woodbury_m_Page_056.txt
8bf29faa3a2eab39993da619c2d8f19e
470dddec7033168cd14189aa14bee824a5917248
F20101219_AACBBM woodbury_m_Page_140.tif
dee96c3debefa4ee48104333a77674fd
b234b4c981a329889ea7eace529601f78813cd52
1966 F20101219_AACBOY woodbury_m_Page_057.txt
ae65c706068ccf3d430309cc728adcf7
e3b9737c2f26b9b2c7c1cef570c0db4c354849a9
F20101219_AACAET woodbury_m_Page_156.tif
11b69b5c5e9dbc637a82df3796eb503d
9da846410eda3eeae113fbd1031f8428d5c9dded
F20101219_AACBBN woodbury_m_Page_141.tif
37d0173d80302d218b2a9f7eb6c72ae3
63c339f13f3affea97c28204eb608d99ebeed83a
F20101219_AACBOZ woodbury_m_Page_058.txt
f6536e76327c46e6846db8a1dde6b578
559d80cdeb71d7560ea100f3de7528c7cc7b0171
49711 F20101219_AACAEU woodbury_m_Page_016.pro
d41493a0d20128f02241101d63b535ce
b1fefddeeab5b023aa7fcdd50383242737acd886
F20101219_AACBBO woodbury_m_Page_142.tif
80509eca8300ff7c28c4ae44b539817b
89212ea23cb71d9e3fdb5d9dca86c53084ac4911
104871 F20101219_AACAEV woodbury_m_Page_151.jpg
3c34ff8caf988908f3ac7a1416ad32fc
da746d1fc6c35014e1118c980919da9259f23b51
51593 F20101219_AACAEW woodbury_m_Page_085.pro
340c6c52252f62211452b30a93afe268
4103412446019fa60ce68b04c27a8b049c0a3ae7
F20101219_AACAXA woodbury_m_Page_018.tif
e1724df2c38b06ec8b061096dbccfa5a
c77b48536cc8f60e80377062dcbe0e7b7e824475
F20101219_AACBBP woodbury_m_Page_143.tif
ad958b1939902c403bc5e6eea9ca3154
4c75c6f19f8d7c28404025922f5786ff28a1c4b1
4267 F20101219_AACAEX woodbury_m_Page_194thm.jpg
1a0cbfbf7574a5fb6c76c03907a1205b
7bc3a2576dd91348baf945153bf5df0de46715e4
F20101219_AACAXB woodbury_m_Page_019.tif
9cc55b2156bc70e8329b7f64278ddbb0
4f36e8aab07ea7f386e23e83d1ec568ef827ff71
F20101219_AACBBQ woodbury_m_Page_144.tif
f90050eacc2bece603a7d622f403634f
d97904cc40e3b5965f75f38abc63fed7a7d1949d
29184 F20101219_AACAEY woodbury_m_Page_123.QC.jpg
3e25939bd5129e2914f9649edbb65cef
4ecdba03ca68ac8b9a4420f46b68b9092454a9da
F20101219_AACAXC woodbury_m_Page_020.tif
a8e3db9af33344de0db819cd46819edc
521ef934fc7c243d0f4f451b25d868a932fb4ccd
F20101219_AACBBR woodbury_m_Page_145.tif
60797586f8d1fdcaa6978d30393013ba
1f246c44ea113482cf2f134b26929fc98612eb78
65290 F20101219_AACAEZ woodbury_m_Page_198.jpg
201a133599dcf805219fcbdff63239e5
92624bd854c8a7850e4027fa4aa67df7cf46abbe
F20101219_AACAXD woodbury_m_Page_021.tif
7053e1fa226f70f911dabd9b47ca1d86
f20d47acd7886ea6a77647126aa12c00b6517b66
F20101219_AACBBS woodbury_m_Page_146.tif
86c1db3c91c6d92fbc81a4e4459e1a7f
53fca74a42338103f25506e1a3cd8ed8256c1c59
F20101219_AACAXE woodbury_m_Page_022.tif
74ba7046d5bf8eafdcf53f463a967853
428e1cacc32455aa3cbd3ad6cbaf9b33ada9701f
F20101219_AACBBT woodbury_m_Page_147.tif
1003c67b057fe17cbe8f0fdde418cd22
197f27aa3905cb5fd9eaaa4ff2b307f4d2f85771
F20101219_AACAXF woodbury_m_Page_024.tif
a2700c0fe790046d3bb021a7115a4417
c0611df6f6320996fce7f33b8795f5d806f89a58
F20101219_AACBBU woodbury_m_Page_148.tif
7835bf2d0a79fd2642ce1a2be71272c4
da0aba46770fe1160dca96ebc60161002b34765e
F20101219_AACBBV woodbury_m_Page_149.tif
391678a3c28bbc49b130a556385dc608
0b17f2f820837108dad2895e982741a03a7ad2ad
F20101219_AACAXG woodbury_m_Page_025.tif
593fec8ca2820dd23ac65350e1534409
759aba940be5de4a8667b89fece8a115f99ab7b1
F20101219_AACBBW woodbury_m_Page_150.tif
34a9dc357ddcc5743787425b5e1263c2
0814d1e9d00af8613057f842a228124ac75ca445
1119 F20101219_AACBUA woodbury_m_Page_198.txt
801ad863c5f21717a005197b9ff5b06d
00c570e11765bc4500e4a3335b48d5c8ad268d29
F20101219_AACAXH woodbury_m_Page_026.tif
c4182b0569c367f50ccb79312b735206
1cfbaf4c6583a4a171da45a76b029e432ee35962
F20101219_AACBBX woodbury_m_Page_151.tif
abbf4aa15c7eccabb63c0df4007b0512
414a8c56d22e13dc7c98ecdb0b37a9c4650f6f5c
1395 F20101219_AACBUB woodbury_m_Page_199.txt
581cf558ea1461e150026447f5b196ff
27f08b4a773263ca8004155cccbde9b3328a7856
F20101219_AACAXI woodbury_m_Page_027.tif
d88965d109d3857be914f07af3b195dd
83531762a8fdea18666866a36577f0ac4204191a
F20101219_AACBBY woodbury_m_Page_152.tif
93857562576169a288354191d107555b
11c9fa2858a5cc7a70ef0d2d60d082dc4956204d
F20101219_AACAXJ woodbury_m_Page_028.tif
c8e851b13ff2488e0ed22d9f32a81c93
1a6b6aea0be39132577bdb2be9d8b74bc7b25503
F20101219_AACBBZ woodbury_m_Page_153.tif
167b08d4d2630faafd42558860e8c40d
37d532437b40b630ad9aacc3f9d9ec24ace4ac8b
F20101219_AACBUC woodbury_m_Page_200.txt
f89fcf00de8ce7bffa480a9c1eb179cd
50394b319be408d60738b369b0c64010c2053815
F20101219_AACAXK woodbury_m_Page_029.tif
1f155ec51a66d957b346fb571a748f81
865f0459dfdb4960fa49f5b5fb22dff1f440eca3
568 F20101219_AACBUD woodbury_m_Page_201.txt
a59f26148e94c1ed43dad5ff234d392e
7aa3e3f2c386f106ead51288bcbaf3476b5b88c1
101005 F20101219_AACAKA woodbury_m_Page_108.jpg
d8c8bbd313dd4b91d60ef0d70dabade8
6b1aed77f06eab48bfee58d90dcdbe6b545921e6
F20101219_AACAXL woodbury_m_Page_030.tif
9d173f1d20cf85dbcd5aade080ba10da
aba057499e70d37c7e2fc080b22e79dcb7c40e8c
1448 F20101219_AACBUE woodbury_m_Page_202.txt
510ed8bfdb21563dc2b6e96cb0da2348
5c5cf5ac3bc937aa0405ffe01b2f0d04baec483e
100642 F20101219_AACAKB woodbury_m_Page_109.jpg
a9348a16f0a4463517475fdbc5a8691e
b55d6cb2f22df29d04ffadf9b07a75dba6c0b5ce
F20101219_AACAXM woodbury_m_Page_031.tif
b7e3f6d14123c201aa67aff3d27aa349
9bb74bed9023225c34b500235ccda2642b54c307
1695 F20101219_AACBUF woodbury_m_Page_204.txt
20cb62e07eb0e162330b30ab6a913636
79a7017711cc050b5f4fa6dcb9e0281a73ba0153
103104 F20101219_AACAKC woodbury_m_Page_110.jpg
6cfdd5608b9b4bed04844a7d12ae04c7
db04b90b331cda0e27b1e6935297f7fa51bdabd4
F20101219_AACAXN woodbury_m_Page_032.tif
b25055b6838d62ce252c2b1fe5bc8d99
a00410f76a94c7beb4af3c01786d61d0967c3cf4
1198 F20101219_AACBUG woodbury_m_Page_205.txt
5fdbd2b9d12d53cd3f3885b41ee4a6cc
55ce60fe9683479de8dcfcc31bc962197566ff92
101679 F20101219_AACAKD woodbury_m_Page_111.jpg
740b743aca998655b4730e9202aeaa29
c225b39355696d6695aeca26678a62607518a5ae
F20101219_AACAXO woodbury_m_Page_033.tif
ded75a3f20f482baf8f4164aed2ef422
2bb3fc1e0b9a4d2a02b9652c80a671d295a30b2b
2200 F20101219_AACBUH woodbury_m_Page_207.txt
06facea19e4308cd17a2fc930452b5d3
897d1d397d8057f4594d0f09b2e14ab0f60f56ff
104278 F20101219_AACAKE woodbury_m_Page_112.jpg
a637a60fe6276686b9e8c36ed81908d8
0864c451167cbc0f527f48ae3fe6a016b8a10af6
F20101219_AACAXP woodbury_m_Page_034.tif
e7cf109766c28110a14ac5a75adaa700
3768bdd0697f65acc94df231e1e0f8ab14e15628
2553 F20101219_AACBUI woodbury_m_Page_208.txt
e5c6870c2083e4a2ff63dd4d760c615e
67d491eaf8c5caf3bec360319a5169b407294dc6
98253 F20101219_AACAKF woodbury_m_Page_113.jpg
e6c048f0564a311115f7dfa2f8d67859
3d2394d9c4786c458c055505ad40f039efe49836
F20101219_AACAXQ woodbury_m_Page_035.tif
0f09e1e2f786fba136b6993c05b0fd37
078230dfb0927298503e95c6dc608a4dac481300
2619 F20101219_AACBUJ woodbury_m_Page_209.txt
c25bb12aaa3e0be04485f1a56f60efff
6a1388a7995def1c1ec61af0e4bbecd1a0946388
52396 F20101219_AACAKG woodbury_m_Page_114.jpg
aee9a201f069f27bbf23caf19a6c616a
c26d3c4df8e84b4005b4ad7cb9520b9e356d840c
F20101219_AACAXR woodbury_m_Page_036.tif
d08c7bf852c8b1c77791410f04b70681
4958b1b1ea286e794dad8a6d9d0f6e22cd5b21a3
2504 F20101219_AACBUK woodbury_m_Page_210.txt
d6847b8441c0495b53ca5a1e65a7c8c3
d0229b234ce4dd022aee2a8db3ffc7062bf0b986
35532 F20101219_AACAKH woodbury_m_Page_115.jpg
bcf1ee3c2624a6391965794c4cebd93e
2f29d1707aaaca3e68028e0f0d0c3c0c1671929b
50845 F20101219_AACBHA woodbury_m_Page_064.pro
1db87b38cfb3e1737ccbac177911be06
b48f26e48ab9e7c51a9bcd805e233ecafa09a780
F20101219_AACAXS woodbury_m_Page_037.tif
31ce2e46ae625a34df163a5090c9faa5
b28b9b356609a7ba349a889ec8cf1ff1ebc7746e
2610 F20101219_AACBUL woodbury_m_Page_211.txt
3e0461411373cfa9ca528b1d7d722fff
63388e4995a0cefaf13d1a2fed9aab383db5f113
86089 F20101219_AACAKI woodbury_m_Page_116.jpg
8912111e92f64a6f7f09fe341ee3ea25
8213698cb25e185fa53558e1a3c83268f0f77da1
47823 F20101219_AACBHB woodbury_m_Page_065.pro
9e998a1335ab427ded648ba36606e878
27020883d4d9b8683a7cfdc5ff699091327104f6
F20101219_AACAXT woodbury_m_Page_038.tif
530c3bea223553d0ea006f836d471546
4d18f69d4a602fbeabec69e82c721fef3fba0cd6
2296 F20101219_AACBUM woodbury_m_Page_212.txt
91130a641ba51a1417846435d333e095
100474a24cce574094465795b0be8da444d13b1c
85621 F20101219_AACAKJ woodbury_m_Page_117.jpg
e016e419d0fc6d8e95cfa055e7c8b108
ea2ac386f11a78c8307cd02733a670e29e831a30
40091 F20101219_AACBHC woodbury_m_Page_066.pro
d869628fef109417eee037b542226092
f69df8005ff3ee29a8b2d0799cfff60da0d0f3a6
F20101219_AACAXU woodbury_m_Page_039.tif
8c7c1e84aa401216344638a1e0755f05
cd526b39ed39b4bacdb5fb80c5a50a4e5312950a
2327 F20101219_AACBUN woodbury_m_Page_213.txt
bc0a1d90bed7c4e8520fd03e013b7785
54ab20feee99a995a50f89b6f971b9580887a01f
62285 F20101219_AACAKK woodbury_m_Page_118.jpg
682cbd5664682927de3c2806471de70c
3422d928dc83bc8efffe30fe7d7cdb441c19dd61
F20101219_AACBHD woodbury_m_Page_067.pro
651d831e4d5ea63bf90bc5c2cf8c3219
a1431f91154712f7a63be958542dfaf221244f22
F20101219_AACAXV woodbury_m_Page_040.tif
600a149bfd9f1f927676ef0d450d1d05
8041d3ba32a31f8c1db65b246cef3e5a94aa2ae1
2354 F20101219_AACBUO woodbury_m_Page_214.txt
93feb7f97862323d942c0eae10116514
b2bbfef62c77c966f3a96b9b2df450c9a7264ee6
96051 F20101219_AACAKL woodbury_m_Page_119.jpg
f7e2ff0f09be7c428ef02474f63d0037
4ff60ad3b0bb7eb029479a90ee4dfddc605269b3
44687 F20101219_AACBHE woodbury_m_Page_068.pro
2d278b171f004d99e0790e2445c150a1
88c2c5b474c511f230c0df89becbf1c38c86ceee
F20101219_AACAXW woodbury_m_Page_041.tif
cf55cfe7cb53495fec14bc4bafb710e6
d5837bf68f6399af674dcc37d7d77512071123dc
2487 F20101219_AACBUP woodbury_m_Page_215.txt
7ceb55dc3e6a355bd8d810aee60195ef
659c7ecf2839eb80639943bbd8cb2a0ea5ee30d5
101096 F20101219_AACAKM woodbury_m_Page_120.jpg
d013fb7f5155c96b93e06fcc61be5d62
08ba6a05830f5ba48971e61493bc2df4e3bbfe2f
53509 F20101219_AACBHF woodbury_m_Page_069.pro
a04fa89e7bf2e69480c6c4e12cd6de7d
987c2bd8de1b0aff129310433c788aa2f3974318
F20101219_AACAXX woodbury_m_Page_042.tif
e9289e31c4b6620d14d04a25f83e6948
040f57cb63698d6053befd9a0e007b5bf9d44e62
2520 F20101219_AACBUQ woodbury_m_Page_216.txt
07cb5e4664b9872825c94f7ba2917642
6dc7ff4179404b88479f41f4e2a0600d0493e368
96346 F20101219_AACAKN woodbury_m_Page_121.jpg
828691fae99763c98b91e4b88d787412
8de5142159dae2714d773b7464ba4e10eeb528c2
55741 F20101219_AACBHG woodbury_m_Page_070.pro
654ada7f0d9c9b8103bbff921af6d5d5
3d55d7315d0d4ab3eb148da5720f75eb02ccafb1
F20101219_AACAXY woodbury_m_Page_043.tif
ac3267c5c5f652bb0bac39bb5d2008e7
8072d35987d69e37c34c9e2d026a18324bb07850
2587 F20101219_AACBUR woodbury_m_Page_217.txt
510e3848dc516ebe0058829906cc1ed4
4c784126a9b5bd819289e54fcdc3b9f38d40a262
105809 F20101219_AACAKO woodbury_m_Page_122.jpg
9732f30f048a1ed7ea7ac8dd90849609
b3cca0648e8f96933ea3fe25989992eab08b74e3
48247 F20101219_AACBHH woodbury_m_Page_071.pro
2e4da63544377b81859a6c61e1260c14
4b7ece02919fa62ae3fd0dee3ad391e747a5c5ad
F20101219_AACAXZ woodbury_m_Page_044.tif
dd6ab4357f8f5bc0a619aa346189da74
125e663e21df675b3bc88a4a8c99267942c51805
7745 F20101219_AACCEA woodbury_m_Page_124thm.jpg
3e54bc5485c31257add2ebf3b62d806b
8082bb997b142442491329b59de730c2cf99b485
2397 F20101219_AACBUS woodbury_m_Page_218.txt
264e05ce793234d631bf630816c90bf7
6e0ca673ec82f2d28dcee476de07335a6869f882
90224 F20101219_AACAKP woodbury_m_Page_123.jpg
2d1b60dc98ec51d263b98a89cc503cd6
c123e794c21c1a12bbdf53aa793a5d11c3d24af9
51181 F20101219_AACBHI woodbury_m_Page_072.pro
5c43ca47bd63c87e26045e2482884686
2a67c1e81b85cadbb536385dece58bb5acf4fe30
30911 F20101219_AACCEB woodbury_m_Page_125.QC.jpg
4998666717de029c174ed22151acd2f7
f572125842741b4bdbefc8a587511eaff0462893
2360 F20101219_AACBUT woodbury_m_Page_219.txt
ec805da6a8e05ed30605c1cdb1edae47
250c243603b77faef8390f4d761c957bd148fded
96371 F20101219_AACAKQ woodbury_m_Page_124.jpg
fb3a4d907f593f710ae70e61ddd168af
f31775d3a8e0d50352d00bf030a8159a16154cf9
50574 F20101219_AACBHJ woodbury_m_Page_073.pro
feb570b02f8b3966a44b5e6baa79fd9d
1921b711c75270ab7b669026193d10860fb3ff8b
7453 F20101219_AACCEC woodbury_m_Page_125thm.jpg
27c2bbbc9fb6926f0e531a13eea981d5
1affc97b2515720870713f068b0f6b6a72446c93
2623 F20101219_AACBUU woodbury_m_Page_220.txt
a963f4332da3ace36bc25057c2518899
746d91fbb43ac212dbd9052db08a7bcecb040a39
94255 F20101219_AACAKR woodbury_m_Page_125.jpg
b8d2a9030d0abe5cfaf362c33c44ae65
7c39438ee8929865f0361723211f9043baa1d72c
51545 F20101219_AACBHK woodbury_m_Page_074.pro
f0a243f4c34bfc3003ca7b3d645f203a
81564d0c4b287eded1389f3a95cba9865c6e7668
35581 F20101219_AACCED woodbury_m_Page_126.QC.jpg
4a57ede928835f878422bf2351abaa44
3c40fe85ec1717b676a8cfee16f7544a041451a1
2662 F20101219_AACBUV woodbury_m_Page_221.txt
04ede286b8426b64b9b42eb0fe17cec8
2655761c35781b67d85970788b62b5a71b335226
116725 F20101219_AACAKS woodbury_m_Page_126.jpg
6361cf509a7f422fae93b52005002fdd
5f51988b0c8d02e7b70c265af6db63cde3e90182
50590 F20101219_AACBHL woodbury_m_Page_075.pro
457e21d2306c7cc3e08ac915bfe13ab4
4b6e09535905146c8a60f7b5af7bf26c3231ef69
8120 F20101219_AACCEE woodbury_m_Page_126thm.jpg
210ceec0aa012059ea8374855bf38ba6
cab2eae87693dca5c1ae535f75c8e8342be7d6ee
2410 F20101219_AACBUW woodbury_m_Page_222.txt
71132705f2813761d60fd065f0bbf701
09a2a5c3cdd87676d6819c7bd843cd71f497ad13
103989 F20101219_AACAKT woodbury_m_Page_127.jpg
3991f38fb454a7f7b6ed93b0ab742119
7c68a8fff5463e74be7381cb85b94b6fa8a8734c
50372 F20101219_AACBHM woodbury_m_Page_076.pro
9c50ed1704049b361829ac2f17f296d1
00490d0f54e3ccf82ca8acdf3e8c8b4ba91cd63c
34190 F20101219_AACCEF woodbury_m_Page_127.QC.jpg
94b144e4accb573169ce1323b8bd54ed
dd29603769eec5a9c821395de90e4256159816bb
2428 F20101219_AACBUX woodbury_m_Page_223.txt
dfcfc22b715be819551f6b4097c67c38
85cc3b9ed77181b0943a205cd0c9800fe80f76bc
105338 F20101219_AACAKU woodbury_m_Page_128.jpg
72ab2fbced8493d87b3de9d585db7d19
2ca07be36c508da020d7d9512e6a852236a16529
52378 F20101219_AACBHN woodbury_m_Page_077.pro
995100a99a436c0557fd1a8694acc722
f677cc0342bc2013857f8023190bc9060129895f
F20101219_AACCEG woodbury_m_Page_127thm.jpg
c808fb8b48554bb4d3f5e020da732f31
3f9e110c062ba98e51b4effa17ab79a4da87c519
2517 F20101219_AACBUY woodbury_m_Page_224.txt
9b4fd08540d1b8e36cd8985f92f64c1d
54804ef7570e82da58aadae10ce5484d52c3b62c
98998 F20101219_AACAKV woodbury_m_Page_129.jpg
083f86a5b1a8054302054f8f927e4121
a4a6ded123bc335326cf224f1cd36d7747348ce0
47123 F20101219_AACBHO woodbury_m_Page_078.pro
d68044fd79b7a3329f3e4b58227b90c2
1a100523abad5387859385296d98a5de1a7016e8
33716 F20101219_AACCEH woodbury_m_Page_128.QC.jpg
76ea82cb632f80bc43bb81de32061e22
91f31dbfddb9d291a36f1a34f709b88ce053406c
2400 F20101219_AACBUZ woodbury_m_Page_225.txt
3d5d5aa87a1ce98c27a079276f54cf37
649f1c2ffdfc5c88ece3d28a89ed070f6178d9de
103757 F20101219_AACAKW woodbury_m_Page_130.jpg
ea7e5d0554fe69684f2985df83203ffa
b196cf2b3ca4cf76a403baaabf675e0060ad8a3c
48180 F20101219_AACBHP woodbury_m_Page_079.pro
0b81a390f5076a7ff158b67be5d47b3d
4d568d959be23cd8edc563f1d92827c03cec4456
8184 F20101219_AACCEI woodbury_m_Page_128thm.jpg
4d41c27bb7d5e0efc91652333ae3aa7c
505c8a2b84a8e92feda2bad1fe5f0a9fe0bebb3b
98717 F20101219_AACAKX woodbury_m_Page_132.jpg
a2bd8c12f1336ec11073133dc19d915d
174550410009a256caeea1c6ee9fe9e7def68859
46017 F20101219_AACBHQ woodbury_m_Page_081.pro
d3ac31b7282a3b166dea0ba20a06da58
b836c975ae10588862ec29cdfee409a61d559fd3
32481 F20101219_AACCEJ woodbury_m_Page_129.QC.jpg
9f2fae070aa1ab63566094b66b24bd32
9802210a6546ef2ff503e10496efa0ed5095855d
97976 F20101219_AACAKY woodbury_m_Page_133.jpg
37fefd3b07c8211fd922253266cc49e3
252e448bebca24933fc7c082cdb5ba0cb7e61fc0
52541 F20101219_AACBHR woodbury_m_Page_082.pro
eaa26c604518c8c85f139f76fbec2705
09e9e20ae26a93cbad604e4a52280e9a9d10998b
7889 F20101219_AACCEK woodbury_m_Page_129thm.jpg
ef7dd35ee319baa3698f46f65a04145f
6dac4156b401c53c9a16007d7d68920f88add751
51438 F20101219_AACBHS woodbury_m_Page_083.pro
5e75d07e74647789941153fba9ce49cb
bc4d38d478264fdb39f55cca6462b768bd4e3aea
34263 F20101219_AACCEL woodbury_m_Page_130.QC.jpg
3e46039acc87c7a1fdb15000daf33dc1
a06e7394c3cdd82535e354c632e2f1305647ee3f
104577 F20101219_AACAKZ woodbury_m_Page_135.jpg
607d5b5bc7a511794ad828066a090a96
22d0d176fde2eae4503a07baf30338d6ef6db611
50681 F20101219_AACBHT woodbury_m_Page_084.pro
e589ed7b7baf9638711b54fe5b044fa5
fc8626688bec3008a12c477343ac3315319c3216
8059 F20101219_AACCEM woodbury_m_Page_130thm.jpg
632f8b4f7c95fdc2bba377628cd5d4ba
228c62c67aa1bd2fa9f5b4d5dfc8434f8c4b2ca8
50731 F20101219_AACBHU woodbury_m_Page_086.pro
357fe0601228efe8d6f8f32051632f0d
a62096ab28953a4c9f728a5d7bcd81d556481789
33043 F20101219_AACCEN woodbury_m_Page_131.QC.jpg
e776041b45b70bf7faf1b4323e820526
8fe2ed50324ca40e1a018fe06d083242afe199fd
8100 F20101219_AACCEO woodbury_m_Page_131thm.jpg
e926afd670990091e32cd365d415ad0c
f7cd064ca4e55375406dba47baeef86c237223dc
48562 F20101219_AACBHV woodbury_m_Page_087.pro
51654772460dbe0e47fff2a574d10728
9232bf15d5191c4c3789cfa7d3bc61f4b0fb6cfc
32395 F20101219_AACCEP woodbury_m_Page_132.QC.jpg
0c836b8b1b85c8858bf80c90f7b57140
6b6c47a26ce0a854fcd48aea2fb8ee5672a6525d
34153 F20101219_AACBHW woodbury_m_Page_088.pro
56d6d570bc82cb3271c6ffe7c7cc5e15
f822775bfbb5dbcba4d4ced84fb0a67fa97f8232
8189 F20101219_AACCEQ woodbury_m_Page_132thm.jpg
6d45fc1f275c0fdf1f90b9252cfacd01
a197635cfa2bc4e767867ed954a690bd1cbd3f42
49253 F20101219_AACBHX woodbury_m_Page_089.pro
b5b5eb3aea8365494a7984d4322ff06a
868c020184b1caf20359ffcd7b179f019e66e7d0
10940 F20101219_AACBHY woodbury_m_Page_090.pro
377f5a82bc78882082e740f54cea6135
87fa662234d9de277fe325a28bfc046db77263af
32909 F20101219_AACCER woodbury_m_Page_133.QC.jpg
558e837d6e91b0f0c120484df9f3fad6
4eb7fff300084f8d288c5c3b04b362a8cce0b82d
34262 F20101219_AACBHZ woodbury_m_Page_091.pro
3421eae25b70941ca405cdae6ee721e1
4a3e5c5454847626f47fc0764e1025cc3c7c9529
7840 F20101219_AACCES woodbury_m_Page_133thm.jpg
d28181e7257f4889e4d4ba49ed1b1445
b4061e4e1882c131d1e1afa22747419beaa96774
31904 F20101219_AACCET woodbury_m_Page_134.QC.jpg
8344abe80442fc44d5135c368925f05d
e8f10f07725d65f5a5aa7ff40e0cb90e9b8ef6d6
34187 F20101219_AACCEU woodbury_m_Page_135.QC.jpg
0c58257b9c4bd2cfd81c72679ec8f7d0
8485dc1e767b6c7f8d86d91ce6aa2c9a8b1cedfa
104117 F20101219_AACAQA woodbury_m_Page_047.jp2
48e585796a66a01aa9300f5ec05edcdb
c359a55d7e4da2562df50a46e86ff627f21eb734
8504 F20101219_AACCEV woodbury_m_Page_135thm.jpg
0bc421fe027b144e27d8a8990b77ed42
f727e66855bd3ac22f6552c440b31966c67219a6
105660 F20101219_AACAQB woodbury_m_Page_048.jp2
ff3ca203762bd938b4599015f38164d1
36cf3a936cedc0dc445f98c84608c927188b1144
8629 F20101219_AACCEW woodbury_m_Page_136thm.jpg
1c24fe9c41693715357153297d0a8d87
d0e9ae458b74adb76f9e565ec8698bbd250f05b1
110564 F20101219_AACAQC woodbury_m_Page_049.jp2
00ad5ac0b637ba9c71048fcc059b640c
0386ded3923b6337872b419509bb2ecbf699c443
34324 F20101219_AACCEX woodbury_m_Page_137.QC.jpg
68a54eb24198034aa54b1e5ab939889c
0ae360e4558530b0ed5f28cd0b9cfd0ec89a2e1b
111348 F20101219_AACAQD woodbury_m_Page_050.jp2
3f9974747c50875e4150d4962a69d2b2
a3fd646711f398b128440763e9a63c39fbc71257
8243 F20101219_AACCEY woodbury_m_Page_137thm.jpg
fa618ed9f560078b82b08710d50edadd
abf4958e8371403e2ce6f44ad1dc3f212c8b15c6
107329 F20101219_AACAQE woodbury_m_Page_051.jp2
2bf1023b9a59f4bf67d14dd90094b14d
ce70f8cb4fa9b73beb5730ebac55489b06046c63
31160 F20101219_AACCEZ woodbury_m_Page_138.QC.jpg
c181748a119e979108566045196d01e7
03216a9476684afd6065b06ea9774f68508201dd
109783 F20101219_AACAQF woodbury_m_Page_052.jp2
c3c330cf390ae160d2952951d1eb408e
619f454964c3f4992d5ebad19095a1604f6cb6a6
109424 F20101219_AACAQG woodbury_m_Page_053.jp2
907135019fd848f1b815889e33551b27
ee6300dd5af2eaeee2e4498a7e806b77411b2121
2891 F20101219_AACBNA woodbury_m_Page_006.txt
8ba3fe4a8f2676707b2cb44ae412d0e6
465c6f6c5e96e48b45fd3142fe140daddc59a277
109778 F20101219_AACAQH woodbury_m_Page_054.jp2
61eda09580615430980650e0309835c0
488529f2471fadcc852f6b168a6adb02e8b1aeb7
3568 F20101219_AACBNB woodbury_m_Page_007.txt
df426c0373a0fc0a35f764e400136187
5661a79cfab6cf0f675da159de71f2bfd0385396
111739 F20101219_AACAQI woodbury_m_Page_055.jp2
d43486edd0353fa3917254f278283b3f
79e27ae07dc155f907797bfd2dfa6388b784d1c0
3956 F20101219_AACBNC woodbury_m_Page_008.txt
52c52f7de1b91e445691ceb9fb25f2d1
644fe61bf0af1dc2843dbdc96f3a4c8d6bacdd7a
111294 F20101219_AACAQJ woodbury_m_Page_056.jp2
9427f0eebcef98b84b3c3003d4222bba
2b31d3a9743a65c73cec1176625d0fbad5d254bd
2860 F20101219_AACBND woodbury_m_Page_009.txt
c1e8306deb83ca1277e76b21a5d5e5d6
61a2bc9cbf0662a8182ca702522072c4a6ba5bfa
108058 F20101219_AACAQK woodbury_m_Page_057.jp2
cd72d4ef99e508ba32bd048311d6bbca
e8a946077115e442fead036911c23f814ec9da2f
2100 F20101219_AACBNE woodbury_m_Page_010.txt
a65eb3812c21f59d3ca0cc2806b972b4
ed60f84f8e8c5a80ae32f13690376f260e28bc53
F20101219_AACADA woodbury_m_Page_092.tif
5d782f5682495c29f7775d5c8ea050c4
947edbe88394df3e733f37358b9074f14ab5dd95
112477 F20101219_AACAQL woodbury_m_Page_058.jp2
3810708d1de65f706499b841983d2b94
39f453ad7e4d05104b50a7901f11f4596b827937
929 F20101219_AACBNF woodbury_m_Page_011.txt
6c39a61e2f1227d9ea26482fc3ac620f
f59aafdebf648e8a770ff3bc9707fb8d4a617b15
8439 F20101219_AACADB woodbury_m_Page_055thm.jpg
df2f33692bccaae9a28fca80ef087f4f
6e00bce83cd000af80155d4c9aaed091476e2172
103621 F20101219_AACAQM woodbury_m_Page_059.jp2
7db0891874823a206efc8b5337a5829e
8b7a31c8752dfe8498d08d3a7bd65becfc577c7e
F20101219_AACBNG woodbury_m_Page_012.txt
bee24bfb27dabe33b55e565462da0257
efc8371eb744084e18a937c8df71700794c1db54
F20101219_AACADC woodbury_m_Page_067.tif
48ea88abdc40a4da82e1fc9fa9182a28
08c4f2f052cb7771f73d946c5f70bfdaa860abdd
106110 F20101219_AACAQN woodbury_m_Page_060.jp2
1d706c6096ccede9f10cf061213355ba
fd9b3d2ac37f6a8c62047613bcf85b3526fce5c7
1762 F20101219_AACBNH woodbury_m_Page_013.txt
ff90d85b1ed523648f4d2c55e6e11b41
f6152dd9e081f61148a4dba27ba1e5d69557f702
8338 F20101219_AACCKA woodbury_m_Page_210thm.jpg
912695a2d409e42700c051c3d8fc57e0
e3c7dfbd81d41538368e82899ffc1596656484e7
36826 F20101219_AACADD woodbury_m_Page_206.jpg
681f0ee80886ce0c8f27d66f91e7af15
4b9f64465cd026b693d43a6af8a1148a99462c36
101583 F20101219_AACAQO woodbury_m_Page_061.jp2
fa86da8f6f6e23c0902527088fe361ce
d536f877dfce6ede9d790fd3431c67f9483208b7
1409 F20101219_AACBNI woodbury_m_Page_014.txt
7d8bcdd298028e75b139088377dc63ae
3f844d553063aca323bf8c38ca1c6ad73a9b0b4b
37047 F20101219_AACCKB woodbury_m_Page_211.QC.jpg
5e41b35b07ac643b13c68cdbe4f0b49e
3f47e4738f9dce2434ba28c2b742308e99656525
35114 F20101219_AACADE woodbury_m_Page_150.QC.jpg
a01e109387eacaece146a87e1a8f6517
829feb452419c89cafa78e65ae48d15c820809c0
114533 F20101219_AACAQP woodbury_m_Page_062.jp2
dd2aeaa9579af652f9cb7f40309c0da0
8f802a30eb960fd393560f50b78e59500157838d
F20101219_AACBNJ woodbury_m_Page_015.txt
e11991dd181b0a5ce523614a07352d98
bb8db63181d0afe7e4d492dba8086dc0fa7d735f
9167 F20101219_AACCKC woodbury_m_Page_211thm.jpg
603032611baca57d3aa36bf32addfcf8
5933990f06b2d2fb417906a9f928198fdb202c09
59924 F20101219_AACADF woodbury_m_Page_223.pro
412c6b00eda72c77ffa3f6917a9173bc
38ddb891f4447ac264649296b1e83ceeeb760acb
103453 F20101219_AACAQQ woodbury_m_Page_063.jp2
d81f8e85d94a195e0109832a47abc646
6c5328425688529c617f97ea8bb84f68335bb566
1957 F20101219_AACBNK woodbury_m_Page_016.txt
287d91e68197476ae247c5701a18317a
174db0db7d2a2b91e116c31a3ecadcc3cf73417d
33658 F20101219_AACCKD woodbury_m_Page_212.QC.jpg
257df0d85c9d37b7457a753acce3a98d
57d45a4a131db7c60cdd98ccf8c3144a5c57b291
1051979 F20101219_AACADG woodbury_m_Page_012.jp2
72855502afe2df4208bbc4fa29d57e62
5c37370257c20330da0ae09f377a6026e3d12258
108363 F20101219_AACAQR woodbury_m_Page_064.jp2
a6beb73f3b192adcbb5d841d74cbc5a8
dc44314eddbe352bca681d7379e4ae5a3f51342e
1940 F20101219_AACBNL woodbury_m_Page_017.txt
6a84b71d2bc5fa8ff3e005dfd2678fe7
7eef125f60b65a26a0e055adb0bf222c63878452
8134 F20101219_AACCKE woodbury_m_Page_212thm.jpg
8a04bfdd4c3778873f4e4fb88a8b781f
49a03c8bc0eb464bee1efdb25b76ad338bb08620
33740 F20101219_AACADH woodbury_m_Page_149.QC.jpg
871c7db3c82b375fc880a51f04cef5f1
12fd08d7a0d558a69d49c3bb598fe9f0bd6ad063
F20101219_AACBAA woodbury_m_Page_101.tif
3903de4dde4c7eb2a647e8ef0a2926d8
c852becb7461855511843ed1e4bdf9bff2e0b1ed
103991 F20101219_AACAQS woodbury_m_Page_065.jp2
39fa6ed9ed326b6faef228e67f1d5d9d
05df1aafed6c74d843076a82e4a096dfa15cac13
2037 F20101219_AACBNM woodbury_m_Page_018.txt
8943e28c9842de754b1655c897db08bf
a6844d9d6b2d47623b2589054e74b22f96c220d1
33713 F20101219_AACCKF woodbury_m_Page_213.QC.jpg
b6ea97750fe06081edbd210b9b66b75a
4ccda01dea17447b9247ae44eb865bf6e1be1772
7978 F20101219_AACADI woodbury_m_Page_041thm.jpg
ac72c9fce35af167edf769bac60a2a66
9184be70d008b8c753b3f485db1d2e302df7e581
F20101219_AACBAB woodbury_m_Page_102.tif
649e55234748c8a8f548dc8b0c38e555
de3885b023c9507abab6a25b99846a3d7d3257b2
914263 F20101219_AACAQT woodbury_m_Page_066.jp2
0dc74e6db025ae128eddc435249ace44
95272a041b98754ca0ed41083f2c2b327fba849a
2057 F20101219_AACBNN woodbury_m_Page_019.txt
50dc6196784c1d4654b2906f8d67f3da
3d09a23c99239f65686f54ab73f0ceded60daa57
8375 F20101219_AACCKG woodbury_m_Page_213thm.jpg
2aa1de73efe7174275ff0e65c760b1b2
29dac05854cca59f48d275635a5346007133cd7f
52482 F20101219_AACADJ woodbury_m_Page_023.pro
c58ec2ced0afdcfed8fe55ffb19b7145
b1d830cef0cba2244c9bc0dbc70621d89807881f
F20101219_AACBAC woodbury_m_Page_103.tif
bf879d5ce293ae9f6ebb095c52bfe25f
0759e01f509832428d1eec90693c32cad7f48e13
22250 F20101219_AACAQU woodbury_m_Page_067.jp2
adea53b9320e93d726eaf530d67e6afd
169a96dc63b2b73f45dc016d41273e6a38c9b4d4
1864 F20101219_AACBNO woodbury_m_Page_020.txt
58dc038c681c1f1efdeea713c6071971
ecc8c94900bd1708554d221ec78de987736d1917
34585 F20101219_AACCKH woodbury_m_Page_214.QC.jpg
96e1666f2a37bc31406dd320cc93e7fb
90c979ae4fdc77eab874247fd251d34a91d215ce
7564 F20101219_AACADK woodbury_m_Page_119thm.jpg
f8dc994b39bd5cc37074a45656498301
804bdb96330948c89b45287b186765ba8dff553b
F20101219_AACBAD woodbury_m_Page_104.tif
951d196078e590c98aede8bf29692b8f
dab13c302df547d34e1946b7e841fdde9e7da48e
96668 F20101219_AACAQV woodbury_m_Page_068.jp2
bb67e55ac96049cb800e807b077aa61e
008879a33a38bf18154932117dd21821e49c04b0
1984 F20101219_AACBNP woodbury_m_Page_021.txt
85d7c3e56f02c6c8353f9552c65d7366
be965a70772c2cae9e873e0fba8d81f34f34461f
8431 F20101219_AACCKI woodbury_m_Page_214thm.jpg
463b0eda86dda2f150fbf12d59824b00
024eb0b45f5075a87b777223642c511a49caebcb
98927 F20101219_AACADL woodbury_m_Page_134.jpg
da306842fccf89922a65ad277ca8059c
e8a2f492220323b51ede3029091e460a1dbaab78
F20101219_AACBAE woodbury_m_Page_105.tif
32be52bb6793ea428034dd69d751204b
f8bfc97f34902a7b8775f67a76fd2111f6f112a4
108546 F20101219_AACAQW woodbury_m_Page_069.jp2
f3616e7ab191fb79f9e138cfc805571d
d03a92cfc1aeb1594bc8dad2e3b3575b4a8157bb
1853 F20101219_AACBNQ woodbury_m_Page_022.txt
8b2b83485cd433bff76057de4ea651dc
883877f912883da5070a708d2f696f909ebd44e0
34842 F20101219_AACCKJ woodbury_m_Page_215.QC.jpg
33381856cbf3844c6ee7da324b69109d
fe6e0002d0189a62c905d85743407af08e9cc509
32094 F20101219_AACADM woodbury_m_Page_174.QC.jpg
4b0c5347897260c3fd04e6d1b37a8b7f
5a784e546b22c483a0498e1e5dcd533d052afc9b
F20101219_AACBAF woodbury_m_Page_106.tif
d1d802dcf6813d3319bf1bfe5a92ca77
f43cb660f23c39828856080507cdddb4a450c320
118935 F20101219_AACAQX woodbury_m_Page_070.jp2
7a572c93e889d7407aafba038240642f
fe72aa3a301c8e12ea4221838e038160237351aa
2065 F20101219_AACBNR woodbury_m_Page_023.txt
7133e68e063ae8a52f973a3e4f6335aa
971eebaac55abe43c68c0013d2913712679858c0
8490 F20101219_AACCKK woodbury_m_Page_215thm.jpg
3886b6b699940c594c1435ca5e34213f
4343835d48060f5adff7e64dc16fb04de1d8249d
F20101219_AACADN woodbury_m_Page_054thm.jpg
d1cb8226de08f893ab64574e8a9fda45
3b2775fe3d4d0af27e9dc2967157c5d30426c568
F20101219_AACBAG woodbury_m_Page_107.tif
90b820bd5a0bea7a6399c1989ca61df8
7a1e4e67933a0c8977d7866f70215f5102d0c3df
103780 F20101219_AACAQY woodbury_m_Page_071.jp2
0a49d704f7bb1e06abaed58df9edd896
0e64d408bb5f7bbc9b25d7487ad4ff21a5dbb864
2202 F20101219_AACBNS woodbury_m_Page_024.txt
05024fdeacb2a902b49bbec21fe619e1
f1b5bcf74ce7aacbee2db0b4a8897050d0a6dc0c
35770 F20101219_AACCKL woodbury_m_Page_216.QC.jpg
6339877516d20cfd89db6446dc6cc0a2
999d6237c139c9ebc4f0fd2475b71f04126ae2ce
31995 F20101219_AACADO woodbury_m_Page_047.QC.jpg
0a9a9ef89b432060966221c3d0afe840
3f4e6f35ea3958f2b2578956616c3f665e2a3e83
F20101219_AACBAH woodbury_m_Page_108.tif
a5cbdcee8b75c20fd17b4adfc0fe1ab1
e086f4320ab764bb39b9d1d7370eefb2ba2036d1
109332 F20101219_AACAQZ woodbury_m_Page_072.jp2
8cd3455ab845aae31b2249d3e9572485
318b9185b15199acc53e355524dac402a1f8ed4c
F20101219_AACBNT woodbury_m_Page_025.txt
09456450e39d80982791b84b62624897
3567367c92a2e5dfc227b89c8851feae08b27309
8821 F20101219_AACCKM woodbury_m_Page_216thm.jpg
ec38ec77cc030823f039a3d01fee52fd
1fae11092ff5dfe09c78a039158d908a9fecf412
F20101219_AACADP woodbury_m_Page_011.tif
0dcb3c0e42d12f763f32208786c876db
3ed7a58abfe88d55b37fa58a72697795babdf56d
F20101219_AACBAI woodbury_m_Page_109.tif
98e904329f0779f12e384f41f5f6e92a
9aff70220fa62aac0c9b9d91d08cd34e344ac8f0
F20101219_AACBNU woodbury_m_Page_026.txt
adc7b3b094fc265e4adb5e6ec65908d0
735a4094a8ef98752892ce30440966fe514f6677
35422 F20101219_AACCKN woodbury_m_Page_217.QC.jpg
a2cf1379e53309685ed105636e3e19dc
7fcd9fe96dfe9d0a4eff7dda7f110173a34e8117
30878 F20101219_AACADQ woodbury_m_Page_022.QC.jpg
fddac3d488ea83a3f4c9c2bfea20bd1e
40affe1127824c09355fb348488503559350ee68
F20101219_AACBAJ woodbury_m_Page_110.tif
0b4ad7a310e8293a47a1b4973b4d0af4
65d01dedc3394576632b416a57104ddf7e039ba8
2016 F20101219_AACBNV woodbury_m_Page_027.txt
6bc333f56d5a4a8d7a15375df3ad8cc5
68833cbda6591209ad437fb4030fbd21d744ad5b
8892 F20101219_AACCKO woodbury_m_Page_217thm.jpg
e07d2079481f35b79a20382b59da9f9c
b82927be559b685affd069b072e66172beaa29cf
93671 F20101219_AACADR woodbury_m_Page_068.jpg
e8dab07b1198a9dfb62d3febaba1691e
2bfdcc1412dc42ce057f8b3c171ced279727ebeb
F20101219_AACBAK woodbury_m_Page_111.tif
e300654da2de0df63aec5f22780c2135
f393f62471c01099ba59e6f188b93cbbeaaa22b3
2083 F20101219_AACBNW woodbury_m_Page_028.txt
7f448c79a58fb6dee4a8a1641f681115
dbc8e94a37f208ed7d39143fb1c9136af51e6442
34958 F20101219_AACCKP woodbury_m_Page_218.QC.jpg
f0c6c6eb8516648672eb784ae2093114
3f9fb1bd4827f1f8ab3519160b5e297e593191ba
F20101219_AACBAL woodbury_m_Page_113.tif
5af035f7aac64f52ed46092003546092
8a9e2334a77ea5069b74441f785ed97d0a0dfac3
2088 F20101219_AACBNX woodbury_m_Page_029.txt
3015fdc5170222ca813b4292064b0d63
ab643b08c3ca3a9d4773a48842664351a4e67e9c
34015 F20101219_AACCKQ woodbury_m_Page_219.QC.jpg
2b899b3312c08fd401f84d41aba5e853
81298a55051d55d81c8cab003d5b75bcb8b41eeb
30215 F20101219_AACADS woodbury_m_Page_176.QC.jpg
cab3daf601c8712ee4cdb5263374fdf6
8d6eb8e884a33630e7bed7cf05facfe5276f27a8
F20101219_AACBAM woodbury_m_Page_114.tif
5b094adf8844fb726bb11cd35662e451
041e4dd1bae50eba09f718783efde80e509a817b
1948 F20101219_AACBNY woodbury_m_Page_030.txt
1c2229150931bf621b8faeaaf07cfd89
6fd814166a2e7243bcdf094c44a8ce6dc619c6ab
8650 F20101219_AACCKR woodbury_m_Page_219thm.jpg
3e5e24673c7bd21d5681e3bd3b32426f
a04db80d35d2dff9a16e0774f8f9154b50d95cb0
814 F20101219_AACADT woodbury_m_Page_158.txt
47acad039b42a0745740390e0023bf74
004fd69ba2da78d5fca7d30c8ecde94bf4c56c85
F20101219_AACBAN woodbury_m_Page_115.tif
66b5d763f125fbf38f351d30ae455eb3
320ce42ef53fab5668533a90c50c510c1e4ea038
F20101219_AACBNZ woodbury_m_Page_031.txt
158d4e54014b25c00921c46423f3793c
e56720f80ad6dcfc9cdf946a628f41899c3ee0f0
37975 F20101219_AACCKS woodbury_m_Page_220.QC.jpg
66e0e3818d95048a018220984f621d7e
1cff30dc37f6bbba56dcf32064320751c47e29eb
F20101219_AACADU woodbury_m_Page_061.txt
ae7464c434c2724e06d4ed8b171eb054
02eb7a941a306675842da69477922144eed0429e
9230 F20101219_AACCKT woodbury_m_Page_220thm.jpg
5962a9462d19873a95236d95d6877c2d
66e714a9f960b16707d99f7b019ed44ca8f97045
100129 F20101219_AACADV woodbury_m_Page_051.jpg
7d627f5155d75ebdc4f376c650d33ebd
144f023ca819906ccd92e8cfff99bf9823cc3e3d
F20101219_AACBAO woodbury_m_Page_116.tif
5d5750f0c6f5b87b48dd4d83e987cdc8
e449a7816d4ccc2b447a5812b11def8ce3e4f11e
36155 F20101219_AACCKU woodbury_m_Page_221.QC.jpg
65ccf4ed7015abe5b9e86efc5b23b61c
b3dc5cb8cced0cf492155d52b8839d5036ee6359
32314 F20101219_AACADW woodbury_m_Page_051.QC.jpg
97d0159960d87f7e95a492a76647da20
83b48c230bf871563922fc805b25a47d8c61c858
127422 F20101219_AACAWA woodbury_m_Page_215.jp2
5470044c828f0ed2904504a5a798ca34
2d1268c4d27afda8fae9357a2d48455b6f3f839f
F20101219_AACBAP woodbury_m_Page_117.tif
9c89c4555be8c10a8f660a09cb3ce13b
c5d742b258a051627cd863d921c4396c06ec4718
8526 F20101219_AACCKV woodbury_m_Page_221thm.jpg
d3587ee279e1158b21fee3664f80e32c
49c192b4b42183f1716e7dfd27725d15148d2626
93027 F20101219_AACADX woodbury_m_Page_123.jp2
145374ba163f9aa1030167b2695c083d
f44a780c0d474dc8485ed78898e6ed4b13461d28
129465 F20101219_AACAWB woodbury_m_Page_216.jp2
e7258f62abd10168620bffb3be581502
a186712c9fdf2f60628bad696214a2586f199e4f
F20101219_AACBAQ woodbury_m_Page_118.tif
36877fc39c0cc198e29151fbfb3fec80
fee6f6cbe49c79302c177dde941d73e32e29a655
35314 F20101219_AACCKW woodbury_m_Page_222.QC.jpg
c8fd9399d348b37427058c334a28a145
b07393fd4f274d1211b89635591394151d388b85
48843 F20101219_AACADY woodbury_m_Page_133.pro
73e7599163c6cf8f4512af498c0e30f7
055ca6e51a8e3aac48d1ec4c482818a0ee0f2a2b
132613 F20101219_AACAWC woodbury_m_Page_217.jp2
a1c466b154d73fb099f6943f9df07da5
c1619901e1a14fdfe7987d3ccb652a3f5d8b4fdb
F20101219_AACBAR woodbury_m_Page_119.tif
691e065f3675850cf302212466e11a12
681d7a8d591227431999ad8793b60fcef611a31d
62572 F20101219_AACADZ woodbury_m_Page_216.pro
d7b32d1f2a9b9729826c9213784f535d
4259ab80c69e96e3cb984cf808ce019bb5e4c2f6
1051954 F20101219_AACAWD woodbury_m_Page_220.jp2
c3abb296ea152d82ab8a4b191d39e698
5b2d8fb4df625b941413b5873611f5398d0f07d9
F20101219_AACBAS woodbury_m_Page_120.tif
586c661c1569e83e3b539e0b9948d65d
cb57eebcb35c80066a61d13150a37dcfd1cf7590
F20101219_AACCKX woodbury_m_Page_222thm.jpg
353135f01428132a2861edd50f872b5f
bb9b28ae8cbc2135620747d5a41eab116835eb05
134772 F20101219_AACAWE woodbury_m_Page_221.jp2
3d33d05cd8e1229e9f6a69c3fc74b9be
81e5ca4e3295e43b35fa06bd7b586674525329f4
F20101219_AACBAT woodbury_m_Page_121.tif
4a6c4d2531bd12ccb0a86b1bdf58ebeb
f84b2b547cb84b0484b4359815f0dc85967f735c
33872 F20101219_AACCKY woodbury_m_Page_223.QC.jpg
e9aa55ac1a35c71d9e24f79fe78c683d
a0d9fd25bea87eda85c8ad4e335d8c8f01dc6767
F20101219_AACBAU woodbury_m_Page_122.tif
6d96fcef99fe694fc01a18c340e97f87
e4dd58f74c0ade17965d5cf60528254759a07cc1
8501 F20101219_AACCKZ woodbury_m_Page_223thm.jpg
5300c7037e7740332f8de058776433d3
e322478ee47776b384af25f756ebff7b8d755feb
126555 F20101219_AACAWF woodbury_m_Page_222.jp2
c6622fe0842e3123ac0abe7a82f89602
f2b9ac47c08289a18973b016f3cfdf9a41ac4ec6
F20101219_AACBAV woodbury_m_Page_123.tif
a89feed7f34294084f30edfdf4a9565e
d9a16bace1ebc14eaee7bed0e5432615b300422d
126496 F20101219_AACAWG woodbury_m_Page_223.jp2
b2364a9fc4a94245a2b43229f18bddf8
2e456f1df8743c285961425a658e4b254c5ee122
F20101219_AACBAW woodbury_m_Page_124.tif
dada9bb66ba3ebbc3df1850e0a38b7e6
779ecfe967cc82abbf9e9abffe912092b5e24a08
1677 F20101219_AACBTA woodbury_m_Page_170.txt
56a3791d37be6aa34180b4c89a3c05f1
0b0a93f3767ddea08b1265bb34a36957f94753f2
1051984 F20101219_AACAWH woodbury_m_Page_225.jp2
5313ecb621eac54270b2c8efdd039ffa
57dd35da01d3d57aed5f64dc21366b5ba13713f9
F20101219_AACBAX woodbury_m_Page_125.tif
e499e0ba32f4a583ee47caac5b57cdb4
d311338ec14b157c2919413997c52c3ba69a48c3
1051941 F20101219_AACAWI woodbury_m_Page_226.jp2
03a0a04a5a5120c73e535bc9b25ce880
948a4e22b587500f748bf91a10989496b4d6e7ea
F20101219_AACBAY woodbury_m_Page_126.tif
e358cf22a830128d6af97646daf6575b
adcdead068743ba90cd784430e3fc3b6f65e6bf4
1785 F20101219_AACBTB woodbury_m_Page_171.txt
8cf91f1629a36373507dfc7749c86f6f
b8c19f16bb1b45aa4222f8e1e5da385c633e7883
120401 F20101219_AACAWJ woodbury_m_Page_227.jp2
759643a40303e6a8db804e181bebadd5
b7af69fdccab53e0fe054157043396ae53cfeba6
F20101219_AACBAZ woodbury_m_Page_127.tif
59c05e83e3feea132737db5e09dfe48a
3a0299630a427bd5ceeb6a491926f70599d5781f
1757 F20101219_AACBTC woodbury_m_Page_173.txt
9b2d63d95b7ee38f422bdf89a5669b64
179af9d5fb70a71163f5d96130043b3b32fbe54c
54747 F20101219_AACAWK woodbury_m_Page_228.jp2
4d81d395abd49f95c99fa6a153ac455d
e4e234f1b304299a1d4851ef5fbf6d000ff13252
1965 F20101219_AACBTD woodbury_m_Page_174.txt
fa9b174bcb70b4c996f17cbfb825a3aa
01b0f1b0c34cc3bad015605f01ab079c1b8dba63
95550 F20101219_AACAJA woodbury_m_Page_078.jpg
2b2ff11c8b11d7ff254bdd574c0d8091
ed879ef4d3949819215597ba8fb4acc2ac4b1c66
F20101219_AACAWL woodbury_m_Page_002.tif
cf5fa89fa845d8aad265b4b87b847b6b
82bb82589fcea8bfbedd7ac4ab3b3ae75eb96f34
1670 F20101219_AACBTE woodbury_m_Page_175.txt
f2a67ea55e3b3dd3701db8fcb169f22e
883b7f5f1eb250551a9ee87717d8a203c27e17f4
95975 F20101219_AACAJB woodbury_m_Page_079.jpg
8165e6831132391263533482ac8d820d
d473ccca1ef5caa8d7882cf82d50b3e6410a66d2
F20101219_AACAWM woodbury_m_Page_003.tif
ee18dff3ebc0a7579cd3aca06a649ac7
0a89da36b5934b1592b1e483230e60b6adcff663
1842 F20101219_AACBTF woodbury_m_Page_176.txt
dc6c358c213721eb4a783357b4b2b68a
5dad2d097161fdec821a3bd84cb510da692159ef
105928 F20101219_AACAJC woodbury_m_Page_080.jpg
0564b8ed87c3d5653e6b1fc31b633ef4
6a2cdf2c988fe6537dd85775f644eb08716d6dbc
F20101219_AACAWN woodbury_m_Page_004.tif
3f9263ef156fc463297ff0ccbb74d52a
5c8380b39db8ceb307eda797eeb25740abc27bbf
1943 F20101219_AACBTG woodbury_m_Page_177.txt
82d77822349fb896fe93394fd2fa020b
5719dc881c97cd0a0f248ff28d1d0828f0c64352
91683 F20101219_AACAJD woodbury_m_Page_081.jpg
b1bc5b473e40446b4932e311859394b7
29aa55c9e25db24a59df2369f974e9596ac4a5f9
F20101219_AACAWO woodbury_m_Page_005.tif
0245c0b32f2c1e18db42182feb4fcaf0
57f544aeec66050e7ebc719feb6ddb20a51ce5f2
1978 F20101219_AACBTH woodbury_m_Page_178.txt
abd69dc2f20974470d7349c82a1bccb3
95b555719f59415fa942511a77a00f8926608912
103861 F20101219_AACAJE woodbury_m_Page_082.jpg
47997155fa59a3ecccd39835504cbc99
8fcdedb97c1940e4441abd8bdaf2551f4cb8aa47
F20101219_AACAWP woodbury_m_Page_006.tif
e72e2e784ea8d78e43bfc399ea33fefe
32c4798ca61b0f21174e1d2fa52942e3f7b89be7
1961 F20101219_AACBTI woodbury_m_Page_179.txt
c86d5cc3d9f235c3ed4de231d2e370e6
a86049923aeead0b55520aa38b3ae38f310505ad
105361 F20101219_AACAJF woodbury_m_Page_083.jpg
618bf4aadecaccc279c27bc40cd2c424
b2e1bab527e47d7d15e179767ebbc8d9a3e02d12
F20101219_AACAWQ woodbury_m_Page_007.tif
bfb9b4d11c45179d9c2f29ead538326e
089047f623f79a2b97ef0b391b3cfb8c6dc745e9
2048 F20101219_AACBTJ woodbury_m_Page_180.txt
f7bbab4c366f17d0cbba85d2bab458f4
a553788fdb174fdacb748d912054c96b57597ad4
103165 F20101219_AACAJG woodbury_m_Page_084.jpg
ca6404fb3e926a85576de45b54e21575
6a2e8f62d1b24ff440f1d570af7752cd62263cc7
F20101219_AACAWR woodbury_m_Page_008.tif
ec6b3a96c424ffe764d30c02a8a2ecd8
66398507d6f4de72fdbdc61ddc412c433eec0b4f
1324 F20101219_AACBTK woodbury_m_Page_181.txt
b53011e8f685ef5d259ac6ef91fee7b3
20bdff7c868f2086abed2271b1f90aa1e41f15f8
103864 F20101219_AACAJH woodbury_m_Page_085.jpg
c5fa0e302f61f809c7f4d752bbd0e55a
428c645b328b37e5e40cb45cdac05c6e94e2f562
52221 F20101219_AACBGA woodbury_m_Page_038.pro
0de7947df472a6254d99d6f62a8ecfa0
c3f3b7337bf82d536a5dcb8a0982b1f53b4dee0b
F20101219_AACAWS woodbury_m_Page_009.tif
50875761b3c8280864f22bf9e9c39d1a
7d3cf31ca9c826b8b67ac87bf7ec0c5b883fe2ab
1447 F20101219_AACBTL woodbury_m_Page_182.txt
2aff75beb54cfe7b8d3f3bee655f758d
e4c25cbc434437b9f9b7edc029dbe0dc26cdb000
98389 F20101219_AACAJI woodbury_m_Page_087.jpg
1180ba6dba0a75d90b87417a017e07c2
f3702f20d3094e9b535f7ee0dddb65610df2d582
50814 F20101219_AACBGB woodbury_m_Page_039.pro
b1d04c2e91d946b689699613b9fe646c
12c43b66658e44e2465272aabb1332c5c13bc56c
F20101219_AACAWT woodbury_m_Page_010.tif
30c840830bce6fab3604eb88dc8dbbc4
81277146ed41996055de4a80ee55b358809d5c24
1407 F20101219_AACBTM woodbury_m_Page_183.txt
8eb8c7ec67b8cd03581d789c0ecde00f
7535758c4c56ea48acbac2b71b3dea660ae7e75a
67851 F20101219_AACAJJ woodbury_m_Page_088.jpg
1b50f4058d11dd152aa6be9c00fc91f2
59403b0fc1eac069ef869aacd46bdf813f407912
48688 F20101219_AACBGC woodbury_m_Page_040.pro
af30c4becaa29acfc6c91b4ecb864470
2cef9189813613b586a60471901af4e73f2254ee
F20101219_AACAWU woodbury_m_Page_012.tif
77e7701a7b38c306a4f7228251ae0768
04f8668186dc14177a203fb3278600b574a66582
2127 F20101219_AACBTN woodbury_m_Page_184.txt
605d97a7b41dd46c85a447ef6622b99b
d2d836ae9fac4f5175938656b31e9cf560312a62
60714 F20101219_AACAJK woodbury_m_Page_089.jpg
edc0399f0e7e597e223b4d3aaa38d65e
cf6893470812047b1ab61d2683c6eb8f635d1ee6
54620 F20101219_AACBGD woodbury_m_Page_041.pro
fab941e7ec757b05cc1ec946be658b57
c6be1c93a38fa298fb8e8360dffb0a23c0975292
F20101219_AACAWV woodbury_m_Page_013.tif
96ae360d53008821de2dc49f68aefa1c
dd603f9f3464f39f411e8d783fbb78affee0a653
916 F20101219_AACBTO woodbury_m_Page_185.txt
219c760831ab3691dcc8d411ba1c6a78
5539a774a48fa5b7ea80067a7cf3545729b3db74
16670 F20101219_AACAJL woodbury_m_Page_090.jpg
544a8b372a9054dac3b362fed3ce8693
960876806ad0577226ed3a15eec3f14bc6ae46a8
50984 F20101219_AACBGE woodbury_m_Page_042.pro
73afee2cb3a280cc85d250c3597050ea
c69d526038e2b3c69aff2cf979e3140531b00851
F20101219_AACAWW woodbury_m_Page_014.tif
e344db5369648d9ac61f6d6a47459f41
553207623290d8a171c1bf103461ea3c8ceb1f7f
2149 F20101219_AACBTP woodbury_m_Page_186.txt
360bb0b86417f4ee006979779a52bf66
ac855eeec8fe3fabc96681f85a332595181be1be
82161 F20101219_AACAJM woodbury_m_Page_091.jpg
c1054b75001461c40341a768e18eff19
95d55532388fed5e5615ebb96dc2290daee223da
43330 F20101219_AACBGF woodbury_m_Page_043.pro
99e58b281d445a7643092ba5d34f52c2
ee25ae12e29e140361c20e9e617cf8c0f9ba4133
F20101219_AACAWX woodbury_m_Page_015.tif
03225cd1aae6f0599ff30d58042a4843
3a656b36279facbad954175baa09c5ae3286bf38
906 F20101219_AACBTQ woodbury_m_Page_187.txt
c5b52fa17c7a0fd2141913f714fcc504
e883d0aac59001f6fb3b699c01ac2eec11873e15
102220 F20101219_AACAJN woodbury_m_Page_092.jpg
a1c6d26ab4e9ac6bcdb671852556ea98
76ef1eafbb634f04d07e9cbb437e3e6668d87e7c
46139 F20101219_AACBGG woodbury_m_Page_044.pro
d194fa8dc3dba71b1f78bc4944086900
ed00ccc66112018a9af07066086160636a8f7d76
F20101219_AACAWY woodbury_m_Page_016.tif
6ae320ed84d3bc1f50b4a4be6a0b0f6b
56b1b1c017d0f92b03db140b4e1f5e69bb1f378d
612 F20101219_AACBTR woodbury_m_Page_188.txt
9b5d2e4e2d6484721b810b6c2ca4a0e6
cc5e652d4bf4ae614c5430f2c59a8003252ad705
42950 F20101219_AACAJO woodbury_m_Page_093.jpg
2fe9431646b908b568256c4275f0c61c
48126674039aa3ee8cbda9371f13ccb817219bbf
47262 F20101219_AACBGH woodbury_m_Page_045.pro
ab918fa79c14eaf9b4c2263540d60ed6
bbd0fa933454b6e269ba71e7b5bc4d193a8c8e2f
F20101219_AACAWZ woodbury_m_Page_017.tif
2e429e3dfd1b5f170578029507032e66
df9031852af85d51e8d3b26eb4412d72fd3a237b
32905 F20101219_AACCDA woodbury_m_Page_110.QC.jpg
b527a8b99814e817f1e508b8a40a82ec
2388f443b57013c06668d2a28ebda60abef4e44f
1870 F20101219_AACBTS woodbury_m_Page_189.txt
ba48c0330cc0e5450b573fd63961dbdc
ad86e5d160c84b993d806cd6fca4c6eee1566d95
74418 F20101219_AACAJP woodbury_m_Page_094.jpg
2dbb15b3fefdaa9a6b6f14fd26ffcfde
7419551c004d88f788365994c386a98b6919dff8
53472 F20101219_AACBGI woodbury_m_Page_046.pro
79e61a4945b7834a0e957bdb2bec7aca
784d30d7c1605f3fc9d4a891acf4dd69ea3af905
8290 F20101219_AACCDB woodbury_m_Page_110thm.jpg
33f4601a1956b73a4b73a9b962dadc55
aef5e7be9ff2635f107bc08590d82eb75e596b7d
1991 F20101219_AACBTT woodbury_m_Page_191.txt
5c211495f16e05964026e4f70514d246
9b0129f0160239eb6563230c8ff1e72416ca44d8
41447 F20101219_AACAJQ woodbury_m_Page_095.jpg
1c59844856326ee39762d144ecd6aacf
9758c79cb0b9d8e9dabd9a80d7d7d50fc1dbf38c
48925 F20101219_AACBGJ woodbury_m_Page_047.pro
c2bbf4023190451c58d7c4b812dd60bc
cba34fc50a19f73fef913ebada7628d3f5744250
F20101219_AACCDC woodbury_m_Page_111.QC.jpg
eb5029c8ca557dc677385051535b3080
806c1b1c1dbfa8f350158f164bedbb67d656316a
1244 F20101219_AACBTU woodbury_m_Page_192.txt
c6c58637164861118cccf1d5b46e3a75
5bc70922b05e9a5538ee4f37e03027ce94c9e38d
95489 F20101219_AACAJR woodbury_m_Page_096.jpg
e9131377484b99bf8b4f83544ee564fd
6e38092fac03254e43130152f81871d8508b5dd7
48774 F20101219_AACBGK woodbury_m_Page_048.pro
53b847c1ab61ae1efd2f84f438849ce2
77552ba765cd39706012e80986fab8a0d925cd0a
F20101219_AACCDD woodbury_m_Page_111thm.jpg
b311c781c82d6a96899a25a74110f4b8
831ce41500d3e3e59ce9fbbed9ab87504375d4c6
992 F20101219_AACBTV woodbury_m_Page_193.txt
b3e366ee7114e85e77fa90d7541eec1e
be8dba97fb29bbde7cebfedd7a2922a497f9312c
101184 F20101219_AACAJS woodbury_m_Page_098.jpg
1aa7543cd41d16572ea7958b6d9dcb08
5d9b06466f18d4bc8effbfd177a5dc2eaac2cff2
52168 F20101219_AACBGL woodbury_m_Page_049.pro
a3d1272541e6db575637ebc4b81f4472
397630c4f707455900dad254eb56fe5a220d601c
33831 F20101219_AACCDE woodbury_m_Page_112.QC.jpg
f5e8b2c51e2706bad4c5ca6468218e49
ed603c529f166b526027a2500cd4600a768d8196
1133 F20101219_AACBTW woodbury_m_Page_194.txt
ecb5ba1a35b7175aef3692e9c5243c52
bcf85fe2f30c9bc3f5c497d67a2948282c9fe1a4
90602 F20101219_AACAJT woodbury_m_Page_099.jpg
c33e631ef9b619477650e2f1e2a4d5e0
21136d4a33b0067351e3f02c905b6efa99921417
53179 F20101219_AACBGM woodbury_m_Page_050.pro
fae1e1b74712455f0ae1f8100de8ed5a
694c45d5a8b9a523ed248a7a36e58481bb5dcf9a
32732 F20101219_AACCDF woodbury_m_Page_113.QC.jpg
2bed9b9c480dfd056b44c002e091ebef
2db531b44b401f1fdbceea52f4f5deb27cc3b6f3
1188 F20101219_AACBTX woodbury_m_Page_195.txt
f6f189bd3a92957cd60153f840801d15
a6ff1139e90088c77c21a76473a2bb470e411f2d
101342 F20101219_AACAJU woodbury_m_Page_100.jpg
e149ef0b9d53df04f314b8689700f3e4
19a205b8dc7f7fead9199c9ef33ef301c6dcd0c8
49340 F20101219_AACBGN woodbury_m_Page_051.pro
03b458c2c28f712fc8d103c165cce621
c29a84f07e0d7331c92c6f744f355eaca00852df
8114 F20101219_AACCDG woodbury_m_Page_113thm.jpg
437b0dc3d5756035e326120b26617357
45379a81c4bbddf3d25ddac3e6e305a4190e508f
1164 F20101219_AACBTY woodbury_m_Page_196.txt
7c513bbf830a373d63224484f6d46b77
3b65687fb728874a3bb115bce51f8948da5c2720
90129 F20101219_AACAJV woodbury_m_Page_102.jpg
177b3971b7b8c3c68502e354ed4a00ee
c4b01a9544951483d0bf6b0e7184e40f07e62e68
51164 F20101219_AACBGO woodbury_m_Page_052.pro
8b220b6f1043cff2c533c1867729e1c8
9f9333be14118741042f7eb77a8a251b7aa916c9
17149 F20101219_AACCDH woodbury_m_Page_114.QC.jpg
69ea644afef766144bb21efb65f993b6
5a7114189bffbab99d75ab8ba902636ffe68997a
1024 F20101219_AACBTZ woodbury_m_Page_197.txt
65cfcc4fc13157530dc0386de9557c46
b57052aee6e4000c3dd6bf1ae531bf56bb65e6ce
100466 F20101219_AACAJW woodbury_m_Page_104.jpg
4b99aaea818531fac02277191cc99a47
613a2d5ef7e771c357722fd3de543891b7656802
51199 F20101219_AACBGP woodbury_m_Page_053.pro
cd6fb51f4f8c4b20c5f93d695931fca7
5845cea1a0919b3407e6c24901620772336c20a4
4193 F20101219_AACCDI woodbury_m_Page_114thm.jpg
b4301c46b2f973e50481e23fbf19a751
b3116a74fbadccda11eeacb1e14abc073dc11a63
104537 F20101219_AACAJX woodbury_m_Page_105.jpg
fa98aea5718d9ac02d5f0d801839fcbc
9785bff471b04ee593ba721ccb10549759b9fbc3
50746 F20101219_AACBGQ woodbury_m_Page_054.pro
8c51c8a41f9d4c0b8b1f08ab9e645246
a7926d303b29fe20b8dad85bcf292cd33d102f41
11700 F20101219_AACCDJ woodbury_m_Page_115.QC.jpg
0bf167061bd8814d1344f37893565d68
7f9e40e67118111eb943d85a2ae8bee9794315f2
51294 F20101219_AACBGR woodbury_m_Page_055.pro
a3578cfcd892b967085c018debc0c77c
d9452675f88357f6d98d110ddf35820ca0bbe6eb
3335 F20101219_AACCDK woodbury_m_Page_115thm.jpg
acf48ab53112c2942fa9d052436730f9
92364d2fa13bb876f31b437a54eb297b0ace2e34
97438 F20101219_AACAJY woodbury_m_Page_106.jpg
d18ece95d24784a3982dc8fb40fe7028
23af18395c9e614a0ee433ca84d6c9a7de53ffaa
51390 F20101219_AACBGS woodbury_m_Page_056.pro
5bef1834821eae9a9319589cf027b1f6
8cab780d3b5ab95f15076e56bb3247b9365c6a0d
24444 F20101219_AACCDL woodbury_m_Page_116.QC.jpg
5fd87f0761b076606b6d302605a43dce
17889bf6cca242063d57a6441330bc5390e7ddf4
103666 F20101219_AACAJZ woodbury_m_Page_107.jpg
db95642c6f0828da2f415df4a8a917ec
1de31804de8fd154ee52142d49bea94655d8fc04
49880 F20101219_AACBGT woodbury_m_Page_057.pro
99d3d51f6629a81b59b6278cc7028381
6670fc2b6940502ec90ca17cc941d7bae8becd5d
5980 F20101219_AACCDM woodbury_m_Page_116thm.jpg
fabea0ec0e74f129470074b2fc2919ec
b4c9e74c45ddbb6416a3fc9c3eacc5c864664aab
25396 F20101219_AACCDN woodbury_m_Page_117.QC.jpg
7430bf163f6f89bf4e97f3be76d97c0c
62e70392ff1dfaedeadd32a5a0d0387205cd7641
52151 F20101219_AACBGU woodbury_m_Page_058.pro
316876bb83cc988ba3309766d4ede10e
3eff833fe0a6bc4cf3982182e0de4872bacb1b27
6295 F20101219_AACCDO woodbury_m_Page_117thm.jpg
1a4c8858d43fcf759b07c2bf3be70feb
d089dfb218736e2468232739456be78ff1c816eb
47134 F20101219_AACBGV woodbury_m_Page_059.pro
94030e7063f52cacaade0eef037769fa
7877aa521d153982036d70561840a2d33abba563
21435 F20101219_AACCDP woodbury_m_Page_118.QC.jpg
54491013df0d93502fad864aee804ca6
e74486679317cca4219b32ed434400436ff89aa6
49351 F20101219_AACBGW woodbury_m_Page_060.pro
4b99a701da7ae7d4d446e7730bee9cb6
507b06a5ac15e4d98c5b97f26d9be727160379df
8159 F20101219_AACBZA woodbury_m_Page_053thm.jpg
008776bee84720a2a467239d8ed5ab15
1efc87834f74f6841ea432b430ab0908ac366009
F20101219_AACBGX woodbury_m_Page_061.pro
3d1ab95eddf1833a436cda21ccfc568f
0d0d4d333bdf9048206efaaf26232936869b9a04
33285 F20101219_AACBZB woodbury_m_Page_054.QC.jpg
8c1a6ffdca986374bcf51309c922007f
08485d680db13dcdff98d5c86a292e05e73b007c
6636 F20101219_AACCDQ woodbury_m_Page_118thm.jpg
5b4458aa82e96677a3a0502123005a26
6d0ec7107e837bfd55ed60ecdd71a737c18579e3
53729 F20101219_AACBGY woodbury_m_Page_062.pro
f2d372f6792725d546aaa5b007f24a8d
fff2ee311351d2ef43e56d1ef33a6d1b8ff26b9f
33509 F20101219_AACBZC woodbury_m_Page_055.QC.jpg
f98753185c055f0312dd9c5557e32300
6a75cf7e0a7662648a48ae9a11e21ed9a9854bf7
30490 F20101219_AACCDR woodbury_m_Page_119.QC.jpg
710c08756119a007ff2b626e5c18e5d8
a83e127bf92787b7a2b2466da360cec6b91ae062
48409 F20101219_AACBGZ woodbury_m_Page_063.pro
ebf94d04aeb6ca12db8abdd60b576d34
2981eacd526721961b72e56ae37701ba939dbea1
33350 F20101219_AACBZD woodbury_m_Page_056.QC.jpg
dcf1a3c25905cd7ebeeca814a0dedaec
5608fe08856ea3ccd2c25df68a9bf603c6470c23
32920 F20101219_AACCDS woodbury_m_Page_120.QC.jpg
a263adecf64579c8f34b6c466a7e0636
98fa64e219c73b34a80d94e669e14986430b7e80
8448 F20101219_AACBZE woodbury_m_Page_056thm.jpg
50d7f586ef859a5abcc028f25cac2640
8ba6db446738a2d3df3cca72c67a99df394e426a
8425 F20101219_AACCDT woodbury_m_Page_120thm.jpg
be36f923c87e8715bef8396a2bfa359d
548d26f45f2ba8a9cc7d6ce3278dfc89c4be1169
33208 F20101219_AACBZF woodbury_m_Page_057.QC.jpg
1033a62cc5fcc408497d2beded57bb45
4df97da6554c3f640befc5973ec0378cafc3984d
31418 F20101219_AACCDU woodbury_m_Page_121.QC.jpg
f349d8537ed8ea94d2fe45fc80de5e9a
6759b56de564ce9db1d0d0885dedae19267ef16b
110582 F20101219_AACAPA woodbury_m_Page_019.jp2
c1cd9faccf93924c6f4a7d9403f4c1ed
4a718f080ccd1cbf7e05b9d99f61e28a9e7586c6
8197 F20101219_AACBZG woodbury_m_Page_057thm.jpg
533cd86ba217e645560a6691d5484544
35aa48e58e4ed4e03f36e162a784fff2071ab237
7940 F20101219_AACCDV woodbury_m_Page_121thm.jpg
cda53748584eb002b5a5bc73eac477ba
5b08d1ec1534c689f59bc339ab18ae8ca3bf5770
102255 F20101219_AACAPB woodbury_m_Page_020.jp2
3f59caad795208a1e7df8f720b24f2f1
2c227ea52ae8c1a094e2b203a90f98be33c1a52c
34135 F20101219_AACCDW woodbury_m_Page_122.QC.jpg
b5c1637ce2b6079dc4e6bd6b2ba008a4
c59f1ae1c514445288fa4baa5f10e54016895bd3
107368 F20101219_AACAPC woodbury_m_Page_021.jp2
adcf494d27c2133d91a05ef50eb9d161
943876c986f9aa7eb0842cfa8cce0d5026fe9374
33815 F20101219_AACBZH woodbury_m_Page_058.QC.jpg
a31bfff7ade3edbc4a084af6e22382b8
ebd225b3342390d06b3f4fca717334b21b6d8fce
8517 F20101219_AACCDX woodbury_m_Page_122thm.jpg
493065b00885558fc5d2098296efcfa0
59a436710ab94193b9872801c47362b277c490f0
101569 F20101219_AACAPD woodbury_m_Page_022.jp2
bb0fba55d1491d551d9fef1a0bb16ea4
8c1b339956f01803abfb370f7001599cbdd8534d
8388 F20101219_AACBZI woodbury_m_Page_058thm.jpg
f10eaa5a0e53cf6ba52828709f95f988
b9fa692e85d70d9c6f31aa8994ca0dd3158f2642
7429 F20101219_AACCDY woodbury_m_Page_123thm.jpg
b19dcf741b5bdf67c45cbd9a1ab48e0a
8ef98478a05639ac1d1edf721244c1552fe093cd
113802 F20101219_AACAPE woodbury_m_Page_023.jp2
b104e2bdb99869a0f3317b812bf48bde
2b28c58683b8a6f8f20caf4a78135b4737654a41
31660 F20101219_AACBZJ woodbury_m_Page_059.QC.jpg
dcfa57ef7f42e2cccd53c27310d7fea7
eb5b19ce7619f255c44b133a068090e49c76a8a1
31440 F20101219_AACCDZ woodbury_m_Page_124.QC.jpg
fb00a5114ea3562f3d9fbdb3d95dc65b
e44fbb9f185ae8180ae9adc3933628d3ccc61666
117098 F20101219_AACAPF woodbury_m_Page_024.jp2
f1e9d2d5742da011b3704281cf83e6ec
107a515c92e56d042d286fd3a36e3c721167d431
7999 F20101219_AACBZK woodbury_m_Page_059thm.jpg
7f1242024d402eb4e668b375263a6ef8
28f02271b45f78ed6c5630f13be10d6cf10c6702
111482 F20101219_AACAPG woodbury_m_Page_025.jp2
cbf44dc30edbdc4950ed848c5edf3df2
446d2f0f09b3bc871b14cc5b54a3dc6553e7d30d
43629 F20101219_AACBMA woodbury_m_Page_204.pro
894b72adce1cd88863192aeee4f4d9ff
cf03d12a8cc5f9601e039e705f76aac05c7f81ad
32399 F20101219_AACBZL woodbury_m_Page_060.QC.jpg
ba8b72949f810e11e53cd4a63694252c
b3d92969951abb9eebe7c5ee1ec45967fc6efcc9
108903 F20101219_AACAPH woodbury_m_Page_026.jp2
a885dd5bffd955b9d23267562331d8eb
f0e86e6cc5786bcc63fd65afc0592ac832ea7fa4
28996 F20101219_AACBMB woodbury_m_Page_205.pro
280b123cba1e59a13b9154a8b626d33a
73c33507684349fa337e4c0f0595878c64aa2e13
8235 F20101219_AACBZM woodbury_m_Page_060thm.jpg
cb6329f528be46aa581fb17a9a4020fe
542062a45a2e2eea735748a9c127c8459d41b2ec
110942 F20101219_AACAPI woodbury_m_Page_027.jp2
8d1aea35a390cbb7e26c6827bb7e62d2
2cdf2814340cb1cc065b077a187aefb0dd601308
13944 F20101219_AACBMC woodbury_m_Page_206.pro
99aa8e947fab72737a25fa050d523775
97c1e75c2cdb4d43da0256fa88cd85e68c9a5fbf
30616 F20101219_AACBZN woodbury_m_Page_061.QC.jpg
b69e3d99d72a1995a64c0cb6e197922f
10ce4887411be341c8bf95432a1dc985fdad91a7
113489 F20101219_AACAPJ woodbury_m_Page_028.jp2
f1e5cb56823f6a8e759774b58a49a019
a4e9bc93916d7beca36cecd668dabe56a8e8b647
53848 F20101219_AACBMD woodbury_m_Page_207.pro
01cc8bc4555fcbb9d45d08128e64c2e5
77d8a4951e262439d09f638d4816a1ac047c1c17
7822 F20101219_AACBZO woodbury_m_Page_061thm.jpg
f1bcc87e32d08d71a2b4f887ee9842c2
b8888affd40b8ac91026f488a30426bbd341b3db
113646 F20101219_AACAPK woodbury_m_Page_029.jp2
603ae270f9b2311ca125b26ab633739d
914a9bcff75c1125e5af99ddd666580b9cca9e3f
62946 F20101219_AACBME woodbury_m_Page_208.pro
1e7bbad7ecdc0a73b0fd1768e09773ee
9e9234d7d0ac56d82021c419f0f76c6ee3a56d0f
62174 F20101219_AACACA woodbury_m_Page_224.pro
7f4f696f415dbdb13cdaa862879c0197
174d5ecb9cd5ded3d894c1b046c41c6c492ebae9
35201 F20101219_AACBZP woodbury_m_Page_062.QC.jpg
338c6e43fa36af5aaea02ca35793f659
d5ca8ffdbd7c230ae84cdfc8dfc2e1802df98504
106140 F20101219_AACAPL woodbury_m_Page_030.jp2
76d25f31d8a68de7ffca08c3c4e69c85
a3292eec74d02b2f2a7f1e5808289d5a8a3ab0a7
64427 F20101219_AACBMF woodbury_m_Page_209.pro
1e884b06729e3a3d3973fd4eee130ae3
6fa3f8ee1c31fc633f93bfa6068d55c48a92b399
8937 F20101219_AACACB woodbury_m_Page_218thm.jpg
b9bac737ddb175c6e5d86548ef0e60c1
030d2628f9914710ab3aaeffdc20fcef78df80a6
8558 F20101219_AACBZQ woodbury_m_Page_062thm.jpg
3d0e65f8dd5a551f2b84e10bd1f8baf3
e48c6138e0caa2b3fbb9b5110de68a048d8e47b5
113212 F20101219_AACAPM woodbury_m_Page_031.jp2
c8700f62150dc9bae45a5b971f5a6817
e1cb73dfc44a6f021d78ec968d26822b86a72388
61064 F20101219_AACBMG woodbury_m_Page_210.pro
1399b8d979fdcb3809d5bfef4ffd754b
042b80c60899c9ca8e72b3cd81fe269b34512f40
77236 F20101219_AACACC woodbury_m_Page_199.jp2
0037388843e025ae864bb8ce65e7a500
cede8052610eb2e48d7f45b0d1bfc4e7e1a32842
31992 F20101219_AACBZR woodbury_m_Page_063.QC.jpg
72d889923d8b04e2030b3521d53fca0d
522dc6c50e09b0849dbe211f7cc3ce1dea7eb277
108530 F20101219_AACAPN woodbury_m_Page_032.jp2
7f35e0d7675ff275b04aabf247648226
aa212780c7eb724cb5a6a627132d4f049e30613c
64357 F20101219_AACBMH woodbury_m_Page_211.pro
8154ad7d7128d87223870aa55ac0d596
2bdcb6a04dfa34975056022edafd164cf52675c8
18302 F20101219_AACCJA woodbury_m_Page_197.QC.jpg
c6ee6b6ad23b38645af78a92e87fc3cf
48f0b66881f919f546900e7c6fe688386f498a73
F20101219_AACACD woodbury_m_Page_228.tif
e119e30e64c4b70b019c97509f76383f
4209b69dc3d561000093e43904ddf973a768ff61
8023 F20101219_AACBZS woodbury_m_Page_063thm.jpg
fbb64f54cf9de92e55ba613c8839a997
81c602c4eb937934e2667d92c3abc215ee6c2ec2
108551 F20101219_AACAPO woodbury_m_Page_033.jp2
63c5ec2082f6a7e474f441768281cc8c
a643c6e64cdeccb2f59f5aa88fe16169d59b2e02
56566 F20101219_AACBMI woodbury_m_Page_212.pro
3514309255b9022cf402caf68f156c91
c4e803d67bd64adb3e0e9b068237ce6ce5543458
4563 F20101219_AACCJB woodbury_m_Page_197thm.jpg
2ba9d584d29ace152b88e044238ea94f
348547a25f73f51722e18c6a80c9f8799cf34ea1
F20101219_AACACE woodbury_m_Page_046.tif
42c1785558808b7dd1383c669ceb0c9b
1090460c0c666e6fc95f0557489ba5e26164354b
33857 F20101219_AACBZT woodbury_m_Page_064.QC.jpg
ee8ea44d240dd53bf6f082cb17c1ac7c
e36ffd78d7c2e7273b5266016ec9d96b19f7eaaa
109395 F20101219_AACAPP woodbury_m_Page_034.jp2
6c42630a1ba90eb03dbab474972d37fb
f4d4b8223615c1bc2a0c781a6ef7e84909d54ed3
56631 F20101219_AACBMJ woodbury_m_Page_213.pro
265538e69c664c3b49a62b3da2f36e7d
8e0504d5b1eafe0d4d44c59340a1db1387ef070c
18530 F20101219_AACCJC woodbury_m_Page_198.QC.jpg
fcfbe6a43ae214b34bab9d31f9f79f20
295ba803afb173c75b569f1980a3594d1c7c6850
613 F20101219_AACACF woodbury_m_Page_206.txt
5f1eaa507496a98739d56d8d4eafcedc
4bc91386ac49208de004aec2cbbc1d8fcd2bf11b
8225 F20101219_AACBZU woodbury_m_Page_064thm.jpg
6c703fdedd59d0de4abfed60860bfdcd
87ea1ffd0d8cd611ebfbbe1008a9630ff3cd97b6
107591 F20101219_AACAPQ woodbury_m_Page_036.jp2
94a1b2014c691c35df292678955cdd49
d8ada5f7d39dd4c8984a390f39cfef9700401093
58130 F20101219_AACBMK woodbury_m_Page_214.pro
70ad4c2fba964780b309586fbe98eca3
d6ad699acf85d6fb41fa1d806b9658fd22aa1553
4601 F20101219_AACCJD woodbury_m_Page_198thm.jpg
b7b87285a7b2789536803285d3697e63
4215589d63b2222399160b8f792708ec25172fb1
F20101219_AACACG woodbury_m_Page_222.tif
79b012c2ec0567bbed97ee728c8fd2db
2507b3cedfe08fbfd01a2be364eb87d251e53367
7818 F20101219_AACBZV woodbury_m_Page_065thm.jpg
2454ad5d090fee5cf0df99d4dedec0ad
aee29eb0e7654b34d0e360d2706c1af61df2c245
104911 F20101219_AACAPR woodbury_m_Page_037.jp2
18d36003cdebf6a083c7dcb40d1d12fa
70d07a78267f0ac0d288434ccd10bfb43cb435d0
61291 F20101219_AACBML woodbury_m_Page_215.pro
ac66b2db797870b10dc194b27ff1465f
897301477f85591fcb47123995d459f96ad2d258
22683 F20101219_AACCJE woodbury_m_Page_199.QC.jpg
3a6066f1db1dc8bce29f8228765f940e
04294d79f7ff825b4f344b8e9396f184f0fac8a5
8317 F20101219_AACACH woodbury_m_Page_084thm.jpg
b8e8c91a474242ef55025cdb4718c7c1
43786df28e490df0c046037c91a3a267f2fbb435
24049 F20101219_AACBZW woodbury_m_Page_066.QC.jpg
b02b7c3204a366737e4c7f35f5cbbb40
b148283bad95837c1dc5b3b9e9ffaadc32a5e484
112191 F20101219_AACAPS woodbury_m_Page_038.jp2
3483837c559cd3b14269c4989c4dd147
e6bf5705307e94fb6544746ec6368cf4acf2e662
63999 F20101219_AACBMM woodbury_m_Page_217.pro
38d3191ab9206ed281d43af2aeddaf3c
8328c079d8a03ba3c3e7a56d5dff3b15c4280687
5181 F20101219_AACCJF woodbury_m_Page_199thm.jpg
3b8cbfdc8a958263c62da1c90f66df71
d8e7e5190124216a5469e0756498fcef3d70f44c
99351 F20101219_AACACI woodbury_m_Page_131.jpg
f6cd6702fc1bf1464fdeb76e36f2981b
fb402fcc57cd383c6204eb258ece8709c7475abc
6080 F20101219_AACBZX woodbury_m_Page_066thm.jpg
f0082f4daf32239377d9e59a249ff739
c25c121af7c8c14239868282df80d2e0a271c8a5
109024 F20101219_AACAPT woodbury_m_Page_039.jp2
805ef32839dc965a750df924d49963e3
cc9d59f6ab63e4fbc583dd18bf8d9651474207e3
59178 F20101219_AACBMN woodbury_m_Page_218.pro
ea05da6363ccbf53b30d0f471aa6406f
9ba70706b092ef1e97a25daa6a84061ba38cc222
F20101219_AACCJG woodbury_m_Page_200.QC.jpg
b01a693f2f38a3675f9eb97126d574b4
9696da3f0f31c09cd7a907f17edf9486f3c0699a
113655 F20101219_AACACJ woodbury_m_Page_153.jp2
c237afe53e177d611bbe016409b4ded2
a03146c702b2fd4a6609e86c2e1824fd6edccf3f
6608 F20101219_AACBZY woodbury_m_Page_067.QC.jpg
e2323fa33813165a4577ada0f71fbef0
fae24a310ec1d304cde25b81eb3090b4d3060b8c
114715 F20101219_AACAPU woodbury_m_Page_041.jp2
0ebc53bfb4a1949b34971bcb997bba29
ef96a5b1e2e3733a662b64ec5fdb867df3a86625
57994 F20101219_AACBMO woodbury_m_Page_219.pro
a0f642c4e97a7dca697902b0a12a8155
8c346c49a99c15d7cbd709a85296cbeec276b486
5090 F20101219_AACCJH woodbury_m_Page_200thm.jpg
54051d59282d624317fc9df2182f217d
eb84db438b0c33c22e2ef6caf540352de2b08789
33143 F20101219_AACACK woodbury_m_Page_177.QC.jpg
e5cf895e009380ec2baffec7165a74e9
391add333ef211b87265d2101353807c47bca87d
1838 F20101219_AACBZZ woodbury_m_Page_067thm.jpg
a512bc42f12d2ae930c3917c46ab76d7
422975bd828e5ffecf9f05e29afabffa8f42dcc6
109472 F20101219_AACAPV woodbury_m_Page_042.jp2
06c040455001596a12ef1ceec37f03ee
ad4b6b2301b29da3cfb73366ecc828969e4ef811
65535 F20101219_AACBMP woodbury_m_Page_221.pro
edfb385d57b24883ffc50b11b8c0a398
f23dc8bb9d098eb8f13d6dea58b2faf767fba145
9752 F20101219_AACCJI woodbury_m_Page_201.QC.jpg
de8cda1362f9e4e4be8b507bc8f9d4e7
e48860fd49e63acb860bf40d34245f7dd9310031
102363 F20101219_AACACL woodbury_m_Page_141.jp2
c4c30c2d830ecf69bee126c7b6bc94e2
19e512b8ad2063466e7cfef8018348748c6d6135
96238 F20101219_AACAPW woodbury_m_Page_043.jp2
115a7e1e21c15a51de2344ac7f8a509a
2e21c939ea321642ee589b7b83f56362abf19d9e
59380 F20101219_AACBMQ woodbury_m_Page_222.pro
a247daea29024cb85101ce62ea97f7ce
dd18da55476f11ddec64e7610cc19dfe7b759e5c
F20101219_AACCJJ woodbury_m_Page_201thm.jpg
587fafef571c117baa609491bd2efaaf
64510e740d640908e2947cd04256c012bc1f242c
7516 F20101219_AACACM woodbury_m_Page_044thm.jpg
c0590907c4a594c155c2aa2ad20d6cbd
f38ae10c727aa51d68a3567266b0e6aa7290edcd
98760 F20101219_AACAPX woodbury_m_Page_044.jp2
1ba21595a6f2fdcc3e90e96d10792d9d
708d0adb1938299d3315027e31da8eb6ef6bfa75
59354 F20101219_AACBMR woodbury_m_Page_225.pro
52587e62012c1cfa5748f602ddd85b68
707a53f6793c9db503382dba3ceb29a79e9b0b18
21104 F20101219_AACCJK woodbury_m_Page_202.QC.jpg
e2d4a0ee47e93bd292eca88f5500b0d4
ac86553531bcda39d845b23456361005d02054aa
F20101219_AACACN woodbury_m_Page_172.txt
d5ed0333fdaf39dc53280f929def2d09
8dac958f8dee85b8b35573bfd4b8364a1ae8ef8d
100322 F20101219_AACAPY woodbury_m_Page_045.jp2
521537b317cb3c208a249c8ab95f6e3a
cbec04eb0ae5bee8a9623b9553ad24239f8a6d03
60722 F20101219_AACBMS woodbury_m_Page_226.pro
d0364749c1aa730e210630c7eb2caa04
8bdbbf95d0c5592d9a719cac096abf8262bf664b
5010 F20101219_AACCJL woodbury_m_Page_202thm.jpg
5186ad638d82783b4d6d70b94915f3ed
84b02dcdf9fb18d5ad3bc4e1c85f6e37c28fbf74
35649 F20101219_AACACO woodbury_m_Page_115.jp2
02c612e63c6c3ac9fb99ede527c1cb3e
833978cd428ffa71b06185f4e46eb4c67ff6727b
112741 F20101219_AACAPZ woodbury_m_Page_046.jp2
dda7e2d5807ffb55aaccdcc30f6b2920
cdc3a1b5b4292a8370637f73dd294d5eafc37b2b
56843 F20101219_AACBMT woodbury_m_Page_227.pro
fa127d41dc3dafb569691b352479169f
0e74d372ff46628f87eb1f2496c0a062bc9d1017
18318 F20101219_AACCJM woodbury_m_Page_203.QC.jpg
691eac2fec42f65bf2330a73a02f10a7
7e0b51241672048ad966c0d21714420f6c5cc5e0
51884 F20101219_AACACP woodbury_m_Page_180.pro
0dd70e8b77038e23dfb372c140def466
ca662ee837fd9b4d3ba6ca77dde2f55ce81a11c5
23755 F20101219_AACBMU woodbury_m_Page_228.pro
74dbca25f4632d2935446566a8824205
59df12b4142907ead60d6753a4ddf90fee286ec8
4266 F20101219_AACCJN woodbury_m_Page_203thm.jpg
bcb7d0ef2cb9f43854bf422c1bf140c5
368bcf413030443a7334a49217cded905f5be651
73837 F20101219_AACACQ woodbury_m_Page_202.jpg
64591bd78ba268ef28f6ac3768828991
83bedf0507d1c59c57a8e554545f242d9b4ed99a
553 F20101219_AACBMV woodbury_m_Page_001.txt
581575d8a533459b52c84b000a776987
dfec6b5111d9e8ade32032f1908ab3815c2db85b
25103 F20101219_AACCJO woodbury_m_Page_204.QC.jpg
b7426b5b0ce229f9f0e876b0e96f7cc6
9244e02d1ad916fc7e10ff8f2b05f4baf30eb341
117 F20101219_AACBMW woodbury_m_Page_002.txt
e3adf4a1891c597401edca345e84fe62
529528f4016f1c6837a3290e3a236b720cb05649
5785 F20101219_AACCJP woodbury_m_Page_204thm.jpg
2e56d777ad0e82d542e082d1557fa406
9f5f76b9e9851e1fd4730c239007c54f5090fce6
134301 F20101219_AACACR woodbury_m_Page_209.jp2
01afc9813a36e2c3506efd40666eb628
bbdac0b075f47778e614f48836caad855014659c
118 F20101219_AACBMX woodbury_m_Page_003.txt
3af46e3eb75f645565ddc6e77f6a494d
8b3bcff8c31516e80e4e01175f60a5c6aea06f71
19728 F20101219_AACCJQ woodbury_m_Page_205.QC.jpg
0599ff3f3e4709833de1ed4aa7e56ebb
d5f5592c4862a142d4463bf71248ebf889e8be75
8223 F20101219_AACACS woodbury_m_Page_098thm.jpg
617a7c4fa8e3b1c75f0145d7c531a8db
6b317e99022146522e47e3c824e53e72051d4e24
1749 F20101219_AACBMY woodbury_m_Page_004.txt
be39c7148f1926decda634feb1b11aee
e755c3baebc590f3b41368b6e4c565af2b0210c5
4533 F20101219_AACCJR woodbury_m_Page_205thm.jpg
81b2e308a7f06625131a2ab11c06ae24
d36696540f9e710ff0baa97d475c1876f2bc0808
F20101219_AACACT woodbury_m_Page_211.tif
3ea1e85cc7e5449d49ab9e0a554bd9d7
3cca1e384d3301eb0e86d32006e8c2b173c3d154
1951 F20101219_AACBMZ woodbury_m_Page_005.txt
4bd02c8880d413a0a8706f8e0ada2cd6
30fc8e1da71777a30795bd367cbe08e95aa047f1
10947 F20101219_AACCJS woodbury_m_Page_206.QC.jpg
dda6a04fcb3cfee4a5632fe1ba47e4bf
a521616c09bbc53ea88ab11573d7635c3f4c8e31
8004 F20101219_AACACU woodbury_m_Page_134thm.jpg
f1aa9a23ac9baf5d4b42b673871ad0e6
da8c2601b091d129362ec20ee1f875a59d5026bd
2618 F20101219_AACCJT woodbury_m_Page_206thm.jpg
5f2dc1523170fcf7f431bf18441ee99a
defaccd393ce0d479d23946946799a9ad43f906a
F20101219_AACACV woodbury_m_Page_214.tif
910cbca99e76d37ad88b3e4f1d273f2e
236cfcf8049d513231dca7c4f9a2c1ab3bece975
7649 F20101219_AACCJU woodbury_m_Page_207thm.jpg
36d06a320b2a36cb2e67aa7475337239
ac84e67938553a92cebe474af3503efdbeea021b
102242 F20101219_AACACW woodbury_m_Page_101.jpg
70f5c8fd0bef6fd2bd8e74567c0c880e
def93cdeaee958f064f793d5e1bab7ca40d126d8
682019 F20101219_AACAVA woodbury_m_Page_185.jp2
8524ce455ed5602172bda11bdfd16b36
0c8ea3e6dbe88843c2021f963381761081167b07
35554 F20101219_AACCJV woodbury_m_Page_208.QC.jpg
a77b99cae1a5f7a55bb2f678cd41546c
ae9a15c217a49db67e43543652b92ffd858feb1c
F20101219_AACACX woodbury_m_Page_023.tif
b9988c18fc43ec24e30d1e448e110eba
4b1651a65dcba6ccab2b454b13493203bfd877c9
993228 F20101219_AACAVB woodbury_m_Page_186.jp2
78209e060e4f0d15360049545a16aa4d
a4908b4312ef8211f108bdc31d8cdc53d3507c9f
53946 F20101219_AACACY woodbury_m_Page_080.pro
bee2994475bb999220584d5d32f75211
a6a5242e19cf955ea6495a468f2f77b840df5f07
635881 F20101219_AACAVC woodbury_m_Page_187.jp2
7d2f173438660a0ce08821c414963234
344c4a996edd06f93ec373dce2ea8bcdc37302d5
8507 F20101219_AACCJW woodbury_m_Page_208thm.jpg
be67366ff6c88efe5549f4059cafd84e
cb88bf8198ef6293059471e5dfe0be25730b0eeb
31292 F20101219_AACACZ woodbury_m_Page_065.QC.jpg
7bd872cc40102d6447898597b950a125
00f81a843c2c4f641477a36231c787c09e15cd51
97738 F20101219_AACAVD woodbury_m_Page_189.jp2
3bcd7db12cdc7f8da6efdd2724fbd56a
d89be685e290728891758c1a0bfe1aa509641d1c
36033 F20101219_AACCJX woodbury_m_Page_209.QC.jpg
6196fb0ef0dcae17241390b4838da44b
429d75659fbddfeeb9d9973719f37ec53c05502d
8923 F20101219_AACCJY woodbury_m_Page_209thm.jpg
faf2ea15ae3ef40910975d3c16cdb5ca
249096fe3f9e366acb57bc1eed600d381910e528
107324 F20101219_AACAVE woodbury_m_Page_190.jp2
bbffb8b007e8e9bd58aede0e332e7d2c
a0fbed70e01e31491bea188f911f71a061409edf
33712 F20101219_AACCJZ woodbury_m_Page_210.QC.jpg
86858c409a279ff97d7560b87a1c1c7d
b2b3c533a9eb760f2ebee37ebe397c7931b29884
109111 F20101219_AACAVF woodbury_m_Page_191.jp2
f32b3deaf2e9c292dc1e4fa2c18e0381
c20cf9efbc85dcb08600ff67b138dac5597c82f9
69682 F20101219_AACAVG woodbury_m_Page_192.jp2
2d7a0154d5bf0246525380719b6372c7
5f97ddeef7ecac6aa70a45546d79189df2da5016
50918 F20101219_AACAVH woodbury_m_Page_193.jp2
ccd4802eecb03105f42bd2d5070729d1
5753890df28498b0ad3f9989b67a6106886cebad
2173 F20101219_AACBSA woodbury_m_Page_142.txt
e36ec79b765bd9b125d69c43631d9043
7031f460e7038c1469c73fd866d19d60f80a5883
57163 F20101219_AACAVI woodbury_m_Page_194.jp2
240db99fc88c5c88a06149a20b6bf61c
9169a1c04ed2baec721328b40bf1610eb8cce69a
1975 F20101219_AACBSB woodbury_m_Page_143.txt
ee6b23c7a968a590412c228c40a3d7ae
b327e11202ee728e006e105aeac3b47470997216
59291 F20101219_AACAVJ woodbury_m_Page_195.jp2
2017e951acc7ba7689a63b0c88837dac
52b5c0ba8f29a1755e8c62fb4d622e5d45ce7aca
F20101219_AACBSC woodbury_m_Page_145.txt
191e2cdba8951d301563dadedf2bc097
703e8206e84f1d45251d24a1b245a193c3d65f1b
61804 F20101219_AACAVK woodbury_m_Page_196.jp2
376330098db28f9e6d67402b843d5e97
9fb64a44c3539cc453f14761e53bb286f963dd3a
F20101219_AACBSD woodbury_m_Page_146.txt
5988ab0d680c3a23480fe2a4631a7603
450d9487879ed07f6e3da6d26cf356ab45d1cbbb
105057 F20101219_AACAIA woodbury_m_Page_050.jpg
cd9b685334110ea07b5a7660c6c916ab
629defbbc7361a7dbe5259646a19ebe251cbe69d
54685 F20101219_AACAVL woodbury_m_Page_197.jp2
6bd6ce05ae7c15afd24048303250d269
0726bd4a9ad009fcf5fcd5a03655cdde23689b55
1893 F20101219_AACBSE woodbury_m_Page_147.txt
ad77444acfb42dc0cfccd91a1043bedf
af53f4fb0f15376efc6f2f0496c06b9b774fa9e4
102681 F20101219_AACAIB woodbury_m_Page_052.jpg
afe1b6c89667ea79cc6019a65748de50
62c8bf7367ffc8d22df62a1436845092713dda22
59935 F20101219_AACAVM woodbury_m_Page_198.jp2
4a0bb5afee2cca268639f0530b58d7dc
82df24d7b63eb87688847706189f2036edc2f3c0
F20101219_AACBSF woodbury_m_Page_148.txt
58f5f8fa01b5a2aac97a5414d2f26cf5
8a33c4f69587dadcb14f4652a94f50ecac36ea39
100722 F20101219_AACAIC woodbury_m_Page_053.jpg
ac1af8320d817f1a058df184daf68bbe
2d606b0a4110595ee7a6b478fbcf85e4826d2d31
75669 F20101219_AACAVN woodbury_m_Page_200.jp2
893f81ccbf73e5086554139987e9649c
ae38870dbbb1f0cab5b19dd9d684891d67917fab
1962 F20101219_AACBSG woodbury_m_Page_149.txt
0ec625e078346123b910775b599c86c1
68d7184c7083f03085738e7a1051f107c79e0e50
103083 F20101219_AACAID woodbury_m_Page_054.jpg
dff1fb9389187864a79ccb85501e286c
45ed765ce5b3549280420e45be534088c92849af
70518 F20101219_AACAVO woodbury_m_Page_202.jp2
05a9fb4c3ca3034f47e46ef8421332c1
e7a619aa2bfea52395bbeef09319b30cd213444a
2052 F20101219_AACBSH woodbury_m_Page_150.txt
5eadeb270cf5f808825af8d71aad0484
9c641b70005e230e791601763304773a66c3175a
104319 F20101219_AACAIE woodbury_m_Page_055.jpg
62e5a19f26b35daac120fdfef480d68b
0c6ee2708fa630d54aac6e2ed7e40fd3a172dc3d
60802 F20101219_AACAVP woodbury_m_Page_203.jp2
2ec32a7dcc9a4a672c4a67a8b19cfa7c
eafb358860382907ea2b13ba3e821e3bd317401d
2039 F20101219_AACBSI woodbury_m_Page_151.txt
2bc76cf029bb4c139cf52ec64af3575d
e6f0fdcac25ab7db8f1ad799981eba6956362e1f
103905 F20101219_AACAIF woodbury_m_Page_056.jpg
fc36df18b068c110d4c2f46439126585
cbbfd9d0d6dee8acbfa4120687409becfab1c9af
86679 F20101219_AACAVQ woodbury_m_Page_204.jp2
4a642082763f0ee13ef88de863c34f3f
1e265477238de080def4f3db5dd9f518cfc2e44f
2003 F20101219_AACBSJ woodbury_m_Page_152.txt
e64330844403b55afd6378e7604d9eb5
9f532d463b5e2937ba7beeaabfe4f21054d1fcea
101360 F20101219_AACAIG woodbury_m_Page_057.jpg
94dc67491efc559146728a7013269fa1
ce2c3a9e870306093bca88f574dadc1e771c7a91
62295 F20101219_AACAVR woodbury_m_Page_205.jp2
1c0ce7c15eb318a4a95910e6aefaf147
9ffad57ae33df7f2608fffe2ab1d257a63193c27
2077 F20101219_AACBSK woodbury_m_Page_153.txt
60a883708a7e29caa00bd91df040c660
697f48c91385f23a9067c8be0cc4cec751ece880
105493 F20101219_AACAIH woodbury_m_Page_058.jpg
73332ee66863c755ffd0579886b31f1b
2dfbd1e75ea75ba137fb2b7294dc969ffc085e64
50442 F20101219_AACBFA woodbury_m_Page_010.pro
e17de3679a6620acb988fd31d48372c4
cfc721a7ed798bc792ef9346ef4cdd716e47a26f
35403 F20101219_AACAVS woodbury_m_Page_206.jp2
ff1b0332e2403ccb4afc69e101235b21
d9a140b6415e5053d9a0bb48aa1117702c543e93
1859 F20101219_AACBSL woodbury_m_Page_154.txt
41a0a2f265417e038d18002c9929f06d
c43e63a8e977229abec371ec39e411306142a5d3
97835 F20101219_AACAII woodbury_m_Page_059.jpg
90dfa21fceca68d28c2f8591741d4e02
d9df57a4f876888857090773d72c6f5484f68070
23151 F20101219_AACBFB woodbury_m_Page_011.pro
6b96193cf60569b4e27e2ac51ff07f25
cf640bde90371914c0df4805958b05740d6d8d16
112367 F20101219_AACAVT woodbury_m_Page_207.jp2
afb2d73b9232b62449c8e4eacbd4c632
bac748028bd11da84ef428a4e2ab0289080a5268
1527 F20101219_AACBSM woodbury_m_Page_155.txt
6219683124cc8e9863de94e7d690e875
3698bb1fa8d9fad040c1311b20221928873ad20d
101759 F20101219_AACAIJ woodbury_m_Page_060.jpg
de455b407c8bd5e3113efbdc7f57fc3e
70cb2631f057ad58d099c46041dc4b478f800a01
46454 F20101219_AACBFC woodbury_m_Page_012.pro
24b4b607c08698f1b8b2efef5a25ffc1
2a7e39b5f9153860a34827e5b86b530c82f55008
130665 F20101219_AACAVU woodbury_m_Page_208.jp2
06e2e3ec10e571fd3f248572c76ec73d
c26a3b3f7d9699db47d8139d852371f7b46a235b
584 F20101219_AACBSN woodbury_m_Page_156.txt
180d42fe6ac8937b6eed0c56993ad670
94711d3370819f18f29df38630e90bf989244627
94483 F20101219_AACAIK woodbury_m_Page_061.jpg
f1e8d34cb0b8625b71ed359b1e0c8f55
bc7c12f07f3a43d0ac86ea355575fffc333dfe33
39811 F20101219_AACBFD woodbury_m_Page_013.pro
be12283abad41ad6b9b271a88aac8b93
c1d80972f0ede1df2d2f8100a84199b3bba081f8
125874 F20101219_AACAVV woodbury_m_Page_210.jp2
6f4403aa72163405ce6b26d7c24f3bee
6cf099e0af6ce73090f3871a86a1fd346140afd1
687 F20101219_AACBSO woodbury_m_Page_157.txt
1c6d1c25d9f189a1511805f9e6314ad1
7cbda88e26ac9128c33e54c1c92268f5f2e0eb53
108117 F20101219_AACAIL woodbury_m_Page_062.jpg
fb7fb4192323aa84b91a91790edd6206
c232febf64962c86be55dc6624b0eff9a93f2212
F20101219_AACBFE woodbury_m_Page_014.pro
62b45b29f142aa2d6c1a244e7ed043f9
46f982b6d041fed1491b9fcdd1d00cc45f634cfb
1051981 F20101219_AACAVW woodbury_m_Page_211.jp2
0a3d65686c00593ce3d90cbed4ecafa8
b0486c6ec065dc2c1d87d417b121fbc5f5c94d66
1559 F20101219_AACBSP woodbury_m_Page_159.txt
7fd2713912569430c8cf1806ca8d23f3
04dabb5aafcde69a9648fee6f93dcb4d37931af0
99736 F20101219_AACAIM woodbury_m_Page_063.jpg
5465a95781ee76420be7dc0a7f8e0983
145ae35483a2e63669a303953cf9b60cde6f1865
45485 F20101219_AACBFF woodbury_m_Page_015.pro
cd706ba0dd7a0e58da703ae1081c914f
c07c9c5e972e6a816f186e79ec816787df484979
119272 F20101219_AACAVX woodbury_m_Page_212.jp2
646d5f9ce697582b61d44414b9d69f1c
5c6d2584e30d28579e1499cab8899d556d3d9cc8
1947 F20101219_AACBSQ woodbury_m_Page_160.txt
4a45be60ed1505c940ee04ef6ae759de
1744564b8764e9725bdef53493659a4b3a4672c0
104425 F20101219_AACAIN woodbury_m_Page_064.jpg
b33d875ffcbd9b725e98b7631a2dcb5a
bb478676a6601fa0797c4921773f0fe3950b22e4
49127 F20101219_AACBFG woodbury_m_Page_017.pro
1adb079a2485c0f979ced1edeab1a55d
ede0e28f83e955d0c6bf8ac4fc9caeb81c8ed4d4
119857 F20101219_AACAVY woodbury_m_Page_213.jp2
e18e9f6702a3b78c854ed4f14cdb48d6
15fd2d0afe5a9ce5b5d8b35643ff6a9aad32d483
1064 F20101219_AACBSR woodbury_m_Page_161.txt
b17252f197a8dae3c1a6289791056113
de72aa925481303c8593f4ad18f80ad35fd9d3d2
97071 F20101219_AACAIO woodbury_m_Page_065.jpg
3d97394db8a6befd0909298f10569840
f2c06028fd689ca34e53cbd53cd329245248bf28
51606 F20101219_AACBFH woodbury_m_Page_018.pro
3f6c5431a0a95f3cc0a2fd22f1b006e5
8d7527856e37eaae6742161d794c9b74d52e7761
121928 F20101219_AACAVZ woodbury_m_Page_214.jp2
1c88805af6a133eea6e8b85af0587225
2823ec78c615eb53ab4a0545424569517fa45fe4
5553 F20101219_AACCCA woodbury_m_Page_094thm.jpg
75735271f71667968013b89baea8287a
052647218a839407fdede212d9dfd06d27d607f5
2806 F20101219_AACBSS woodbury_m_Page_162.txt
c8af06d9045c593b69f7ad489dafd171
d25b4085d332a226b536e6f6d7d243a9738cc7eb
90046 F20101219_AACAIP woodbury_m_Page_066.jpg
405bb384132eca9a055c80f2ac4c5d3f
2e53056e8a96871c50fc10b3275bb51580c766ea
51810 F20101219_AACBFI woodbury_m_Page_019.pro
df1ec146a6ef1010fa3aafce410e9012
512a936fee6e9ffe01ab84a746a211196098ea18
11883 F20101219_AACCCB woodbury_m_Page_095.QC.jpg
3a7d7467e986ed2d97e78673e669bf69
156df63ca2c8772c082bcf4a4a74d42b8d7966a4
2757 F20101219_AACBST woodbury_m_Page_163.txt
797d114fa93e6cf590bea1598becef52
dbc6e425d1871f1c0701aac43f7e96e03ced54cf
22586 F20101219_AACAIQ woodbury_m_Page_067.jpg
93faa60875e0c1eb7b32f39c8337cca1
dcf446030776bc19476b53351bb10338777f761d
47131 F20101219_AACBFJ woodbury_m_Page_020.pro
8ae138349f17538e78f5b96a242b8b28
b0f4a52d3e3c0004dbda9b9ea954ec8beea7ce70
3218 F20101219_AACCCC woodbury_m_Page_095thm.jpg
8cf39ea58e8b225a0b08ae037f9d16e8
bde39266a2a7eb8a33e64054c2f82253b014d7ae
1611 F20101219_AACBSU woodbury_m_Page_164.txt
ec03325c5fb7ac456ac5959a125b1620
be1f0db62a5fdb1a03e1371acaa063ebab0b24ef
100476 F20101219_AACAIR woodbury_m_Page_069.jpg
e6d70c3f68e263bfcd1ca535ee2a52af
7159876d642e5c16469f5adac3fd175691b76b52
49327 F20101219_AACBFK woodbury_m_Page_021.pro
5df4b104ee53603f1e81a2db06de4913
ec2b0b2e806285144d6b31f135c9dbe2db9b7e74
30415 F20101219_AACCCD woodbury_m_Page_096.QC.jpg
d04208e86d8cf05a6c565fd63236af97
f79d9db911ae4d7f0aaa38cea3df1ac1a8c61bf4
556 F20101219_AACBSV woodbury_m_Page_165.txt
7e81c9561beecb20500294f96c3467f7
3b155180cb426624ff1ef24b7e308b2ca0b5f708
113543 F20101219_AACAIS woodbury_m_Page_070.jpg
2e7634e4dc1c5c4296e10230da137951
837b207470eadc1bb35f31a9f8202869561aabe3
46885 F20101219_AACBFL woodbury_m_Page_022.pro
eaeb56fc58886be16d8da75ca71ab6c6
800e2325e272fbc562fbcafeaf3d46d56dce6bfa
7268 F20101219_AACCCE woodbury_m_Page_096thm.jpg
3e9fe9649ff77ae0b09f124b22739ff9
2595e53abe42c182a622290d5224ef069d332084
1281 F20101219_AACBSW woodbury_m_Page_166.txt
ce6b07548e6006c91ce38fe3924c9659
8c8b2a3c99fbeb7e929beeffaaaf6c1866044de8
98427 F20101219_AACAIT woodbury_m_Page_071.jpg
c00d1b26ec629dd89470f4a42d95550d
002b8129005ccb95c4b72ff88b505c0eb6e7d9b4
54932 F20101219_AACBFM woodbury_m_Page_024.pro
73a55a7e6473e0a39fb45e2ce304dd50
830026f78368a306bd204bfa2800e51004636e0d
31504 F20101219_AACCCF woodbury_m_Page_097.QC.jpg
0e3d143d2c82805608aa687921c3960e
1578840dd7b58e9c7a5ec292a60d9bf844701a1c
285 F20101219_AACBSX woodbury_m_Page_167.txt
8eabe0151a91614fa5176b67aa426ee7
27e437c8fde0b72687be2c711cf8ffaf006a0eb6
103223 F20101219_AACAIU woodbury_m_Page_072.jpg
03ca2ddfd63a2a48f8a5de9c8658c54f
3a001ae2a223d1e97ae66b85ab65f4adc5219cb9
51442 F20101219_AACBFN woodbury_m_Page_025.pro
548f032e643bdd562e3583489a6b1b59
07db2c64c55934da57368c0697bfef4659fbf93e
7796 F20101219_AACCCG woodbury_m_Page_097thm.jpg
3d8e84d04dbfc53626bbdee331b0fcc5
abecf272d5fefb4431c0f19d2e3b3636f4ca1038
1887 F20101219_AACBSY woodbury_m_Page_168.txt
89579a8e92315c1400fc1286b51e837f
f683ff9ca57cdb4ea3cef630d335ca5b9c7fddb9
101691 F20101219_AACAIV woodbury_m_Page_073.jpg
965de760c42c4b25d1e1cae8d2c03b85
6f706908d33581158572909a1ad97816405d35e3
50280 F20101219_AACBFO woodbury_m_Page_026.pro
b06f0ef94d0c0517ad2a48a233105040
9679b2d4a731ae0070f2b3371cd34cf9a01d8b15
33825 F20101219_AACCCH woodbury_m_Page_098.QC.jpg
6fb1469580e40970efa14e91ffa06335
2a3843d9ea0b6d941d2ebce9c1987f4d3e2014cf
1964 F20101219_AACBSZ woodbury_m_Page_169.txt
5d2b85e58f1152b82636783b9147d3f7
ca2c4a1634c1bff7890230965caa1fb73f1525e1
103484 F20101219_AACAIW woodbury_m_Page_074.jpg
3b3036c1675eb2eecf71a4807f32e593
08ca13bb2bfb840b94e695d50755d8f23e0d3d4e
51360 F20101219_AACBFP woodbury_m_Page_027.pro
4fd0d05d045ce5179af02c3a7e797cc3
6f6d2ceeb4d19d0f26ed97488ed364d9c7923a6d
29704 F20101219_AACCCI woodbury_m_Page_099.QC.jpg
c1a882b210eacf5afb5ebdc6fc70e35d
8c2b819624764126c20e083f7073b682366eb322
53115 F20101219_AACBFQ woodbury_m_Page_028.pro
3af2ae9b89383aaf866deaf13d148be1
6771d340840d7fbebcd648f6e0cfa2394116c9c0
7468 F20101219_AACCCJ woodbury_m_Page_099thm.jpg
2b25cd1fd0b47de896fed8eeb7b02a85
39d116db8fae188646cbcadfbb8a38e1d601de46
102244 F20101219_AACAIX woodbury_m_Page_075.jpg
629c09fd5468eb0b20ba0f22abf97d73
c80dffd93acdaa85e4a4525d5ce4cfacb4fac031
53078 F20101219_AACBFR woodbury_m_Page_029.pro
36928d896341eaaff6bee61bd2c84a46
bcb74b0485e498f7556a8446c9ea0e6bcf23e8b6
8018 F20101219_AACCCK woodbury_m_Page_100thm.jpg
fd3041d46916e01a25cad7b880801432
7d140257ee03cd3dce759b5f78f1c07d67b44985
100025 F20101219_AACAIY woodbury_m_Page_076.jpg
53251dd365f3a24bfcc4ea29e1c81463
8808866b4fbea0762a09b7e64e7a6419ba4194e2
49356 F20101219_AACBFS woodbury_m_Page_030.pro
daf50595e78fca372b30cd8c0d7ac3d4
5f90afab30c2e7fbcc53c49bcc81e51f7a27b647
8176 F20101219_AACCCL woodbury_m_Page_101thm.jpg
a2a1d8eaa8aef77ab50259b6482c51d6
b56695276de1e10d82f7c9a30bf81456d4e0c426
103896 F20101219_AACAIZ woodbury_m_Page_077.jpg
ff8f0cf0cb098d6feddc7499296c9633
a80d5f4eaed9853bff370953376c6c5e207a85e0
7116 F20101219_AACCCM woodbury_m_Page_102thm.jpg
b1aa10fc791e21d735230399a23e9e7b
8dae12bbdde9014fd89ea6e009f7f08d185910a5
50973 F20101219_AACBFT woodbury_m_Page_031.pro
9d2952386ba4f6e4845c9c91a0d75e0d
17a982a9958997aa3a6106cca126e59f760515ad
28963 F20101219_AACCCN woodbury_m_Page_103.QC.jpg
03a9a73cfa5ca8dc22efba0ddd095256
605c6397c1f553d86e9152ae67791a9189a477d3
49494 F20101219_AACBFU woodbury_m_Page_032.pro
68db15d1ceee596356d235999f3364a3
0075d1f8deb4a33a494a6235d105eb55aa70a851
7325 F20101219_AACCCO woodbury_m_Page_103thm.jpg
49976f9f8511fc417d1f691612b49ec9
69e7d4ba7bb7a4dad43adc36ffd36c6d8ed020ab
48732 F20101219_AACBFV woodbury_m_Page_033.pro
794360fa0e17b31ce4a94b72fd5f01db
097f8e18c06d18f7671ef7068f8f10377cfbcf79
50251 F20101219_AACBFW woodbury_m_Page_034.pro
f06b8870e65f3a0474a2e074d3728f50
f003265efa9985101f07f695fdea43d7178cb2db
34941 F20101219_AACBYA woodbury_m_Page_038.QC.jpg
111fae39f746a41450960f94214b0448
72d672bd7112f3a19ddfe8a5060d7b8a008f5b11
F20101219_AACCCP woodbury_m_Page_104thm.jpg
a05d7947387d681ab48cc860d1e15662
af03f9a94156835abbc494ee2fa08d24d05334a3
44409 F20101219_AACBFX woodbury_m_Page_035.pro
02c1f60b581f106b41a2f0fa49e298e4
4983d2d1d809bc98f3fe88c69c34cdde0b844a1c
F20101219_AACBYB woodbury_m_Page_038thm.jpg
79ba2ea6808af0fe032ff31dd7055c76
82c697a1af6edfe140c1cdedd79417029ab754c3
F20101219_AACCCQ woodbury_m_Page_105.QC.jpg
131e8f5f8eb7a2686a808a581843986f
c928a1934106ad9fc075ffa42ef898e8401f87b2
50075 F20101219_AACBFY woodbury_m_Page_036.pro
9c32e5bde8c45fb6630eb7d8cc5d41b4
7feba7301003ebc8f410bcd7db94c0ed854ea49f
32892 F20101219_AACBYC woodbury_m_Page_039.QC.jpg
b7ad7e19355949e00620e05110fcc266
b8758ddcc24c8dc2d541e3f17cf04b99a31f0136
8424 F20101219_AACCCR woodbury_m_Page_105thm.jpg
dc955fa235c25f6d137af0f91b13572c
2920f13a2e9c470aee3ec668053286116a9a8419
49162 F20101219_AACBFZ woodbury_m_Page_037.pro
eedb10d46781e1dc8a107aa7a9ada965
3e752cae2e95f070ac7e354a4ba15d63e790378d
8199 F20101219_AACBYD woodbury_m_Page_039thm.jpg
cf4f1330b3e08b544eb4651bd80970e4
e95a3b478896c06f93d859ff6cedb14922a3e2c9
32843 F20101219_AACCCS woodbury_m_Page_106.QC.jpg
74669e0fe6bf0fb40b87e57ac99a9d74
b735b26a6e41e0b4c77df73ed5fd74ff4dc34b43
33977 F20101219_AACBYE woodbury_m_Page_040.QC.jpg
f877e441bf4fee3f81e1274c22ce268e
4350b575fa36f9ae589fa6f5f5c10a350650a2e0
8210 F20101219_AACCCT woodbury_m_Page_106thm.jpg
f9710f0304d6c2b7e4f7c4d66c67dc73
c86858350c8895640716eea1cc3563683e3212d0
F20101219_AACBYF woodbury_m_Page_040thm.jpg
bff7424b6375ee19e44cb786e3adc2c8
818ffcce847fce57ea760ca130436e6bad535ca1
33296 F20101219_AACCCU woodbury_m_Page_107.QC.jpg
94349eefa875feca077718dc02c29dac
554591ba898166141e0de4fb06a5e818616d4778
139695 F20101219_AACAOA woodbury_m_Page_220.jpg
e95e93d0e1187ffb129da87c31357ab3
8a4cbc7c768f08c4f232719cd53d0e3906049bcc
8575 F20101219_AACCCV woodbury_m_Page_107thm.jpg
0447a083c623258a5df71d1f98739629
0114ae2ee0cbd0785459629b54c6d5758e0acff7
133854 F20101219_AACAOB woodbury_m_Page_221.jpg
99396462491af4f8e5ea9ab8fa7ea65c
064e59e5e6edee079e7336be0499a866eb8716f3
33787 F20101219_AACBYG woodbury_m_Page_041.QC.jpg
55ad488b742505a9db51d83819080ea3
acd7691661b122adf5c0b7aadf9fcdbad8e4fe7e
32650 F20101219_AACCCW woodbury_m_Page_108.QC.jpg
f1ed135a75cb047d6451fe57a96e61d7
97114e486803643a0d4f2d615617c6e955284f5e
125677 F20101219_AACAOC woodbury_m_Page_222.jpg
f599ebc7843830ddd97e348ff1a68a22
c92c9bb8bc4e11352acc065c9bd92bc264fe0171
33647 F20101219_AACBYH woodbury_m_Page_042.QC.jpg
8adc6b1691ded4244de2e17d73ffab6e
7ee582bb51f78a0d800f6e13be1297d7034212ec
8514 F20101219_AACCCX woodbury_m_Page_108thm.jpg
4475849b6d77c037bbaf462343fab945
0f6e27f3dd90b063ebc17317763aba1940ce0afe
118363 F20101219_AACAOD woodbury_m_Page_223.jpg
f0974b17df7991acebc3b1283189a50e
97d193538cd6a128865db939cd0fa30b7df7caa6
8381 F20101219_AACBYI woodbury_m_Page_042thm.jpg
f88a24b0a94d3d16200477be23d70199
dc6faaaafcc7b35ce4069c3adfd4ed712c1c5090
32321 F20101219_AACCCY woodbury_m_Page_109.QC.jpg
98462bcb6fb22ec71a97d179fea5091e
d9159daa7c320e143f4b451500359fce7941fc91
127642 F20101219_AACAOE woodbury_m_Page_224.jpg
34abbfafbdb2d2d7fd3baeb221f58a1d
b1bfacfc707462263d2e8ade31fec8fcadf931b4
29191 F20101219_AACBYJ woodbury_m_Page_043.QC.jpg
b5556d3bdc89039f275ecd7f4deecc35
e284ae2475d1717de99a1451274554f3ffdc20af
8250 F20101219_AACCCZ woodbury_m_Page_109thm.jpg
861b633743c5e923ff8b0f9e2ad7a632
16724672122d3d8d26397bddeb6f1d48e6262bc7
128570 F20101219_AACAOF woodbury_m_Page_225.jpg
ff15b36f9bf3b1353a529bdf6951ad54
32ea25b2f6cbfe84df273435e4c4d0650fdd32fc
7191 F20101219_AACBYK woodbury_m_Page_043thm.jpg
a5ed45fb4b61fa7c45ac32ab880be3d8
9477ee39773ea1ff31f490448d071c579bbd26b5
127772 F20101219_AACAOG woodbury_m_Page_226.jpg
83ad86921411131cbdbc46c4467e2ff8
30783240cc79aaf3cae1ff38c924cd77659c630f
45343 F20101219_AACBLA woodbury_m_Page_176.pro
ba9967faf6bb3b610ad2c1afc36deb74
5c156a69e831fbb07049248b4de68f9fe0fcd881
29735 F20101219_AACBYL woodbury_m_Page_044.QC.jpg
579769bce913b68c98970aca3ca653d3
9be355aef8b846b2943df24ca474098e2dcfd796
118951 F20101219_AACAOH woodbury_m_Page_227.jpg
9e14b15628628271ba7900c81aa8c08b
afc8bd343b70112547ff76f73a936f099e347549
49027 F20101219_AACBLB woodbury_m_Page_177.pro
2667528e769ceb9966882112490fb750
55592703fbc651803aae39919411060f271102f2
31555 F20101219_AACBYM woodbury_m_Page_045.QC.jpg
59baad694bdae4b69fddc186444fe00e
0192b4487c2ed84e4e05efe65c34199f420b6589
52759 F20101219_AACAOI woodbury_m_Page_228.jpg
8a7afe0a1b3a6f8b7805755c67fd23f2
828b2dce737359db7f05c27f773082c4f52bdf55
49975 F20101219_AACBLC woodbury_m_Page_178.pro
e240f4f6a7c4386931254ccb5697cf3a
0d6f2e967b51fa12112303ffb34a19c688966aba
7748 F20101219_AACBYN woodbury_m_Page_045thm.jpg
44bc0ac4cb2cfee89f789160bad7a999
2ab2936722342f802f7a51d14e8d80589bb7df57
31679 F20101219_AACAOJ woodbury_m_Page_001.jp2
06205a75f7c4319460d0cacec2be5a2c
40df3299a54b506d6b1b916ba15a445b514bda3a
49744 F20101219_AACBLD woodbury_m_Page_179.pro
afefb149eb2aa3785db1b11b8bcd8fc1
0ab8aa08fb6fac6007dab451eb4b5d0b0cf878f9
34514 F20101219_AACBYO woodbury_m_Page_046.QC.jpg
331aa0ad73ec61ef3f28156d318548fd
73b2db660b4336b8176537b6922f7e20f7fcda2d
6344 F20101219_AACAOK woodbury_m_Page_002.jp2
73628320cea9fa2af1a82b558bc8f838
0c419b5f568ff89d59cfa2553a820d1ef63fc273
33165 F20101219_AACBLE woodbury_m_Page_181.pro
368160d9708d54f560e55a01b6e3b28f
b1b94306d099ca5766b15ef3bf7b103adc5af53c
8494 F20101219_AACBYP woodbury_m_Page_046thm.jpg
e509c5571951b34178b93d36950ea8f1
255c5703b9fbac60e1a64b6e8b040b9ca2dc5872
6447 F20101219_AACAOL woodbury_m_Page_003.jp2
dd9404b9196191af02c75f81e167ef81
a0a481e051441668f1e87b6b9ab7859bded6e531
33581 F20101219_AACBLF woodbury_m_Page_182.pro
0653c20ea9f227835742940e218d5860
27f5d21222a6a29376dbb883246d41d2c7e1b6cd
7838 F20101219_AACBYQ woodbury_m_Page_047thm.jpg
4b661bd70d67e94eb2bb19676c47734e
82a9968776c840c893d9d93d6f180739cf4847b2
95152 F20101219_AACAOM woodbury_m_Page_004.jp2
b52a87ba7f525f34b9b3bb1bfde83d4a
8c546379af7a1a53b49e138ee76be9c7ecda3b25
33528 F20101219_AACBLG woodbury_m_Page_183.pro
a88a82ae49a93200922646ad3c2ee793
b8d18927251a4a61dc0091359b637f9c45db14ee
7761 F20101219_AACBYR woodbury_m_Page_048thm.jpg
049b8d8e9de15bf4c9c7ed5e97d463ba
bbdf690fb9cfcf299eefc49992bff762dab81425
107162 F20101219_AACAON woodbury_m_Page_005.jp2
ed3123e38448a57d979dd2a01718a69b
6b470980167dd8c0e7d10d1d5134277d3ffeb15f
48443 F20101219_AACBLH woodbury_m_Page_184.pro
1cf4bfe3351d279b509142ca57a94c1a
a9f101ad7671d6ccd9810f1dfb0dcaaae6460b7b
20628 F20101219_AACCIA woodbury_m_Page_183.QC.jpg
f18e855615ac7ac6c0ff6d0d694f4836
dd0afa7d73993af73ce359ffd059d2243f975850
33349 F20101219_AACBYS woodbury_m_Page_049.QC.jpg
c0c18bc5ee6a08ba240bb349e61b30fb
c404b366a012d8315080ed4e28da755bd4536fab
1051982 F20101219_AACAOO woodbury_m_Page_006.jp2
f3a0b7cccbb452bec72ed11603b485fc
bd5fee59f61e06e2dce3567cb1a5cb938a6194ff
15388 F20101219_AACBLI woodbury_m_Page_185.pro
123a42672c03e72993bbcf4ebd88fd3c
100b77a605dc2ce92b72b1f98988da3a4192f1bb
4794 F20101219_AACCIB woodbury_m_Page_183thm.jpg
2a8c8aa2081b3364a62d015950569d84
480fa42265e5584f5a0cdc4f48dd33c77dc1f295
8347 F20101219_AACBYT woodbury_m_Page_049thm.jpg
69476c9073e7b87b7b9f0b7f01778f61
d61bc6e5fbabed1e1b7beb97d5b9aeece4947aa4
F20101219_AACAOP woodbury_m_Page_007.jp2
08fb89f2407f8f872d5636c66331b8c6
0f875176eb632a20ac88f58a662dcc7874903a28
49185 F20101219_AACBLJ woodbury_m_Page_186.pro
8a2d3e5b2b0c3c795cafc12e248f9b38
433ad4cd135ee0d8919cae61e770f71190fc67cd
26994 F20101219_AACCIC woodbury_m_Page_184.QC.jpg
12669d69fa01e2d08fd4223ac508e25c
cccd7a3ebb38069c541882c75eb4f8f3ae3fc90b
34841 F20101219_AACBYU woodbury_m_Page_050.QC.jpg
a4f90c729b68f945c580c68fac99aba8
a01be5ca364687496dcc72b433d5dd001e0e7c90
1051977 F20101219_AACAOQ woodbury_m_Page_008.jp2
0dfd2c469edfa01c34ae4d00e8ea402f
d67060e9087c18ebfa9ce5f9337920005db67b07
18379 F20101219_AACBLK woodbury_m_Page_187.pro
ff25f7513f4ce7e8976bca33281ea4e1
ac49fd7b30d6607ede22dd5b94e2de2bd8f5806c
6042 F20101219_AACCID woodbury_m_Page_184thm.jpg
75f4b1a52bb50b1b339c13befb81edc3
cde5d87bb630208d2ae3382ac1b7baa845b22831
8491 F20101219_AACBYV woodbury_m_Page_050thm.jpg
7bd8f2e0f5b8262392801ad21117ed57
65927de464c400677c1f36d7a0ee82e2d8e7cbc9
1051972 F20101219_AACAOR woodbury_m_Page_009.jp2
69e18b7b33e2a3764fe2a7ea61804069
013fb5c2c86f039d9ff7c416b007b63cd99633e6
10612 F20101219_AACBLL woodbury_m_Page_188.pro
ea940af03aacffa1db4306b80fb28153
64b63817a8e5c7874f2758855de083129df230f8
20458 F20101219_AACCIE woodbury_m_Page_185.QC.jpg
0cda151c5d36282ebf134703e656c9d9
478fdcab3928c120733ba485aa52f494f5a42021
8113 F20101219_AACBYW woodbury_m_Page_051thm.jpg
b8f311c261749e0ae31e832008789a95
08dbcdb6845a1c29ae7e7319afc9e2ad5a7d733f
1051968 F20101219_AACAOS woodbury_m_Page_010.jp2
6da2bf30d081b859e7fbc3faf7a0eb56
cbdf7f731250df97cfe546716765628ae2c66702
45764 F20101219_AACBLM woodbury_m_Page_189.pro
dbd553c8c5ec298d32b4bda0df320568
4ecb79c69619be2ce20c5926f964ae3b5cbdd748
5761 F20101219_AACCIF woodbury_m_Page_185thm.jpg
ab90ef26ae170cafc86454cf6d19cb2f
f25f9c3b10d87a6b15354b530395e2a95cb59588
33690 F20101219_AACBYX woodbury_m_Page_052.QC.jpg
b8867edc78e6d3caea5634592042ee8f
0be1c58ac0f27901a7903299435ef538eba48802
1016822 F20101219_AACAOT woodbury_m_Page_011.jp2
3c246d300d86179e90b30415ad5fd766
805b7580cde6ec019560f51adef0934a8d2bb12f
49703 F20101219_AACBLN woodbury_m_Page_190.pro
5507bc0fc231385df28fbbff16aab9b9
a1358914379e4c622ef9d392892581927cf998e5
26788 F20101219_AACCIG woodbury_m_Page_186.QC.jpg
bb3515ce0255bc38a075b5a1dde9edab
c687051f8e0c60cad9162b1a4109ebdc5ac7c9f0
8264 F20101219_AACBYY woodbury_m_Page_052thm.jpg
6839ac448f88c8334f1c6706e1a7b485
d92ddd690d6eb61897b9814f5a9e1426fe301285
88723 F20101219_AACAOU woodbury_m_Page_013.jp2
6cb44858391d647d80553b627599da71
48e84bbeb04f93768ab8f60bd9f990b02096a441
50592 F20101219_AACBLO woodbury_m_Page_191.pro
d45935034e8049758ece3487834d92a4
4213d3152162b5456c926961a7ab508a28d639a1
6020 F20101219_AACCIH woodbury_m_Page_186thm.jpg
74294f0f7e54425bba132c3a97bc2735
2e07b992cf1779db859077943a6f3614161e2e13
33161 F20101219_AACBYZ woodbury_m_Page_053.QC.jpg
4d18d19275fd963b18965644cad7d757
9436af4c7848ceac73122c6277fbb1102fd6262b
78552 F20101219_AACAOV woodbury_m_Page_014.jp2
c5463c56651f544a2158ee58325ba3d8
c5e4636a1f5fe26e30780fa8073f816b6aba8320
31212 F20101219_AACBLP woodbury_m_Page_192.pro
78944c1740cf4872b25e4d12053aa685
6baf1f154ec568626543e0202cb23ed166c4597d
19376 F20101219_AACCII woodbury_m_Page_187.QC.jpg
ed113485fbcd196068074ec50bcf690c
ec2c11c86e12bd33309fa922773c64fdbef342ba
96742 F20101219_AACAOW woodbury_m_Page_015.jp2
8e8875a2ede44ac02982e9f7064a2651
4019c8f7cfc9e11f7946dba63c3892d7da307eea
23877 F20101219_AACBLQ woodbury_m_Page_193.pro
bbcd11213580b5b9585019353ee30f8a
faecec9a54a00ed9e42ab226c88d73bc7188543e
5546 F20101219_AACCIJ woodbury_m_Page_187thm.jpg
49019b6037aad4957a4523a5f2e957a4
d9ee892578edc93a07e4dda26afa2f166a8ac147
105443 F20101219_AACAOX woodbury_m_Page_016.jp2
d96d320afe3296325dcc1d9abf5759c9
984106376be8b3ff90735177348812b5f0243fab
25531 F20101219_AACBLR woodbury_m_Page_194.pro
19289f4d022dc977f80d71a2a230901f
806fb0d67a3bc3da8d29f3cdd401093d9566b057
14891 F20101219_AACCIK woodbury_m_Page_188.QC.jpg
6a5acf3d01b8d56a257fddbe64a9fd35
ed73fbfb22825de286bab4e84eebc6ed2f5b2436
106846 F20101219_AACAOY woodbury_m_Page_017.jp2
1d24987b4f7447128aa0fb0b2cd308bc
7dec9f414f47d8092cbdad9c9f97dbf737dbd430
26349 F20101219_AACBLS woodbury_m_Page_195.pro
f9a731c070274087e6bd2af763d03cef
ce8c508a0b0a37fdd0a02ca53fc00432bd9dad88
5017 F20101219_AACCIL woodbury_m_Page_188thm.jpg
53fd5a08357a85d677d38aba07875558
5288adb0d776107260f039c6ba5e56cffebd25df
107599 F20101219_AACAOZ woodbury_m_Page_018.jp2
a9172085604576669babc32744591f61
06bfd4efae2bf3114ddd3459d2112f526511e21a
28731 F20101219_AACBLT woodbury_m_Page_196.pro
dace657a70a7417954ce132503858326
073fbc8bbabe32f06455b844cb40b9b3534d6be6
30522 F20101219_AACCIM woodbury_m_Page_189.QC.jpg
4af2cb9947d5bbba33e59cd548dff664
e1c4ac5ea835431e38d0bcac025d77b2a141192c
25696 F20101219_AACBLU woodbury_m_Page_197.pro
8f2a238c074ea922c54a0a4e6ff1bb39
25dd80d9066be5c5feb9ae0cfd77690472d77cca
7461 F20101219_AACCIN woodbury_m_Page_189thm.jpg
373e4ba6d0abea047a08e4f2a0021307
fea325ba74846018fc8ad6890010e3aa3d62bbd7
27693 F20101219_AACBLV woodbury_m_Page_198.pro
9a0ecb6a59193b5a3887a23ceb14348f
e48878cf0f7439463b63cf78090eea2f8bbff402
32653 F20101219_AACCIO woodbury_m_Page_190.QC.jpg
ac700c3112947da3a646fe55261e37e7
f357aa7593d6119252c4856848dc591b39d3eac2
34965 F20101219_AACBLW woodbury_m_Page_199.pro
a81224951e0189255d1471636e68cde2
cb07dd5f0c4745d8a5c06cc197cd3832dd273396
33277 F20101219_AACCIP woodbury_m_Page_191.QC.jpg
2ad03e7d480f75f480666c1bbcb379bc
0db48bfedd4b0f6f595f68f2450ca7426e90b038
F20101219_AACBLX woodbury_m_Page_200.pro
7488c6675e89e954996f6109254a0a5c
380466ee9eab09e9db4a08e23bf91c7a7347ed78
8407 F20101219_AACCIQ woodbury_m_Page_191thm.jpg
88f71e54a95ff674e43de671e3c35695
509fb35ccedf238724693f82d2c7e243d102787c
13585 F20101219_AACBLY woodbury_m_Page_201.pro
61d450e6f046bbccb2523fb2a8805144
2a58cd639daaf419c591228aa14bbf4a2df91ec9
21106 F20101219_AACCIR woodbury_m_Page_192.QC.jpg
3b05ce269381a3a586506b86494f8f0f
dde66b8e409cc97e86c843b537d28b687dc8249e
5192 F20101219_AACCIS woodbury_m_Page_192thm.jpg
dc27604843b3ba5d2b053d670bc8f154
75cfd82ab9d2833d5b62ad5fe0c65ae507d22067
28574 F20101219_AACBLZ woodbury_m_Page_203.pro
44974a77c63b3e998bc44d03810123ec
ba3b968b61e0aeec2712f58a1324f6aef792235b
16177 F20101219_AACCIT woodbury_m_Page_193.QC.jpg
119832cde3066b8e589432b7da0d67f4
bda3d6c68e51eec22bed947fc3b956c63046eda8
3963 F20101219_AACCIU woodbury_m_Page_193thm.jpg
3c7a8d606316f20a2bde38119847e698
f9f6d50493eadf4a24d9d30a2cf192fe227200f2
25121 F20101219_AACABW woodbury_m_Page_114.pro
6b5eaf62b991e27902c9513b19f69c03
6286dcad7184b07dbcf2fff4b89024f64cd0b915
671771 F20101219_AACAUA woodbury_m_Page_159.jp2
7fee296311eee5f3cb9610d4a869a8e8
ac26ee04c0da6169050362bf02609fe216020bb4
F20101219_AACABX woodbury_m_Page_112thm.jpg
1c8b545d027f86aa302daab2856964e4
b06b8f33e04b7b6ef8209563da687e6f43c91eec
888573 F20101219_AACAUB woodbury_m_Page_160.jp2
e5e2f7ff2805e9776b0184142e20efa3
129c0734422aff625fcda55b31997e8d7ba01489
18058 F20101219_AACCIV woodbury_m_Page_194.QC.jpg
eac825167adad5e53202ac9e4c4cb72b
56568a67178a9373291d4611d5e36d77777a7999
108498 F20101219_AACABY woodbury_m_Page_041.jpg
2e822bc85bee3f601d080fa0f6b16b10
487dbc117895808b912d33db41ae6d9952330307
436706 F20101219_AACAUC woodbury_m_Page_161.jp2
2a65726eefc9a5c9c751356aa6d3101a
1285cd5c6d28a035d65febae80a78242afa24509
17591 F20101219_AACCIW woodbury_m_Page_195.QC.jpg
b55fada03eb4301afe7eb8af4bbd20fd
e3d20668f0965d53f7687af7b16f2d40823612ac
46956 F20101219_AACABZ woodbury_m_Page_121.pro
e566228ea5cc25699b4e21805ea2a62c
5d89323bba4ac5c111e098915799557a7180512c
4408 F20101219_AACCIX woodbury_m_Page_195thm.jpg
78e1777a195da3d7cb496a140e3c0c77
66fec7bc0dd2b377cc3f6767d99de7bd46d4324e
1051958 F20101219_AACAUD woodbury_m_Page_162.jp2
4b40d498f0b00a4058263229e343a982
1afb418d202a90133e71a8345b9ee2f1cd94fc5c
18535 F20101219_AACCIY woodbury_m_Page_196.QC.jpg
af34b93d5272ad90bf4e743078547dc4
fc0a95864e4b6d22567bf650caf7f832f039b7ec
1027779 F20101219_AACAUE woodbury_m_Page_163.jp2
94e4630197a7eb77ad2092f440a81263
880966014b61951091d0f04537e582b74d342f37
4480 F20101219_AACCIZ woodbury_m_Page_196thm.jpg
18d7d48aa5d4fda7353cb8f355111374
26d135952e73e1e33bf74a47da2673615b36c669
637642 F20101219_AACAUF woodbury_m_Page_164.jp2
b5b28b4a336cea7b3db4a15e1437a9f9
774a352bc0a6646bea251169aa9f9e1fa52b1910
236400 F20101219_AACAUG woodbury_m_Page_165.jp2
2a0234146456471ddda1f6eb32885f48
c709ffd0c7a3977abfb3cf7f96f988c35e8c7425
484019 F20101219_AACAUH woodbury_m_Page_166.jp2
6bf577ac27846fc5982aff9b67abe3be
e9dbe4a58776aaff1999c5c4e50a743328fa0de6
2233 F20101219_AACBRA woodbury_m_Page_116.txt
2688557b7b53a778e377627d73e2f8f5
7b6975d6be556fdbe3003d092f83df8459f80d17
343886 F20101219_AACAUI woodbury_m_Page_167.jp2
fe6bacebc40eec414ba329dcb9ffb130
6fd27a1c4bd4f8d29f439312bb14541390e6a7b3
F20101219_AACBRB woodbury_m_Page_117.txt
3ccaf1d0b22d691fd75932948c926b90
669e5d9c6ef788af8f6d5b599a911873057316ef
101495 F20101219_AACAUJ woodbury_m_Page_168.jp2
a36e67a563cd2e470b8d52745ee5c149
e2b8de6de500985ce6041337bbfdd08797c3dd13
1092 F20101219_AACBRC woodbury_m_Page_118.txt
a207d019d3783828c9151669becd3461
2d045709a20af302c893740dd9b95676e1ea1d5f
107175 F20101219_AACAUK woodbury_m_Page_169.jp2
362c585687b1a64a9c5e2e0a1c4301cf
371a10a39454c1ec072eafa2a916535de23b6eef
F20101219_AACBRD woodbury_m_Page_119.txt
871dd3bd203b8d490fb7eefb9b8abc29
d4b7a2d637a679aabdb52bd3a5a864551b2423be
106059 F20101219_AACAHA woodbury_m_Page_023.jpg
d26ebb8d9e6a941ebbd1325712a4d708
6c00a7255f155c0f79f67f51e053431516b97dc6
89696 F20101219_AACAUL woodbury_m_Page_170.jp2
1bc069e93f365e9aea31ba3c39b1b810
b4bc961491dbd773f4673feae6c6d4b549ed7b1a
F20101219_AACBRE woodbury_m_Page_120.txt
338aa509dae5b833ba82a414e234d9c4
4c93c256ab211d065a0cdabc33777b14b7494040
108890 F20101219_AACAHB woodbury_m_Page_024.jpg
3c25eae78ed0c808bed93e8f017130f5
8a2d670f654e70395953a443d33046129e16fcf2
97578 F20101219_AACAUM woodbury_m_Page_171.jp2
ece34c0833b5c8c87c23d8d2318107da
a24943eef7d6f9d9710bda2f5adc3534e477c648
1856 F20101219_AACBRF woodbury_m_Page_121.txt
3813d0433515ef6413086869ee6ba10d
2a0623c329a7d79ce7ea5abe9c127e5847000aec
104801 F20101219_AACAHC woodbury_m_Page_025.jpg
8e8d5cc322f2a4130e83234a5e6f80b7
81462c0bfe1013cf0b586162773b5c4259ca3824
96498 F20101219_AACAUN woodbury_m_Page_172.jp2
b8695ab8ac06d8a87187f067a27b1115
2021dfca7e38c56aef6d7180f1a41e9f3c11a3f8
F20101219_AACBRG woodbury_m_Page_122.txt
58eebc12209dc8c6736196b9badc3932
d58dd713faba75917e46c1625785581bff48875d
104091 F20101219_AACAHD woodbury_m_Page_026.jpg
fc6839e9310a2971a549755d81f15632
ad58211fd3016d4dc895e8bb2aa174bf13dd578e
97372 F20101219_AACAUO woodbury_m_Page_173.jp2
f30dedc8e725ba60b904ff2fbc795257
0d713a7b6bfe0f1bcdd3953d9e350dcc4163b477
1683 F20101219_AACBRH woodbury_m_Page_123.txt
f11e4c4c7250d62268c4a7583aa0fb82
17cd2a16401d46015fbfa68a1b705c113c8a0575
103499 F20101219_AACAHE woodbury_m_Page_027.jpg
4dd93a562a982c0ca9f62631e2844bb4
7c458000a4fd628a4972d85156122a3babd4807c
105587 F20101219_AACAUP woodbury_m_Page_174.jp2
ee4176f49a02ee6070c1c00eb93d8bd9
3eb62bc0adc82b1f9dec5a6811553a2ea2a0b658
1904 F20101219_AACBRI woodbury_m_Page_124.txt
c8502c8e8c532f2cf9d49997915da46a
bcdd7d39f5d4a907828b8af3df3791202c3a2eb6
106027 F20101219_AACAHF woodbury_m_Page_028.jpg
dc14641e0f28cc192c3282777db4bb05
628624dc08af5042f3e06761f921d845da92fb3e
90145 F20101219_AACAUQ woodbury_m_Page_175.jp2
d5a6edc1bde37034ecadb727903c66ca
9f4ba8a11e8645e5299b89c4f14ef69e42d7c66c
1815 F20101219_AACBRJ woodbury_m_Page_125.txt
c055cf29d4aa116e1dd06675fce43c12
0f601cbbb97ad78d623259961bc0028638c0afc3
105578 F20101219_AACAHG woodbury_m_Page_029.jpg
6dd50fef33c060026c028bdb74a9e0e3
c89a217dc135d95d47e0844d308133b62527d554
97871 F20101219_AACAUR woodbury_m_Page_176.jp2
5a9b2e2367783d138d7f21519a6edbc0
4e9300a08664b2585c4ae7877e61c982a668440d
2356 F20101219_AACBRK woodbury_m_Page_126.txt
fd37d20234f3de0ab28d28bc405c63a4
442ffb4f16c7b059558dd50b34e9723be9ad2253
98964 F20101219_AACAHH woodbury_m_Page_030.jpg
03f09c7df121678e7f57c40e8b0e016b
b11e531fb5aa6110fc4be253e6dc03981ed68747
F20101219_AACBEA woodbury_m_Page_208.tif
1aa109991fcb60635388204916173652
9df7d36e2de9fff3338c8008165f6ea8e074942c
107366 F20101219_AACAUS woodbury_m_Page_177.jp2
9113d99b34bbd45f225ca8281251a7ad
9ea8120d97dda0ced7cc5f0827e2e4b53aad34e1
F20101219_AACBRL woodbury_m_Page_127.txt
bca88e21c573b9aa143ba403f960fce1
2a92e12f4bcf6de352b8d0064d48bc7a384825bf
105269 F20101219_AACAHI woodbury_m_Page_031.jpg
02116f387199f069f1c333fb3d23043c
44d29f3d9a26d3d8d05e0049ed93c01aeb44232b
F20101219_AACBEB woodbury_m_Page_209.tif
490b8bfd908765ff7e0a842975220b41
fa1280296d05a24d3e0891e5a86beb7507bb1273
108222 F20101219_AACAUT woodbury_m_Page_178.jp2
f57353608942c302316de555a173bd87
6da3e0497ff7ebfa5cd56de8a0580f854fe8d05d
2073 F20101219_AACBRM woodbury_m_Page_128.txt
048e0cd66ccf5ca5490265b8602e880f
50f801da33ddbaa376fc8b1476821c8e5ae014ac
99884 F20101219_AACAHJ woodbury_m_Page_032.jpg
cfbfae2754753d3be77bf6d342504b97
55afef74b31e1ccf939817c49b4ee4d9d96e4b8b
F20101219_AACBEC woodbury_m_Page_210.tif
35fc4e4c3f5fb228b971cb7a2c4e1638
046e16d84febe84600a61dfe3f49cfa194a01099
106979 F20101219_AACAUU woodbury_m_Page_179.jp2
eb3c04e32111bc8f88b31f91bf9d4008
e04c94bb14af74ecbc1f7209a4aa67540e19d441
F20101219_AACBRN woodbury_m_Page_129.txt
8ed18717d604173c9b8f8942fd4805a5
77b961f51b9c97bb00712b94da1671cb6475dd11
99398 F20101219_AACAHK woodbury_m_Page_033.jpg
8144cd38a3c0af520933ac06d0337b80
c5153a6c82389450a558ba3c3ed13a8b4fa74781
F20101219_AACBED woodbury_m_Page_212.tif
aeb35964681cff6388bd0c540f7d493e
db5e1a6d035baa330ac6d6970a1b2ee5815161f2
110096 F20101219_AACAUV woodbury_m_Page_180.jp2
725e6c9ac07beaae715ad81a085fd968
a2ecbdf247695b930ead0fa0dc5a1bcd4e33e689
1997 F20101219_AACBRO woodbury_m_Page_130.txt
6bc88bd7ceea5c6df2fee196a6ed7051
5793f58149d35f692a6a2603366b0b0b5c32dcdb
102291 F20101219_AACAHL woodbury_m_Page_034.jpg
f3510ddff4407074ead0c6d6fb1291c3
ee8843b7d7ffe1d44fd9e35e93599836032f638b
F20101219_AACBEE woodbury_m_Page_213.tif
1fbfd658b7239516eb57c651c19c13ac
3677e7e8539c3117e40e42f270144c9c69e935a5
74838 F20101219_AACAUW woodbury_m_Page_181.jp2
f0f0498c1acccdd9976475674166fd54
7ef9dbc1947076bcc5b7222be65caf95a140e93d
1922 F20101219_AACBRP woodbury_m_Page_131.txt
153c31c62551f10e0690394410b32047
73b4fd7ff141e083716de98d3dc44142fd499760
90459 F20101219_AACAHM woodbury_m_Page_035.jpg
298dc9f8d59ee16a92a79fb9cd2aaa8d
1dcd66bc4ba9abe29c8c6d94219cd72e244927ca
F20101219_AACBEF woodbury_m_Page_215.tif
49d038cac0008ce11bf061460e584f49
cef3b73b1be8ab75a341e139efcd558edeca1bb4
790081 F20101219_AACAUX woodbury_m_Page_182.jp2
2849d5fd00aff6ac0951342518bae49b
eacfce720be068a4125928cd502b074927fd7112
1888 F20101219_AACBRQ woodbury_m_Page_132.txt
b2275273092595baf5bfe3160cbe283c
87a1baf45383653ec45cde466ba149ad887b7ca3
100926 F20101219_AACAHN woodbury_m_Page_036.jpg
e397994ed8fb01ab5151455c76c8846b
aec47ec4bc31a747c9f1974d0c3bcd596132f89f
F20101219_AACBEG woodbury_m_Page_216.tif
0591bdd69c0168ffafe6be2a55296444
1c9b0a5776fe76a3695451fcac5c05fae97c283f
722482 F20101219_AACAUY woodbury_m_Page_183.jp2
3aaa86e465eda3ded862e8a30454f002
22c05324b93dee66a8211d227173a75a187370c8
1967 F20101219_AACBRR woodbury_m_Page_133.txt
c33fbc73720ac37b1af8768e7b24b30a
40187e87f86a3584dcd91a9232a5277a2553db68
98444 F20101219_AACAHO woodbury_m_Page_037.jpg
834987e6c441dc45021ba0d9b1065e7e
f5be8d42eb43f5292fde25817dc1a299654153e9
F20101219_AACBEH woodbury_m_Page_217.tif
55cbf1f1c00ce621a90ddafa81f09535
71de8064e8a53f234d3c8fd1982a032d06d6d99a
984961 F20101219_AACAUZ woodbury_m_Page_184.jp2
af773140e33904c12f47fc64fad3edbb
c63d7d58ecf47388ea396747937092e7b7f1e20e
30007 F20101219_AACCBA woodbury_m_Page_081.QC.jpg
c66867701ab0501e8e51b8cd2bcf4ad7
230e97c4925bfc764c7673e95c62426446907d52
F20101219_AACBRS woodbury_m_Page_134.txt
33eda2ee1018ed0bba0e12c0e4d5e921
5e96c29c4b0a2ecac5f2810284328c57522e9934
106884 F20101219_AACAHP woodbury_m_Page_038.jpg
42d7b38d4392c03d0a57ba920f767a5d
af907d3395a5c20608603ac3a682f186d778ab74
F20101219_AACBEI woodbury_m_Page_218.tif
e371454e7461f2a49ff1980617461af7
e9da71ca9295ae26bb8e1b0155d3457567f00cc2
7539 F20101219_AACCBB woodbury_m_Page_081thm.jpg
8c60fa35092e93b938f811e6a5ade202
13f29bb417ebf377a41ad748c8ba406c4080d1b5
2080 F20101219_AACBRT woodbury_m_Page_135.txt
3cfc553441c86d408feebe4cc214f47d
eeba14f7dc93a7469701eb8045834518d46fd5da
102456 F20101219_AACAHQ woodbury_m_Page_039.jpg
c50020ef4ecf4c9199f5ece5579de67a
fe8c5819f515843340f7c6b23ed56a8634b7386f
F20101219_AACBEJ woodbury_m_Page_219.tif
0040517c625f9e0d298d5362aa0dcad2
77dc456d97928ca5f6c1f148422a94e464932702
33841 F20101219_AACCBC woodbury_m_Page_082.QC.jpg
d9f2b20599c2053b09edcb7c7439c7e0
eb712fd1bca001bf8df47de779f729c7f1371f06
2118 F20101219_AACBRU woodbury_m_Page_136.txt
0f36e9eb9391e4bfa51e5878029bec77
4b7a4a20f823c43e7098793e5d09212279a80974
101517 F20101219_AACAHR woodbury_m_Page_040.jpg
82786e6dfce05e0ea2b475139d28a3a5
c1f6156f4a6ef96872b3c2ae4719998b6f4e5a6f
F20101219_AACBEK woodbury_m_Page_220.tif
7ab736f8c3d63bec53271d8f0488f421
9583c3d744c63134169b210d47916a80d0402c98
F20101219_AACCBD woodbury_m_Page_082thm.jpg
2d86565a2c8434bcb7f5593faa8d33e7
9ff0e3bf3f34e5957ee1c788a58c572554ea9fcc
2041 F20101219_AACBRV woodbury_m_Page_137.txt
78acae30a485d3b06384dac7d8432b57
3d2d91d8fbc5faf8cdcaae7d20194418a9b78df3
103742 F20101219_AACAHS woodbury_m_Page_042.jpg
c6e92ac8a33667a92b74eae0ac1fc7dd
ab6ca6d435a5fe7072fc0d215f85e6b0f0f873cb
F20101219_AACBEL woodbury_m_Page_221.tif
87121f152e5d9459bd4c74b057ef26df
f3db82b97bf3d471c0a1094276bedb8f4088caea
34283 F20101219_AACCBE woodbury_m_Page_083.QC.jpg
726a9544e8c9fc0ccf594e731f6844f3
8a795b57d6f690745ff813577d16fa5e9c2397f0
1898 F20101219_AACBRW woodbury_m_Page_138.txt
8b6630318bfd6ae3a04f2ee95996008d
b4f4928d089332f1fd2669e948fbee69888bc3af
89011 F20101219_AACAHT woodbury_m_Page_043.jpg
aa0efdbbedbe8bf1ec78e2a72d0bcec4
f1753e44ea70946de900c0deecbfc77cae5a01d9
F20101219_AACBEM woodbury_m_Page_223.tif
f6e0359cb47969e74cb3c2e402899aca
4cf6eb1d889bc31a648b247d29958fafc7e57cdb
8537 F20101219_AACCBF woodbury_m_Page_083thm.jpg
48425e8e50a1280d8fbb8a6d5825455b
cca000d682d51ef7811d8ac5078264d1c80be74f
1869 F20101219_AACBRX woodbury_m_Page_139.txt
9f8f22466c11909c1e84b9bd49ed78ef
41508e8bf95c806fa0be723e506a152c1d9f9299
92294 F20101219_AACAHU woodbury_m_Page_044.jpg
cd41159c778cddd61201b08be45c7f78
d05b6be18293fd3cc089a2dab6ed76e32fabbd8a
F20101219_AACBEN woodbury_m_Page_224.tif
88ce4d86c0c481757cfd1de721a227cb
ef2c1d6e5bc7b852c8a9d07a155ede841ef9389a
33526 F20101219_AACCBG woodbury_m_Page_084.QC.jpg
9e92995861a66cd016ba942af58a440d
b0a31b856b515df61a9786789d2b0bb20b0d250e
F20101219_AACBRY woodbury_m_Page_140.txt
19a46abd93df4174d2d7bf546dd2cd85
3e7fc71a60b46bcfbc2cd07ce4cc7a44f9ca9438
96698 F20101219_AACAHV woodbury_m_Page_045.jpg
9dd9c8aea77ee79fb4f2916d24b19691
1562f19c0fc919e472ddd00f548a622ef214cffa
F20101219_AACBEO woodbury_m_Page_225.tif
a38091e6210b3661d9c00e5b24a1a136
eaecdce4673069060f23a1585acca0cea5b6bb33
33597 F20101219_AACCBH woodbury_m_Page_085.QC.jpg
907f25c0b1675ab857e4147871720407
66a6fcce14a991bce70e45438aecac781b6bd9da
1874 F20101219_AACBRZ woodbury_m_Page_141.txt
e6d2c81c92451ed3ad31cfaa44214762
e363d7cc492f95676689c6fcdfd743c6b4ef7bc5
F20101219_AACBEP woodbury_m_Page_226.tif
620746f98087bc604d0e4dffbf8315e9
d6e2c3318c2fd803c9a7f5d505d2041d2be52f2c
8411 F20101219_AACCBI woodbury_m_Page_085thm.jpg
fe9053d60d5b43c38a8be83644fa0ad8
a63b958cfacb1ff4ba6710ff5fa5ffb6b45bea97
105814 F20101219_AACAHW woodbury_m_Page_046.jpg
8c41796f14456f78a60b077b3b15f7fe
42b7e269ebd524cdb8c06c308d69d7b8217f5e5e
F20101219_AACBEQ woodbury_m_Page_227.tif
4c8931ab7cf58ddd164e5aadee4bc092
8b8ecf401de5ed016e42863367849092f3124615
F20101219_AACCBJ woodbury_m_Page_086.QC.jpg
d065b55e9f033d4da536e2b6a7ad2800
5dacee6980a38a6fcde756b32d8ede91453be713
98711 F20101219_AACAHX woodbury_m_Page_047.jpg
a5fb3396ae57495b73f962acdd4bfa9a
dcacce0a56a44ac9f28c3fe05547769c67199b7c
10807 F20101219_AACBER woodbury_m_Page_001.pro
a3553475c8d445a654e0a43a04d60865
144ed08124ae0a04ff74513f348b0fa66503fac4
8177 F20101219_AACCBK woodbury_m_Page_086thm.jpg
903ddf280d9b730fedcbe54e230abd34
c58dac2bcffbbab255ab46423845f9c8b0ebecfd
98135 F20101219_AACAHY woodbury_m_Page_048.jpg
f62694ddebb654a50cc034051dc44171
9d01e60f91d9c1a9cfef6a553a08560901c5f405
32064 F20101219_AACCBL woodbury_m_Page_087.QC.jpg
5f886d005146b8b5a193da67a3fc0336
399d2bf6cb84484cf606d44c169ad5b6379c0330
103714 F20101219_AACAHZ woodbury_m_Page_049.jpg
df3c27ef2cacd480bc26cf72bd0aad70
1bb67dfbeeff06efbab6b470d5ad8abc2cc4a706
1382 F20101219_AACBES woodbury_m_Page_002.pro
be6a01a455970b65f89b23e9c866e861
1c2c9b2cfd66dbcd0cbb7a168acdca11d754123b
8062 F20101219_AACCBM woodbury_m_Page_087thm.jpg
365d911547f8ff1f468b49f8fa253d67
7f0ccddaf3fefc5e6aad28bc74954d22f4a21437
1508 F20101219_AACBET woodbury_m_Page_003.pro
e9d8a0939ace242a986421428ce7ac4c
681a7e61ad13ecbecc8296245aac3c2502942554
23250 F20101219_AACCBN woodbury_m_Page_088.QC.jpg
77922c83196fb65ca956e3021c718d67
c997a1ecfeb598e75ca16efa3ff5d27c9ebb5819
43252 F20101219_AACBEU woodbury_m_Page_004.pro
250720568412b5f8777993845520b332
2ebf481d370ff96991c841ec1f8b5fa3f9ff52bd
49215 F20101219_AACBEV woodbury_m_Page_005.pro
6a181ee6cad776998759da62104c2d11
08adf53ca5d1120d2bbaa1879147c7cf7bfb1adc
6337 F20101219_AACCBO woodbury_m_Page_088thm.jpg
caf5daa0b529a908983a44db7a633d16
2b09c5f4ca95edaea53af2beb290b6e5db5bf1f3
63209 F20101219_AACBEW woodbury_m_Page_006.pro
43ef0c0f6c2650bdbbab632e54a8543f
43f00884e2805631ed74768ad4ad927e839c7ae4
34844 F20101219_AACBXA woodbury_m_Page_025.QC.jpg
595e47a42360f886db05e03737d0625a
5898dc7c01150a3f9f415c9b67e2d74b7b4ad1ca
17657 F20101219_AACCBP woodbury_m_Page_089.QC.jpg
edfa162a5af0d03d27cda236bd5078d0
e1e436a86258757942b207f1e4522963a0dcf757
82793 F20101219_AACBEX woodbury_m_Page_007.pro
a2a7f3f0f83adf989e12c64f0a9bd34d
5bc14f4013e7d98ed89b758d8a044020814ecea6
F20101219_AACBXB woodbury_m_Page_025thm.jpg
f786c38ccfb0fc2f6cb6507cfa1636f3
a34c679263a122de2572b93cd29bc5c6bf2ac54f
4707 F20101219_AACCBQ woodbury_m_Page_089thm.jpg
657164c451aea450e5a6132b81996b1d
375784221da5fccb6bb0270f9bb1d6a0813acbcc
91036 F20101219_AACBEY woodbury_m_Page_008.pro
66eb5638fb88fdfdfb9a2fa71c12d87d
94c7e4e50af91d58e91bd68b26aec76d3ae8e414
34122 F20101219_AACBXC woodbury_m_Page_026.QC.jpg
9c02cf881a6ee13e3c048a759081e148
da682935b29a31e3df25eee62bdbde742f2e7a5c
4743 F20101219_AACCBR woodbury_m_Page_090.QC.jpg
4cf26518f1ae8124c68181ae7a72a0b2
de67f4bff307bf5a4912b563ea40095615851f90
65565 F20101219_AACBEZ woodbury_m_Page_009.pro
5f87bf4ac07b5d6279dc597d44c793fa
2a9ed28b75861a3c2b0fa4a39b85cd27a21f5d5e
8276 F20101219_AACBXD woodbury_m_Page_026thm.jpg
8d9da1877f873afd52539f40955f479b
5077c07d22453cf801ef74373c781d868bfd3c70
1704 F20101219_AACCBS woodbury_m_Page_090thm.jpg
37dae1dd08ebdf75599e4a01c7034f86
80a53665802afa88b559af88edcb3e36fcd9294b
33084 F20101219_AACBXE woodbury_m_Page_027.QC.jpg
77671f4ee087bd5c6c59e5c54853ce4a
0dd0cedcf373cf9a292813979fca26af5fbada26
22939 F20101219_AACCBT woodbury_m_Page_091.QC.jpg
21d811f626de940359c31445d02eedab
ab14b119fab794631ff2aa1de6daf73f8f8cdbcd
100805 F20101219_AACANA woodbury_m_Page_190.jpg
a13cbd98a6080043db5f615b9dc4c05b
38badd597f3fc307ecab0809d64d61a22216d8ab
5906 F20101219_AACCBU woodbury_m_Page_091thm.jpg
cbc0d2e71831e5251146af49342ef553
78ad5c20533c319487dafa89a7ccdf7ade3a7704
102909 F20101219_AACANB woodbury_m_Page_191.jpg
77d0a12b18ff1d97a000f8eb49b4c399
1931c9b6a0c8d56e3c81c428e5b7ea23b5e6aaa4
8423 F20101219_AACBXF woodbury_m_Page_027thm.jpg
a605217dce5130504ddfad81e5888820
6a95526a213781c6de53297aee95967dd1141687
28907 F20101219_AACCBV woodbury_m_Page_092.QC.jpg
a75ff8856f97b87910159674ca10211a
11e0a6e496be9e65bd8e4c42063c1785de918f99
65485 F20101219_AACANC woodbury_m_Page_192.jpg
c0de1edbf9f8a660bb003b4f069ab5b6
6284cae111bd703020c60be3c241a39cb7b824ca
34325 F20101219_AACBXG woodbury_m_Page_028.QC.jpg
fb1e732f77bd2112fa080ac7cb9cef18
5748db4eb5e5a21831375b3d827d6ae8689b37e9
6388 F20101219_AACCBW woodbury_m_Page_092thm.jpg
41cd7586e674a4c9342856bc6ef62aa7
0c6ad65c59836e6e0c17c7b0237bc6c498c736bf
52842 F20101219_AACAND woodbury_m_Page_193.jpg
6b060226003acad1981a86c82fd715f2
3d2d4c7aa0fdf3380f0d196676f17c0d29f9ce23
8462 F20101219_AACBXH woodbury_m_Page_028thm.jpg
a53887444be0533f421a10f08c5464f2
694d83ebfdef8d0f5080380ce8d898f155a322b0
12768 F20101219_AACCBX woodbury_m_Page_093.QC.jpg
06668a4826bcd9a5301d7d1515cd8197
c5abce13ebaa39236a08d6e3e2a85c99a2a72606
59084 F20101219_AACANE woodbury_m_Page_194.jpg
f40d68a93dfa8d53a81d90728fc50386
cf42d02390f79788f1f226b5e5dfe79124eae9fc
34346 F20101219_AACBXI woodbury_m_Page_029.QC.jpg
e16329885bd5bbad9600806d10bc464c
91667dae6a22dcc393fc6e06f5df0a70867137e0
2936 F20101219_AACCBY woodbury_m_Page_093thm.jpg
5272221e5ac73ffee9198d42d84c3e4e
b86aedea0e2ff8f987267f573d836130ee8aa0eb
8458 F20101219_AACBXJ woodbury_m_Page_029thm.jpg
64cfbc644fa3cc763a310eb1dd4455df
3447f840eb297adc565180a8c8dd3efb72716b08
22100 F20101219_AACCBZ woodbury_m_Page_094.QC.jpg
57983bfcf056ae56a99747ec81c9d090
518201337b2d2e571ccec9ca75bfca0f11f19c70
61256 F20101219_AACANF woodbury_m_Page_195.jpg
05e835d0187f744473029f57de66f9e2
17045b5b6e1ad3367e40dbbfcf1c17d0a141bcac
32289 F20101219_AACBXK woodbury_m_Page_030.QC.jpg
778c85b5b82b577eacaaae898fc69fae
b4424fa74df55910b2f1cc0a72b602ad8ffd792c
63245 F20101219_AACANG woodbury_m_Page_196.jpg
ce05ae8cbd56f13761d0ea25aec4da00
0e6c89ddf2b653df95b8b1370f4b24fd2e08180e
52189 F20101219_AACBKA woodbury_m_Page_150.pro
ddd1b9ea557c4fd1fc7f120d35f1258f
f7135b5588b5d7f82366c8906bdc5fafa03f8ddd
8221 F20101219_AACBXL woodbury_m_Page_030thm.jpg
d53eb3c66838a7808a9afa30448ff932
4b5359f1d36aebd5da066bf58b8f1be08a7b6c7d
57838 F20101219_AACANH woodbury_m_Page_197.jpg
81fbb7c6249b70ce1f258566af01f2ea
75d68878a5e7fcd2c107ca256b534aa0f5988a3f
51832 F20101219_AACBKB woodbury_m_Page_151.pro
44129e4561d90071809885ffda5df0a6
d56dce7ffdad2d49e85f56179f2b445cf3efe884
33617 F20101219_AACBXM woodbury_m_Page_031.QC.jpg
89f7168d2a7e2194bffa50b488b588a9
fc12186728f6f3f7560ee80b23e967839d3e881a
80009 F20101219_AACANI woodbury_m_Page_199.jpg
eb5043c7fbbb805d315b354ba3e64804
b131f79c07acf8a30ed0f6809e2a51ad302ec6f3
50548 F20101219_AACBKC woodbury_m_Page_152.pro
854a2674088e1101d4cb67e2d61d565f
e6b4e344a1b0429910d9486ff61f72a7885f8b7d
8371 F20101219_AACBXN woodbury_m_Page_031thm.jpg
4c6adc31c4f711da8db394213eb498e6
1997d3c9fcd64ee0b7dffd5c237c1ac52de0ca7d
79362 F20101219_AACANJ woodbury_m_Page_200.jpg
4164ec40f5e96146eb655c957912e0b2
fba7091d849f67698fb67ea82dd05afe72071e2c
52985 F20101219_AACBKD woodbury_m_Page_153.pro
16587edb89201b065bf9b4a4b7feee21
18eb2fc66fb7abd65929202db8a15e51e2db154d
32299 F20101219_AACBXO woodbury_m_Page_032.QC.jpg
0a8e1602eb59079b0fbeeda99ab93064
ac3266b8af0e62ee3c6bb24fbe11375d248fdbdd
33783 F20101219_AACANK woodbury_m_Page_201.jpg
af84cfc1113fb1309393787c46423f6d
edba36a11d79f264e295bfe5e7bcb5b7775db809
47037 F20101219_AACBKE woodbury_m_Page_154.pro
773b3e2a31ac45c46dfaea7f8b620b05
ff0da8eccf75ae5b91d13f54e802a081c21480b4
8048 F20101219_AACBXP woodbury_m_Page_032thm.jpg
2a06b5aea7111e2e6707ac14389181db
4b6c4a0b09b8186a1935263129356a67f0efcca6
61306 F20101219_AACANL woodbury_m_Page_203.jpg
a803c516ca561a37b85e1c23c0b7f411
7a710c8d49ada9abe92eb1f15b64be47cb16c05b
37378 F20101219_AACBKF woodbury_m_Page_155.pro
573c5898fbad2dd932a32eeaf50c42fc
bfe2baa578797af9d514ba90dc90cc048f7b16bd
32280 F20101219_AACBXQ woodbury_m_Page_033.QC.jpg
c83a2fec1abbcdfc4168701b15800e61
534ee98b1f4799019d1f10115cf63c736d603c6f
66916 F20101219_AACANM woodbury_m_Page_205.jpg
f692045fb9ba8e91108e0a9dfdc886ce
6bf3601925db0646eda9b57c779282ac3d174fbe
13858 F20101219_AACBKG woodbury_m_Page_156.pro
12cada64824ce908c9bfb1389287b126
40d062f8642d7a7f5f12c4ec262a9c5ce25c46f4
7905 F20101219_AACBXR woodbury_m_Page_033thm.jpg
615162071f3c421671310346cce88840
6d0e514d9682983cadb810773b9c853a6b127e35
105074 F20101219_AACANN woodbury_m_Page_207.jpg
953fb33438fcd925e7119d721cefe09e
4b30b22a22d1467fdfacadbf054ad5c7a17df6d5
12383 F20101219_AACBKH woodbury_m_Page_157.pro
1ffec19f6a67a9b7200b96f9fbcac05c
39e1a30ff3dfb24c0ed500409e6518c84dac9225
9973 F20101219_AACCHA woodbury_m_Page_167.QC.jpg
1dcabddec7a3db56e861c6d1f78e0e25
d025d320bbb68ae2049ba8da3891b77de48eca65
33013 F20101219_AACBXS woodbury_m_Page_034.QC.jpg
10305662d6ded344df9f3efebfce0c11
ec1165b0d418fae4a890223412e2c414fb003a82
127142 F20101219_AACANO woodbury_m_Page_208.jpg
e663e050fa04397e0686c80d79d954c7
dc261d4ab5c203e03d47e1cb895fc93bfdf8edfb
19182 F20101219_AACBKI woodbury_m_Page_158.pro
2bfb561ed2826a30969f68d826ca9447
78698e25f2ebc235187847e066ae5cf1e21f65f8
3141 F20101219_AACCHB woodbury_m_Page_167thm.jpg
05b81cba65a9ab51e08fd2c7f261d070
4dca25c31bd557702ecf6bdbebda56f2be275205
8314 F20101219_AACBXT woodbury_m_Page_034thm.jpg
5106c7d3b614734c47a67d7be2b3938b
ad8a09ec9c0c5b8da467a2c23292769e162acd11
126662 F20101219_AACANP woodbury_m_Page_209.jpg
4ff581ba65ff76a2a49f220e0dd06f34
e033c3559b5bc07186817cbebae156972412880d
33771 F20101219_AACBKJ woodbury_m_Page_159.pro
081820cda00aa1acf555b0421bd3d509
4306723a0ef85d10e222bf48e53314e396ff135f
31631 F20101219_AACCHC woodbury_m_Page_168.QC.jpg
2df3c201679c8f860670061f6a95a8bf
72c2219073b4e536f34438ca4ac7fb48ac6899db
29251 F20101219_AACBXU woodbury_m_Page_035.QC.jpg
d55ec105d89de38af3bb513629916c90
08df177ec1fa2e1589cc2dd078c8554d61e112b3
122009 F20101219_AACANQ woodbury_m_Page_210.jpg
f954adadbb82f04d5ab8f6ed7dd5766c
9a977c5f9d8aca0b82ab391588fec5def31c7874
43261 F20101219_AACBKK woodbury_m_Page_160.pro
c5f6c153b4305458b7691e3ebe83b7e5
a8617029d28be3b9265c099acd687d42d78d4991
7779 F20101219_AACCHD woodbury_m_Page_168thm.jpg
24bec0ae860810d0f6a6511bc9872587
b82b71fc1d87573a056c471b1a7c95b75a4fa75e
7386 F20101219_AACBXV woodbury_m_Page_035thm.jpg
cee10b046b5808a89aeeaf022b2c53e1
82e66725ad2cea5758b407ec5c8e976058c7d4ce
133185 F20101219_AACANR woodbury_m_Page_211.jpg
3141a6d40d649ecca33cf438a5bbf8f7
97597bd3759b0cbf2ec7d977e4b6375a2d078790
20822 F20101219_AACBKL woodbury_m_Page_161.pro
b00e2472ce596d613ba5f26a65deb05e
c9677fcf763a58a33442b2e146cb79f79c04db1a
F20101219_AACCHE woodbury_m_Page_169thm.jpg
8e593705e7c018304736be8fbc2e5034
18a4aca2f399c415fc4142cff74ab78c43cb11ea
32678 F20101219_AACBXW woodbury_m_Page_036.QC.jpg
2f55e9fd729dd962dd2500928095d46b
0d6ba5040099d458dcd587be9a9a5844684f8302
114211 F20101219_AACANS woodbury_m_Page_212.jpg
400460fb979e0ff629a5233b7ee0a98b
c6a1e20bb72050e4e5a5ea0d1cb6c75c79c21260
60085 F20101219_AACBKM woodbury_m_Page_162.pro
21fbe14d9ff195ec723ef44af89b438b
69e6f9ebcf45f7602db75fd473af7635068be1b9
27461 F20101219_AACCHF woodbury_m_Page_170.QC.jpg
ae1d53e863916ed65896614da00eb6d1
59579cc96e037f954cf556b6ddcc3bb9f54a02e0
7996 F20101219_AACBXX woodbury_m_Page_036thm.jpg
8bdef92654860e77daec7ea28dafc306
28c8499ae3c4277929ddd3dd8caf54038f6c8ed8
116230 F20101219_AACANT woodbury_m_Page_213.jpg
dd5201858cca8c74fbc6596e292e6c7e
0f6bad1683d6eb8ef49a70b8904115ea7d0c5bd1
62586 F20101219_AACBKN woodbury_m_Page_163.pro
6bfabb3a257f190d78e6f20da3fe69e5
dd56996a60a2f57030dd89b1bc700e7631e23f3d
7279 F20101219_AACCHG woodbury_m_Page_170thm.jpg
55b50612406b2c068372b3403c7f5136
74dafb12aeee6afd5179c838717e0b27a7997144
32092 F20101219_AACBXY woodbury_m_Page_037.QC.jpg
32b7394d8b6a8257ec027d157262ca0f
fba4a044abf3fc677e2f8a4abc528b15bca9151d
122752 F20101219_AACANU woodbury_m_Page_214.jpg
a773dd391b612a1fa9529704ebd09f80
2751640da32863e967a376e2b27eff6ae1ac9380
38942 F20101219_AACBKO woodbury_m_Page_164.pro
2123105aeee38b1d2e5617a2208ec36a
cdcc431175448ee1875c540ebc968fee45f43924
29520 F20101219_AACCHH woodbury_m_Page_171.QC.jpg
c3817f51f88ed81464597d3769781ebd
cd1dde93865a8de752731d115f5900d72ce79e9e
8140 F20101219_AACBXZ woodbury_m_Page_037thm.jpg
4248c50a11cc8e93112aa027332e539a
b70635ab4db7fac18fb725eebf29afb68ad2c6da
126202 F20101219_AACANV woodbury_m_Page_215.jpg
d4914e050ca63b2bcea1fc820aa4b2e9
a6b913958f4cecb70478e2515a88fde2f9c31609
7189 F20101219_AACBKP woodbury_m_Page_165.pro
6df1fa0877cd90cc8adfd68de6da17f5
ff8310783c180993bcb44a1a60cfdd0d08616ee7
7637 F20101219_AACCHI woodbury_m_Page_171thm.jpg
c28a58f303c8236335d59fdc2b31441c
63ac669539cc301438f9027cbdee9437f0fb8906
128712 F20101219_AACANW woodbury_m_Page_216.jpg
edeca65e8c46fa04235a206da97d5138
8b89e2b234492abefd5e9925b54da7f625bf9ce7
21736 F20101219_AACBKQ woodbury_m_Page_166.pro
be11768d6c182fd788a7abd91b32e755
f089611a393469e06abd2474fa1fd00506420e2f
29052 F20101219_AACCHJ woodbury_m_Page_172.QC.jpg
8f79853e8eb49f69a0312a7fbb9c5d8d
132e0ae6e3c549faebb95620f0101b3b2dd1a5fa
125086 F20101219_AACANX woodbury_m_Page_217.jpg
3831b65d8daba587f49e9dd3d5417870
1a961e9ca8d06be25bfab9f7c40bd818930c38aa
5272 F20101219_AACBKR woodbury_m_Page_167.pro
35812ead925a0511f706ad50819d82f8
458c54c2f05ff5042a1ba394d875ae67178512f2
7764 F20101219_AACCHK woodbury_m_Page_172thm.jpg
346e0fbfb3821b5120019eee6ac5a522
f6eb28a6a843c4a7ee528238a7b6cc2871dc56e1
119696 F20101219_AACANY woodbury_m_Page_218.jpg
dffe4f3b7690f29da38c1cef4487db19
eb7ce02414b304659372412d103c2613d2c9a558



PAGE 1

MEASURING POST-STROKE ARM MOTOR ABILITY: MEASUREMENT PROPERTIES OF THE FUGL-MEYER UPPER EXTREMITY ASSESSMENT EXAMINED WITH AN ITEM RE SPONSE THEORY MEASUREMENT FRAMEWORK By MICHELLE LYNNE WOODBURY A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2006

PAGE 2

Copyright 2006 by Michelle Lynne Woodbury

PAGE 3

To Bernard Earl Watrous and Leah Michelle Smith

PAGE 4

iv ACKNOWLEDGMENTS I must first thank my dissertation co-cha irs for their help and guidance throughout the process. Lorie Richards, a fellow Elizab ethtown College alumnus, has been a steady source of encouragement. When I have needed her, she is there with enthusiasm, empathy and ideas. Craig Velozo is a model mentor. His integrity, clarity of thought, humility and witty demeanor make it possible for him to expect, and get, excellence from his research team. I also must thank the other members of my committee, Dena Howland and Kathye Light. Dena, an occupational therapist turned neuroscientist, somehow is able to balance her family life with a highly demanding research career. I am grateful that in the midst of chaos she always had time for me, gave me her full attention, and was able to clearly provide guidance and encouragement. Kathye is a highly intelligent wonderful teacher who has shown me how to mix academic cont ent with a healthy dose of real world application and clinical intuition. I am deeply grateful to Pam Duncan. Th is dissertation idea be gan with her; she graciously allowed access to her databases, and she has been dangerously enthusiastic about the project from the beginning! My family has been a constant source of support. Many thanks go to my sister and brother-in-law, Melissa and Christopher Sm ith; my very cute niece, Leah Michelle Smith; my mother and step-father, Bonnie a nd Red Beuglass; my dad and step-mother, Gary and Robin Woodbury; my crazy cousin Rosemary Watrous; my wonderful neena Ruth Watrous; and my clever and funny gra ndfather Bernard Watrous whom I miss.

PAGE 5

v I must also thank my RSD family, the fellow students that are the source of 24/7 support and wisdom. Many thanks go to the Ve lozo research team (Leigh, Inga, Jia-Hwa, Pey-Shan) for their assistance with Rasch a nd 2PL analyses. Thanks go also to Arlene, Dennis (Steve), and Mike for forcing me to choose UF. I am grateful for the OT RSD students who make me laugh: Megan, Jessica Roxanna, Rick, Bhagwant, Eric, Kezia, Sandy W., Sandy P., Patricia, Christine, and Mindy. It is easy to loose perspec tive in the academic rehabilita tion research context. Three groups have helped me remember the truly important big picture. Prior to coming to PhD-school I was privileged to sit at th e feet of amazing teachers: Drs. Mullen, Hulbert, Priest, Larkin, and Davidson (Colum bia International University). They modeled academic and personal excellence, becoming r eal life illustrations of wisdom and knowledge. Waleed-Al Oboudi is an OTR w ith stunning talent whom I have been privileged to call mentor and friend. His passionate desire to advance clinically relevant stroke rehabilitation by traini ng therapists to pursue excellence has been the goal to which I have aimed both academically and clini cally. Finally, I have very clear memories of two former patients, Luke Harrell (14 year s old) and Caitlin Hill (9 years old). I see them smile and laugh despite devastating brai n damage. Their memories remind me, with crystal clarity, that improving outcomes for patients like Luke and Caitlin (and their families) is unquestionably the only importa nt reason for rehabilitation research. This dissertation work was partially supported an NIH T-32 Neuromuscular Plasticity Training Grant Fellowship (T32 HD043730). The Kansas City Stroke Study was funded by the National Institute on Ag ing, Claude D. Pepper Center Older Americans Center (5P60AG14635).

PAGE 6

vi TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES...............................................................................................................x LIST OF FIGURES..........................................................................................................xii ABSTRACT.....................................................................................................................xi ii CHAPTER 1 INTRODUCTION AND LITERATURE REVIEW....................................................1 The International Classificati on of Function and Disability.........................................5 Motor Control Theory...................................................................................................7 Traditional Theories of Mo tor Control and Recovery...........................................7 Challenges to the Traditional Theories of Motor Control and Recovery............10 Role of reflexes............................................................................................11 The pattern of upper extremity behaviors evidencing recovery...................12 Contemporary Theories of Mo tor Control and Recovery...................................18 UE motor impairment...................................................................................18 Post-stroke UE recovery...............................................................................22 Post-stroke recovery of f unctional UE motor abilities.................................24 Clinical significance of the FMA-UE score.................................................27 Summary......................................................................................................29 The Rasch Measurement Model.................................................................................29 Measurement.......................................................................................................29 Item Response Theory.........................................................................................30 Potential Advantages for Using the Rasch Model to Explore the FMA-UE Measurement Properties...................................................................................31 Validity of the item content..........................................................................31 Item difficulty hierarchy...............................................................................36 Reproducibility of the item structure...........................................................38 Functional interpretation of the FMA-UE score..........................................43 Criticisms of the Rash Measurement Model.......................................................47 Sample size...................................................................................................47 Should data fit the model or visa versa?......................................................48 Conclusion..................................................................................................................50

PAGE 7

vii 2 USING THE RASCH MEASUREMEN T MODEL TO EXAMINE POSTSTROKE UPPER EXTREMITY RECOVERY.........................................................54 Methods......................................................................................................................57 Dimensionality....................................................................................................57 Construct Validity...............................................................................................60 Results........................................................................................................................ .61 Participants..........................................................................................................61 Fit Statistics.........................................................................................................62 Principal Components Analysis..........................................................................62 Item Hierarchy.....................................................................................................64 Keyforms.............................................................................................................65 Discussion...................................................................................................................67 Dimensionality....................................................................................................68 Item Hierarchy.....................................................................................................69 Limitations...........................................................................................................73 Conclusion..................................................................................................................74 3 LONGITUIDNAL STABILITY OF THE 30-ITEM FUGL-MEYER UPPER EXTREMITY ASSESSMENT ITEM DIFFICULTY HIERARCHY.......................82 Methods......................................................................................................................84 Participants..........................................................................................................84 The 30-item Fugl-Meyer Upper Extremity Assessment.....................................85 Data Analysis.......................................................................................................85 Step 1: Dimensionality.................................................................................86 Step 2: Equivalence of item difficulties.......................................................87 Step 3: Differential item functioning...........................................................87 Step 4: Impact on assessment of arm motor function..................................88 Results........................................................................................................................ .89 Participants..........................................................................................................89 Step 1: Dimensionality.................................................................................89 Step 2: Equivalence of item difficulties.......................................................89 Step 3: Differential item functioning T1 vs. T2...........................................90 Step 4: Impact of DIF on person ability measurement.................................91 Discussion...................................................................................................................92 Study Limitations................................................................................................98 Conclusion...........................................................................................................99 4 LINKING THE FUGL-MEYER TO TH E WOLF MOTOR FUNCTION TEST: ASSESSING THE CONTINUUM OF IMPAIRMENT TO FUNCTION...............105 Methods....................................................................................................................110 Participants........................................................................................................110 Instruments........................................................................................................110 Data Analysis.....................................................................................................112 Step 1: Dimensionality......................................................................................112

PAGE 8

viii Principal Components Analysis (PCA)......................................................113 Rasch analysis............................................................................................114 Item correlations.........................................................................................115 Step 2: Longitudinal Stability............................................................................115 Step 3: Linking and Concurrent Cali bration of the FMA-UE + WMFT Item Pool................................................................................................................116 Step 4: Dimensionality of the Common Item Pool...........................................117 Step 5: Longitudinal Stability of the Common Item Pool.................................117 Step 6: Interpretation of Change Scores............................................................118 Functional threshold...................................................................................118 Functional interpretation of a FMA-UE change score...............................119 Results.......................................................................................................................1 19 Participants........................................................................................................119 Conversion of WMFT Performance Time Data to Performance Categories....119 Step 1: Dimensionality of the WMFT...............................................................123 Step 2: Longitudinal stability of the WMFT.....................................................124 Step 3: Linking and Concurrent Ca libration of the FMA-UE + WMFT..........125 Step 4: Dimensionality of the FMA-UE + WMFT Item Pool...........................125 Step 5: Longitudinal Stability of the FMA-UE + WMFT Item Pool................128 Step 6: Interpretation of Change Scores............................................................129 Functional threshold...................................................................................132 Functional interpretation of a FMA-UE change score...............................132 Discussion.................................................................................................................134 Linking as a Method to Expl ore Significant Change.....................................134 Linking as a Method to Measure Body Function and Activity Performance....135 Linking as a Way to Conceptualize the Motor Rehabilitation Treatment Process...........................................................................................................136 Limitations of this Study: Conversion of Performance Time data to Ordinal Data................................................................................................................137 Limitations of this Study: Sample Characteristics............................................138 Limitations of this Study: Possible Mu ltidimensionality of the Item Pool.......139 Limitations of this Study: Sample Size.............................................................140 Conclusion................................................................................................................141 5 A COMPARISON OF THE RASCH MODEL AND A TWO PARAMERTER ITEM RESPONSE MODEL TO DERIVE ITEM AND PERSON PARAMETERS........................................................................................................154 Methods....................................................................................................................156 Participants.................................................................................................156 Dimensionality...........................................................................................156 Preparing the data for 2PL analysis............................................................156 Model fit.....................................................................................................157 2PL analysis...............................................................................................157 Comparison of parameter estimates...........................................................157 Item discrimination....................................................................................157 Person ability and item difficulty...............................................................158

PAGE 9

ix Results....................................................................................................................... 158 Model Fit..................................................................................................... 158 Item Discrimination D512........................................................................... 158 D377 Item Discrimination........................................................................... 159 D512 Item Difficulty................................................................................... 160 D377 Item Difficulty................................................................................... 161 D512 Person Ability.................................................................................... 162 Discussion................................................................................................................. 162 Which IRT Model, 2PL or Rasch, Best Fits the FMA-UE Data?............... 162 Within Each Dataset, Do FMA-UE Item Discriminations V ary?................163 Does the Choice of IRT Model Affect Either Item or Person Parameters?.164 What Does Item Discrimination M ean?.......................................................164 Conclusion......................................................................................................... 167 6 SUMMARY AND CONCLUSION.........................................................................175 APPENDIX A RASCH ANALYSIS CONTROL FILES.................................................................179 Control File for Project 1 (Chapter 2)........................................................179 Control File for Project 2, T1 (Chapter 3)..................................................180 Control File for Project 2, T2 (Chapter 3)..................................................181 WMFT Control File for Project 3, T1 (Chapter 4).....................................182 WMFT Control File for Project 3, T2 (Chapter 4).....................................183 FMA-UE + WMFT Control File fo r Project 3, T1 (Chapter 4).................184 FMA-UE + WMFT Control File fo r Project 3, T2 (Chapter 4).................185 B SAS AND MULTILO G CONTROL FILES............................................................188 SAS PCA Control File for FMA-UE Data Project 1 (Chapter 2)..............188 SAS PCA Control File for WMFT Data Project 3 (Chapter 4).................189 SAS PCA Control File for FMA-UE + WMFT Data Project 3 (Chapter 4).............................................................................................................190 Multilog Control File for D512 (Chapter 5)...............................................191 Multilog Control File for D377 (Chapter 5)...............................................192 LIST OF REFERENCES.................................................................................................193 BIOGRAPHICAL SKETCH...........................................................................................214

PAGE 10

x LIST OF TABLES Table page 1 The Fugl-Meyer Upper Extremity Assessment........................................................52 2 Descriptions of the traditiona l upper extremity limb synergies...............................53 3 Characteristics of the 512-person sample.................................................................74 4 Item measures, fit statistics and scor e correlations for the full 33-item FMA-UE (n = 512)...................................................................................................................75 5 Eigenvalues for the principal components analysis.................................................77 6 Item correlations with first principal component.....................................................77 7 Rotated factor loadings (oblique rotation) for 4 principle components...................78 8 Item measures, fit statistics, and corr elations for revised FMA-UE (n = 512, 30 Items)........................................................................................................................7 9 9 Item difficulty hierarc hy revised 30-item FMA-UE................................................80 10 Sample characteristics (n = 377)............................................................................101 11 Rasch derived infit statis tics T1 and T2 (n = 377).................................................102 12 Differential item functioning analys is (n= 377, df = 594, ** p<0.0017, p<0.05)103 13 Sample characteristics (n = 91)..............................................................................142 14 Example of rating scale statistics...........................................................................143 15 Principal components analysis eigenval ues of the Wolf Motor Function Test......144 16 Principal component factor loadings......................................................................144 17 Wolf Motor Function Test item measur es, fit statistics, and point measure correlations.............................................................................................................145 18 Wolf Motor Function Test it em difficulty hierarchy.............................................146

PAGE 11

xi 19 Wolf Motor Function Test diffe rential item f unction analysis..............................146 20 Eigenvalues of the FMA-UE + WMFT linked assessment....................................147 21 FMA-UE + WMFT item pool rotated factor matrix..............................................148 22 FMA-UE + WMFT item measure, fit stat istics, and point measure correlations..149 23 Comparison of item discrimination valu es calculated by Multilog to those calculated by Winsteps in the first dataset.............................................................168 24 Comparison of item discrimination valu es calculated by Multilog to those calculated by Winsteps in the second dataset.........................................................169 25 D512 comparison of item parameters de rived with the 2PL and Rasch models....170 26 D377 comparison of item parameters de rived with the 2PL and Rasch models....172

PAGE 12

xii LIST OF FIGURES Figure page 1 Keyform recovery maps for th ree individuals with stroke.......................................81 2 item difficulty measures of T1 (0 14 days post-stroke) and T2 (6 months poststroke).....................................................................................................................103 3 scatter plot of 30-item FMA-UE item difficulties at T1 versus T2........................104 4 Comparison of mean person ability with and without DIF adjustment.................104 5 Scree plots of the FMA-UE + WMFT linked assessment......................................147 6 FMA-UE + WMFT differential item functioning scatter plot..............................151 7 FMA-UE + WMFT linked item map.....................................................................152 8 FMA-UE person ability measures ove rlaid onto the FMA-UE + WMFT item map.........................................................................................................................153 9 D512 comparison of item hierarchies ge nerated by Multilog (p robits) to the item hierarchy generated by Winsteps (logits converted to probits by multiplying by 0.59). Error bars 2 SEM.........................................................................................171 10 D512 scatter plot of item difficu lty estimates (n = 512, r = 0.99)..........................171 11 D377 comparison of item hierarchies ge nerated by Multilog (p robits) to the item hierarchy generated by Winsteps (logits converted to probits by multiplying by 0.59). Error bars 2 SEM.........................................................................................173 12 D377 scatter plot of item difficu lty estimates (n = 377, r = 0.98)..........................173 13 D512 scatter plot of person abil ity estimates (n = 512, r = 0.99)...........................174 14 D377 scatter plot of person abil ity estimates (n = 377, r = 0.99)...........................174

PAGE 13

xiii Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy MEASURING POST-STROKE ARM MOTOR ABILITY: MEASUREMENT PROPERTIES OF THE FUGL-MEYER UPPER EXTREMITY ASSESSMENT EXAMINED WITH AN ITEM RE SPONSE THEORY MEASUREMENT FRAMEWORK By Michelle Lynne Woodbury August 2006 Chair: Lorie Richards Cochair: Craig Velozo. Major Department: Occupational Therapy There is a pressing need to ascertain wh ether assessment tools commonly used in post-stroke upper extremity motor rehabilita tion research and clinical practice are accurately quantifying impairment, characteri zing recovery and producing a score that has clear functional interpretability. Th e Fugl-Meyer Upper Extremity Assessment (FMA-UE) is the most widely used clinic al assessment of post-stroke UE motor impairment. Because of the importance of th is assessment in post-stroke rehabilitation research, it is critical to examine, and pe rhaps improve, the quality of its measurement properties. The overall purpose of this di ssertation was to examine the measurement properties of the FMA-UE. An item-oriented measurement perspective was used to investigate (1) the validity of the FMA-UE ite m content, (2) the re producibility of its item structure, and (3) interpretation of its sc ore. Rasch analysis, a type of item response theory statistical method, was applied to FMAUE data collected from 512 individuals in

PAGE 14

xiv the Kansas City Stroke Regi stry and Kansas City Pepper Center exercise study. In the four studies comprising this di ssertation, it was determined th at the items of the FMA-UE contribute to the measurement of multiple c onstructs and demonstrate a difficulty order that is not consistent with the item order proposed by Fugl-Meyer. The reflex items were removed, resulting in a modified 30-item FMA-UE, a unidimensional measure of poststroke UE motor ability. It wa s found that the Rasch-derived item difficulty order of the 30-item FMA-UE remained consistent, inde pendent of person ability, longitudinally across two testing occasions. The modified 30-item FMA-UE was linked to the Wolf Motor Function Test. Linking is a novel method to improve the interpretability of the FMA-UE score and clearly illustrated how ch anges in motor impairment can translate to improvements in activity performance. Th e item and person parameters estimated by Rasch analysis were validated with the two-parameter item response theory graded response model. This project is significant in th at it not only establis hes the quality of the 30-item FMA-UE as an instrument to meas ure post-stroke UE motor impairment, but also establishes the 30-item FMA-UE as an in strument with which to gather information that will advance rehabilitation science motor control theory.

PAGE 15

1 CHAPTER 1 INTRODUCTION AND LITERATURE REVIEW Each year over 700,000 individuals in th e United States have a stroke [1]. Approximately three-fourths of these individuals will experience immediate flaccid paralysis of one arm and hand [2]. This flacc id period is followed by rapid recovery of arm motor function over the first 3-months poststroke [3]. However, full recovery of arm and hand function is rare. At four years pos t-stroke 85% of str oke survivors report residual arm or hand motor impairment [4]. Fo r many individuals with severe stroke the arm never becomes useful [5] and for many indi viduals with less severe stroke, residual arm motor impairment interferes with perfor mance of daily tasks [4]. Stroke is the leading cause of long-term disabil ity in the United States [1]. In recent years, progress in neuroscien ce and rehabilitation science has advanced the understanding of the potential for neuronal recovery after stroke -related brain damage [6]. In turn, this has led to the development of new rehabi litation interventions [7-9] and revision of traditional rehabi litation interventions [10] to restore post-stroke upper extremity (UE) motor function. The rehabilitati on disciplines have adopted an evidencebased practice paradigm [11, 12], necessitating accurate measurement of motor recovery. There is a pressing need to ascertain whet her assessment tools commonly used in poststroke UE motor rehabilitation research and clinical practice are accurately quantifying impairment and characterizing recovery [13, 14]. The Fugl-Meyer Upper Extremity Assessmen t (FMA-UE) [15] is the most widely used clinical assessment of post-stroke UE motor impairment [14, 16]. It has excellent

PAGE 16

2 intra-rater reliabili ty (r = 0.995 [17]), inte r-rater reliability (r = 0.992 [17]), test-retest reliability (Intraclass correlation coefficients 0.94 0.99 [18]), and internal consistency (r = 0.97 [19]. It has been used as the standard fr om which to establish the construct validity of other commonly used tests of post-stroke UE motor function such as the Wolf Motor Function Test and Ashworth Scale [20-26]. Re searchers consistently use the FMA-UE as a descriptor of functional stat us following stroke [4, 20, 27]. To this end, FMA-UE scores have been used to stratify research study part icipants into categorie s of stroke severity [28], predict long-term functional particip ation [29], and to describe arm motor impairment [30-32], arm movement quality [21], and residual ar m motor function [3335]. Furthermore, the FMA-UE is the prim ary criterion for evaluating the success of novel upper extremity interventions such as rhythmic bilateral m ovement training [36, 37], neuromuscular stimulation [38], electromyogra phic triggered electr ical stimulation [39, 40], Botox [41], home based exercise [ 30], community based exercise [42], robot aided therapy [43, 44], virtual reality [45], imagery [46], fo rced use [47], and modified constraint induced movement therapy [4851]. Because of the importance of this assessment in post-stroke rehabilitation resear ch, it is critical to examine, and perhaps improve, the quality of it s measurement properties. Table 1 presents the FMA-UE. The evaluato r observes the client s performance of 33 items, scoring each item by applying a 3-po int ordinal rating scale (0=the movement cannot be performed at all, 1=the moveme nt can be partially performed, and 2=the movement can be faultlessly performed). Ra tings are summed and the FMA-UE score is reported out of the maximum 66 points. Highe r scores indicate more UE motor ability.

PAGE 17

3 Recently the Canadian Stroke Network identified 88 stroke-related UE motor function assessments [52]. The obvious questio n is, which is the best assessment? Salter and colleagues (2005) [14] argue that evidence based practice is advanced only when researchers and clinicians choose assessments with superior measurement properties. The question then b ecomes, what are the essentia l elements of a high quality assessment? The quality of an assessment is usually judged by its essential measurement properties: reli ability, validity, and responsiv eness [14]. These criteria assure that an assessment measures the skill that it is purported to measure (validity), produces a reproducible score (reliability), and is sens itive to change over time (responsiveness). Psychometric evaluation of an assessments measurement properties is usually accomplished with classical test theory statistical methods su ch as correlating or comparing assessment scores. The item response theory (IRT) measurement framework, e.g., Rasch measurement, offers another approach for judging the quali ty of an outcome measure. As the name implies, IRT methods allow one to evaluate the measurement properties of items rather than whole tests [53]. One can examine an asse ssments item content (i .e., does a specific item contribute to the measurement of an inte nded construct?), item reproducibility (i.e., do items measure the purported construct in the same way each time the assessment is administered?), and item-level interpretation of the assessments scor e (i.e., with respect to item content, what does an assess ments change score mean?). Although the measurement properties of the FMA-UE have been extensively studied using classical test theory [15, 17, 18, 20, 24, 54-57], presently there are no published studies applying IRT approaches to the FMA-UE.

PAGE 18

4 The overall purpose of this dissertation is to examine the measurement properties of the FMA-UE. I will assume an item-orient ed viewpoint from which to investigate (1) the validity of the FMA-UE item content, (2) the reproducibility of its item structure, and (3) interpretation of its score. In the subsequent literature review I will develop the background to address the studys purpose. First, I will review the World Health Organizations International Classification of Function and Disability (ICF). The ICF provides an operational definition of the UE skills of interest to post-stroke UE rehabilitation researchers and models a way to organize measurement of th ese skills. Next, I will review traditional and contemporary concepts of UE motor control. This is important because exploring FMAUE item-level construct validity will serve to focus our attention on motor control theory. Third, I will review the Rasch measuremen t model because it permits testing the theoretical claims about a construct. Finally, I will briefly review the critical differences between the Rasch model and other IRT mode ls as they pertain to this project. In the chapters that follow I addressed the purpose of this di ssertation in four studies. The aim of study 1 was to test the va lidity of the traditional post-stroke UE motor control theory that underlies the FMA-UE by examining its item-level dimensionality and item difficulty hierarchy in a cross-sectional analysis. The aim of study 2 was to test the longitudinal reproducibility of the FMA-UE item structure. The aim of study 3 was to test a method for enhancing the inte rpretability of a FMA-UE change score by linking to another assessment with more functionally or iented items. Because Rasch analysis was the primary method used throughout these three studies, the final project was to compare the results obtained with the Rasch model to those obtained with another IRT model to

PAGE 19

5 assure that my choice of st atistical method was valid. The overarching purpose of this dissertation and the aims of each project are consistent with that of an overall emphasis within the field of rehabilita tion research to assure that clinical assessment tools are optimal for measuring targeted outcomes [58-60] and translate to functionally interpretable measuremen ts of change [61]. The International Classification of Function and Disability The World Health Organizat ions International Classi fication of Function and Disability (ICF) [62] offers a systematic approach to classify and describe post-stroke UE motor function according to a theory of health and disability. The ICF provides a useful taxonomy and standard language to define UE motor function and describe its measurement. As a practical research applica tion, the ICF describes se ts of behaviors that should be included in an instrument in tended to measure given domain [63]. The purpose of the ICF is to describe components of health in which a persons functioning is conceived as a dynamic multi-di mensional interaction between individual health attributes and contextual factors. The ICF identifies three primary health dimensions of the individual: the functioning of the body or body part s, the functioning of the individual in his/her daily activiti es, and the functioning of the individual in relationship to his/her participation in the socio-cultural community. These health dimensions can also be expressed in th e negative to describe disability: body function/structure impairment activity limitation and participation restriction The ICF models the concept that stroke -related UE motor deficits affect the individual along a continuum of health di mensions; from impaired body functions to limited performance of activities and rest ricted participation in the community. Accordingly, post-stoke arm/hand motor func tion can be measured in each dimension.

PAGE 20

6 Furthermore, the model suggests that altere d arm motor function as measured in one dimension (e.g., body function) conceivably tr anslates to altere d arm motor function when measured in an adjacent dimension (e .g., activity performance). The purpose of this dissertation centers on UE motor body function impairment and activity restriction; thus the subsequent discussion is limited to these domains. In series of papers [14, 64, 65] research ers from the Canadian Stroke Network organized the most commonly used post-str oke assessments according to ICF domains for psychometric review. The FMA-UE is classified as an assessment of body function impairment because its item content is cons istent with the ICF operational definition of this domain [14]. For example, neuro-muscu loskeletal and movement related functions illustrate one aspect of body function. More sp ecifically, the functions are described as control and coordination of single and mu lti-joint motions (p. 99-100). FMA-UE items fit this description. A valuable assessment for measuring upper extremity outcome after stroke is one that provides information about the recovery spectrum, i.e., movement impairment and the functional consequences of movement impairment. The ICF models a relationship between body function impairment and activity pe rformance. According to the ICF, it is conceivable that body function impairment im pacts behaviors in the adjacent activity performance domain such as the ability to carry, move and handle objects. More specifically, reaching, grasping, and lifti ng, (p. 1413). It is conceivable that an UE assessment intended to capture the func tional aspects of reduced body function UE impairment should be comprised of item s with this variety of content.

PAGE 21

7 Although the ICF offers a broad framewor k to define FMA-UE item content and organize its structure, the IC F does not specify what UE mo tor behaviors are affected by stroke, nor does it detail e xpectations regarding post-s troke recovery. For that, a discussion of motor control theory is warranted. Motor Control Theory Motor control theory provides a conceptu al framework of how the CNS organizes and controls motor functi on by addressing potential rules governing movement organization. Motor control theory models nor mal motor behaviors, defines injury-related motor behaviors, and predicts a post-injury course of motor recove ry [66]. One way in which an abstract motor control theory become s tangible to the rehab ilitation professional is in the design, structure and scoring of a clinical assessment [67]. The theory defines normal and abnormal movement behavior s therefore, govern ing the content of assessment items. Moreover, the theory desc ribes behaviors that evidence less and more motor control and, in doi ng so, suggests an appropriate array of items to measure a range of motor skills. Barecca et al. ( 2006) [68] argues that stroke rehabilitation knowledge is advanced only if the tools chosen to measure motor control are consistent with contemporary motor control theoretical ex pectations of deficit and recovery. Recent advances in movement science challenge the motor control theory upon which the FMAUE was based. Prior to a review of these challenges, I will review the theoretical foundation of the FMA-UE. Traditional Theories of Motor Control and Recovery The characterization of post stroke recovery has been the focus of theoreticians, clinicians and researchers and represents one of the histor ic underpinnings of modern rehabilitation. Many of the trad itional stroke rehabilitation frames of reference emanate

PAGE 22

8 from Jacksons late 19th century work with persons with epilepsy and other nervous disorders [69]. Jackson, a neuroscientist influenced strongly by Darwin, applied the doctrine of evolution to his observations and ultimately posited a theoretical order of the CNS. According to his framework, the co mplex motor abilities seen in humans are evidence that higher motor centers have evol ved out of lower centers. In the healthy nervous system these more advanced centers or chestrate and modify the reflex chains of lower centers. Jacksons observations of indi viduals with epilepsy led him led him to hypothesize that injury to the cortex resulted in dissolution of th e CNS and reversal of its evolutionary structure. That is, wit hout input from the advanced motor centers, movement control defaults to primitive lowe r centers evidenced in gross undifferentiated reflexive motions. In the early 20th century Charles Sherrington and his colleagues performed a series of elegant experiments to show the existence of the reflex and descri be the interaction of afferent feedback with voluntary movement. In a classic primate experiment Sherrington performed complete upper limb sensory denerv ation. This procedure appeared to abolish all voluntary upper limb movement. The limb wa s held close to the body and not used for daily tasks. Sherrington concluded that a fferent information was a necessary building block of voluntary movement [70]. The work of Sherrington and hi s colleagues formed the experimental foundation of the classic reflex chain hypot hesis [71]. The results of Sherrington and Jacksons experiments and observations were drawn together and are often referred to in the clinical literature as the reflex-hierarchical theory of motor control [66, 72].

PAGE 23

9 With the advances in medicine and impr oved post-stroke survival in the mid 20th century, clinician-researchers applied the refl ex-hierarchical theory to their clinical observations [73-75]. In a classic paper Dr. Thomas Twitchell (1951) [75] described an orderly progression of phenomena characteri zing the course of post-stroke UE motor recovery. Initial UE flaccidity was followed by emergence of gross flexor and extensor voluntary movement synergies, gradual sepa ration of synergy movement patterns and finally distal fine motor coordination. Twitche ll attributed abnormal motor behaviors to cortical dissolution and emerge nce of default reflexive behaviors as per Jacksons model of the CNS. Twitchell suggested that recovery was a re-evol ution of this hierarchy as the cortex reassumed its natural inhibition of primitive proprioceptive reflexes. Renowned physical therapist Signe Brunnstr om furthered Twitchells work, and made an important contribution to stroke rehabilitation by creati ng an assessment tool that synthesized clinical observation with motor control theory [76, 77]. She mapped post-stroke recovery into a six-stage sequen ce. Beginning with flaccidity (stage 1), motor recovery was first evident as return of reflexive respons es (stage 2). With further recovery individuals gained voluntary motion within whole-l imb flexion and/or extension synergies (stage 3), movement requiring a comb ination of the synergy patterns (stage 4), followed by movement independent of synergy patterns (stage 5), a nd finally individual joint motion and normal coordination (stage 6). The Hemiplegia Classification and Progress Record [76] detailed specific motor behaviors typifying each stage. The hierarchical ordering of move ments documented in the Progress Record was useful to the clinician for assessment and treatment planni ng. Using it, therapists could identify a patients stage of recovery by observing the movements the patient could or could not

PAGE 24

10 perform. Once the clients location along the recovery continuum was known, the Progress Record enabled the clinician to map tr eatment to advance the client to the next stage. Noting the dearth of quantitative assessments available to stroke rehabilitation clinicians and researchers, Fugl-Meyer et al (1975) [15] created a rating scale assessment based, at least in part, on Brunnstroms Pr ogress Record and Twitchells observations. Fugl-Meyers theoretical framework broadl y reflects the assumptions of the reflexhierarchical motor control theory. Fugl-Meyer stated: The form has been constructed following the hypothesis that the restoration of motor function in hemiplegic patients follo ws a definable stepwise course. Thus for a patient with hemiparalysis, recurrenc e of reflexes always precedes volitional motor action. Thereafter through initia l dependence on syne rgies, the active motion will become successively less de pendent upon the primitive reflexes and reactions and finally complete volunta ry motor function with normal muscle reflexes may be regained. [15] p. 14 Fugl-Meyer chose items to exemplify the construct of motor recovery including reflex items and voluntary movement items. Moreover, because the assessment was intended to measure recovery, Fugl-Meyer arra nged the items in a hierarchical order to illustrate the recovery process. He expected motor recovery to proceed in a reflexive-tovoluntary, and synergy-to-isolated progression as the cortex resumed its ability to integrate reflexive movement componen ts into coordinated limb motions. Challenges to the Traditional Theori es of Motor Control and Recovery In recent years there have been ma ny advances towards understanding how movement is controlled, organized and rec overed after injury. Th ere has been a shift away from a purely neurophysiologic explanatio n of motor control/recovery to a taskoriented view of motor cont rol/recovery [66]. C ontemporary motor control science posits that movement emerges from a dynamic, comp lex interaction of the individuals neural

PAGE 25

11 and musculoskeletal apparatuses with speci fic tasks and environmental demands [78]. Recovery is viewed as re-ac quisition of an individuals ability to successfully conquer task-specific contextual demands rather than neurophysiologic re-evolution of the cortical motor control hierarchy [79]. Recent advances in movement science challe nge the definable stepwise course of motor recovery upon which the FMA-UE is base d in two ways: (1) the place of reflex behaviors in an assessment of motor recovery, (2) the pattern of behaviors that evidence the course of recovery. These challenges will now be expanded upon. Role of reflexes Contemporary views of the central nervous system indicate th at reflexes and voluntary movements are uniquely different type s of motor behavior. A relatively simple spinal neural circuit gover ns the tendon tap muscle c ontraction, while a complex supraspinal and spinal neural network govern s voluntary goal-directed movement [80]. Afferent input is not necessary for voluntar y movements to occur [81], instead reflexes (e.g., the stretch reflex) appear to be importa nt adjuncts of voluntary motor control [80, 82, 83]. The reflex motor control system is complex as evidenced by research showing that the sensitivity of the of the stretch reflex threshol d is modulated according to previous muscle activity [84, 85]. The research of Wolpaw (1985) and Wolf and Segal (1986) has demonstrated that the magnitude of the stretch reflex is amendable to operant training [86, 87]. Burne and colleagues (2005) [8 8] found that spasticity (measured as the amplitude of the biceps stretch reflex) was different in a resting limb versus a moving limb. The researchers suggest th at reduced ability to dynamically modulate reflex activity contributes more to an unde rstanding of voluntary movement impairment than assessing reflex activity in a static condition. The re search evidence suggest s that the FMA-UE

PAGE 26

12 reflex items may be assessing a motor behavior that is uniquely different from other items that assess voluntary movement. The presence of the three reflex items in the FMA-UE has been long questioned. Lindmark and Hamlin (1988) [89] removed the items, altered the rating scale, and added items to form the Lindmark Functional Capacity Evaluation. Gladstone et al. (2002) [55] stated that reflexes were over represented in the FMA-UE total score. Whitall et al. (2004) [13] suggested that asse ssment of tendon-tap reflex behaviors does not contribute to an understanding of post-stroke UE motor ab ility. It is reasonable to question whether the presence of the three tendon tap FMA-UE reflex items confound the interpretation of the FMA-UE score and threaten the assessments construct validity. The pattern of upper extremity behaviors evidencing recovery The extent to which post-stroke individua ls exhibit abnormal UE flexor and/or extensor synergy as per Fugl-M eyers descriptions [15] (ada pted from Brunnstrom [76]) is debated in the literature. Table 2 reminds the reader of limb movements characterizing traditional synergy patterns. Note that these movements are items 3 11 on the FMA-UE (see table 1). Recently, Welmer and colleagues (2006) [90] prospectively observed 64 consecutive patients with st roke at one week and three months post-stroke. The researchers tested the hypothesi s that stroke recovery is consistent with Brunnstroms sequence, specifically that UE flexor and extensor synergie s emerge early in recovery while non-synergistic isolated movements em erge later in recovery. Subjects were evaluated by trained, study-sp ecific, occupational and physical therapists. One-week post stroke 15% of the sample was unable to follo w testing directions and 22% of the sample exhibited flaccid hemiplegia. Among the s ubjects who participat ed in voluntary UE

PAGE 27

13 movement assessment, no participant exhibited only (i.e. pure) flexor/extensor synergy movements. Welmer found that 63% of th e sample demonstrated at least one UE movement inconsistent with synergy (e.g., wa s able to partially extend the elbow while abducting the shoulder). Three months post stro ke, 13% of the sample remained unable to move. Among the others, 16% exhibited at le ast one UE movement typically considered part of the flexor/extensor synergy pattern (e.g., tended to flex the elbow while flexing the shoulder to or above 90 degrees) but 77% exhibited no evidence of synergy-related movements. The results suggest that some flexion/extension synergy movements occur early in recovery, but some isolated movements also o ccur early in recovery. The researchers concluded that individu als with stroke rarely exhibit pure manifestations of the flexor or extens or synergy patterns. Welmers work does not discount longstandi ng clinical observati ons [75, 76] that individuals with stroke exhibit impaired intra-limb control. Instead, Welmers study suggests that the abnormal patt ern is not manifest as an a ll or nothing obligatory link between all movements comprisi ng the traditional definition of the synergy (table 2). Indeed, contemporary experimental paradigms ha ve elucidated the sp ecific parts of the abnormal pattern that are most likely to be pr esent. In a series of studies Beer et al. (1999) [91] and Dewald et al. (2001) [92] have shown that in dividuals with stroke were unable to maintain maximal joint torque of a one joint in a si ngle direction without producing secondary torques of an adjacent jo int in an unintended direction during an isometric task. In this study, ei ght individuals with chronic st roke and moderate spasticity (Ashworth scores 3-5) were directed to attain and then maintain maximal shoulder flexion or shoulder abduction force. Forces were measured with load cells on the

PAGE 28

14 humerus and forearm. In both the shoulder fl exion and abduction conditions, all subjects concurrently produced unintended elbow flex ion torque. In a similar experimental paradigm, Lum et al. (2003) [93] showed that post-stroke subjects also produced unintended humeral internal ro tation torque. The results suggest some individuals with stroke exhibited abnormally constrained intra-limb motor control evident by abnormally linked shoulder flexion/abduction, elbow fl exion, and internal rotation torques. Abnormal linkages of elbow flexion with the shoulder flexor or abductor musculature are also evident during supporte d and unsupported reach. B eer et al. (2004) [94] studied eight persons with chronic stroke and moderate spasticity. In this study subjects performed rapid movements in the ho rizontal plane from a central starting point to five targets located at th e outer limits of the reachable workspace. The targets were placed to require different combinations of flexion and extension ranges of motion at the elbow and shoulder. In one condition the he miparetic limb was supported in a frictionless arm trough, in another it was not. In both conditions subjects had difficulty generate concurrent shoulder flexion and elbow extensi on torques necessary to reach the targets, however the deficit was greater in the uns upported versus supported condition. Kamper et al. (2002) [95] showed similar interjoint co ordination deficits in an unsupported reaching condition. In this study, sixteen participants with chronic stroke were directed to point to toward a screen containing 75 targets displayed side to side across the arm reachable workspace from waist to head. Kinematic anal ysis showed that participants exhibited abnormally constrained elbow flexion and shoulder flexion evident in curved hand trajectories when attempting to reach to ta rgets the farthest from the body. The deficit was similar irrespective of target locations in the ipsilesional or c ontralesional workspace.

PAGE 29

15 Dewald, Beer and Kampers re sults are consistent with th ose of Levin et al. (1996) [35]. In this study, the resear chers studied reaching trajector ies of 10 participants with moderate post-stroke UE impairment (moderate spasticity measured on a scale designed by the investigator and baseline FMA-UE sc ores 19 66). Targets were arranged on a table so that subjects were required to utilize muscle activation patterns within and outside of the traditi onal synergy configurations. For ex ample, to reach a target placed contralateral to the affected limb, the partic ipant was required to activate the extensor synergy musculature; shoulder adductors, internal rotators and elbow extensors. The researchers reasoned if particip ants movements were strictly constrained into a flexor or extensor synergy pattern, part icipants would be less able to reach targets outside this pattern. That is, if extensor synergy were dom inant, reach to a contralateral target would be less impaired than reach to an ipsilateral target. If the flexor synergy were dominant, reach to a near target would be less impaire d than reach to a far target. The results showed that this was not the case. Reachi ng trajectories were less smooth and more curved for individuals with stroke when comp ared to healthy individuals, however poststroke movements within a typical synergy pa ttern were no less impaired (i.e., were not smoother and straighter) than movements to targets outside th e synergy pattern. In an earlier study Trombly (1992) [96] rea lized a similar phenomena. In this study, 5 subjects with left hemiparesis were directed to reach to three targets placed so as to require movement into an extensor synergy pa ttern or away from this pattern. Kinematic and electromyographic (EMG) data were reco rded. All clients demonstrated impaired kinematics (less smooth trajectories, more movement units, increased movement time, and earlier time to peak velocity) and a ltered EMG (higher percentage of maximum

PAGE 30

16 voluntary output) in the impaired UE as comp ared to the unimpaired UE. However, the biceps-anterior deltoid coactivity indices we re not significantly different between the limbs indicating that an abnormal synerg istic constraint was not hampering the movement. The work of Trombly, Levin, Kamper, De wald, Lum and Beer provide evidence that the post-stroke reach is impaired by an abnormally constrained inter-joint coordination pattern involving a tendency to flex the elbow and internally rotate the humerus when attempting to lift the humerus ag ainst gravity. Tromby [96] suggested that this observable pattern is not directly attributable to abnormal co-activity of the flexor synergy musculature, but may evidence a motor control strategy employed to reach targets despite muscle weakness. When put toge ther with the work of Welmer et al. [90], it is apparent that the observed post-stroke UE motor deficit is not fully explainable by traditional definitions of synergy. Full flexor or extensor synergy pa tterns are not always present, at least in the subjects studied by th e above labs. It is possible that the other muscles typically considered part of the s ynergy patterns, specif ically the scapular retractors, scapular elevators, shoulder exte rnal rotators and for earm rotators have not been studied due to limitations in coll ecting surface EMG from deep muscles and limitations in UE kinematic and kinetic modelin g. However, at least to date, the evidence suggests pure synergies (as per the traditiona l definition) do not appear to be obligatory patterns for all individuals with stroke. The importance of this research lies in its potential impact on the interpretation of the FMA-UE score. Fugl-Meyer designed the scale with an expect ed regularity of recovery in mind [15]. The evidence only pa rtially supports this expected synergistic

PAGE 31

17 recovery pattern. Therefore, it is possible that the measurement properties of the FMAUE with regards to the interpretatio n of its score may be affected. Researchers use the FMA-UE score to characterize post-stroke recovery. One interpretation of the FMA-UE score is consis tent with traditional expectations regarding the pattern of recovery; specifically low sc ores represent synergy-dominated movement abilities whereas higher scores represent more isolated movement abilities. For example, Chae et al. (2003) [97] interprete d the FMA-UE score in this way it (the FMA-UE) measures evolving synergy patterns with subj ects with voluntary movements limited to proximal joints and synergy patterns exhibi ting the lowest scores and with subjects demonstrating isolated movements in distal joints, with normal reflexes exhibiting the highest scores (p. 7). Some researchers have used cat egories as a context from which to interpret the FMA-UE score. For example, Michaelson et al. (2001) [98] stated that FMA-UE scores of 0 points represented severe UE motor impairment, FMA-UE scores of 20 points represen ted moderate UE motor impairment, and FMA-UE scores of 65-66 represented normal UE motor ability. Pang [42] stated that FMA-UE scores of 0 27 points represented severe UE moto r impairment, FMA-UE scores of 28 57 points represented moderate UE motor impairment, and FMA-UE scores of 58 66 represented mild UE motor impairment. Sometimes re searchers designate a FMA-UE score to indicate the boundary betw een synergy-dominated and non-synergy dominated movements. In the original article, Fu gl-Meyer provided evidence that a score approximately equal to 60% of the total s houlder/arm score, a score of 13 out of 22 points, was the boundary between synergy and n on-synergy movements [15]. Finley et al. (2005) [44] utilized a maximum FMA-UE score of 15 as inclusion cr iteria for a study of

PAGE 32

18 robotic training because the score represented severe motor impairment in the paretic arm, as demonstrated by limited movemen t within synergy patterns and no voluntary wrist or hand function (p. 684). The ranges of FMA-UE sc ores and their relationship to the categories are frequently used in clinical research as a reference point for describing participants motor behavior and interpret outcomes. Howeve r, whether the scores are being interpreted corr ectly is not known. Contemporary Theories of Motor Control and Recovery To this point I have argued that the va lidity of the FMA-UE item content and the interpretation of the FMA-UE score may be th reatened by challenges to traditional motor control theory upon which the assessment wa s founded. The purpose of this dissertation includes enhancing the measurement propert ies of the FMA-UE if necessary and possible. This necessitates further exploration of contemporary theory to provide a better understanding of the UE motor recovery c onstruct than was provided by traditional theories. UE motor impairment Bernstein (1967) [99] observed that any given motor task could be achieved using many different combinations of muscle activ ation patterns and limb configurations. The motor system chooses from the vast number of degrees of freedom available to it depending on the contextual demands of the movement task (e.g., influence of gravity, inter-segmental joint torques). Bernstein argued that it is unlikely that the motor control system exerts executive control over each m ovement possibility (i.e., one to one control of each degree of freedom) because a vast am ount of storage capacity would be required of the cortex. Instead, Bernstein proposed functional synergies as the more probable motor control strategy. Functional synergies are defined as highly evolved task-

PAGE 33

19 specific ensembles of neuromuscular and skel etal components constrained to act as a single unit so as to reduce the computati onal burden associated with controlling voluntary movements in multisegmented limbs [99] p. 47. In the healthy CNS, functiona l synergies appear to be flexibly organized according to specific task requirements and goals. Us ing kinematic assessment, researchers have shown that distinctive motor patterns emerge in response to task and contextual demands [100-105]. For example, Tromblys lab dem onstrated that neurologically intact individuals displayed unique kinematic patterns of movement when asked to scoop actual coins from a table versus scooping imagined coins [100] and smoother reaching trajectories when reaching to a food choppe r containing actual versus imagined food [101]. Trombly suggested that th e perceptual and cognitive e nvironments in the imagined versus real conditions were different sets affording movement responses particularly patterned to the respective context. Functional synergies also appear to be dynamically organized according to certain task-specific mechanical factors. For ex ample, Murray and Johnson (2004) [106] found that healthy individuals scal ed UE joint torques according to the number of joints involved in a task. For example, healthy i ndividuals generated le ss joint torque during functional tasks requiring primarily a singl e joint movement such as feeding, when compared to tasks requiring multiple joint m ovements such as simulated hair brushing. This research suggests that healthy indivi duals dynamically, and flexibly organize and execute motor programs in accordance with visuo-perceptual and mechanical taskdemands. Do individuals with st roke retain the capacity to flexibly organize and apply functional synergies in task-specific ways?

PAGE 34

20 Latash and Nicholas (1996) [107] sugge sted that an abundant motor control system enables an individual to adapt to specific environmen tal conditions or states of injury. He argued that the abnormal syne rgy is actually a normal attempt to overcome muscle weakness and stroke-related neural damage in order to successfully accomplish a task (i.e., lift the arm). Latash notes that the changed movement pattern reflects the admirable adaptabili ty of the human CNS [108]. Research supports the presence of po st-stroke UE weakness as a possible mechanism underlying the abnormal flexor syner gy pattern. In a paradigm similar to that of the Dewald and Beer lab, Lum et al. (2003) [93] found that individuals with stroke producing unintended secondary to rques in patterns consistent with traditional definitions of flexor synergy. However, unlike the Dewa ld and Beer studies, Lum collected EMG recordings of shoulder and elbow agonists/an tagonists during the experiment. Subjects producing the largest secondary to rques (i.e., less able to di ssociate elbow flexion from shoulder abduction) also exhibited the greate st force production deficits. According to Lum, the results suggested that abnormal post-st roke synergy patterns ar e, at least in part, as a consequence of stroke-related UE weakness. Mercier et al. (2005) [109] examined single and multi-joint isometric muscle activity requiring different combinations of shoulder and elbow flexion/extension. Subjects were included in the study if thei r UE movements were consistent with ChedokeMcMaster Stage 3 [23] (similar to Brunnstroms flexion synergy pattern; participants demonstrated ac tive UE movement, but mostly in flexion). Joint torques were measured with force transducers placed at the forearm and humeral shaft. The results showed that subjects demonstr ated overall decreases in forc e production at the shoulder.

PAGE 35

21 However, the deficits were similar irrespect ive of the direction to which the force was aimed. In subjects with stroke, Zackowski et al. (2004) [110] f ound that a movement requiring simultaneous shoul der flexion and elbow flexion was performed faster and more accurately than an arm movement requiring shoulder flexion and elbow extension Zackowski suggested that activat ion of the biceps to flex th e shoulder might be a strategy to compensate for weakness of the anterior deltoid. McCrea and colleagues (2005) [111] demonstrat ed that the anterior deltoid reached maximal activation (saturation) in persons with stroke much sooner than healthy controls while abducting the shoulder. As sa turation was achieved, subjects flexed the elbow and internal rotated the humerus to reach the target. McCrea found that the impairments were worse in individuals with mo re severe stroke compared to less severe stroke. Taken together, these studies suggest th at shoulder weakness contributes to the observed flexor synergy movement pattern. Following stroke the motor control system has reduced resources with which to successf ully interact with (i.e., adapt to) the environment [112]. Cortical stroke damages the corticospinal pathway and may unmask spared secondary motor trac ts (e.g., vestibulospinal and rubrospinal) [113]. Stroke reduces the capacity to recr uit motor units [114], there by reducing strength [115] altering patterns of agonist a nd antagonist recruitment [33] and reducing interjoint control [116]. Latash and Anson (1996) [108] sugge sted that flexing the elbow (i.e., the observed abnormal flexor synergy pattern) may afford a mechanical advantage by

PAGE 36

22 reducing the length of the lever arm and allowi ng for the individual to adapt to the poststroke state yet still accomplish a task goal (i.e., flex the shoulder). Extending this argument, Carr and Shepherd (2002) [78] propos e that post-stroke recovery can be viewed from a similar task-specific perspect ive. Individuals with strokerelated motor control deficit are able to accomplish tasks if their motor resources are sufficient to satisfy the task-specific demands placed on the motor system. Individuals with less UE motor ability would be more likely to accomplish an easy versus a more challenging task. As these individuals recove r they gain the capaci ty to accomplish both easy and more challenging motor tasks. Post-stroke UE recovery According to Carr and Shepherd, motor recovery following stroke is fundamentally a process of relearning how to move [78]. Most contemporary rehabilitation frameworks are predicated on th e assumption that in order for a patient to improve his/her motor skill (i.e., recover from stroke-related hemiparesis) the patient must practice motor skills [79]. Pr inciples of neural plasticity first researched in animal models translated to humans s uggest that healthy brain neuron al connections and cortical maps are dynamically remodeled by experience [6, 117]. Post-stroke UE recovery may be enhanced with intense task specific-practice that engages the system to actively solve motor problems [118, 119]. Note the conceptual dissimilarity betw een this contemporary understanding of recovery and its more tradit ional counterpart. Traditionall y post-stroke recovery was viewed as a passive (from the viewpoint of the patient) process. Rehabilitation interventions were provided to patients who la rgely assumed the role of passive recipient of the intervention, in order to facilitate (or inhibit) an internal neurophysiological

PAGE 37

23 process [120]. In contrast current motor cont rol researchers and theorists view post-stroke recovery as an active process in which the in dividual, given his/her internal neural and muscular deficits, re-learns how to interact with the external environment to satisfy the demands of the activity [70, 121]. This may signal that recovery is charact erized by enhanced ability to execute movement patterns with the appropriate tr ajectory, acceleration and relative muscle forces necessary to meet the temporal and spatial requirements of a task [66]. Roher (2002) [122] demonstrated that with one m onth of robotic training, individuals with chronic stroke were able to reach to a targ et with a smoother hand trajectory velocity profile, suggesting increased ab ility to move independently of the elbow-shoulder flexor synergy. Ellis and Dewald (2005) [123] f ound that with practice, individuals with chronic stroke were able to modify the abnormal coupling of isometric shoulder and elbow joint torques evidenced by increased ability to isolat e joint torque production at a single joint during a multijoint task. Mi chaelson et al. (2006, 2004) [9, 124] and Thielman et al. (2004) [125] have shown that with training, individuals with severe to moderate stroke demonstrate improved shoulde r-elbow inter-joint coordination evidenced by decreased trunk recruitment and shorter (straighter) hand path trajectories during reaching tasks in various parts of the reachable workspace. Taken together, the above studies suggest th at post-stroke recove ry is characterized, at least in part, by the patien t relearning the rules that gove rn motion [107]. Individuals with stroke become more able to contro l each joint during a multijoint task. Patients improve, using ICF terminology, the ability to coordinate single and multiple joint motions [62].

PAGE 38

24 Post-stroke recovery of functional UE motor abilities The research presented to this point supports the view that post-s troke recovery is a process in which individuals gain more UE mo tor ability with regards to overcoming task demands. More functionally oriented activitie s requiring interaction with objects and at increased speeds of movements. Task-specifi c increases in neural, mechanical and/or contextual requirements concomitantly increase the computational demands on the motor system [112]. Interacting with an object not only increases the mechanical load on the limb but also increases the perceptual proces sing requirements of the task, e.g., attention to the characteristics of th e object [126], feedforward adju stments of posture [127] and increased reliance on feedback to correct moveme nt errors [70]. It is likely that these tasks are more difficult and recover later than movements without functional endpoint. This appears to be the expectation of F ugl-Meyer who found a strong correlation between the FMA-UE score and ADL performance [15] However, evidence suggests that the progression from reduced body function im pairment (i.e., improved intra-limb coordination) to recovery of activity performa nce is neither linear nor clearly understood [128]. The association between motor function im pairment assessed by the FMA-UE and functional assessments of activ ities of daily living is both supported and debated in the literature. For example, both Filiatrault et al. (1991) [21] and Platz et al. (2005) [56] have shown a low correlation (Spearman r = 0.04 and 0.08) between the FMA-UE score and the score on the Barthel Index [129], an assessm ent of ADL performance. Francisco et al. (1998) [39] found that four individuals with stroke, all of whom exhibited palpable wrist extension at baseline, dem onstrated concurrent gains in FMA-UE and Functional Independence Measure (now known as the FIM) motor scores following inpatient

PAGE 39

25 rehabilitation with usual care augmented by wr ist EMG electrical stimulation. In contrast, Winstein et al. (2004) [130] showed no re lationship between gains on any impairment measure (e.g., the FMA-UE) and changes in the FIM after a tw o week intervention. Shelton (2001) [131] showed that admissi on motor impairment (measured by the FMAUE) predicted discharge disability (measur ed by the FIM), however the researchers note that much of the gain in FIM score ca me as a result of increased use of the less affected limb to perform the tasks. Sheltons observation is consistent with Winsteins argument [59, 130] that the FIM like other global assessments of ADL su ch as the Barthel Index, are not sensitive to impairment level changes in UE motor ab ility. This is because neither the FIM nor Barthel directly assess the f unction of the hemiparetic arm. A client is able to use compensatory one-handed strategies to acco mplish tasks with the unaffected limb and still receive a high score on these assessmen ts. The association between the FMA-UE score and scores of functionally oriented a ssessments where the assessment requires use of the hemiparetic arm is stronger. The FMA-UE was highly correlated (Spearman r = 0.92) with scores on the Action Research Arm Test [56], the Wolf Motor Function Test (Spearman r = 0.57 0.68) [26], the Arm Moto r Ability Test (Spearman r = 0.92) [97] and the Lindmark (1988) [89] scale. Each of these assessments requir es the patient to use the hemiparetic arm for functiona l reaching or grasping tasks. Mercier et al. (2004) [132] notes that it is possible that the relationship between body function impairment and functional may not be linear across different levels of ability. For example, she suggests that a task requiring grip strength (e.g., lifting a pitcher filled with water) will be performed at a lower stage of the recovery process than a task

PAGE 40

26 requiring a precision grip (e.g., picking up a sm all object from a table). Winstein (2004) [130] found that partic ipants in the less severe group (s tratified by Orpington Prognostic Scale score [133]) showed greater improve ments in the Functional Test for the Hemiparetic UE (an assessment comprised of 17 graded tasks assessing integrated function of the UE) [134] than clients in the more severe group. Feyes et al. (1998) [135] found that clients with more severe motor impairment demonstrated significant gains on the FMA-UE but the re sults did not generalize to concurrent changes in the Action Research Arm Test or Barthel Index. It is possible that individuals with less UE motor ability do not have enough motor ability to perform even the simplest functi onal tasks. Chae et al. (2003) [97] found that individuals with higher scor es on the FMA-UE hand subscale demonstrated a strong correlation with the overall AMAT score. Chae suggests that these individuals most likely had enough UE motor ability that enab led them to successfully complete AMAT items (e.g., cutting meat, dialing a telephone). In contrast individuals with less UE motor ability (a lower FMA-UE score) likely di d not have enough motor capacity to accomplish the more functional AMAT items. Pang et al (2006) [42] found a similar relationship between participant ability a nd functional performance. In this study, more severely impaired clients showed significant pre to post intervention changes in FMA-UE scores (2.5 2.3 points), but no change in WMFT score (-0.3 0.5 seconds). Moderately impaired clients showed concurrent ch anges in both assessments (FMA-UE 7.4 3.8 points and WMFT .5 6.2 seconds). The least severely impaired clients demonstrated little increase in FMA-UE score (7.4 3.8 points) but the greatest gains on the WMFT

PAGE 41

27 test (-6.6 1 6.4 seconds). In general, the evidence suggests that indivi duals with greater ability will show changes in both domains, less able people will not. Clinical significance of the FMA-UE score The importance of the studies reviewed above is that the FMA-UE score is difficult to interpret with regards to its functional relevance. That is, although individuals with stroke may recover the ability to coordinat e single and multiple joint movements as evidence by improved FMA-UE score, the functi onal meaning of this improvement is not always clear. Issues of clinically significant or clin ically meaningful changes in assessment scores have become paramount in rehabil itation research [61, 136, 137]. The relationship between statistically significan t research results, and cl inically significant clientoriented functional improvement is not alwa ys apparent [136]. B onnifer and colleagues (2005) articulated this concern; Although this study showed significant incr eases in FMA-UE (scores) after the constraint induced movement therapy (C IMT) intervention, these findings do not suggest that a 3-week CIMT program restores motor ability to pr e-stroke levels. It is not clear how increases in scores on any of these measures translate to real-world functional abilities. For instance, a peak score of 33 on the FMA-UE score does not suggest that a person ca n perform functional task s using the involved upper extremity. On the contrary, the arm is stil l quite impaired and most likely used as an assist at most for some activities. [51] Conceptually, there are two approaches for determining what constitutes significant change: distribut ion and anchor (or criterio n) based approaches [61, 137, 138]. Distribution-based methods define an important change score as one that exceeds a pre-defined level of error. These techniques used to calculate significant change include minimal detectable change, th e minimal important difference, effect size and/or one of numerous responsiveness sta tistics. The drawback of all of the se

PAGE 42

28 approaches is that the value obtained in th e analysis is not a fixed property of an assessment, instead varying w ith the setting of the study and characteristics of the sample. Different values of significance might be obtained in different groups of patients based on their initial level of impa irment. Ceiling or floor effects can mask change and effect the calcula tion of a responsiveness or e ffect size value. Moreover, these methods require a judgment of what is important change. But, who should be the judge, client, therapist, physic ian, payer or test developer? Therefore it is unlikely that there exists a single score to define significant change ac ross all samples and contexts. In contrast, anchor-based methods rela te the change score to a functionally relevant criterion pre-determined by client or clinician. This cr iterion anchors the meaning (i.e., significance) of the change. Both approaches have been applied to interpretation of FMA-UE change scores [61]. According to Van der Lee et. al. (1999) [ 47] the literature cont ains no criteria for interpreting a FMA-UE change score with re gards to its minimal clinically important difference (MCID). Van der Lee et al. (2001) [24] determined that a 10% change in the overall FMA-UE score made measurement error unlikely thus allowing for the conclusion that real change occurred. Using Van der Lees criteria, McCombe-Waller and Whitall (2005) [139] interp reted their 8% gain in FM A-UE score as approaching clinical relevance. Based on an effect size calculated in a pilot study, Feyes et al. (1998) [135] determined that a 10% change in the overall FMA-UE score was clinically relevant meaning that particip ants moved from a more impair ed to a less impaired stage of recovery. Pang et al. (2006) [42] explored the clinical significance of a FMA-UE change score using external criterion. In this study the researchers administered a

PAGE 43

29 satisfaction survey to participants in a commun ity based exercise program. Clients in this intervention showed a 0.71 0.81 effect size fo r pre to post test FMA-UE score change, and concurrently reported increased ability to perform simple functional tasks such as now able to switch on/off lights. Pang not es that the intervention related changes appear to be clinically significant from the patients perspective. Summary In the section above I have identified three concerns w ith the FMA-UE. First, it is possible that some items, specifically the re flex items, are not consistent with the construct represented by the other items (i .e., voluntary UE motor ability). Second, many assumptions have been proposed regarding the meaning of the FMA-UE score. These assumptions have gone untested. Finally, th e relationship between the FMA-UE score and performance of functional reaching, gr asping, lifting activities with the more affected UE is not obvious. Item Response Theory offers a method to test specific hypotheses addressing each of these concerns. The following section will review Item Response Theory in general, the Rasch measurement model specifically, to establish the bac kground for using this measurement framework to explore these areas. The Rasch Measurement Model Measurement Measurement is defined as determining the amount or quantity of something. Mass, length, and time are considered funda mental quantities of measurement in the physical sciences [140] and are measured with devices accepted as objective measurement tools. An objective measurem ent remains constant and unchanging across

PAGE 44

30 the persons measured and is described in a unit that maintains its size as the unit is repeated [141]. Ideally the process of measuring a clients UE motor function should be similar to the process of measuring lengt h, temperature or distance. That is, UE motor function should be quantifiable using an assessment tool that is structured to measure UE motor skill in the same way that a ruler is structured to measure length or that a thermometer is structured to measure temperature. Bond a nd Fox argue that in order for a clinical assessment to objectively measure a skill, e .g., UE motor function, the assessment tool should display markings (e.g., assessment items ) that clearly describe less and more of the skill. The order of the markings shoul d represent the acquisition of the skill in a pattern consistent with it s recovery or development [142]. Also, an objective measurement of UE motor function shoul d connect the numbers produced by an assessment with its content [143]. Item Response Theory The increasing need for meaningful a nd psychometrically sound rehabilitation assessments has led many to apply modern measurement theoretical statistical methods (IRT) to the construction, re-design, a nd scoring of clinical assessments. IRT measurement models are comprised of a fam ily of mathematical models [53]. Each model estimates an item difficulty parameter. The two-parameter logistic models (2PL) also estimate an item discrimination paramete r, and the three-parameter logistic models (3PL) estimate item discrimination and gue ssing parameters [144]. The Rasch model is usually considered a one-parameter IRT model, although the technicali ties of this are the subject of ongoing debates in the literature.

PAGE 45

31 IRT, as its name implies allows for an item-oriented rather than test-oriented analysis of an assessments measurement prope rties. The appeal of IRT is that personability and item-difficulty ar e calculated on the same m easurement scale which means that a score can be direct ly related to the probabilit y of item responses [53]. Potential Advantages for Using the Rasch Model to Explore the FMA-UE Measurement Properties Rasch analysis has been widely utilized to construct and/or refi ne clinical health care assessment tools [for example [145-152]]. The model is a powerful framework with which to address the measurement properties of the FMA-UE: specifically to test (1) the validity of the FMA-UE item content, (2) the re producibility of its item structure, and (3) functional interpretati on of its score. Validity of the item content Using Rasch analysis an item-by-item analys is is conducted to test whether each item contributes to the measurement of UE motor ability. This is accomplished by examining the dimensionality of item set and the item difficulty hierarchy. In order to create a legitimate measure, all items must contribute to the same construct, that is, the assessment must be unidimensional [143]. Unidimensionality means that a single trait or skill accounts for performance of assessment items [153, 154]. Unidimensionality should remain invarian t each time the assessment is used because without unidimensionality it is not clear what co nstruct the items define or what the score means [153]. All IRT models assume th at the data undergoing analysis are unidimensional [53]. There is no single satisfactory method to asse ss the unidimensionality of an item set and so researchers often use a variety of techniques to assess it [154-156]. Principal

PAGE 46

32 Components Analysis (PCA), a cl assical test theory statistic al method, is often partnered with Rasch analysis for this purpose. The intent of PCA is to reorganize multivariate data into a limited number of components (factors or dimensions) so that each component captures a substantial amount of the overall variance within the dataset [157]. PCA can be applied to either the original raw scor e data or the Rasch-derived residuals [154]. PCA produces an index of variance statisti c for each factor, an eigenvalue, to describe how well items measure a single co nstruct. Techniques for interpreting the eigenvalue(s) vary. The Kaiser rule [157] is typically employed as the criteria for determining the number of factor s to retain. According to th e Kaiser rule, factors with eigenvalues > 1.0 are retained, and each is t hought to represent a distinct construct. Cattells scree test is also used to interp ret eigenvalues [157, 158]. The scree plot is a graph of all eigenvalues in their decreasing order. The plot looks like the side of a mountain, and "scree" refers to the debris fall en from a mountain and lying at its base. According to the scree test, factors are retained if their data points occur prior to the point at which the mountain ends and the deb ris begins. Eigenvalues of scree are considered trivial because they are thought to detect noise (e.g., measurement error) rather than actual multidimensionality. Only points that markedly deviate from a linear trend are valuable to the an alysis [157]. McHorney and C ohen (2000) [159] suggest that if greater than 20% of the variance is acc ounted for by the first factor in a PCA, unidimensionality is established. However, C ook et al. (2003) [160] argued that this often cited criterion is a perpetuat ion of an incorrect interpre tation of an earlier work concerning the stability of item parameter estim ates in multidimensional tests rather than suggesting a criterion for esta blishing unidimensionality. Fi nlayson and Mallinson (2005)

PAGE 47

33 [161] suggest that a factor must contribute a minimum of 10% of e xplanatory variance to be considered a unique factor in modeling the data, while Haley and colleagues [162] suggest that meaningful factors are those that each account for 5% of the variance. Parallel analysis provides an alternative, empi rically derived criterion [163]. In parallel analysis dimensionality is determined by retaining only the number of eigenvalues from the real data that exceed the eigenvalues esti mated from a parallel data set of the same size that is randomly simulate d from a Rasch model [162]. Once the number of factors has been identi fied the meaning of the factors must be determined. PCA typically involves a second st ep, factor rotation. Factor rotation is a mathematical technique intended to re-expre ss factors in a more simple structure to facilitate their interpretation. A factor is interpreted (i.e., subjectively labeled as per the construct it represents) by determining what trait is shared among the items that load strongly onto the factor [164]. An application of PCA to the Rasch-derive d standardized residuals is also intended to elucidate potential multidimensionality in the data. Residuals are the data not explained by the Rasch model. If the da ta are unidimensional there should be no dimensionality remaining in the residuals once the Rasch model has extracted the dominant factor [154]. The pres ence of distinct factors within the data is determined by visually inspecting the factor structure plot produced by the analysis [165]. This diagram cross plots the standardized residual loading on each factor against the item difficulty calibration. If the data is unidimensional the graph should show a random distribution of the items. Multidimensionality is evident when items cluster together in obvious groupings [166].

PAGE 48

34 Smith (2002) [153] acknowledges that defini ng unidimensionality as the existence of one (and only one) skill accounting for perf ormance on assessment items is likely too strict of a definition. Smith recognizes the likelihood of many f actors (e.g., cognitive, psychological or physical) influe ncing test-takers responses during assessment. He states that unidimensionality will hold as long as the same factors affect the performance on each item consistently throughout an assessm ent. Muraki and colleagues (2000) [167] agree, and state that it is difficult to sa tisfy unidimensionality when IRT models are applied to performance assessments, becau se performance assessments, by their very nature, require examinees to complete tasks th at demonstrate ability to apply skills to a real life situations. These situations are co mplex, thus it is likely that the data is influenced by construct-irrelevant varian ce such as environmental effects, practice effects and examinee fatigue. Murakai argues that unidimensionality is violated to some extent in every dataset, and even more in data from performance assessments. The authors suggest that as long as the violations are not extreme IRT analysis appears to be robust. One way to interpret the apparent presence of more than one factor in data was exemplified in a recent study by Hart and colleagues (2006) [168]. In this study the researchers used factor analysis to exam ine the dimensionality of the Moss Attention Rating Scale (MARS). The analysis suggested th e presence of 3 factors in the data. The authors interpret the results as meaning th e instrument measures a single overarching construct, attention, but is comprised of 3 inter-correlated sub-dimensions; initiation; restlessness/distractibility; and sustained/consistent attenti on. The authors argue that the

PAGE 49

35 sub-dimensions are not unique constructs, bu t reflect different ways of expressing the overarching construct. Rasch analysis takes the analysis of unidi mensionality one-step farther in the sense that it tests the extent to which persons have responded to assessment items in an expected manner. Using Rasch analysis, the unidimensionality of an item is evaluated with infit and outfit statistics. Under the Rasch model, each person with a given level of UE motor ability is expected to do well (score higher) on rela tively easy tasks, and have more difficulty (progressively lower scores) on tasks of in creasing difficulty [142]. The model expectations are asse ssed through item goodness-of-fit st atistics. Fit statistics are reported as the mean square (MnSq) of th e item residuals, where a residual is the difference between a persons actual respons e to an assessment item and the response expected by the Rasch model. Consequently th e MnSq statistic has an ideal value of 1.0 with departures from this value indicating potential departures from unidimensionality. Fit statistics are also reported in a standardized form (ZST D), which is the transformation of the mean square of the resi duals to a statistic with a z-distribution [153]. Two types of fit statistics are reported. The infit statistic is sensitive to items th at depart from model expectations when the item difficulty is of a similar level as person ability. The outfit statistic is sensitive to items that depart fr om model expectations when the item is much harder or easier than a persons ability leve l (i.e., outliers). The acceptable criteria for unidimensionality depends on the intended pur pose of the measur e and the degree of rigor desired. For clinical observations, Wright and Linacre [169] suggest reasonable ranges of MnSq fit values between 0.5 and 1.7 associated with standardized z-values (ZSTD) less than 2.0 for clinical observati ons. High values indica te that scores are

PAGE 50

36 variant or erratic, suggesting that an item belongs to a construc t that is different from that represented by the other items of the instrument. High values c ould also indicate that the item is being inaccurately scored. A low MnSq value suggests that an item is failing to discriminate individuals with different levels of ability (i .e., with different amounts of motor ability) or that an item is redundant (i.e., other items represent similar motor challenge). Items with high MnSq values repres ent a greater threat to construct validity. Item difficulty hierarchy An assessment intends to measure a skill. Rasch analysis paints a picture of the skill by arranging the assessment items along a hierarchical continuum (an item map) from least difficult to most difficult. Th e item difficulty hierar chy is therefore a visual representation of the abstract skill the assessment is intended to measure. Ryall et al. (2003) [170] argues that examining the agreement between the a-priori predicted hierarchical ordering of items and the empirical item order is one way to test the validity of the theoretical construct underlying the a ssessment. The empirical disordering of one or more items may indicate that the item(s) contain(s) flaws, or that the theoretical construct governing item orde r is deficient [171]. Rasch analysis places person ability and item difficulty on the same linear continuum measured in logits. A logit is the natural logarithm of the odds of a person being able to perform a particular task to the probability of being able to perform the task [172]. Logits of greater magnitude repres ent increasing item difficulty (and person ability). Because items and people are measured on the same scale, person ability scores reflect what a person ca n or cannot do on the basis of item content. The item hierarchy affords a method to clea rly interpret the assessment score which is beneficial for measuring person ability a nd informing theory [ 173]. Because the items

PAGE 51

37 are ordered according to their difficulty level, the expected next st ep of the clients progress is clearly visible. The Gross Motor F unction Test (a test of motor development) item map presented in Russell et al. (2000) [174] provides an exam ple of how the item map connects a score to a description of beha vior. In this example, a GMFM score of is located towards the higher end of the developmental continuum. Because developmental behaviors (i.e., assessment items) are also positioned on the continuum, the score of marks the location at which the child is able to accomplish the behaviors below that level (i.e., easier items) but is not yet capable of accomplishing the behaviors above that level (i.e., harder items). Similar item maps (also termed keyforms) are used for scoring the Pediatri c Evaluation and Disabi lity Inventory [175] and the ABLEHAND [176] questionnaire. The item hierarchy is useful to ascertain whether an assessment score is interpreted in the way expected. White and Velozo ( 2002) [145] applied Rasch analysis to the Owestry Low Back Pain Disability Questionnair e to determine if the often used Owestry classification scheme was empirically supporte d. By matching assessment scores to the item hierarchy, the researchers learned that the Owestry classification categories did not describe actual client beha viors in the way intended by the classification taxonomy. The researchers suggest re-categor ization of the resp onse scale to improve the meaningfulness of the score. Sabari and colleagues (2005) [177] a pplied Rasch analys is to the Motor Assessment Scale [178]. The researchers found that the Rasch-derived item hierarchy both supported and challenged the underlying co nceptual foundations of the assessment. For example, the item hierarchy revealed that overhead reach was less difficult in supine

PAGE 52

38 than while standing, a finding consistent with contemporary motor control principles that movement difficulty is governed by extern al mechanical factors, e.g., gravitational torque. The item hierarchy also showed di screte, proximal hand movements were of similar difficulty as some functional activ ities requiring simultaneous control over multiple movements at a variety of joints, a finding that challenged theoretical expectations. Chiu and colleagues (2006) [179] applied Rasch analysis to the Dynamic Gait Index. The hierarchical orde ring of locomotor behaviors wa s consistent with clinical expectations. Penta et al. (1998) [176] utilized Rasch analysis to develop the ABILHAND questionnaire. Th e item hierarchy defined manual ability, a trait previously not well explored in motor cont rol theory. For example, the most difficult items elucidated characteristic behaviors (stre ngth, mental effort, dext erity) that added to an understanding of the construct. For the construct underlying the FMA-UE, we might expect flexor synergy items to represent easy items along the continuum of UE motor ability, while hand items which involve more intricate prehension patterns and isolated joint m ovements to represent more challenging items across the conti nuum. Comparing the Rasch-derived item hierarchy to Fugl-Meyers expect ed course of recovery will pr ovide an opportunity to test traditional theoretical expectations of recovery. Reproducibility of the item structure Rehabilitation researchers are often inte rested in whether an intervention is effective. To this end, researchers choose an appropriate assessment to score an outcome of interest and compare scores between and/or within groups. Regard less of the statistic used to assess effectiveness (e.g., comparison of group means, correlatio n test, etc.) it is implicitly assumed that the measurement prope rties of the assessment are identical each

PAGE 53

39 time the assessment is used. Gluck (2001) [ 180] points out that this often untested assumption may negatively impact the reliability of an assessment. Bingenheimer et al. (200 5) [181] argues that measurement equivalence is necessary for valid interpretation of assessment scores. Specifically, an inst rument should display item and scalar equivalence meaning that an asse ssments items should assess the same trait each time it is given (item equivalence), and identical scores s hould always have the same interpretation (scalar equivalence). The items of an assessment are the operational definition of the trait the assessment is intended to measure, and the score quantifies how much of the trait the test-taker possesses. If an item varies with regards to its meaning or the score varies with regards to its interp retation, the assessment is not an accurate measure of the trait. Bingenheimer argues that without convincing evidence of item and scalar equivalence, there is no basis for co mparing scores on an assessment between or within groups. Differential Item Functioning (DIF) is a stat istical procedure in which the items of a test are examined, one at a time, to ascer tain whether test-item characteristics are consistent irrespective of test-taker characteri stics. The presence of DIF means that a test item measures a unique feature (trait) of one group of people when compared to another group of people, or that the test item was not administered correctly [182]. There are various approaches for examini ng DIF (see [183] for review), the most commonly used is the Mantel-Haenszel (MH) procedure. MH begins with the assumption that the ratio of correct to incorrect responses is the same between the groups for each item. To test this hypothesis, test-takers are split into two groups (e.g., time 1 and time 2) stratified by ability level (tes t score). Item difficulties betw een identical strata of each

PAGE 54

40 group are compared, item-by-item, using the odds ratio statistic An odds ratio value of 1 means that the item displays no DIF. Based on the magnitude of the values deviation from 1, items are rated as having negligible, intermediate or large DIF. Proponents of the MH procedure support excluding items showing large DIF, while leaving to clinical judgment decisions regarding the inclusion or ex clusion of items with intermediate DIF. Rasch analysis employs a relatively stra ightforward DIF procedure in which the item difficulty parameters between two gr oups (or two time points) are directly compared. This procedure is similar to the above MH procedure, but is independent of sample-dependent person ability (i.e., i ndividual raw scores are not used in the mathematical calculation [184]). The Rasch procedure has reliably detected DIF with small sample sizes (N < 200) [183]. In the healthcare liter ature DIF analyses have been useful to study item equivalence (i.e., to detect potential test item bias) am ong persons of different cultures and diagnostic groups. For example, using DIF analyses, Az ocar and colleagues (2003) [185] learned that Latino-Americans were more likely to receive a higher score on some items of the Beck Depression Inventory than non LatinoAmericans, potentially flagging a distinct Latino cultural interpretation of the items. Az ocar recommended a reconsideration of the assessments Spanish translation in order to facilitate cross-cultural comparisons of assessment scores. Tennant and colleagues (2004) [186] found that certain Functional Independence Measure self-car e items displayed DIF. The researchers learned that culturally unique ways of bathing and gr ooming influenced the way the items were administered among European countries. To facilitate cross-country comparisons of scores, the assessment scoring procedure wa s adjusted for those items. Roorda and

PAGE 55

41 colleagues (2004) [187] found that the hierarch ical item ordering of a stair climbing assessment was different for amputees versus non-amputees. The researchers recommend using a shortened assessment, with DIF items removed, if the intention is to compare scores between the two diagnostic groups. Da llmeijer and colleagues (2005) detected DIF in the Functional Independence Measur e [188]. When comparing Functional Independence Measure scores among individual s with various neurological diagnoses (stroke, multiple Sclerosis, and TBI) DIF was detected in 7of 11 motor items and 4 of 5 cognitive items. For example, persons with MS found the ite m dressing lower body more difficult than did persons with stroke and persons with st roke found the item problem solving much more difficult than did persons with MS The authors caution that FIM scores cannot be compar ed across these patient groups. Instead of testing the measurement equiva lence of an assessment between cultures or diagnostic groups, it is also useful to employ DIF analyses to compare the measurement equivalence of an assessment between testing occasi ons within a single group. The use of DIF analyses for this purpose has precedence in the literature. Tesio et al. (2003) [182] unexpectedly detected DIF among several FIM mobility items (transfer to a chair, locomotion, and stairs) when admission scores were compared to discharge scores. The items were easy for a group of orthopedic patients with mobility restrictions at admission (e.g., no lower extremity weight bearing due to orthopedic precautions). However, the items became mor e difficult for this group at discharge after the mobility restrictions were removed. This counterintuitive result led the staff at this facility to examine the manner in which the assessment was scored. They learned that the admission score was based on a prediction of the patients expected outcome, while

PAGE 56

42 the discharge score was based on actual observa tion of patient performance. In essence, the assessment was measuring a different cons truct at patient admission (e.g., ability of the staff to predict patient stat us) versus patient discharge. Finlayson and Mallinson (2005) [161] a pplied DIF analysis to the primary functional outcome measure used in a la rge longitudinal stud y of aging and ADL performance in various health care settings. Rasch analysis of a 24-item ADL assessment was conducted at three time points over th e 13-year study. Although DIF was detected among 4 items, the researchers were unable to discover a specific re ason for it. Because statistically significantly differences in item difficulty do not al ways translate into practical differences in person ability scores [189], the researchers measured ADL ability using an assessment in which the 4 unstabl e items were removed and compared the results to measurement using the full assessm ent. The analysis suggested that DIF had only minimal impact on the Rasch-ability meas ures over time or acr oss settings, leading the researchers to continue using the unadjusted entire assessment. The presence of DIF can have significant impact on the validity of an assessment for measuring person ability. Linacres (1994) [190] study provided a classic example of the impact of item instability on the meas urement of patient change. Using Rasch analyses, Linacre found that when the 13 mo tor and 5 cognitive items of the Functional Independence Measure were combined into a single 18-item assessment, the admission and discharge item difficulty ca librations of the items were not equivalent. He suggests that the presence of variant item difficulty measures indicated that a single pattern of disability could not be iden tified when comparing patient admission scores to discharge scores. That is, the meaning of the scor e was dependent on the time point (admission

PAGE 57

43 versus discharge) at which the assessment was administered. Linacre demonstrated that the statistical validity of th e scale was improved if the motor and cognitive items were separated into sub-scales. Functional interpretation of the FMA-UE score As previously noted, one of the primary problems facing uses of the FMA-UE is the functional interpretation of its score. IRT provides the theory and method for linking the FMA-UE to a more functionally orie nted assessment [143, 191, 192]. Linking is a method that re-calibrates separate assessmen ts onto a common measurement scale. The theory behind this process has been termed scale free measurement [143]. Scale free measurement means that what one measures is independent of the instrument used to make the measurement. Applied to post-str oke UE rehabilitation, scale-free measurement theorizes that an abstract tr ait (e.g., UE motor ability) can be measured by an infinite number of items representing the trait. Any assessment used to measure this trait is simply a subsample of this item pool. Item pools (termed an item bank[193] ) are the foundation of Computerized Adaptive Testing. This testing procedure is co nsidered adaptive because different testtakers answer different sets of questions de pending on their level of skill with regards to the content (construct) being tested. The co mputer selects question sets from the item pool, which are items that have been previous ly validated and calibrated. A persons test score is derived from responses to the items ad ministered meaning that test takers are not required to complete the same items to r eceive comparable scores. Item banking and computerized adaptive testing is widely used in educational research (the Graduate Record Exam is a computerized adaptive test) and has been increasing ly applied to health care outcomes research [159, 194].

PAGE 58

44 Items within an item pool are calibrated to a common metric using one of two processes. The first involves combining the items together for a single analysis that concurrently calibrates difficulty parameters. A variation of this method is to estimate difficulty parameters for one set of items firs t, then use these difficulty estimates to fix (anchor) this set of items in a combined an alysis with the remaining items [193]. Hanson et al. (2002) [195] suggests that at least for 2PL analyses, co ncurrent calibration realizes less error in item parameter calculation than separate calibrations. Kim and Cohen et al. (1998) [196] suggest that the two methods yiel d nearly identical re sults in situations where a large percentage of items are common to the participant sample. Linking two existing assessments is anothe r way to create a common item pool. If two assessments measure the same underlying tra it, then the items of each assessment are derived from the common item pool and therefore the assessments can be mathematically linked. Linking establishes the relationship between different assessments, therefore allowing direct comparison of items and sc ores [167]. Fisher (1995) [192] linked two assessments of physical function to a common scale with Rasch analysis. Thirteen motor FIM items and Patient Evaluation and C onference System (PECS) items were cocalibrated to single scale that allowed a pati ents score on one assessment to be directly translated to a score on the other assessment. Fisher suggested that the assessment scores could be reported in rehabits, a un it of measurement representing the common measurement scale to which the assessments were linked. Costner and colleagues (2004) [148] linked five existing widely used ADL and IADL assessments (FIM, Minimum Data Set (MDS), MDS-post acute care, OASIS and Physical Functioning ) for the purpose of tracking recovery across the c ontinuum of care. The conceptual basis

PAGE 59

45 underlying the link was that the assessments each contain daily living items sharing a common requirement for upper limb and hand skills. The researchers concurrently calibrated the items from the existing assessments, retaining 62 items in a unidimensional item pool. Fisher and Eubanks (1997) [197] concurre ntly calibrated the physical functioning subscales of the Medical Outcomes Study S hort Form 36 physical functioning subscale (PF-10) and the Louisiana State University Health Status Instruments Physical Functioning Scale (PFS) to a common metric using Rasch analysis, demonstrating that the method was feasible. Segal and colleag ues (1997) [198] used Rasch methodology to link the PF-10 and Functional Independence Measure motor items. Six PF-10 and 4 FIM items were co-calibrated to form a ne w assessment for post-stroke outpatients. As separate assessments floor effects in th is population hampered the PF-10 and ceiling effects hampered the FIM. However, the linked assessment demonstrated excellent measurement properties for measuring long te rm physical functioning in this population. The ICF provides a conceptual framewor k that identifies UE motor behaviors relevant to a single domain (i .e., body function and/ or activity performa nce) and specifies a relationship between domains. This ope rationalizes health care dimensions and provides a framework for rehabilitation outcome measurement within and across dimensions [199]. Linking a post-stroke UE motor assessment from the body function domain to an assessment from the activity dom ain sharing similar UE motor behaviors is a way to test whether the domains are distin ct or related constructs in post-stroke recovery. Moreover, linking is a way to direct ly investigate the f unctional relevance of

PAGE 60

46 reduced body function impairment because linking connects the assessments to a common metric. The Wolf Motor Function Test (WMFT) [ 26, 200] is a logical choice for linking to the FMA-UE. Both the FMA-UE and the WM FT measure UE motor ability: the FMAUE is a measure of UE body function impairme nt, while the WMFT is a measure of UE activity performance restricti on. The WMFT is an assessmen t of post-stroke UE motor function [201]. It is intended to measure a wide range of func tional UE tasks in individuals with mild to moderate stroke-r elated UE motor impairment [201]. Its items are arranged in a sequential order according to joints involved (shoulder to hand) and level of difficulty (gross movements to fine motor movements) [26]. The first seven items involve simple limb movement s without functional endpoint (e.g., extend elbow to the side on a table top, lift hand to table ), and the remaining eight items involve performance of common tasks (e.g., lift soda can, flip cards ). The WMFT has been shown to have high interrater reliabil ity (intraclass correlation coefficients, 0.88), internal consistency (Cronbach 0.86), and test-ret est reliability ( r 0.90) [26, 201]. Placing the FMA-UE and WMFT items on the sa me ruler as client ability allows the individuals Rasch ability score to be directly related to hi s/her probability of performing a particular set of movements or tasks. This permits one to describe a persons motor ability in terms of specifi c movements or tasks that he can/cannot perform. Furthermore, changes in score can be directly linked to acquisition of more difficult movements or tasks. By linking the FMA-UE and WMFT in this manner one should be able to convert improvement sc ores into performance of specific UE movements and tasks (e.g., moving from elbow extension to grasping a soda can) thereby

PAGE 61

47 identifying a critical threshold at which im pairment reduction translates to functional ability. Translating improvements in UE ability from impairment to function is the foundation of documenting the clinical effec tiveness of restora tive rehabilitation. Criticisms of the Rash Measurement Model Sample size The purpose of a sample size justificati on is to predict how many subjects are needed to satisfactorily answer the research objective. In traditional statistics, a power analysis guides the researcher in determ ining the appropriate sample size for the objectives of the study. With IRT analyses sample size guidelines are not as straightforward. Linacre (1994) [202] suggests that as few as 50 subjects may be an appropriate sample size for a valid analys is using the one parameter Rasch model, however Hambleton (1989) [203] suggests th at at least 200 subjects are necessary for Rasch analyses. Embretson and Reise (2000) [ 204] indicate that the two-parameter model may require 250 to 500 participants while Ha mbleton indicates that a minimum of 500 subjects are necessary. Hambleton also sugge sts a minimum sample size of 1000 subjects if the researcher intends to use a three-parameter IRT model. The purpose of the study affects the sa mple size recommendation. According to Reeve and Fayers (2005) [205], if the aim of a study is to examine the properties of an existing assessment, one does not need a larg e sample size to describe it clearly. However if the study purpose is to calibrate ite ms for test development then larger (over 500 subjects) sample sizes are needed. Lai et al. (2003) [206] recommends that test equating requires a sample size of 400 for the Rasch model and 1,500 for a threeparameter model.

PAGE 62

48 There are very few empirical stud ies influencing IRT sample size recommendations. Holmans (2003) [207] si mulation study results suggests 450, 90, and 40 participants are needed in each arm of a ra ndomized trial in order to detect an effect size of 0.2, 0.5, or 0.8 respectively with a signi ficance of 0.05 and a power of 80% using a two parameter IRT analysis. In comparison, a t-test powered to detect the same effects would require 394, 64, or 26 patients per arm of the study. Holman also indicates that the number of informative items in the assessmen t affects the sample size. That is, more items with more precise information require lower sample size. Obviously these sample size guidelines may make using IRT methods prohibitive for many health care researchers. In Rasch m easurement, the researcher seeks to obtain stable person, item and rating scale estimates A stable parameter is one that has low standard error. Linacre indicated that 0.3 logits is the best that can be expected for most variables and that the sample size needed to have 99% confidence that no item calibration is more than 1 logit away from its stable value is 50 [202]. In a Monte Carlo simulation study, Wang and Chen (2005) [208] found that w ith sample sizes of 5 20 individuals, item parameters were unstable at either th e higher or lower end of the item difficulty hierarchy in short tests (10 to 20 items). However, the parameters were stable with longer tests and the same sample size. The research ers conclude that longe r assessments contain more information and thus stabilize item pa rameter generation in a small sample size. Should data fit the model or visa versa? The Rasch model is criticized because it assumes equal item discrimination. Item discrimination refers to an items ability to di fferentiate between test takers of different levels of ability. As stated, the two-paramete r IRT model contains an item discrimination parameter while the one parameter Rasch m odel does not. The Rasch model constrains

PAGE 63

49 the item discrimination value at . Acco rding to McHorney [209] it is critically important to examine the item discrimination capabilities of assessment items. It is possible that equally valid items will discriminate in various ways. To this end, the Rasch model software calculates an item discrimination index [148]. However, in contrast to the 2-paramete r IRT model, this index does not enter into the original person scoring. The index describes whether items are close to or depart from the ideal discrimination value of . In ot her words, the index re veals how consistent items are with the unidimensional expectations of the Rasch model. Low values indicate that an item fails to discriminate as well as other items; high values may indicate that the item is idiosyncratic in some way. Misfitting items are reexamined for redesign if necessary. The Rasch model uncovers possible anomalies in item design thereby enabling further investigation of and modi fications to the test [210]. The underlying conflict between the Rasch model and other IRT models has to do with views of how data provide informati on [211]. In short: should a measurement model explain data, or should a measurement model disclose information perhaps hidden in data? Andrich (2004) [ 211] frames this controversy using the language of mathematician-philosopher Thomas Kuhn as a clash of incompatible paradigms. Kuhn stated that the purpose of measurement is to disclose anomalies in data that cannot be explained by the theory underlyi ng the measurement model with which the data is analyzed. With this approach one is able to ask questions as to why data does not fit a given theory thereby potentially lear ning new information that contributes to advancement of the theory. Kuhns view is in stark contrast to traditional measurement

PAGE 64

50 theory that asserts data s hould fit a model so as to s upport the theory underlying the model. This issue is notably evident in c ontroversies surrounding the Rasch model [211]. According to CTT and non-Rasch IRT models, data is best interpreted according to a mathematical model found to explain the larg est amount of variance in the data. Models are tried and abandoned depending on how well the model explains the data. For example, a recent article eloquently advocates IRT methods to measure clinical cognitive change [212]. The researcher describes how various models were tried and discarded until he settled on a 2-parameter IRT model b ecause it provided a better fit than the other models. Kirsci et al (2006) [213] employed signifi cance testing of chi-square statistics as a method for comparing one and tw o parameter IRT models as best fitting the data. The point is that the a-priori assumption in the above paradigms was that a mathematical model should fit the data. The Rasch model has the inverse assumption. That is, data should be viewed with the assumption that it either fits or does not fit the model. If it does not fit, then, rather than abandoning the model in favor of anothe r, the data is examined for anomalies [211]. According to Andrich [ 211], the process of examining misfitting data (either items or persons) uncovers information that w ould have been hidden otherwise. New information may lead to advances in theory. As Wright states: Misfitting items can be redesigned. Misfitting populations can be reassessed [214]. Conclusion There exists an interdependent link betw een theory and measurement. The theory guides the design of the measurement tool. In turn, the information gathered from the measurement tool advances the theory. This concept was recently articulated by Dennis Tate in the John Stanley Memorial Lecture to the American Congress of Rehabilitation

PAGE 65

51 Medicine [215]. He stated t he process by which we measur e rehabilitation domains and test our theories about real ity shapes what constitute s rehabilitation knowledge. As recent advances in neuroscience and rehabilitation science have increased knowledge about the potential fo r upper limb motor recovery after stroke, recognition of the need to adequately measure impairme nt and the functional implications of impairment has also increased. Stroke rehabili tation professionals face an urgent need to assure that assessments used to establish the efficacy and effectiveness of translational intervention are of the highe st quality for measuring clie nt ability, change with intervention, informing the clinical deci sion-making process and providing valuable quantitative information about impairment and recovery. The FMA-UE is considered the gold sta ndard clinical assessm ent of post-stroke UE motor impairment, however its item-level measurement properties have not been well studied with contemporary measurement m odels, e.g., item response theory. Several deficiencies potentially exist in FMA-UE item content and structure that could impair its use as an outcome measure; (1) some items may not be consistent with current motor control theoretical expectat ions of post-stroke impair ment and recovery, (2) the reproducibility of the construc t recovery illustrated by its item hierarchy has not been established across subjects of various ability le vels, nor within subjects over time, and (3) in its current format, the FMA-UE score has no clear functional meaning. The aims of the subsequent studies address th ese deficiencies and serve the purpose of exploring and potentially enhancing the measurement prope rties of this important and long-standing stroke rehabilitation measurement tool.

PAGE 66

52 Table 1: The Fugl-Meyer Upper Extremity Assessment FMA-UE Item Number FMA-UE Item Description Shoulder/Elbow/Forearm Item 1 Biceps reflex elicited Item 2 Triceps reflex elicited Item 3 Scapular elevation Item 4 Scapular retraction Item 5 Shoulder abduction Item 6 Shoulder external rotation Item 7 Elbow flexion Item 8 Forearm supination Item 9 Shoulder adduction with internal rotation Item 10 Elbow extension Item 11 Forearm pronation Item 12 Hand to lumbar spine Item 13 Shoulder flexion to 90 degrees with elbow extended Item 14 Pronationsupination of forearm with elbow at 90 Item 15 Shoulder abduction to 90 with elbow extended Item 16 Shoulder flexion to 90 -180 with elbow extended Item 17 Pronationsupination of forearm with elbow extended Item 18 Normal reflex activity Wrist Item 19 Wrist stable with elbow at 90 Item 20 Wrist flexion-extension with elbow at 90 Item 21 Wrist stable with elbow extended and shoulder at 30 Item 22 Wrist flexion-extension with el bow extended and shoulder at 30 Item 23 Wrist circumduction Hand Item 24 Finger mass flexion Item 25 Finger mass extension Item 26 Hook grasp (MPs extended, PIPs and DIPs flexed) Item 27 Lateral prehension (t humb adduction to hold paper) Item 28 Palmar pinch (thumb to index finger to hold pencil) Item 29 Cylindrical grasp (hold small can) Item 30 Spherical grasp (hold tennis ball at fingertips) Coordination/Speed Item 31 Movement without tremor Item 32 Movement without dysmetria Item 33 Movement with normal speed

PAGE 67

53 Table 2: Descriptions of the trad itional upper extremity limb synergies Flexor Synergy Extensor Synergy Shoulder retraction Shoulder adduc tion with internal rotation Shoulder elevation Elbow extension Shoulder abduction to 90 degrees Forearm pronation Shoulder external rotation Elbow flexion Forearm supination

PAGE 68

54 CHAPTER 2 USING THE RASCH MEASUREMENT MO DEL TO EXAMINE POST-STROKE UPPER EXTREMITY RECOVERY The characterization of poststroke upper extremity (UE) motor recovery has long been of the focus of rehabilitation clinicians and researchers. An assessment tool that quantifies UE motor impairment can identify the amount of UE mo tor recovery, i.e., locate a patient along the motor recovery progression. Accurate measurement of UE motor impairment is important to capture the effects of translational interventions, explore optimal training parameters of existi ng interventions, and to predict future UE motor function. There is a pressing need to ascertain whether asse ssment tools commonly used in post-stroke UE motor recovery re search are accurately quantifying impairment and characterizing recovery. The Fugl-Meyer Upper Extremity Assessmen t (FMA-UE) [15] is the most widely used clinical assessment of post-stroke uppe r extremity motor impa irment [16]. It has excellent inter-rater reliab ility (Pearson Product-Moment Correlations r = 0.97-0.99 [17]) and test-retest reliability (Intraclass correlati on coefficients 0.94 0.99 [18]). It has been used as the standard from which to establis h the validity of other commonly used tests of upper extremity motor function such as the Wolf Motor Function Test and Ashworth Scale [20-26]. Researchers c onsistently use the FMA-UE as a descriptor of functional status following stroke [4, 20, 27]. To this end, the FMA-UE scores have been used to stratify research study participan ts into categories of stroke severity [28], predict longterm functional participation [29], and to describe arm mo tor impairment [30-32], arm

PAGE 69

55 movement quality [21], and re sidual arm motor function [ 33-35]. Furthermore, the FMAUE is the primary criterion for evalua ting the success of novel upper extremity interventions such as rhythmic bilatera l movement training [36, 37], neuromuscular stimulation [38], electromyographic triggered electrical stimulati on [39, 40], Botox [41], home based exercise [30], community based exercise [42], robot aided therapy [43, 44], virtual reality [45], imagery [46], forced use [47], and modified constraint induced movement therapy [48-51]. Because of the im portance of this assessment in post-stroke rehabilitation research, it is critical to ex amine, and perhaps improve, the quality of its measurement properties. The measurement properties of an assessment tool are usually studied with traditional psychometric methods such as re liability, validity, and responsiveness. The item response theory (IRT) measurement framework offers another way to examine an assessments measurement properties. Stroke re habilitation researchers have increasingly used IRT methods, e.g. Rasch analysis, to develop new assessments [176] or redesign existing assessments [177]. Rasch analysis offers the advantage of examining an instrument at the item level rather than as a whole. With this a pproach one can explore item content (i.e., does a speci fic item contribute to the measurement of an intended construct?), and item structur e (i.e., does the item difficulty hierarchy progress from less of to more of the intended trait?). A lthough the measurement properties of the FMASome authors use the terms Item Response Theory (IRT) and Rasch analysis interchangeably. However, the measurement models are different. The Rasch model requires that items have equal discrimination, whereas IRT models include a parameter for items to have different discriminations. The merits of each model are the subject of ongoing debates. For the purposes of this article the simpler of the models, the Rasch model, is used.

PAGE 70

56 UE have been extensively studied using traditional methods [15, 17, 18, 20, 24, 54-56], presently there are no published studies ap plying IRT approaches to the FMA-UE. McDowell and Newell (1996) suggest that he alth care instrume nts be founded on a specific theoretical framework [67]. Fugl-M eyer et al.s (1975) [15] theoretical framework broadly reflects th e observations of Twitchell (1951) [75] and Brunnstrom (1966) [76] regarding the stag es of post-stroke UE sensorimotor recovery. Fugl-Meyer stated: The form has been constructed following the hypothesis that the restoration of motor function in hemiplegic patients follo ws a definable stepwise course. Thus for a patient with hemiparalysis, recurrenc e of reflexes always precedes volitional motor action. Thereafter through initia l dependence on syne rgies, the active motion will become successively less de pendent upon the primitive reflexes and reactions and finally complete volunta ry motor function with normal muscle reflexes may be regained (p. 14). Fugl-Meyer chose items to exemplify the cons truct of motor recove ry including reflex items and voluntary movement items. Moreover, because the assessment is intended to measure recovery Fugl-Meyer arranged the items, from easy to hard to map the recovery process. It is clear that Fugl-Meyer expected motor recovery to proceed in a reflexive-tovoluntary, and synergy-to-isolated progressi on. We may therefore expect the FMA-UE items to be ordered according to this hypothesis. Recent advances in movement science challe nge the definable stepwise course of recovery upon which the FMA-UE is based. Contemporary views of the central nervous system suggest that the reflex items measur e a different behavior than other items. A tendon tap reveals the integrity of a relatively simple spinal ne ural control circuit, while a goal directed voluntary movement reveals the integrity of a more complex supraspinal neural network [80]. Furthermore, recent st udies of UE motor control suggest that coordinated arm motion and patterns of mu scle recruitment are influenced by task-

PAGE 71

57 specific mechanics [98, 216-218] and environmental demands [100]. For example, Michaelson et al. (2001) [98] suggested that trunk restra int normalized post-stroke reaching kinematics. Wu and Trombly (2000) [100] found that characteristics of the task-specific target influenced reaching kinematics in both healthy individuals and individuals with stroke. While Fugl-M eyers recovery model may provide a neurophysiological description of post-stroke UE motor impairment, task-specific interactions of neural, biom echanical and contextual fact ors may also influence the course of post-stroke UE motor recovery. The purpose of this study is to investigat e the dimensionality and construct validity of the FMA-UE using Rasch analysis. The aims of this study are to (1) determine if all items of the FMA-UE contribute to the measurement of a single construct (dimensionality), and (2) determine if the items are ordered according to Fugl-Meyers expected stepwise sequence (construct validity). Methods Dimensionality In order to create a legitimate measure, all items must contribute to the same construct [143]. Using Rasch analysis, the extent to which items contribute to a unidimensional construct is evaluated employing infit and outfit statistics. The infit statistic is most sensitive to ratings on items that are closely matche d to subjects ability, the outfit statistic is most sensitive to ra tings on items that are much easier or much harder than subjects ability [142]. Fit statistics ar e reported as mean square standardized residuals (MnSq) produced for each item of the instrument. MnSq represents observed variance divided by expected variance [172]. Consequently, the desired value of MnSq for an item is 1.0. The acceptable criteria fo r unidimensionality depe nds on the intended

PAGE 72

58 purpose of the measure and the degree of rigor desired. For surveys using ordinal rating scales, Wright and Linacre [169] suggest reas onable ranges of MnSq fit values between 0.5 and 1.7 associated with standardized Z values (ZSTD) less than 2.0. High values indicate that scores are variant or erratic, suggesting that an item belongs to a construct that is different from that represented by th e other items of the instrument. High values could also indicate that the item is being inac curately scored. A low MnSq value suggests that an item is failing to discri minate individuals with different levels of ability (i.e., with different amounts of motor ability ) or that an item is redundant (i.e., other items represent similar motor challenge). In this analysis we focused on items with high MnSq values, because they represent a greater threat to construct validity. Fit statistics alone are inadequate to determine unidimensionality [154-156]. A principal components analysis (PCA) was performed using the SAS Factor procedure (SAS v 8.2) with oblique rotation. The inte nt of this PCA was to reorganize the multivariate data (participants FMA-UE item ratings) into a limited number of components (factors) so that each component ca ptures a substantial amount of the overall variance within the dataset. That is, the purpose of the PCA was to detect obvious factors within the FMA-UE. The PCA was carried out in two steps: an initial analysis to examine the dimensionality of the assessment, and an oblique rotation to examine if traditional subscale divisions were empirically supported. We hypothesized that the assessment items illustrated a single factor, UE motor ability. To test our hypothesis we exam ined PCA eigenvalues and factor loading statistics. We employed the Kaiser rule [ 157] as the criteria fo r retaining principal components, i.e., we examined only the components with an eigenvalue > 1.0. We

PAGE 73

59 expected one eigenvalue would explain the majority of the variance in the data. Furthermore we expected that the majority of FMA-UE assessment items would correlate with a single principal compone nt as evidenced by factor lo adings on the first component > 0.40 [219]. Fugl-Meyer suggested dividing the assessment into four subscales: arm (items 1 18), wrist (items 19 23), hand (items 24 30), and coordination/speed (items 31-33) [15]. Each of Fugl-Meyers s ubscales may represent a unique motor control construct. To examine if there was empirical evidence for this subscale division, we executed a second step in the PCA analysis. We performed an ob lique rotation (obvarimax) of four principal components. We chose an oblique rotati on because we anticip ated inter-factor correlations, i.e., movements of the hand were expected to correlate with movements of the arm. Factor rotation is a mathematical technique in which the multi-dimensional axes of initial PCA factor loadings (the number of axes matches the number of retained factors) are rotated to model the data. The rotation re-expresses the factors as a less complex structure to facilitate interpretati on [164]. This procedure produces a factor matrix with factor loading values (equivalent to regression coefficients) indicating the weight given to the item wh en interpreting the factor [ 157]. We considered factor loadings meaningful if the absolute valu e exceeded 0.40 [219]. We interpreted the factors (i.e., we labeled the factor as per the construct it represents) by subjectively determining what trait was shared among the items loading strongly onto that factor. Because the purpose of our factor rotati on was to explore whether Fugl-Meyers suggested subscale pattern was present in our data, we examined the groups of items comprising a factor to determine if they matched Fugl-Meyers suggested item groupings.

PAGE 74

60 Construct Validity The foundation of objective measurement is to connect the numbers produced by an instrument with its content [141]. Th rough Rasch measurement, items represent difficulty markings (calibrated in log-equivalent units called logits ) along the continuum of a construct. For example, for the construct underlying the FMA-UE, we may expect flexor synergy items to repres ent easy items along the continuum of motor ability, while hand items which involve more intricate prehension pa tterns to represent more challenging items across the continuum. For purposes of this study, we postulated that the item hierarchy would illustrate the tr ait UE motor ability. Analysis of data from a cross section of individuals with different abilities post-stroke may provide an initial description of the hierarchy. By placing item difficulties and person abilities on the same continuum, Rasch analysis can be used to match the difficulty of the assessment items to the ability of the sample tested. In general, the analysis will tell us what items are easy enough for individuals with poor arm motor skills (e.g. individuals who have little arm movement) and what items are challenging enough for i ndividuals with good arm motor skill (e.g. individuals who have fine-motor coordina tion). The placement of item difficulty and person ability on the same continuum also provides information on how specific individuals are expected to re spond to particular items. For example, if a person receives a logit measure of 6.0 and the item shoulder flexion to 90 degrees with elbow extended also receives a logit measure of 6.0, then it would be expected that that person would have a 50% probability of being able to pe rform this movement successfully. This individual would be expected to have a greater than 50% probability of being able to accomplish easier items such as those re presenting the flexor synergy items.

PAGE 75

61 Furthermore, he/she would be expected have less than 50% probability of being able to accomplish more difficult types of movement, such as palmar prehension (grasping a pencil with the pads of the thumb and index finger). Results Participants A secondary analysis was performed on da ta pooled from two studies. The first dataset consisted of 100 persons enrolled in a randomized cl inical trial of therapeutic exercise [220]. FMA-UE data was collect ed prior to the exer cise intervention by therapists trained in standard administration of the assessment. Persons with stroke were included in this study if they met the following inclusion criteria; ( 1) Stroke within 30 to 50 days, (2) ability to ambulate 25 feet i ndependently, (3) mild to moderate stroke deficits defined by a total Fugl-Meyer scor e of 27-90 (upper and lower extremities) and Orpington Prognostic Scale score of 2.0 to 5.2, (4) palpable wris t extension on the involved side, and (5) Folstein Mini-Mental Stat us exam score of greater than 16 [220]. The second dataset consisted of 459 individuals enrolled in the Kansas City Stroke Study [221]. Participants were included if they met the following criteria; (1) stroke onset within 0 14 days prior to en rollment, (2) 18 years of age or older, (3) ischemic stroke as diagnosed by physician, (4) lived in the commun ity prior to stroke onset, and (5) able to participate in baseline tes ting. Study staff trained in the administration of the FMA-UE evaluated patients. Pooling of these data resulted in a 559-person dataset. However, because lesion location may effect neural reorganization and c linically observed patterns of UE recovery [222], subject s with non-cortical (e.g., brai nstem or cerebellum) stroke were not selected from the pooled dataset. This resulted in a final dataset of 512 participants. Characteristics of the 512person sample are presented in Table 3.

PAGE 76

62 Fit Statistics Item Fit Statistics from the initial an alysis of the entire 33-item FMA-UE are presented in Table 4. Two items ( biceps reflex and triceps reflex) shaded in gray, show infit values that are beyond the acceptabl e ranges described by Wright and Linacre (1994) [169]. These items have infit sta tistics that are above 1.7 with a standard deviation greater than 2.0. These items also showed abnormally high outfit statistics. In addition, the two items showed relatively lo w score correlations (0.36 and 0.26). Infit and outfit statistics are within accepta ble ranges for all other items. Principal Components Analysis The PCA retained 4 factors having eigenva lues > 1.0 (Kaiser rule [157]). See Table 5. A single component explained 68% of the variance in the data. Three other factors explained an additional 5%, 4%, and 3% of the variance. Table 6 displays the factor loadings onto the firs t principal component. Thirty of 33 items loaded strongly onto the first component (factor loading va lues ranging from 0.75 0.91). Three items (biceps reflex, triceps refl ex, normal reflex activity) had a poor to moderate loadings (r = 0.14 0.53) with this factor. Rotated factor loadings for the 4 component s with eigenvalues > 1 are presented in Table 7. Primary loadings (factor loadings with an absolute value > 0.40) for each component are shaded in gray. The values ar e sorted into descending order of magnitude and grouped into columns to facilitate inte rpretation of the item groupings. Several items do not load strongly onto any factor. For example, pronation-supination with elbow at 90 and movement without dysmetria do not display any factor loading greater than 0.40. Five items cross-load onto more than one factor. For example, wrist flexion-extension with elbow extended and wrist stable with elbow extended each load onto factor 2 (0.41

PAGE 77

63 and 0.46) and factor 3 (0.54 and 0.52). Items with no factor loadings and those that crossload are grouped near the bottom of the chart and mark ed with a double asterisk. Examination of the shared traits among ite ms that load onto each factor may illustrate the motor behavior represented by each factor. For example the items, elbow flexion, shoulder abduction, scapul ar elevation, scapular retrac tion, shoulder external rotation and forearm supination load onto the first factor. In addition, this factor contains shoulder adduction with internal rotation and forearm pronation These items are consistent with the traditional description of the post-stroke UE flexor and extensor synergies as described by Twitchell and Brunns trom. Therefore this factor may pick up a synergy trait underlying the FMA-UE. The grasp and prehension items load onto the second factor. Therefore this f actor may detect a hand trait evident in the data. This item grouping roughly matches the hand subsca le division suggested by Fugl-Meyer. Items requiring movements away from the body using elbow extension (e.g., shoulder flexion to 180 with elbow extended, pr onation-supination with elbow extended ) load strongly onto the third factor. However, it is difficult to interpret th is factor because it contains five cross-loaded items. Finally, triceps reflex and biceps reflex load onto the fourth factor, perhaps detecti ng a reflex trait. Although th e items located in factors 1 and 2 are roughly similar to Fugl-Meyers arm and hand subscales, the other item groupings do not match Fugl-Meyer s FMA-UE subscale suggestions. Although the PCA suggests that four motor control constructs (four eigenvalues > 1) may be present, the majority of the va riance in the data is explained by a single principal component. However, poor factor lo adings onto the first principal component, and abnormal fit statistics suggest that the reflex items do not fit with the intended

PAGE 78

64 measurement construct of the assessment; therefore these items were removed from subsequent Rasch analysis. Fit statistics, th e large first eigenvalue and strong principal component factor loadings support the unidime nsionality of the remaining items. Because the intent of our study is to look at the relationship of items across the entire upper limb we chose not to subdivide the assessment. Item fit statistics the revised 30-item FMA-UE (reflex items removed) are presented in Table 8. All 30 items fit th e unidimensional assumptions of the Rasch model. The revised 30-item FMA-UE show ed good internal consistency. The person reliability index, analogous to coefficien t alpha, was 0.96. The instrument divided the sample into 7 statistically significant strata (Number of Strata =[4(Gp = 4.67) + 1]/3, where GP= person separation [223]) indi cating a good person to item match. Item Hierarchy Table 9 presents items in order of de creasing challenge. FMA-UE items at the bottom of the center column indicate the le ast challenging items; those at the top represent most challenging items. Elbow flexion (measure = -1.76 0.13 logits) and shoulder adduction with internal rotation (measure = -1.56 0.12 logits) were the easiest items for this sample to perform, and wrist circumduction (measure = 1.67 0.10 logits), and hook grasp (measure = 1.33 0.10 logits) were the most difficult items for this sample to perform. Five of the nine flexor synergy and ex tensor synergy items (items 3 through 11) group towards the easy e nd of the hierarchy, wh ile some flexor and extensor synergy items span the item difficulty hierarchy. For example, elbow flexion is the easiest item in the hierarchy (.76 0.13 logits), scapular elevation is slightly more

PAGE 79

65 difficult (.40 0.12 logits), and forearm supination is moderately difficult (0.65 0.10 logits). Movements requiring combinations of s houlder flexion and elbow extension are much more difficult than movements de manding no shoulder motion with the elbow positioned at 90 degrees. For example, p ronation-supination with the elbow at 90 degrees wrist flexion-extension with the elbow at 90 degrees and wrist stable with the elbow at 90 degrees are among the moderately difficult items. Shoulder flexion to 180 with the elbow extended wrist flexion-extension with the elbow extended and pronationsupination with the elbow extended are among the most difficult items. In addition, although the majority of the hand items (items 24 through 30) group at the difficult end of the hierarchy, some of these items also span the item difficulty hierarchy. For example, h ook grasp (1.33 0.10 logits), spherical grasp (1.25 0.10 logits), and lateral prehension (1.08 0.10 logits) are among the most difficult items, while finger mass flexion (-1.44 0.12 logits) is among the easiest items and palmar prehension (0.06 0.10 logits) is calibrated near th e middle of the scale. Surprisingly, finger mass extension (-1.25 0.12 logits) calibrates as an ea sy item for this sample to perform (see discussion). Keyforms While the above results show the overall hierarchical pattern for the sample under study, a critical question is th e consistency of th is pattern across i ndividual subjects. Similar to the fit statistics that are produced fo r items are fit statistics for persons. Ninetyeight percent of the sample showed acceptable infit statistics (MnSq <1.7 and ZStd <2.0), suggesting that individual subjects in the sample are responding similarly to the item

PAGE 80

66 difficulty hierarchy. Figure 1 demonstrates th e item difficulty hierarchy relative to the scoring pattern of three client s representing different UE abil ities. Items are listed in terms of increasing difficulty level (i.e., elbow flexion being the easiest item at the bottom of each panel, wrist circumduction representing the hardest item at the top of each panel). The 3-point rating scales for each item are presen ted to the left of each panel. As the item difficulty increases, the rating scale stair step s to the right. The bottom scale represents item difficulty. For example, for the item elbow flexion a rating of is at a difficulty of approximately .0 logits. This bottom scale also reflects person ability (i.e., the solid vertical line crossing the scale represents the individuals person ability measure, the dotted vertical lines represen t the 95% confidence interval around this ability measure). Circled numbers represent the actual ratings for the individual. The triangle shapes represent ratings that are stat istically erratic (p=0.05). A person of low ability (-1.25 logits) is re presented on the top left panel (Person A), of moderate ability (0.20 logits) is repr esented on the top right panel (Person B), and of high ability (3.88 logits) (Person C) is represented on the bottom left panel. The individual of low ability (Person A) has a tendency to receive ratings of (faultless performance) and (partial performance) on the easy items ( elbow flexion, shoulder adduction with internal rotation, finger mass flexion ), and receive ratings of (unable to perform) on the most difficult items ( shoulder flexion to 90 degrees with elbow extended, hook grasp, and wrist circumduction ). The individual of moderate ability (Person B) has a tendency to receive ratings of (faultless performance) on the easy items, and receive ratings of (partial pe rformance) on the moderately difficult items ( scapular retraction, shoulder external rotation ), and receives ratings of or

PAGE 81

67 (unable to perform) on the most difficult items The individual of high ability (Person C) has a tendency to receive rati ngs of (faultless performan ce) on the easy items, and the same the moderately difficult items. However this person receives ratings of (partial performance) on the most difficult items ( shoulder flexion to 180 degrees with elbow extended, wrist circumduction ). In general, while these individuals are of different abilities their scoring pattern is the same; a ll three score higher on easier items and lower on harder items. The pattern re tains its structure when measuring a low ability, moderate ability, and high ability person. There are exceptions to this pattern as evidenced by the ratings denoted with a triangle. Person A with lower ability receives a higher than expected rating of on a moderately challenging item ( scapular retraction ). Person C of high ability receives a lower than expected rating of on a mode rately difficult item (lateral prehension). Overall, 12 out of 512 people (2 %) showed erratic scores ba sed on infit statistics. Among these 12 individuals, only 2-5 item s were statistically erratic. Discussion The purpose of this study was to investigate the dimensionality and construct validity of the FMA-UE. Using item response th eory statistical techniques we determined that the items of the FMA-UE contribute to the measurement of multiple constructs and demonstrate a difficulty order that does not re flect with the item or der proposed by FuglMeyer. Furthermore, we determined that this item difficulty order remains consistent, independent of person ability. The above result s will be discussed rela tive to traditional and contemporary motor control theories.

PAGE 82

68 Dimensionality The initial finding from both the Rasch An alysis and the more traditional PCA revealed that the reflex items empirically a ppear to be disconnected from the remainder of the FMA-UE. While, according to Fugl-Mey ers suggested stepwise course of recovery, it would be expected that persons of little to no UE motor ability would be the only individuals without reflex be havior, the high infit and outfit statistics indicate that at least some individuals of high ability also show this behavioral response. Multiple statistical findings (i.e. low PC A weights, low correlations an d high fit statis tics) suggest that the reflex items are contributing little explanatory variance for the FMA-UE and support our decision to remove these items from further analysis. The three reflex items are exceptional to the assessment in two ways. First, the items evaluate a passive response rather than active motion. Specifica lly, the behavior of the phasic stretch reflex is assessed through tendon taps of the biceps (items 1 and 18), triceps (items 2 and 18) or finger flexors (i tem 18). Secondly, the rating scales for the reflex items are different than the rest of the assessment. For example, items 1 and 2 utilize a two-point scale ( 0 = No reflex activity elicited and 2 = Reflex activity elicited ), and item 18 utilizes a uniquely defined three-point scale ( 0 = At least two of three phasic reflexes are hyperactive or 1 = One reflex is hyperacti ve or two reflexes are lively or 2 = No more than one reflex is li vely and none are hyperactive). These procedural and scoring differences from that of the major ity of the assessment may cause the items to misfit and correlate as a separate factor. While the above statistical findings may be a product of administration procedures (passive vs. active) or the rati ng scale structure, they also may indicate that the FMA-UE reflex items reflect a neurological mechanism that is different than that of the other

PAGE 83

69 assessment items. A relatively simple spinal neural circuit governs the tendon tap muscle contraction, while a complex supraspinal a nd spinal neural ne twork governs voluntary goal-directed movement [80]. Rather th an a hardwired precursor to voluntary movement (as suggested by traditional motor control theory), stretch reflexes are important adjuncts of voluntary motor cont rol [80, 82, 83] and the magnitude of the stretch reflex is amendable to operant trai ning [86, 87]. Given that the FMA-UE is an assessment of voluntary movement ability, our results suggest the three reflex FMA-UE items diminish the construct validity of the instrument. As suggest ed by Gladstone et al. (2002), the reflex items may confound interpre tation of the FMA-UE total score [55]. Item Hierarchy Fugl-Meyer attributed the assessment struct ure and item choice, at least in part, to the earlier works of Twitc hell (1951) [75] and Brunns trom (1966) [76]. Twitchell described an orderly progression of phenomena characterizing the course of post-stroke UE motor recovery. Initial UE flaccidity was followed by emergence of gross flexor and extensor voluntary movement synergies, gradual separation of synergy movement patterns and finally distal fine motor coordi nation. Twitchell attributed this recovery sequence to neurophysiological mechanisms, i.e., the cortex reassuming its natural inhibition of primitive proprioceptive reflexes [75]. Brunnstrom, furthering Twitchells work, documented specific movements that typi fied each stage of motor recovery thereby mapping the recovery progression. Fugl-Meyer app ears to have, at least in part, applied a numerical rating scale to Br unnstroms Hemiplegia Classification and Progress Record [76] to create the FMA-UE [15]. The Rasch item difficulty hierarchy may illustrate the pattern of post-stroke UE recovery. The keyform provides evidence that th is hierarchy is replic ated across different

PAGE 84

70 individuals in this sample. This hierar chy however, does not follow the expected stepwise item arrangement of the FMAUE. Contemporary motor control theoretical models suggest that UE movement is dynami cally organized in response to the demands of a task rather than in a strict reflex ive sequence [224]. Following stroke, UE muscle activation patterns and reaching kinematic s for a task depend on contextual and mechanical factors such as target location [217, 218], characteristic s of the environment [100] and external constraint [98]. Mu rray and Johnson (2004) found that healthy individuals scaled arm joint to rques according to the number of joints involved in a task [106]. In addition, post-stroke UE motor contro l is also influenced by descending inputs from secondary non-cortical moto r areas to axial musculature ( likely spared from infarct) and inputs directly from primary cortical mo tor areas to forearm and hand muscles (likely damaged with infarct) [225]. Taken together, these studies suggest that for individuals with and without stroke, UE motor contro l is dependent on a dynamic interaction of neural, mechanical, and contextual factor s. Our results suggest that the FMA-UE assessment items are arranged in a Rasch-deri ved difficulty order compatible with the task-specific complexity of motion. That is, arm movements are easier or more challenging based on the inherent demands of the movement. Elbow flexion may be easier than wrist circumduction because the movement is inherently less complex and the movement may be orchestrated via spared secondary motor areas. Shoulder flexion to 180 degrees with the elbow extended may be difficult beca use of the influence of gravity on a long lever arm. Our task-specific interpretation of the item hierarchy may affect the interpretation of the FMA-UE score. Our re sults indicate that the FMA-UE aggregate score should not

PAGE 85

71 be interpreted in a strict proximal-to-distal or synergistic-to-isola ted manner. This is particularly important when the FMA-UE scor e is used to describe patient movement characteristics. Chae and colleagues (2003) [ 97] suggest that a low FMA-UE aggregate score indicated recovery of only proximal UE motor function, while higher FMA-UE aggregate score indicated recovery of bot h proximal and distal UE motor function. Chaes interpretation is consistent with Fugl -Meyers proposed stepwise item order in which movements assumed to recover early in the process appear firs t on the assessment, while those to recover later in the proce ss appear further alon g the assessment. Our results indicate that an individual with a low FMA-UE score is likely to exhibit some proximal arm motions characte ristic of an abnormal syne rgy pattern (e.g., the easy items; elbow flexion, shoulder adduction with in ternal rotation, sc apular elevation ), but is unlikely to exhibit other proximal syner gy-based movements (e.g., the more difficult items; scapular retraction, shoulder exter nal rotation, and forearm supination ). An individual with a low FMA-UE score is likely to exhibit some easy distal, nonsynergistic, hand movements (e.g., finger mass flexion, finger mass extension, and cylindrical grasp ). Participant groups formed using FMA-UE scores may be less homogenous than desired. Similarly, our item hierarchy, if assume d to illustrate the pattern of post-stroke UE recovery, shows th at recovery does not proceed in a strict proximal-to-distal sequence as was traditiona lly endorsed by some therapists [226] and suggested by some therapy textbooks [227, 228]. Our results do not entirely contradict the or iginal clinical observ ations of Twitchell and Brunnstrom. Indeed, the results provide at least partial support for their classic works. For example, both clinicians observe d flexion synergy and extension synergy

PAGE 86

72 movements to be among the first motions to return after stroke. Our results show that three of the six flexor synergy movement s and two of the three extension synergy movements are among the least chal lenging items of the FMA-UE ( elbow flexion, shoulder abduction and scapular elevation ; and forearm pronation and shoulder adduction with internal rotation ). Furthermore both clinicians described movements deviating from synergy occurring later in re covery. The Rasch hierarchy also partially supports this observation. Items requiring wr ist motion are easier when the shoulder is neutral and the elbow is bent, than when the s houlder is flexed and the elbow is extended. For example forearm pronation-supination, elbow at 90 and wrist flexion-extension, elbow at 90 are moderately challenging items, while pronation-supination, elbow extended and wrist flexion-extensi on, elbow extended are among the hardest items. In addition, Brunnstrom proposed separa te hand and proximal limb evaluation scales. Several researchers [31, 34] have separated the hand items to focus their assessment on an area interest. It is feasible that different motor control constructs underlie arm versus hand movements give n the different supraspinal areas likely influencing arm versus hand motorneurons [225]. Although the PCA rotated factor pattern provided some statistic al support for this division, other results (e.g., Rasch fit statistics and PCA item correlations) s upport keeping the arm and hand voluntary movement items together as a unidimensional scale. One surprising finding in the present study was that the finger extension item calibrated as an easy item. Studies have s hown finger extension to be highly impaired after stroke [31, 229]. Our results may be explai ned in two ways. First, the vast majority of participants in our sample (90%) sustained mild or mode rate stroke as defined by the

PAGE 87

73 Orpington score. It is possible that our sample was not severely impaired and therefore did not exhibit impaired finger extension. Secondly, the FMA-UE defines finger mass extension as gross re lease of the mass flexion grasp [15]. Finger extension as defined by the FMA-UE may actually reflect the na tural tenodesis mechanism of grossly extending the fingers by flexing the wrist. It should be noted that some researchers have more rigorously evaluated finger ex tension by controlling for wrist position and defining finger extension as return to met acarpophalangeal joint neutral [31, 229]. One weakness of the FMA-UE is that there exis t no standard administration guidelines [18, 56]. It is feasible that if the assessmen t were administered such that the wrist configuration during th is item was controlled, the item w ould have a different difficulty level. Limitations There are a number of sta tistical and demographic limita tions to this study that could limit the generalizability of our results. Traditional PCA was designed for continuous rather than ordinal data. Flora a nd Curran (2004) [230] suggest using factor analysis approaches designed specifically for multivariate categorical data when analyzing ordinal data (e.g., FMA-UE ratings ). While these approaches are not yet common in the health care lite rature they should be considered for future studies. Additionally the generalizability of the findings is confined to sample characteristics of the study. Although our participants represented a range of stroke severity, the majority, 90%, of the sample experienced mild or mode rate stroke. In order to generalize these findings across the broader range of stroke seve rity the study needs to be replicated with individuals with more severe motor impairme nt. Finally, the most critical limitation of,

PAGE 88

74 the present study is that we used cross-se ctional data to challenge Fugl-Meyers recovery paradigm. We are currently replica ting this study using longitudinal data. Conclusion In summary the Rasch analysis of the FMAUE challenges the use of reflex items to measure UE movement and challenges the st rict definable stepwise structure implied by the instrument. A better understanding of th e post-stroke UE recovery progression and possible revision of the theore tical mechanisms underlying this recovery may be critical for clinical intervention. Instruments clearl y reflecting the recove ry pattern may more accurately identify the clients stage of recovery and in form the clinician of the appropriate treatment progression. Table 3: Characteristics of the 512-person sample Table 1: Characteristics of the 512 person sample Years of Age (Mean SD) 69.8 11.12 Gender Males 242 Females 270 Race White 411 African-American 85 Other 16 Stroke Type Ischemic 474 Hemorrhagic 38 Stroke Location Right Hemisphere 250 Left Hemisphere 262 Days Since Stroke (Mean SD) 16.88 31.23 Range of Days Since Stroke Minimum 0 Maximum 145 Stroke Severity Minor (Orpington Score <3.2) 191 (37.3%) Moderate (3.2 <= Orpington <= 5.2) 269 (52.5%) Severe (Orpington > 5.2) 52 (10.2%)

PAGE 89

75Table 4: Item measures, fit sta tistics and score correlations for th e full 33-item FMA-UE (n = 512) FMA-UE Item Measure Error Infit MnSq Infit ZSTD Outfit MnSq Outfit ZSTD Score Correlation Normal reflex activity 3.40 0.11 1.34 3.8 1.10 0.4 .57 Wrist circumduction 1.55 0.09 0.79 -3.2 0.92 -0.3 .77 Hook grasp 1.26 0.09 1.29 3.9 1.08 0.5 .72 Shoulder flexion to 180 elbow extended 1.20 0.09 0.70 -5.0 0.57 -2.7 .80 Spherical Grasp 1.15 0.09 0.96 -0.6 1.02 0.2 .77 Lateral prehension 1.05 0.09 0.96 -0.5 1.21 1.2 .76 Wrist flexion-extension, elbow extended 1.03 0.09 0.65 -5.8 0.57 -2.8 .82 Pronation-supination, elbow ex tended 0.98 0.09 0.57 -7.4 0.44 -4.1 .83 Wrist stable, elbow extended 0.93 0.09 0.96 -0.5 0.69 -2.0 .78 Movement with normal speed 0.88 0.09 1.10 1.4 0.93 -0.4 .76 Forearm supination 0.68 0.09 0.70 -4.8 0.73 -1.9 .84 Shoulder abduction to 90 elbow extended 0.37 0.09 0.72 -4.3 0.53 -4.0 .83 Movement without dysmetria 0.36 0.09 1.37 4.7 1.21 1.4 .76 Shoulder external rotation 0.34 0.09 0.68 -5.0 0.63 -3.0 .85 Wrist stable, elbow at 90 0.29 0.09 1.01 0.1 0.71 -2.3 .80 Wrist flexion-extension, elbow at 90 0.24 0.09 0.58 -6.9 0.52 -4.3 .86 Palmar prehension 0.19 0.09 0.95 -0.7 0.83 -1.2 .81 Scapular retraction 0.16 0.09 0.89 -1.5 0.96 -0.3 .82 Pronation-supination, elbow at 90 0.00 0.10 0.59 -6.5 0.48 -4.8 .86 Shoulder flexion to 90 elbow extended -0.03 0.10 0.78 -3.2 0.52 -4.4 .84 Hand to lumbar spine -0.19 0.10 0.72 -4.0 0.62 -3.3 .85 Shoulder abduction -0.32 0.10 0.79 -2.9 0.77 -1.8 .84 Elbow extension -0.39 0.10 0.57 -6.6 0.53 -4.2 .87 Forearm pronation -0.57 0.10 0.60 -5.9 0.64 -2.9 .86 Movement without tremor -0.60 0.10 1.19 2.3 0.98 -0.1 .80 Cylindrical grasp -0.74 0.10 0.88 -1.5 0.76 -1.7 .84

PAGE 90

76Table 4 continued Finger mass extension -0.86 0.10 0.70 -4.1 0.58 -3.3 .86 Scapular elevation -0.97 0.11 0.93 -0.8 0.87 -0.8 .83 Finger mass flexion -1.00 0.11 0.79 -2.7 0.70 -2.0 .85 Shoulder adduction with internal rotation -1.09 0.11 0.65 -4.8 0.49 -3.8 .87 Elbow flexion -1.23 0.11 0.67 -4.4 0.81 -1.1 .85 Triceps reflex -3.66 0.13 5.20 9.9 9.90 9.9 .26 Biceps reflex -4.40 0.14 3.51 9.9 9.90 8.2 .36

PAGE 91

77 Table 5: Eigenvalues for the principal components analysis Table 3: Eigenvalues for the Principle Components Analysis Factor Eigenvalue Proportion of Variance Cumulative Variance 1 22.48 0.68 0.68 2 1.60 0.05 0.73 3 1.31 0.04 0.77 4 1.04 0.03 0.80 Table 6: Item correlations w ith first principal component Item DescriptionComponent 1 Elbow Extension 0.91 Forearm Pronation 0.91 Pronation Supination with Elbow 90 0.91 Wrist Flexion Extension Elbow 90 0.90 Sh Flex to 90 Elbow Extended 0.90 Shoulder Adduction with IR 0.90 Finger Mass Extension 0.90 Hand to Lumbar Spine 0.89 Sh Abd to 90 Elbow Extended 0.88 Finger Mass Flexion 0.88 Elbow Flexion 0.88 Cylindrical Grasp 0.88 Shoulder Abduction 0.88 Shoulder External Rotation 0.88 Pronation Supination with Elbow Extended 0.87 Forearm Supination 0.86 Wrist Stable Elbow at 90 0.86 Scapular Elevation 0.85 Palmar Prehension 0.85 Wrist Flexion Extension Elbow Extended 0.85 Movement without tremor 0.84 Scapular Retraction 0.84 Sh Flex to 180 Elbow Extended 0.83 Wrist Stable Elbow Extended 0.83 Movement with normal speed 0.81 Wrist Circumduction 0.80 Lateral Prehension 0.80 Spherical Grasp 0.80 Movement without dysmetria 0.78 Hook Grasp 0.75 Biceps Reflex 0.24 Triceps Reflex 0.14 Normal Reflex Activit y 0.53

PAGE 92

78 Table 7: Rotated factor loadings (obli que rotation) for 4 principle components Item DescriptionFactor 1Factor 2Factor 3Factor 4 Scapular Elevation 0.780.040.100.15 Elbow Flexion 0.770.19-0.020.15 Scapular Retraction 0.77-0.080.250.10 Shoulder Abduction 0.750.040.180.11 Shoulder Adduction with IR 0.670.200.100.14 Shoulder External Rotation 0.610.060.300.13 Elbow Extension 0.610.210.190.13 Forearm Pronation 0.590.300.080.16 Movement without tremor 0.490.340.110.10 Hand to Lumbar Spine 0.460.260.230.17 Forearm Supination 0.400.310.230.15 Lateral Prehension -0.040.770.150.09 Palmar Prehension 0.140.730.100.05 Spherical Grasp 0.090.710.100.06 Hook Grasp 0.010.690.180.01 Cylindrical Grasp 0.380.63-0.030.07 Finger Mass Extension 0.370.61-0.010.14 Finger Mass Flexion 0.370.550.040.13 Wrist Stable Elbow at 90 0.130.510.310.14 Movement with normal speed 0.210.480.250.04 Wrist Flexion Extension Elbow 90 0.210.460.330.13 Normal Reflex Activity -0.13-0.170.990.02 Sh Flex to 180 Elbow Extended 0.280.000.720.04 Pronation Supination with Elbow Extended 0.1 7 0.200.640.11 Triceps Reflex -0.23-0.10-0.060.96 Biceps Reflex 0.00-0.16-0.110.92 Pronation Supination with Elbow 90 ** 0.370.350.290.13 Movement without dysmetria ** 0.320.370.180.09 Wrist Flexion Extension Elbow Extended 0.020.41*0.54*0.12 Wrist Stable Elbow Extended -0.040.46*0.52*0.11 Wrist Circumduction -0.050.47*0.50*0.08 Sh Abd to 90 Elbow Extended 0.45*0.080.50*0.05 Sh Flex to 90 Elbow Extended 0.44*0.180.42*0.09

PAGE 93

79 Table 8: Item measures, fit st atistics, and correlations fo r revised FMA-UE (n = 512, 30 Items) FMA-UE Item Measure Error Infit MnSq Infit ZSTD Outfit MnSq Outfit ZSTD Score Correlatio Wrist circumduction 1.67 0.10 0.90 -1.5 0.96 -0.1 .77 Hook grasp 1.33 0.10 1.43 5.6 1.28 1.7 .73 Shoulder flexion to 180 elbow extended 1.26 0.10 0.84 -2.5 0.80 -1.4 .80 Spherical Grasp 1.20 0.10 1.08 1.1 1.17 1.1 .78 Lateral prehension 1.08 0.10 1.10 1.5 1.30 2.0 .78 Wrist flexion-extension, elbow extended 1.06 0.10 0.77 -3.7 0.78 -1.6 .82 Pronation-supination, elbow extended 1.00 0.10 0.70 -4.8 0.63 -3.1 .83 Wrist stable, elbow extended 0.95 0.10 1.13 1.9 0.96 -0.2 .78 Movement with normal speed 0.89 0.10 1.25 3.4 1.13 1.0 .77 Forearm supination 0.65 0.10 0.82 -2.6 0.88 -1.0 .85 Shoulder abduction to 90 elbow extended 0.28 0.10 0.86 -2.1 0.71 -2.7 .85 Movement without dysmetria 0.27 0.10 1.62 7.4 1.52 3.9 .77 Shoulder external rotation 0.24 0.10 0.83 -2.5 0.82 -1.6 .86 Wrist stable, elbow at 90 0.18 0.10 1.20 2.7 0.96 -0.3 .82 Wrist flexion-extension, elbow at 90 0.12 0.10 0.69 -4.8 0.66 -3.2 .88 Palmar prehension 0.06 0.10 1.10 1.3 1.01 0.1 .83 Scapular retraction 0.03 0.10 1.08 1.2 1.29 2.4 .84 Pronation-supination, elbow at 90 -0.17 0.11 0.72 -4.2 0.67 -3.1 .88 Shoulder flexion to 90 elbow extended -0.21 0.11 0.94 -0.8 0.71 -2.7 .86 Hand to lumbar spine -0.40 0.11 0.90 -1.4 0.85 -1.2 .87 Shoulder abduction -0.56 0.11 0.99 -0.1 1.02 0.2 .87 Elbow extension -0.64 0.11 0.70 -4.2 0.70 -2.4 .90 Forearm pronation -0.87 0.11 0.76 -3.1 0.81 -1.4 .89 Movement without tremor -0.91 0.11 1.51 5.4 1.36 2.2 .83 Cylindrical grasp -1.10 0.12 1.09 1.1 0.94 -0.3 .87 Finger mass extension -1.25 0.12 0.89 -1.3 0.76 -1.4 .89 Scapular elevation -1.40 0.12 1.25 2.7 1.31 1.6 .86 Finger mass flexion -1.44 0.12 1.04 0.4 1.03 0.2 .89 Shoulder adduction with internal rotation -1.56 0.12 0.88 -1.4 0.74 -1.4 .90 Elbow flexion -1.76 0.13 0.92 -0.8 1.18 0.8 .89

PAGE 94

80 Table 9: Item difficulty hier archy revised 30-item FMA-UE FMA-UE item number FMA-UE item description Measure Error 23 Wrist circumduction 1.67 0.10 26 Hook grasp 1.33 0.10 16 Shoulder flexion to 180, elbow extended 1.26 0.10 30 Spherical grasp 1.20 0.10 27 Lateral prehension 1.08 0.10 22 Wrist flexion-extension, elbow extended 1.06 0.10 17 Pronation-supination, elbow extended 1.00 0.10 21 Wrist stable, elbow extended 0.95 0.10 33 Movement with normal speed 0.89 0.10 8 Forearm supination 0.65 0.10 15 Shoulder abduction to 90, elbow extended 0.28 0.10 32 Movement without dysmetria 0.27 0.10 6 Shoulder external rotation 0.24 0.10 19 Wrist stable, elbow at 90 0.18 0.10 20 Wrist flexion-extension, elbow at 90 0.12 0.10 28 Palmar prehension 0.06 0.10 4 Scapular retraction 0.03 0.10 14 Pronation-supination, elbow at 90 -0.17 0.11 13 Shoulder flexion to 90, elbow extended -0.21 0.11 12 Hand to lumbar spine -0.40 0.11 5 Shoulder abduction -0.56 0.11 10 Elbow extension -0.64 0.11 11 Forearm pronation -0.87 0.11 31 Movement without tremor -0.91 0.11 29 Cylindrical grasp -1.10 0.12 25 Finger mass extension -1.25 0.12 3 Scapular elevation -1.40 0.12 24 Finger mass flexion -1.44 0.12 9 Shoulder adduction with internal rotation -1.56 0.12 7 Elbow flexion -1.76 0.13

PAGE 95

81 Individual with Severe Upper Extremity Motor Impairment FMA-UE Score = 18/60, Rasch Ability Measure = -1.25 (SE 0.34) logits Rating Scale Item Description 0 1 2 Wrist circumduction 0 1 2 Hook grasp 0 1 2 Shoulder flexion to 180, elbow extended 0 1 2 Spherical grasp 0 1 2 Lateral prehension 0 1 2 Wrist flexion-extension, elbow extended 0 1 2 Pronation-supination, elbow extended 0 1 2 Wrist stable, elbow extended 0 1 2 Movement with normal speed 0 1 2 Forearm supination 0 1 2 Shoulder abduction to 90, elbow extended 0 1 2 Movement without dysmetria 0 1 2 Shoulder external rotation 0 1 2 Wrist stable, elbow at 90 0 1 2 Wrist flexion-extension, elbow at 90 0 1 2 Palmar prehension 0 1 2 Scapular retraction 0 1 2 Pronation-supination, elbow at 90 0 1 2 Shoulder flexion to 90, elbow extended 0 1 2 Hand to lumbar spine 0 1 2 Shoulder abduction 0 1 2 Elbow extension 0 1 2 Forearm pronation 0 1 2 Movement without tremor 0 1 2 Cylindrical grasp 0 1 2 Finger mass extension 0 1 2 Scapular elevation 0 1 2 Finger mass flexion 0 1 2 Shoulder adduction with internal rotation 0 1 2 Elbow flexion -6 -4 -2 0 2 4 6 Measure (Logits) A) Individual with Moderate Upper Extremity Motor Impairment FMA-UE Score = 32/60, Rasch Ability Measure = 0.20 (SE 0.32) logits Rating Scale Item Description 0 1 2 Wrist circumduction 0 1 2 Hook grasp 0 1 2 Shoulder flexion to 180, elbow extended 0 1 2 Spherical grasp 0 1 2 Lateral prehension 0 1 2 Wrist flexion-extension, elbow extended 0 1 2 Pronation-supination, elbow extended 0 1 2 Wrist stable, elbow extended 0 1 2 Movement with normal speed 0 1 2 Forearm supination 0 1 2 Shoulder abduction to 90, elbow extended 0 1 2 Movement without dysmetria 0 1 2 Shoulder external rotation 0 1 2 Wrist stable, elbow at 90 0 1 2 Wrist flexion-extension, elbow at 90 0 1 2 Palmar prehension 0 1 2 Scapular retraction 0 1 2 Pronation-supination, elbow at 90 0 1 2 Shoulder flexion to 90, elbow extended 0 1 2 Hand to lumbar spine 0 1 2 Shoulder abduction 0 1 2 Elbow extension 0 1 2 Forearm pronation 0 1 2 Movement without tremor 0 1 2 Cylindrical grasp 0 1 2 Finger mass extension 0 1 2 Scapular elevation 0 1 2 Finger mass flexion 0 1 2 Shoulder adduction with internal rotation 0 1 2 Elbow flexion -6 -4 -2 0 2 4 6 Measure (Logits) B) Individual with Mild Upper Extremity Motor Impairment FMA-UE Score = 57/60, Rasch Ability Measure = 3.88 (SE 0.62) logits Rating Scale Item Description 0 1 2 Wrist circumduction 0 1 2 Hook grasp 0 1 2 Shoulder flexion to 180, elbow extended 0 1 2 Spherical grasp 0 1 2 Lateral prehension 0 1 2 Wrist flexion-extension, elbow extended 0 1 2 Pronation-supination, elbow extended 0 1 2 Wrist stable, elbow extended 0 1 2 Movement with normal speed 0 1 2 Forearm supination 0 1 2 Shoulder abduction to 90, elbow extended 0 1 2 Movement without dysmetria 0 1 2 Shoulder external rotation 0 1 2 Wrist stable, elbow at 90 0 1 2 Wrist flexion-extension, elbow at 90 0 1 2 Palmar prehension 0 1 2 Scapular retraction 0 1 2 Pronation-supination, elbow at 90 0 1 2 Shoulder flexion to 90, elbow extended 0 1 2 Hand to lumbar spine 0 1 2 Shoulder abduction 0 1 2 Elbow extension 0 1 2 Forearm pronation 0 1 2 Movement without tremor 0 1 2 Cylindrical grasp 0 1 2 Finger mass extension 0 1 2 Scapular elevation 0 1 2 Finger mass flexion 0 1 2 Shoulder adduction with internal rotation 0 1 2 Elbow flexion -6 -4 -2 0 2 4 6 Measure (Logits) C) Figure 1: Keyform recovery maps for three individuals with stroke.

PAGE 96

82 CHAPTER 3 LONGITUIDNAL STABILITY OF THE 30-ITEM FUGL-MEYER UPPER EXTREMITY ASSESSMENT ITEM DIFFICULTY HIERARCHY The primary goal of many post-stroke uppe r extremity (UE) mo tor rehabilitation providers is to observe change in UE mo tor function attributable to the antecedent intervention. Assessing change requires repeated measurem ent of function before and after intervention. Furthermore many rehabili tation researchers and clinicians desire to understand the time course of post-stroke UE motor recovery. This also necessitates repeated measurement of function longitudi nally over time. An assessment used to analyze clinical change and/or describe recovery must meas ure in a valid and consistent manner each time the assessment is administered [231]. The Fugl-Meyer Upper Extremity Assessment (FMA-UE) [15] is the most widely used clinical assessment of post-stroke UE mo tor function impairment [16]. It is often used to assess changes in UE motor functi on with rehabilitation. For example, the FMAUE was the primary criterion for evaluating re sponse to interventions such as rhythmic bilateral movement training [36, 37], neurom uscular stimulation [38], electromyographic triggered electrical stimul ation [39, 40], Botox [41], home based exercise [30], community based exercise [42], robot aided th erapy [43, 44], virtual reality [45], imagery [46], forced use [47], and m odified constraint induced movement therapy [48-51]. Furthermore, the FMA-UE is often used in longitudinal character izations of the poststroke UE motor recovery process [2, 29, 232, 233]. When studied as a whole assessment, the FMA-UE has shown excellent psychometric propert ies in test-retest

PAGE 97

83 situations [17, 18, 56]. However, ther e are no published analyses examining how individual FMA-UE items perfor m in test-retest situations. Because the FMA-UE is often used as a longitudinal measure of UE motor function, it is critical to test whether the items forming this construct remain longitudinally stable. Item Response Theory, e. g., the Rasch measurement model, offers a method to examine assessment stability at th e item level. Rasch analysis has been increasingly applied to health-care outcom es research to examine the stability of assessment items across testing occasions [161, 162, 182, 190]. In a previous cross-sectional study we used Rasch analysis and Principal Component Analysis to test th e dimensionality and construc t validity of the FMA-UE in a participant sample of 512 individuals 2 w eeks to 3 months poststroke. We found that three tendon-tap reflex items (biceps reflex, triceps reflex, normal reflex activity) were empirically disconnected from the other vol untary UE movement items. We removed the reflex items to form a modified 30-item FMA-UE. The 30-item FMA-UE items were arranged along a hierarchical continuum from least difficult to most difficult. This item hierarchy is a visual representation or map of the post-stroke UE recovery progression. Th is item difficulty hier archy was consistent with contemporary motor control science sugge sting that UE movements are more or less difficult depending on the task-specific comp lexities of the movement. The Raschderived item hierarchy was replicated across 98 % of our sample with mild to severe UE motor impairment. Ideally, the assessment should have the same item hierarchy each time it is administered. A stable item order assures that individuals with equivalent UE motor

PAGE 98

84 ability will have equal probability of receiving a score on a given FMA-UE item irrespective of the o ccasion at which it was assessed. While we would expect that individual UE motor abilities would change over time, we would not expect the item hierarchy (i.e., the post-stroke UE recovery map) to change over time. An unstable item order may threaten the assessments constr uct validity making it difficult to compare FMA-UE scores from one testi ng occasion to another testing o ccasion. It is possible that an assessments item difficulty order may une xpectedly reorganize with a repeated test even if the overall consiste ncy of the assessment as a whole remains high [180]. The purpose of the current study is to test whether the item difficulty hierarchy of the modified 30-item FMA-UE is equivale nt at subsequent testing occasions. Specifically, we explored whether FMA-UE items display the same statistical characteristics, specifically item difficulties, in one sample of poststroke individuals to whom the FMA-UE was administered at 2-w eeks (T1) and 6 months (T2) post-stroke. Testing the longitudinal stabil ity of the item hierarchy addresses the important issue of whether it is valid to make comparis ons of repeated-measure FMA-UE scores. Methods Participants A secondary analysis of FMA-UE scores from 459 individuals enrolled in the Kansas City Stroke Study [221] was conducted. Participants included in the Stroke Study met the following inclusion criteria; (1) st roke onset within 0 14 days prior to enrollment, (2) 18 years of age or older, (3 ) ischemic stroke as diagnosed by physician, (4) lived in the community prior to stroke ons et, and (5) able to participate in baseline testing. The FMA-UE was administered to part icipants at admission to the study, i.e., 0 14 days post stroke (T1) and 6 months post stroke (T2). Patien ts were evaluated by staff,

PAGE 99

85 all of who were health care professionals (e.g., nurses, physical therapists) and had undergone at least 2-weeks of training in the administration of the FMA-UE. For the purposes of our study we included only partic ipants with FMA-UE scores at both T1 and T2. This resulted in exclusion of 82 particip ants from the larger database for a final sample size of 377 participants. The 30-item Fugl-Meyer U pper Extremity Assessment The modified 30-item FMA-UE, produced in a previous cross-sectional study, is identical to the traditional assessment excep t that the 3 tendon-tap reflex items (biceps reflex, triceps reflex, and normal reflex activ ity) present in the traditional 33-item FMAUE [15] have been removed. Each of the 30-items are rated on a 3-point ordinal scale ( = unable to perform, = partial performance, and faultless performance). The items of the modified 30-item FMA-UE are arra nged in a difficulty order consistent with contemporary motor control framework in which post-stroke UE recovery is viewed as a process of re-acquiring the ability to meet task-specific UE movement demands. Movements that were easier (i.e., less complex) for most people with stroke (shoulder adduction with internal rotation, elbow flexi on, mass finger flexion) are located towards the easier end of the item hierarchy, and m ovements that were more challenging (i.e., more complex) for most people with stroke (wrist circumduction, lateral prehension, spherical grasp) are located towards the more difficult end of the item hierarchy. Ideally, item locations should be reproducible on different testing occasions. Data Analysis The analysis proceeded in four steps.

PAGE 100

86 Step 1: Dimensionality In order to create a legitimate measure, all items must contribute to the same construct, that is, the assessment must be uni dimensional [143]. We ascertained that all items of the 30-item FMA-UE illustrated a single underlying trait using the WINSTEPS [234] software program. Rasch analysis produces infit and ou tfit statistics to determine the extent to which items contribute to a unidi mensional construct. The infit statistic is most sensitive to ratings on items that are cl osely matched to partic ipants ability, the outfit statistic is most sensitive to ratings on items that are much easier or much harder than participants ability [142]. Fit statistics are reported as m ean square standardized residuals (MnSq) produced for each item of the instrument. MnSq represents observed variance divided by expected variance [172]. Consequently, the desired value of MnSq for an item is 1.0. We defined the acceptabl e criteria for unidimensionality as MnSq infit values < 1.7 associated with standardized Z values (ZSTD) < 2.0 as suggested by Wright and Linacre for assessments employing clinical observation [169]. High values (i.e., infit MnSq > 1.7, ZSTD > 2.0) indi cate that scores are variant or erratic, suggesting that an item belongs to a construct that is different fr om that represented by the other items of the instrument, or that the item is being inaccu rately scored. Low values indicate that the item is too predictable in its measurement. In this analysis we focused on items with high infit MnSq values, since this misfit repr esents a threat to construct validity. Examining the point-measure correla tion values of each item further dimensionality. This value is the correlation of an individual item with the other items of the assessment, omitting that item. The value indicates the extent to which the item contributes to measurement of the overall construct [235].

PAGE 101

87 Step 2: Equivalence of item difficulties To assess the longitudinal equivalence of the FMA-UE item structure, two procedures were followed. First, item difficulty parameters we re derived with separate T1 and T2 Rasch analyses. Individual item measur es were compared across the two testing occasions. Second, as suggested by Chang a nd Chan (1995) [231], an overall indication of the reproducibility of the item difficulty calibrations was obtained using an intraclass correlation coefficient (ICC) (model 2,1) [236]. Step 3: Differential item functioning A more robust comparison of T1 item difficulties to T2 item difficulties is accomplished through Differential Item Functioni ng (DIF) analysis. DI F is a statistical procedure commonly used to ascertain whet her test-item measures are consistent irrespective of test-taker char acteristics [181]. For the purpo ses of this study we used DIF to determine the consistency of item calibrations across testing occasions. The presence of DIF may indicate th at an item measures a unique trait at one testing occasion when compared to another testing occasion or may signal problems with the administration of the assessment [182]. There are various approaches for examini ng DIF (see [183] for review). The Rasch model employs a relatively straightforward DIF procedure (DIF=) in which each items difficulty is directly compared between each ti me point using iterative t-tests [166]. The correct alpha value for this analysis is deba ted in the literature. Kothari and Haley (2003) [237] suggest an appropriate crit ical value of 1.96 to detect differences in item difficulty ( =0.05). This value does not prot ect against a Type I error (i.e., the probability of falsely detecting a significant difference) during multiple comparisons, but it does protect against a Type II error (i.e., the probability of overlooking a significant difference if one exists).

PAGE 102

88 Small differences may occur by chance, theref ore necessitating a corrected alpha value. We utilized a Bonferroni adjustment based on the 30 items in the test (0.05/30, =0.0017). Using a technique similar to one recently used by Crane and Hart (2006) [238] we explored small DI F and large DIF. Small DIF detects marginal item difficulty differences by comparing the probabi lity of the difference to the unadjusted alpha level ( =0.05) and large DIF detects robus t item difficulty differences by comparing the probability of the diff erence to the adjusted alpha level ( =0.0017). DIF was visualized in a scatter plot of ite m difficulty values. A close fit of the item difficulty calibrations, within the 95% confidence interval measurement error boundaries, at each time point verifies ite m stability [239]. Items whose difficulty calibration logit values were more than 2 standard errors apart (falling outside the 95% confidence interval) were considered statistically different. Step 4: Impact on assessment of arm motor function The goal of examining the stability of the FMA-UE item difficulty order is to assure that the assessment is valid as a m easure of UE motor ability changes over time. The presence of DIF does not always transl ate to poor measurement of person ability [189]. Therefore to empirically determine if the presence of DIF had any impact on the FMA-UE ability estimates, the assessment was adjusted for DIF by excluding potential items displaying DIF. UE motor ability was th en measured at T1 and T2 using corrected and uncorrected assessments. The measurem ents obtained from the assessments were compared for similarity.

PAGE 103

89 Results Participants Characteristics of the 377-person sample are presented in Table 10. The sample was 69.2 11.2 years of age, 54% female and 80% white. The majority of the sample sustained ischemic stroke, with the location of the stroke equally di stributed right versus left hemisphere. The sample primarily sustai ned minor (42%) or m oderate (49%) stroke severity as defined by the Or pington Prognostic Scale [133]. Step 1: Dimensionality Table 11 presents Rasch derived item infit statistics for the 30-item FMA-UE at each time point. All items in the assessment demonstrated accep table fit statistics at T1, however hook grasp demonstrated an infit Mn Sq value > 1.70 with ZSTD > 2.0 at T2. Furthermore, all items demonstrated high poi nt-measure correlations ranging from 0.76 0.91 at T1 to 0.75 0.88 at T2. Step 2: Equivalence of item difficulties The relationship between T1 item difficulty calibrations and T2 item difficulty calibrations is illustrated in Figure 2. A separate line represents the item difficulties at each testing occasion. The dotted line illustra tes T1 item difficulties; the solid line illustrates T2 item difficulties. This figure demonstrates the relative invariance of the parameters. Despite some variations, the it em ordering and item difficulty values are fairly consistent. Shoulder adduction with inte rnal rotation, elbow fl exion, and scapular elevation were among the easiest items at both T1 and T2. Wrist circumduction, hook grasp, and lateral prehension were among the most challengi ng items at both T1 and T2. The ICC across item difficulties at T1 and T2 was 0.95.

PAGE 104

90 The reliability of comparing T1 and T2 item difficulties was assessed using an intraclass correlation coefficient (ICC) as suggested by Chang and Chan (1995) [231]. The present comparison yielded an ICC = 0.95 (model 2,1 [236]) indicating a high overall reliability between T1 item difficulties and T2 item difficulties. Step 3: Differential item functioning T1 vs. T2 Table 12 presents DIF analysis of the 30item FMA-UE item order. FMA-UE items are arranged in descending order, from more difficult items (top) to less difficult items (bottom). Rasch derived item difficulties for T1 and the standard error of the value are listed in the left two columns. Rasch deri ved item difficulties for T2 and the standard error of the value are listed in the next tw o columns. The DIF analysis compares the T1 and T2 measures of each item. The calcu lated t-value for each comparison and its associated probability level are listed in the right two columns. Two items: shoulder flexion to 180, elbow extended and movement with normal speed, show large DIF (using a p value corrected for multiple comparisons, p<0.0017). Six items: hand to lumbar spine, shoulder abduction to 90 degrees with the elbow extended, pronation/supination with the elbow extended, wrist stable with elbow bent movement without dysmetria and shoulder abduction, show small DIF (using a p-value not corrected for multiple comparisons, p<0.05). Of these eight items, four (shoulder flexion to 180, elbow extended, pronation/supination with the elbow extended, shoulder abduction to 90 degrees with the elbow extended, and hand to lumbar spine) display lower T1 item difficulty values than T2 values. The remaining four items (movement with normal speed, wrist stable with elbow bent, shoulder abduction, and movement without dysmetria) display higher T1 item difficulty values than T2 values.

PAGE 105

91 Figure 3 is a visual illust ration of DIF. The item difficulty parameters of the assessment calculated at T1 (x-axis) are cro ss-plotted with the item difficulties of the assessment calculated at T2 (y-axis). The solid lines represent the 95% confidence interval for these values. The items outside the error bands are labeled. The two items furthest outside the e rror interval are shoulder flexion to 180 degrees with elbow straight and, movement with normal speed. Six items difficulties are on or slightly beyond the error interval (hand to lumbar spine, shoulder abducti on to 90 degrees with the elbow extended, pronation/supination with the elbow extended, wrist stable with elbow bent, movement without dysmetria and shoulder abduction). Four items (shoulder flexion to 180, elbow extended, pronation/supination w ith the elbow extende d, shoulder abduction to 90 degrees with the elbow extended, and hand to lumbar spine) are above the error band indicating that the item was less difficult at T1 compared to T2. Four items (movement with normal speed, wrist stable with elbow bent, shoulder abduction, and movement without dysmetria) are below the error band indicating that the items were more difficult at T1 than at T2. Step 4: Impact of DIF on person ability measurement To compare the possible impact of DIF on the assessments validity as a measure of UE motor ability, the FMA-UE was adjusted by excluding items with DIF. T1 and T2 UE motor ability measures were calculated using three forms of the FMA-UE; (i) the full 30-item assessment, (ii) a 22-item assessment in which all DIF items (large and small) were removed, and (iii) a 28-item assessm ent in which only large DIF items were removed. Mean T1 and T2 person ability measures and their standard error of measurement (error bars) calculated using the three forms of the assessment are presented in figure 4. The pair of bars to the left illu strate the measures with no DIF items excluded

PAGE 106

92 from the assessment, the bars in the middle illustrate the measures with all DIF items (eight items) excluded from the assessment, and the bars to the right illustrate the measures with only large DIF items (two items) excluded from the assessment. No significant differences in the measurement of participants UE motor function are evident in the 30-item assessment compared to the 22-item and 28-item assessments neither at T1 nor at T2. The effect sizes for the change in UE motor function from T1 to T2 are virtually identical with and w ithout DIF items (effect size = 0. 52, 0.51, 0.51) for the three assessment versions. Discussion The purpose of this study was to test the equivalence of the FMA-UE item hierarchy across two testi ng occasions. We examined data from 377 post-stroke individuals to whom the FMA-UE was admini stered at two points during their recovery process. We tested whether FMA-UE items di splayed invariant difficulty calibrations at each time point in the longitudinal study. Ou r data shows that the item difficulty order for eight items varies significan tly, but that this DIF has no practical consequences for measuring UE motor function. Rehabilitation researchers are often inte rested in whether an intervention is effective. To this end, researchers choose an appropriate assessment to score an outcome of interest and compare scores between and/or within groups. Regardless of the statistic used to assess effectiveness (e.g., comparison of group means, correlatio n test, etc.) it is implicitly assumed that the measurement prope rties of the assessment are identical each time the assessment is used [180]. In the present study, the majority of the FMA-UE items have equivalent T1 and T2 difficulty measures. However, the measures are not entirely identical. The scatter plot and

PAGE 107

93 DIF statistical testing shows that eight item show significant DIF. Specifically, two items displayed difficulty estimates that were sign ificantly different with a corrected alpha value (large DIF) at the two time points, and six other items displayed difficulty estimates that were significantly differe nt with an uncorrected alpha value (small DIF) at the two time points. Item instability may warn that an item is functioning in an unexpected way, i.e., the item measures a different trait on one testing occasion than another. Applied to the FMAUE, item instability may reveal that the pos t-stroke UE motor ability recovery map illustrated by the item difficulty hierarchy at T1 is fundamentally different than the map illustrated by the item order at T2. Four items (shoulder flexion to 180, elbow extended, pronation/supination with the elbow extended, shoulder abduction to 90 degrees with the elbow extended, and hand to lumbar spine) displayed lower item difficulty values at T1 when compared to T2. That is, these items relocated from the easy end of the item difficulty hierarchy (at T1) towa rds the more challenging end of the item hierarchy (at T2). Four other items (movement with normal speed, wrist stable with elbow at 90, movement without dysmetria, and shoulder abduction) showed the opposite trend. That is, these items relocated from the more chal lenging end of the item difficulty hierarchy (at T1) towards the easy end of the item hierarchy (at T2). According to item response theory, an indivi duals response to an assessment item depends on the individuals ability (with regard s to the construct be ing assessed) and the items difficulty. Test-takers with equal levels of UE motor ability should have the same probability of responding correctly to an a ssessment item irrespective of the time poststroke at which the assessment is administ ered. A stable item (i .e., displaying no DIF)

PAGE 108

94 maintains a consistent location on the pos t-stroke UE motor recovery map and therefore affords the same proba bility of correct response at each test situation. An unstable item (i.e., displayi ng DIF) does not maintain a constant location on the poststroke UE motor recovery map, and therefore does not afford the same probability of correct response at each test situation. The occurrence of DIF among the FMA-UE ite ms suggest that an individual tested at 0 14 days post-stroke is more likely to successfully perform hand to lumbar spine, shoulder flexion to 90 with elbow extended, pronation-supination w ith elbow extended, and shoulder flexion to 180 with elbow extended than an individual with equivalent UE motor ability tested at 6 months post-stroke An individual tested at 0 14 days poststroke is less likely to successfully perform shoulder abduction, wrist stable with elbow bent, movement without dysmetri a, and movement with normal speed than an individual with equivalent UE motor ability tested at 6 months post-stroke. Of course individuals with more skill (i.e., greater UE motor abi lity) are more likely to successfully perform FMA-UE items when compared to individuals with less skill. However, the presence of DIF suggests that the assessments underlying co nstruct has shifted. This may threaten the assessments construct validity. Bingenheimer et al. (2005) [181] argues th at measurement equivalence is necessary for valid interpretation of assessment scores. Specifically, an inst rument should display item and scalar equivalence meaning that an assessments items s hould assess the same trait each time it is given (item equivalence), and identical scores should always have the same interpretation (scalar equivalence). The items of an assessment are the operational definition of the trait the assessment is intend ed to measure, and the score quantifies how

PAGE 109

95 much of the trait the test-taker possesses. If an item varies with regards to its meaning or the score varies with regards to its interp retation, the assessment is not an accurate measure of the trait. Bingenheimer argues that without convincing evidence of item and scalar equivalence, there is no basis for co mparing scores on an assessment between or within groups. In the healthcare liter ature DIF analyses have been useful to study item equivalence (i.e., to detect potential test item bias) among persons of different cultures and diagnostic groups. For example, using DIF analyses, Az ocar and colleagues (2003) [185] showed that Latino-Americans were more likely to receive a higher score on some items of the Beck Depression Inventory than non LatinoAmericans, potentially flagging a distinct Latino cultural interpretation of the items. Az ocar recommended a reconsideration of the assessments Spanish translation in order to facilitate cross-cultural comparisons of assessment scores. Tennant and colleagues (2004) [186] found that certain Functional Independence Measure self-car e items displayed DIF. The researchers showed that culturally unique ways of bathing and gr ooming influenced the way the items were administered among European countries, To facilitate cross-country comparisons of scores the assessment scoring procedure wa s adjusted for those items. Roorda and colleagues (2004) [187] found th at the hierarchical item ordering of a stair climbing assessment was different for amputees versus non-amputees. The researchers recommend using a shortened assessment, with DIF items removed, if the intention is to compare scores between the two diagnostic groups. In the present study we used DIF analyses for a different purpose. Instead of testing the measurement equivalence of an assessment between cultures or diagnostic groups, we

PAGE 110

96 examined the same group of people and compared the measurement equivalence of the assessment between testing occasions. The us e of DIF analyses for this purpose has precedence in the literature. Tesio et al (2003) [182] detected DIF among several Functional Independence Measure mobility items (transfer to a chair, locomotion, and stairs) when admission scores were compared to discharge scores. The items were easy for a group of orthopedic patients with mob ility restrictions at admission (e.g., no lower extremity weightbearing due to orthopedic precautions). However, the items became more difficult for this group at discharge, af ter the mobility restrictions were removed. This counterintuitive result led the staff at th is facility to examine the manner in which the assessment was scored. They learned that the admission score was based on a prediction of the patients expected outcome, while the discharge score was based on actual observation of patient performance. In essence, the assessm ent was measuring a different construct at patient admission (e.g., ability of the staff to predict patient status) versus patient discharge. Finlayson and Mallinson (2005) [161] a pplied DIF analysis to the primary functional outcome measure used in a la rge longitudinal stud y of aging and ADL performance in various health care settings. Rasch analysis of a 24-item ADL assessment was conducted at three time points over th e 13-year study. Although DIF was detected among 4 items, the researchers were unable to discover a specific reason for it. Because statistically significantly differences in item di fficulty do not always translate to practical differences in person ability scores [189], the researchers measured ADL ability using an assessment in which the 4 unstable items we re removed and compared the results to measurement using the full assessment. Th e analysis suggested that DIF had only

PAGE 111

97 minimal impact on the Rasch-derived ability measures over time or across settings leading the researchers to continue using the entire assessment. How can DIF be interpreted in the presen t study? The presence of DIF may suggest that the FMA-UF is multidimension al [181]. It is possible that shoulder flexion to 180 degrees with elbow extended is a less demanding task for individuals earlier in the poststroke recovery process than later in the pr ocess because spasticity and/or soft tissue shortening may occur over time. However, th ere does not seem to be an obvious trait shared by the DIF items that would lead to the hypothesize that the motor control construct represented by the items is conceptu ally different at 0-14 days post-stroke than at 6 months post-stroke. It is possible that our results coul d be attributed to potential differences in item administra tion at T1 compared to T2. Perhaps poor postural (e.g., decreased trunk control at T1 versus T2) and/or seating system (e.g., seated in a wheelchair at T1 versus a chair at T2) conf ounded the administration of the items. It is also possible that clients in terpreted the item directions differently at each time point. These possibilities merit further investigation. The presence of DIF can have significant impact on the validity of an assessment for measuring person ability. Linacres (1994) [190] study provided a classic example of the impact of item instability on the measurement of patient change. Using Rasch analyses, Linacre found that when the 13 mo tor and 5 cognitive items of the Functional Independence Measure were combined into a single 18-item assessment, the admission and discharge item difficulty ca librations of the items were not equivalent. He suggests that the presence of variant item difficulty measures indicated that a single pattern of disability could not be iden tified when comparing patient admission scores to discharge

PAGE 112

98 scores. That is, the meaning of the scor e was dependent on the time point (admission versus discharge) at which the assessment was administered. Linacr e demonstrated that the statistical validity of th e scale was improved if the motor and cognitive items were separated into sub-scales. The relationship between the potential it em bias detected in our study and the measurement of person ability was undertaken in a way similar to th at of Finlayson and Mallinson (2005) [161] and recommended by Roznowski (1999) [189]. That is, we excluded DIF items, measured person-ability with the shortened assessment, and then compared the ability measures of the full scale to the shortened scale. There was essentially no difference in person ability m easured by assessments with and without DIF items excluded. Furthermore, the effect si zes were identical fo r each assessment. The results suggest that the amount of DIF in the FMA-UE is negligible had no practical consequence on estimates of UE motor functi on. Therefore we suggest keeping all 30 items in the assessment for longitudinal measurement of UE motor ability. Study Limitations There are several limitations that might limit the generalizability of this study. First, although all items of the modified 30-item FMA-UE demonstrated acceptable fit statistics at T1, hook grasp wa s erratic at T2. The item disp layed infit MnSq (1.82) and ZSTD (7.80) values which are beyond the acceptable ranges suggested by Wright and Linacre (1994) [169]. Smith ( 1996) [240] argues that Rasch an alysis cannot proceed if its assumptions of unidimensionality are not met. High infit statistics, as displayed by hook grasp, may signal that the item is measuri ng a construct different than the construct measured by the other assessment items. We did not explore removing the item prior to proceeding with the DIF analysis in the pres ent study, as would be a common procedure.

PAGE 113

99 We did not do so for two reasons. The item, a lthough erratic in the T2 analysis of this study, was not erratic in the T1 analysis or in the analysis we pe rformed in a previous study with a larger sample size. Also, there is no motor control theory that would support a single prehension pattern su ch as hook grasp, belonging to a unique motor control construct. The second limitation to this study is that we did not inves tigate whether other forms of DIF are present in the FMA-UE. Fo r example, non-uniform DIF is a special type of item instability in which item difficulty characteristics change across clients of different ability levels. The Rasch method for detecting DIF used in this study has been method is criticized because of its inability to detect non-uniform DIF [183]. The presence of non-uniform DIF in the FMA-UE s hould be examined in future studies using appropriate methods for its detection. Finally, in this study we examined the longitudinal stability of the 30-item FMAUE rather than the traditional 33-item FMA-UE. Therefore ou r findings cannot be directly applied to the traditional 33-item FMA-UE. However, as we argued in a previous study, the 30-item FMA-UE is a unidimensi onal assessment with an item hierarchy consistent with contemporary motor control theory and evidence. The results of our previous study suggest that the three reflex items reflect a different motor control construct than the other items of the assessm ent. We purposely did not include the items in the present analysis to avoid DIF resulti ng from multi-dimensionality (i.e., DIF items do not consistently assess a single trait). Conclusion In test-retest situations it is important to examine, rather than assume, the longitudinal stability of te st item difficulties. Alth ough it is common to check the

PAGE 114

100 construct validity of an assessm ent at the beginning of a longi tudinal study, and check the reliability of the assessment at various points along the study, it is not common to check the stability of the assessment items [180]. Item stability is important since the items represent the construct an assessment intends to measure. Without assurance of item stability it is difficult to know whether the assessment measur es the same trait across each testing occasion. The present study illustrates a procedure fo r examining the longitudinal stability of assessment items. The presence of item instabi lity may threaten repeated measures score comparisons and thus, this procedure should be a routine check of an assessments measurement properties. DIF te sting might also be a method useful to inform motor control theory. For example, testing item stab ility could yield valuable insights into the way motor control is theori zed to recover over time.

PAGE 115

101 Table 10: Sample char acteristics (n = 377) Years of Age (Mean SD) 69.2 11.2 Gender Males 174 (46%) Females 203 (54%) Race White 301 (80%) African-American 63 (17%) Other 13 (3%) Stroke Type Ischemic 355 (94%) Hemorrhagic 22 (6%) Stroke Location Right Hemisphere 173 (46%) Left Hemisphere 204 (54%) Stroke Severity Minor (Orpington Prognostic Score <3.2) 160 (42%) Moderate (3.2 <= Orpington <= 5.2) 184 (49%) Severe (Orpington > 5.2) 33 (9%)

PAGE 116

102 Table 11: Rasch derived infit statistics T1 and T2 (n = 377) FMA-UE Item Description T1 item MNSQ T1 ZSTD T1 rPM T2 item MNSQ T2 ZSTD T2 rPM Hard Wrist circumduction 0.93 -0.8 0.79 0.93 -0.8 0.83 Hook grasp 1.52 5.4 0.76 1.82 7.8 0.75 Lateral prehension 1.18 2.0 0.80 1.39 4.1 0.79 Spherical grasp 1.18 2.1 0.80 1.03 0.4 0.83 Shoulder flexion to 180, elbow ext 0.82 -2.3 0.84 0.83 -2.1 0.84 Wrist stable, elbow extended 1.10 1.2 0.80 1.04 0.5 0.84 Wrist flexion-extension, elbow ext 0.76 -3.1 0.86 0.67 -4.3 0.87 Movement with normal speed 1.27 2.9 0.80 0.85 -1.7 0.86 Pronation-supination, elbow ext 0.64 -4.8 0.87 0.72 -3.6 0.87 Forearm supination 0.77 -3.0 0.88 1.02 0.3 0.86 Wrist stable, elbow at 90 1.11 1.2 0.84 1.04 0.5 0.84 Movement without dysmetria 1.58 5.7 0.80 1.40 3.7 0.81 Wrist flexion-extension, elbow at 90 0.72 -3.5 0.89 0.69 -3.7 0.88 Shoulder external rotation 0.88 -1.4 0.88 0.81 -2.0 0.88 Scapular retraction 1.09 1.0 0.86 1.19 1.9 0.84 Pronation-supination elbow at 90 0.68 -4.0 0.89 0.72 -3.0 0.88 Shoulder abduction to 90, elbow ext 0.77 -2.8 0.88 0.80 -2.3 0.86 Palmar prehension 1.16 1.7 0.85 1.05 0.5 0.85 Shoulder flexion to 90, elbow ext 0.86 -1.6 0.88 0.81 -2.0 0.86 Shoulder abduction 1.00 0.1 0.88 1.05 0.5 0.85 Elbow extension 0.76 -2.6 0.90 0.87 -1.2 0.87 Forearm pronation 0.77 -2.6 0.90 1.15 1.3 0.83 Hand to lumbar spine 0.83 -1.7 0.89 1.14 1.3 0.84 Movement without tremor 1.62 5.3 0.83 1.32 2.9 0.82 Finger mass extension 0.99 -0.1 0.89 0.88 -1.1 0.86 Cylindrical grasp 1.06 0.6 0.88 0.91 -0.7 0.84 Finger mass flexion 1.12 1.0 0.89 0.85 -1.3 0.86 Scapular elevation 1.10 0.9 0.89 1.21 1.6 0.83 Elbow flexion 0.94 -0.5 0.90 1.08 0.7 0.82 Easy Sh adduction with internal rotation 0.79 -1.9 0.91 1.06 0.5 0.84

PAGE 117

103 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5Shoulder add with int rotation Finger mass flexion Elbow flexion Scapular elevation Finger mass extension Cylindrical grasp Movement without tremor Hand to lumbar spine Elbow extension Forearm pronation Shoulder abduction Sh flexion to 90, elbow extended Palmar prehension Sh abd to 90, elbow extended Pronation-supination, elbow at 90 Scapular retraction Wrist flex-ext, elbow at 90 Shoulder external rotation Wrist stable, elbow at 90 Movement without dysmetria Forearm supination Pronation-supination, elbow extended Wrist flex-ext, elbow extended Movement with normal speed Sh flex to 180, elbow extended Wrist stable, elbow extended Lateral prehension Spherical grasp Hook grasp Wrist circumduction Item Difficulties T1 Item Difficulties T2 Figure 2: item difficulty measures of T1 (0 14 days post-stroke) and T2 (6 months poststroke) Table 12: Differential item functioning an alysis (n= 377, df = 594, ** p<0.0017, p<0.05) FMA-UE Item Description T1 item difficulty (Logits) T1 error T2 item difficulty (Logits) T2 error Calculated tvalue Probability Hard Wrist circumduction 1.76 0.12 1.97 0.12 -1.24 0.2155 Hook grasp 1.41 0.12 1.49 0.12 -0.48 0.6301 Lateral prehension 1.21 0.12 1.35 0.12 -0.80 0.4247 Spherical grasp 1.31 0.12 1.17 0.12 0.83 0.4054 Shoulder flexion to 180, elbow extended 1.04 0.12 1.61 0.12 -3.36 0.0008 ** Wrist stable, elbow extended 1.11 0.12 0.92 0.13 1.07 0.2831 Wrist flexion-extension, elbow extended 0.93 0.12 1.20 0.12 -1.53 0.1259 Movement with normal speed 0.96 0.12 0.38 0.13 3.26 0.0012 ** Pronation-supination, elbow extended 0.82 0.12 1.26 0.12 -2.56 0.0106 Forearm supination 0.80 0.12 0.75 0.13 0.30 0.7654 Wrist stable, elbow at 90 0.36 0.13 -0.18 0.14 2.89 0.0040 Movement without dysmetria 0.48 0.12 0.06 0.14 2.29 0.0226 Wrist flexion-extension, elbow at 90 0.23 0.13 0.31 0.13 -0.45 0.6545 Shoulder external rotation 0.25 0.13 0.14 0.13 0.61 0.5445 Scapular retraction 0.09 0.13 0.19 0.13 -0.56 0.5753 Pronation-supination elbow at 90 0.00 0.13 -0.22 0.14 1.19 0.2363 Shoulder abduction to 90, elbow extended -0.06 0.13 0.35 0.13 -2.23 0.0262 Palmar prehension -0.08 0.13 -0.13 0.14 0.24 0.8126 Shoulder flexion to 90, elbow extended -0.36 0.13 -0.20 0.14 -0.80 0.4247 Shoulder abduction -0.39 0.13 -0.88 0.15 2.40 0.0167 Elbow extension -0.69 0.14 -0.68 0.15 -0.06 0.9557 Forearm pronation -0.61 0.14 -1.02 0.15 1.96 0.0499 Hand to lumbar spine -0.71 0.14 -0.24 0.14 -2.36 0.0186 Movement without tremor -0.80 0.14 -0.51 0.14 -1.48 0.1396 Finger mass extension -1.28 0.15 -1.02 0.15 -1.22 0.2213 Cylindrical grasp -1.25 0.15 -1.62 0.16 1.67 0.0958 Finger mass flexion -1.62 0.15 -1.28 0.16 -1.51 0.1315 Scapular elevation -1.52 0.15 -1.88 0.17 1.55 0.1223 Elbow flexion -1.55 0.15 -1.88 0.17 1.44 0.1493 Easy Shoulder adduction with internal rotation -1.79 0.16 -1.79 0.17 0.02 0.9848

PAGE 118

104 30-item FMA-UE-3 -2 -1 0 1 2 3 -3-2-10123T1 (0 -14 days post-stroke)T2 (6 months post-stroke) shoulder flexion to 180 degrees with elbow extended pronation/supination with elbow extended movement with normal spee d wrist stable with elbow bent shoulder abduction to 90 degrees with elbow extended hand to lumbar spine shoulder abduction movement without dysmetria 30-item FMA-UE-3 -2 -1 0 1 2 3 -3-2-10123T1 (0 -14 days post-stroke)T2 (6 months post-stroke) shoulder flexion to 180 degrees with elbow extended pronation/supination with elbow extended movement with normal spee d wrist stable with elbow bent shoulder abduction to 90 degrees with elbow extended hand to lumbar spine shoulder abduction movement without dysmetria Figure 3: scatter plot of 30-item FMA-UE item difficulties at T1 versus T2 Mean person ability score1.82 1.93 1.88 -0.19 -0.05 -0.12-0.25 0.25 0.75 1.25 1.75 2.25no correctionall DIF items excudedonly large DIF items excludedlogits T1 T2 n = 377, error bars +/2 SEM Figure 4: Comparison of mean person abil ity with and without DIF adjustment

PAGE 119

105 CHAPTER 4 LINKING THE FUGL-MEYER TO THE WOLF MOTOR FUNCTION TEST: ASSESSING THE CONTINUUM OF IMPAIRMENT TO FUNCTION Each year over 700,000 individuals in the United States have a stroke [1] with three-fourths of these indi viduals experiencing some de gree of upper extremity (UE) paralysis [2]. Although UE motor function im proves rapidly within the first 3-months post-stroke [3], it rarely recovers completely At four years post-s troke 85% of survivors report residual arm or hand motor impair ment severe enough to interfere with performance of daily tasks [4]. Clearly, th ere is an ongoing need to aim post-stroke rehabilitation interventions at reducing UE motor impairment. Furthermore, rehabilitation researchers must assure that interventions e ffective for reducing impairment also improve the clients ability to use his/her ar m in real life daily activities. The Fugl-Meyer Upper Extremity Assessmen t (FMA-UE) is the most widely used measure of post-stroke UE motor impair ment [16]. The FMA-UE is the primary assessment for evaluating the success of post-st roke rehabilitation interventions such as rhythmic bilateral movement training [ 36, 37], neuromuscular stimulation [38], electromyographic triggered el ectrical stimulation [39, 40] Botox [41], home based exercise [30], community based exercise [42], robot aided therapy [43, 44], virtual reality [45], imagery [46], forced us e [47], and modified constrai nt induced movement therapy [48-51]. However, one of the primary probl ems facing users of the FMA-UE is the difficulty determining the functional relevance of its impairment-level score [24, 51]. That is, its score has no obvious functional interpretation. What does it mean that a client

PAGE 120

106 has shown a 10-point improvement in his/her FMA-UE score? Is this client better able to accomplish activities of daily living as a resu lt of the intervention? Although the 10-point FMA-UE gain may be statistically significant, what is the clinical significance of this change [136]? Conceptually, there are two approaches for determining what constitutes significant change: distributi on and anchor (or criterion) based approaches [61, 138]. Distribution-based methods define a signi ficant change score as one that exceeds a pre-defined level of error [ 24, 138, 241, 242]. However, the results of distribution-based approaches depend on the characteristics of the sample. Therefore it is unlikely that there exists a single change scor e that defines significant change across all samples and contexts. On the other hand anchor-based methods relate a change score to a functionally relevant criteri on pre-determined by client or clinician. This criterion anchors the meaning of the change. Both approaches have been applied to interpretation of FMA-UE ch ange scores [24, 42, 242]. Feyes et al. (1998) [135] compared 50 individuals with st roke who received sensorimotor stimulation to the affected lim b to 50 individuals with stroke who did not receive the intervention over the first year pos t-stroke. The FMA-UE was administered at 6 and 12 months. The researchers found that the FMA-UE change score was approximately 10% greater for those who recei ved intervention compared to those who did not receive intervention. The researcher s suggested the magnitude of the difference between the mean change scores represented the distinction between more impaired and less impaired stages of recovery, and therefor e was a clinically relevant difference. Van der Lee et. al. (1999) [47] found no literature guiding the specification of a FMA-UE

PAGE 121

107 minimal clinically important difference, but noted that a 10% change in the overall FMA-UE score was of a magnitude that made measurement error unlikely [24]. Using Van der Lees criteria, McCombe-Waller and Whitall (2005) [139] interpreted their 8% gain in FMA-UE score as approaching clinic al relevance. Pang et al. (2006) [42] took a criterion-based rather than distribution based approach. In this study the researchers simultaneously administered the FMA-UE and a client-satisfaction survey to 63 participants in a community based exerci se program. Following the intervention, treatment effect sizes (assessed by th e FMA-UE) ranged from 0.71 0.81, and participants reported increased ability to pe rform simple functional tasks such as now able to switch on/off lights. Pang concluded that the treatment effects appear to be clinically significant from the patients perspective. Item Response Theory offers another appro ach to anchor-based interpretation of the FMA-UE change score. IRT provides the th eory and method for lin king the FMA-UE to a more functionally oriented assessment, th ereby embedding functiona l criterion into the FMA-UE structure [143, 191, 192]. Linking is a method that re-calibrates separate assessments onto a common measurement scale. The theory behind this process has been termed scale free measurement. Scale free measurement means that what one measures is independent of the instrument used to make the measurement. Applied to post-stroke UE rehabilitation, scale-free measurement theorizes that an abst ract trait (e.g., UE motor ability) can be measured by an infinite number of items representing the trait. Any two assessments are simply sub-samples of this item pool a nd therefore can be mathematically linked.

PAGE 122

108 Linking establishes the relationship between different assessments therefore allowing direct comparison of items and scores [167]. The Wolf Motor Function Test (WMFT) [26, 201] is a logical choice of assessments to link with the FMA-UE. The World Health Organizations International Classification of Functioning a nd Disability (ICF) [62] provi des a conceptual framework to define UE motor behaviors characterizi ng a specific health do main and provides a framework for measuring UE motor behavior s across health domain s [199]. According to the ICF, body function includes control a nd coordination of single and multiple joint movements. The FMA-UE is an assessmen t of UE body function impairment [14]. The first seven items of the WMFT may also represent this domain because these items require simple limb movements without functional endpoint (e.g., place forearm on table, place hand on table) and appear to be logical extensi ons of FMA-UE items. For example, the FMA-UE requires an indivi dual to perform shoulder ab duction to 90 degrees with elbow extended and flex shoulder to 90 degrees with elbow extended. The WMFT requires an individual to extend the elbow to the side by slid ing it across a table top and lift the hand from a table to place it on a box placed at shoulder height. The items appear to require similar UE motor ability. The ICFs activity performance domain includes reaching, grasping, and lifting. The last eight WMFT items represent this domain because the items require performance of functional tasks (e.g., lift soda can, lift a pencil) all of which require reaching to, gr asping and lifting an object. These items can be viewed as logical extensions of the FM A-UE hand items. For example, the FMAUE requires an individual to perform cylindr ical grasp (i.e., maintain the shape the hand around a small can against minimal re sistance), and palmar prehension (i.e.,

PAGE 123

109 maintain ones hold on a pencil against mi nimal resistance). The WMFT items are functional applications of these hand positions requiring the individual to lift a soda can and lift a pencil. Again, the FMA-UE and WMFT items appear to require similar UE motor ability. Therefore, it is reasonabl e to postulate that the two assessments measure a common underlying UE motor ability trait. If the FMA-UE and WMFT are both measur es of UE motor ability they can be linked to a common scale thereby creating a single item hierarchy comprised of impairment-level FMA-UE items and mo re functional-level WMFT items. Linking would provide a more comprehensive range of behaviors eviden cing post-stroke UE motor recovery. IRT, specifically, Rasch analysis places person measures and item difficulty on the same linear continuum. Th erefore, a FMA-UE score (measure) can be connected to the likelihood of successfully achieving (or not achieving) particular items on the assessment. Furthermore, if the impair ment-level items of the FMA-UE are linked to the functional-level items of the WMFT, the combined assessment allows the score (measure) to be connected to the likeli hood of successfully performing non-functional and/or functional tasks The overall purpose of this study is to link the FMA-UE and WMFT using Rasch analysis. To do so, we will test the hypothe sis that the FMA-UE and the WMFT measure the same UE motor ability construct. We w ill also test the hypothesis that there is a threshold value at which the impairment-lev el FMA-UE score transitions to the more functional-level WMFT score.

PAGE 124

110 Methods Participants Secondary analysis of data from 100 persons enrolled in a random ized clinical trial of therapeutic exercise was performed [220]. Persons with stroke were included in the exercise study if they met the following inclusion criteria; (1) stroke within 30 to 50 days, (2) ability to ambulate 25 ft independently, (3) mild to moderate stroke deficits defined by a total Fugl-Meyer score of 27-90 (uppe r and lower extremiti es), (4) Orpington Prognostic Scale score of 2.0 to 5.2, (5) pa lpable wrist extensi on on the involved side, and (6) Folstein Mini-Mental Status exam score > 16 [220]. The FMA-UE and WMFT were administered to the sample prior to and immediately following the three month, inhome, progressive exercise inte rvention targeting balance, flexibility and strength (the WMFT was known as the Emory motor assessm ent [200] at the time of the exercise study). The assessments were administered according to a standardized procedure by trained staff. Because the purpose of the pr esent study was ultimately to explore the usefulness of linking the two assessments fo r interpreting change scores, we included only participants who had completed bot h assessments pre (T1) and post (T2) intervention. This resulted in the exclusion of 9 individuals who were lost to follow up, or failed to complete one assessment. Thus the final sample size for the present study was 91 participants. Instruments The WMFT is comprised of two scales the Functional Abilities Scale and performance time scale. The F unctional Abilities Scale is an observational rating scale in which the test administrator rates the patien ts quality of movement on a 6-point ordinal rating scale ( patient does not attempt to use more affected arm for task to affected

PAGE 125

111 arm is used for the task and its movement appears normal). Performance time is the number of seconds, recorded with a hand-held stopwatch required by the client to complete each of the 15 WMFT tasks. If the cl ient is unable to complete a task within two minutes, the task is consid ered unachievable and a score of 120 seconds is given. The final time score reported is the median tim e required for all timed tasks as per WMFT protocol [201]. A lower score (faster performa nce) indicates more UE motor ability. The WMFT has been shown to have high inte rrater reliability (intraclass correlation coefficients, 0.88), internal c onsistency (Cronbach 0.86), and test-retest reliability (r 0.90) [26, 201]. The Functional Abilities Scal e rating scale data was not available, so for the purposes of the present study we used the WMFT performance time data. To facilitate comparison of the FMA-UE and WMFT with Rasch analysis, the WMFT performance time data was re-coded into a rating scale. The overall intention of the procedure used to recode the data was to capture the connection between performance time (seconds) and UE motor ability thereby pr eserving as much information contained in the data as possible. Participants performan ce time data was divided into performance categories, which were ordered to reflect in creasing amounts of UE motor ability (slower time = less ability). Each category was assi gned a rating scale rank (lower value = less ability). Category boundaries were established using a deductive process in which cut-off times were checked against crite ria, and if the criterion wa s not achieved, other cut-off times were selected. This procedure was c ontinued until the criteria were achieved [243246].

PAGE 126

112 The intent of the criteria was to insure that the rating-scale selected preserved the measurement characteristics of the original time data. The following criteria is based on those outlined by Linacre (2002) [ 247] and Hands (1999) [244]: 1. At least 10 observations per rating scale category for all it ems to assure the stability of item difficulty and pe rson ability parameters. 2. Monotonically increasing item step calib rations to assure that increasing performance on each item is reflected as increasing progressi on thorough the rating scale steps. 3. Person separation index value > 2.0. A person separation index describes how well the instrument differentiates i ndividuals in the sample. It is desirable to have a high person separation index (persons spread out along the measurement scale) because the intention of the WMFT is to discriminate between persons of various levels of UE motor ability. 4. A strong correlation between estimates of UE motor ability measured with the traditional WMFT (summed time to perform each item) and the Rasch person measures (logits) that were derived usi ng the converted-WMFT data. This step assures that the score conversion captures the information originally obtained using the WMFT performance times. Data Analysis Once the WMFT data was recoded, a six-stage methodology was employed. We tested the WMFT for unidimensionality (step 1) and longitudinal stability (step 2). Once the validity of the WMFT item content and st ructure was established, we combined the FMA-UE and WMFT into a common item pool and concurrently calibrated item and person parameters (step 3). The newly cr eated FMA-UE + WMFT item hierarchy was tested for unidimensionality (step 4) and l ongitudinal stability (step 5). Finally, we explored interpretation of FMA-UE change sc ores using the linked assessment (step 6). Step 1: Dimensionality In order to create a legitimate measure, all items must contribute to the same construct [143]. The first step towards test ing the dimensionality of the combined FMAUE + WMFT item pool was to ascertain wh ether each assessment functioned as a unidimensional measure when examined independently. Once unidimensionality of the

PAGE 127

113 separate assessments was established, the it ems were pooled and the dimensionality of the combined assessment was examined. In previous studies, we established the uni dimensionality and longitudinal stability of a modified 30-item FMA-UE. Therefore th e initial focus of the present study was to establish the unidimensionality and l ongitudinal stability of the WMFT. Principal Components Analysis (PCA) A principal components analysis (P CA) was performed (SPSS v. 11) on categorized WMFT data collect ed at two time points: prior to the exercise intervention (T1) and following the exercise interven tion (T2). The intent of the PCA was to reorganize the multivariate data (participants WMFT ordinal item scores) into a limited number of components (factors) so that each component captur ed a substantial amount of the overall variance within the dataset. That is, the purpose of the PCA was to detect obvious factors within the WMFT. The WMFT is comprised of two types of UE motor tasks; single/multiple joint motions (items 1 7) and functional tasks (items 8 15) [26]. To date, there are no published studies that ha ve examined whether the items represent a single trait or separate constructs. We hypothe sized that the assessment items illustrated a single dimension: UE motor ability. To test the hypothesis we examined PCA eigenvalues and factor loadi ng statistics. We employed the Kaiser rule [157] as the criteria for retaining principa l components, i.e., only the components with an eigenvalue > 1.0 were retained. We expected one ei genvalue would explain the majority of the variance in the data and considered a co mponent to be a unique dimension if it contributed > 10% of explanat ory variance [161, 248]. Furthe rmore we expected that the majority of WMFT assessment items would co rrelate with a single principal component as evidenced by factor loadings on th e first component > 0.40 [219].

PAGE 128

114 Rasch analysis A Rasch partial credit analyses was applie d to the WMFT data using the Winsteps software program [234]. The part ial credit analysis allows items to have unique, rather than consistent, response structures. The pa rtial credit model was considered the most appropriate for the data because the rating scale was uniquely defined for each item based on item-specific cut-off times. Dimensionality of the data was further ex amined with Rasch-derived fit statistics. Under the Rasch model, each person with a given level of UE motor ability is expected to do well (score higher) on relativ ely easy tasks, and have more difficulty (progressively lower scores) on tasks of increasing difficu lty. The model expectations are assessed through item goodness-of-fit statistics. Fit statis tics are reported as the mean square (MnSq) of the item residuals, where a residual is the difference between a persons actual response to an assessment item and the response expected by the Rasch model. Consequently the MnSq statistic has an ideal value of 1.0 with departures from this value indicating potential departures fr om unidimensionality. Fit statis tics are also reported in a standardized form (ZSTD), which is the tr ansformation of the mean square of the residuals to a statistic with a z-distribution. Two types of fit statistics are calculated with Rasch analysis. The infit statistic is wei ghted according to th e variance in response patterns (i.e., the pattern of ratings that the sample exhi bited on a given item) on items with difficulties similar to the ability of th e sample. The outfit statistic is an unweighted value based on the sum of the mean squares of the residuals and therefore is sensitive to items that depart from model expectations wh en the item is much harder or easier than a persons ability level (i.e., outliers). Th e acceptable criteria for unidimensionality depends on the intended purpose of the meas ure and the degree of rigor desired. For

PAGE 129

115 surveys using ordinal rating sc ales, Wright and Linacre [169 ] suggest reasonable ranges of MnSq fit values between 0.5 and 1.7 asso ciated with standardized z-values (ZSTD) less than 2.0 for clinical observations. Th is study focused on items with high infit statistics, i.e., those items that depart from model expectations with regards to person ability, because they pose the greatest threat to unidimensionality. Item correlations Two types of item correlations were cal culated using the Winsteps program, Cronbachs alpha and point measure correlati ons. Cronbachs alpha is an index of internal consistency or homogeneity of an assessment [249]. When all items of an assessment measure a single attribute the item s correlate with one another, i.e., the assessment is internally consistent. Nunnally (1978) suggests Cronbachs alpha values > 0.70 to evidence satisfactory in ternal consistency [250]. Po int-measure item correlations (rPM) are the correlation of an item with the other items of the assessment, omitting that item. The value indicates the extent to which an item contributes to measurement of the assessments overall construct [235]. Point-meas ure correlations were calculated in the same manner as point-biserial correlations, exce pt that Rasch analys is replaced total raw scores with Rasch-derived para meters [165]. The rPM coefficien ts were interpreted in the traditional manner; coefficients below 0.50 s uggested poor reliability, coefficients from 0.50 0.75 suggested moderate reliability, a nd values greater than 0.75 suggested good reliability [236]. Step 2: Longitudinal Stability When an assessment is used to measure change, the longitudinal stability of the item hierarchy should be examined to determin e whether it is valid to make comparisons of repeated-measures [180]. In other words, it is important to assure that all items

PAGE 130

116 measure the same trait at each testing occas ion. A robust longitudinal investigation of item difficulties is accomplished through Diffe rential Item Functioning (DIF) analysis. DIF is a statistical procedure in which item difficulty estimates are examined, one at a time, to ascertain whether items display the sa me statistical characte ristics irrespective of test-taker characteristics (e.g ., gender, culture) or testing o ccasion. The presence of DIF may indicate multidimensionality or problems with the administration of the assessment [182]. Following a procedure that we had prev iously applied to the FMA-UE, we performed a DIF analysis of th e WMFT with Winsteps [234]. The analysis calculated the differences in WMFT item difficulty estimat es between T1 and T2. The significance of the differences was tested with iterative pair ed t-tests. We made a Bonferroni correction for multiple comparisons based on the number of items in the WMFT, thus setting the alpha value at 0.05/15 = 0.0033. Step 3: Linking and Concurrent Calibrati on of the FMA-UE + WMFT Item Pool The goal of this study was to develop a general UE motor ability scale, and specifically to relate the lo cation of item content across the two assessments. After establishing the dimensionality and longitudina l stability of each assessment separately, we proceeded with linking the assessment s by merging the FMA-UE and WMFT items into a common item pool. Fisher (1997) s uggested that concu rrent calibration of instruments with two types of rating scales is best accomplished by analyzing the data using two Rasch models simultaneously [197]. Following these guidelines, we concurrently applied a Rasch rating scale an alysis (Groups = 1) to the FMA-UE items, and a Rasch partial credit analyses (Groups=0 ) to the WMFT items, combining the items of both instruments into a single analysis. Applying the two Rasch models to the data

PAGE 131

117 allowed the items of the FMA-UE to share one rating scale structure, and the items of the WMFT to share another. The concurrent analysis simultaneously estimated item difficulty and person ability parameters in th e overall subject and item pool so that the two assessments were linked to th e same measurement scale. Step 4: Dimensionality of the Common Item Pool We tested the hypothesis that the item pool represented a single trait, UE motor ability, using the same procedures describe d above. We applied a traditional PCA using the same PCA criteria (eigenvalues > 1.0, unique dimensions defi ned by contributing > 10% of explanatory variance [161, 248], and f actor loadings > 0.40 [157]) as described above. In addition we conducted a factor rotation of the principal components to further explore our hypothesis. Factor rotation is a mathematical technique in which the multidimensional axes of initial PCA factor load ings (the number of axes matches the number of retained factors) are rotated to model the data. The rotation re-expressed the factors as a less complex structure to f acilitate interpretation [164] This procedure produced a factor matrix with factor lo ading values (equivalent to re gression coefficients) indicating the weight given to the item when interpre ting the factor [157]. We considered factor loadings meaningful if the absolute valu e exceeded 0.40 [219]. Finally, we conduced a Rasch analysis and examined Rasch-derived fi t statistics with the same criteria for acceptable infit statistics (MnS q < 1.7 with ZSTD < 2.0 [169]) and examined item correlations. Step 5: Longitudinal Stability of the Common Item Pool We examined DIF statistics to test whethe r all items of the overa ll UE motor ability scale were longitudinally stable with regards to their item difficulty calibrations using a

PAGE 132

118 corrected alpha value for multiple comparisons based on the number of items within the combined item pool (0.05/45 = 0.0011). Step 6: Interpretation of Change Scores Once the dimensionality and longitudinal st ability of the item hierarchy derived with concurrent calibration was established we addressed the final hypothesis for this study; that there exists a t hreshold FMA-UE score at which changes in a FMA-UE score translate to changes in the WMFT score. To accomplish this, the combined assessment was plotted onto an item map. The item map displayed the items of each assessment along a common measurement metr ic. The FMA-UE + WMFT item map also displayed the relationship between the assessments with regards to the overlap in the range of the trait measured by each asse ssment. The boundaries of the range were determined by examining the extreme category es timates for the easiest and most difficult items of each assessment [148, 251]. A categor y estimate is the average ability measure for the individuals receiving either the lowest rating ( for the FMA-UE or for the WMFT) on the easiest item, or the highest ra ting ( for the FMA-UE or for the WMFT) on the hardest items. The Rasch analysis was used to determine these values for each assessment, and then the values were plotted onto the item map. Functional threshold The point at which the functional items of the WMFT (i.e., items 8 15) overlap the FMA-UE was defined as the functional th reshold. The functiona l threshold is the zone in which individuals have enough UE mo tor ability to be ab le to successfully perform the more functional items of the WMFT. The lower boundary of this zone was defined as the logit value corresponding to the lowest category es timate of the easiest WMFT functional item.

PAGE 133

119 Functional interpretation of a FMA-UE change score FMA-UE person ability measures were calcul ated in separate T1 and T2 analyses with FMA-UE item difficulties anchored to the values derived from the concurrent calibration. To explore interpretation of FMAUE change scores, person ability measures at each time point were overlaid onto th e item map. The item map displayed the functional threshold. Thus, the overlay of pe rson ability measures onto this map visually illustrated the relationship between item difficu lties and person abilities. This allowed for comparison of change scores between individu als with more severe versus less severe stroke with regards to the likelihood that indi viduals were able to perform UE movements (i.e., FMA-UE items) and/or simple UE func tional tasks (i.e., WMFT items), i.e., gain enough UE motor ability to move past the functional threshold. Results Participants Characteristics of the 91-person sample are presented in Table 13. Participants were 69.4 10.3 years of age with slightly more males than females represented in the sample. A majority of the participants were Caucasian. Participants were 76.1 27.1 days post stroke at admission to the study with an approxi mately equal distributio n of right and left hemispheric involvement. The majority of th e sample had a minor (54.9%) or moderate (39.6%) stroke as defined by th e Orpington Prognostic Score [252]. Conversion of WMFT Performance Time Data to Performance Categories A rating scale should meaningfully separate participants into ability levels. The greater the number of rating scale categorie s created, the greater the potential for increasing the measurement precision of the scale which leads to finer discrimination between individuals [244]. However, if too many rating scale categories are created, it is

PAGE 134

120 likely some ratings will not be well represented, i.e., the width of a particular rating category may be too narrow to correspond to a distinct participant ability level [253]. We compared two ways of creating rati ng scale performance categories from time data based on examples cited in the lite rature. We first divided each item into 5 performance categories ( = very severely impaired performance to = near normal performance) with cutoffs at the 20th, 40th, 60th, and 80th pe rcentiles. Percentiles were calculated separately for each item. This pe rcentile division was employed by Hands et al. (2001) [243] to convert childrens moto r performance data obtained from serial assessments each with distinct units of m easurement (e.g., time to walk a distance, number of skips) to a common rating scale for Rasch analysis. We also recoded the WMFT time data into a 4-point performa nce category rating scale based on quartile divisions (25th, 50th, and 75th percentiles). Percentiles were calculated separately for each item. Williams (2005) [246] and Reeves (2004) [245] categorized continuous data (walking speed) and interval data (financi al assets) into ratin g-scale performance categories with quartile divisions for Rasch analysis. Rasch analysis was applied to each dataset and results were compared. We considered four criteria outlined by Linacr e (2002) [247] and Hands (1999) [244] to determine which performance category rating scale was optimally functioning. Results from two items, forearm to box and turn key, were chosen to represent our reasoning process. These results are presented in Ta ble 14. Results from the other 13 items are similar. Criteria 1: at least 10 obser vations per rating scale categ ory for all items. Table 14 presents two examples of the rating scale st atistics, comparing results from the 4-point

PAGE 135

121 scale (left side of the table) to results from the 5-point scale (right side of the table). The number of observations pe r rating scale category for forearm to box ranges from 20 23 for the 4-point scale, and 5 25 for the 5-poi nt scale. The number of observations per category for turn key ranges from 17 31 for the 4-point scale, and 10 26 for the 5point scale. The 5-point scal e does not have the recommended number of observations per category in the rating scale category for the item forearm to box, and barely meets the criteria in the rating scale category for the item turn key. Inspection of other items (data not presented) reveals that 15/15 items me t Linacres criteria when the 4-point scale was employed, but only 4/15 item s met Linacres criteria when the 5-point scale was employed. Among the 11/15 items no t meeting Linacres criteria with the 5point scale, the rating of either or (sometimes both) had less than 10 observations per category. This means that these rating scale categories, representing the least amount of UE motor ability (most severely impaired ), were not well repr esented by the 5-point rating scale solution. As the in clusion criteria for the exerci se study was mild to moderate stroke deficits defined by a total Fu gl-Meyer score of 27-90 (upper and lower extremities), and Orpington Prognostic Scale score of 2.0 to 5.2 (mild to moderate impairment), poor representation of individuals with severe stroke in the sample is reflected in the lack of observations for the lower categories. The 4-point rating scale appears to meet the first criteria better than the 5-point scale because it has a wider range of cut off boundaries for th e lower ability ratings. Criteria 2: monotonically incr easing item step calibrations An item step calibration (also termed item step threshold [165]) can be conceptualized as the boundary between adjacent rating scale categories, i.e., the bounda ry between = severely impaired motor

PAGE 136

122 performance and = impaired motor pe rformance. Table 14 presents the step thresholds for the two representative items. St ep thresholds calculat ed with the 4-point rating scale increase monotonically for both item s shown in the table. For example, under the 4-point scale the item forearm to box exhibits an increasing pr ogression of thresholds (.44, 0.00, 0.44) as does the item turn ke y (-0.97, 0.11, 0.86). U nder the 5-point scale the item forearm to box exhibits increasing steps from .92, -0.03, 1.11 to 1.84. However, the item turn key exhibits a disordered step threshold. The first step is .85 but the second step decreases to .07 before increasing to 0.65 and 1.26. All other items (data not presented) exhibit monotonically in creasing step threshol ds under the 4-point scale. Turn key is the only item with disordered thresholds under the 5-point scale. Criteria 3: person separation index value > 2.0. The person separation index for the 4-point scale were 2.93 (reliability = 0.90), a nd for the 5-point scale was 3.78 (reliability = 0.93). The person separation values indicate that 4-point scale split the sample into 4 statistically distinct strata (as per the equatio n: strata = (4Gp + 1) / 3 where Gp is the person separation index value [254]), and the 5point scale split the sample into 5 strata. Criteria 4: strong correlation between estimates of UE mo tor ability measured with the traditional WMFT (summed time to pe rform each item) and the Rasch analysis (logits). Prior to calculating the correlati on between the Rasch-derived ability estimates and the original performance time data, we examined the performance time data for outliers to identify any extreme scores that ma y distort the statistical analyses. An outlier was defined as a performance time greater th an or equal to three standard deviations beyond the mean performance time for that it em [236]. Because the outliers were actual true data (i.e., were not aberrant) all outlier s were retained, and adjusted to one decimal

PAGE 137

123 point outside the closest non-outlying score [246]. The correlation between the sum of each participants performan ce time and the Rasch-derived ability measure showed an inverse relationship as exp ected because higher performance time means less UE motor ability and visa versa. Using the 4-point ordina l scale the absolute value of the correlation between WMFT time scores and Rasch-de rived person ability measures was 0.77 indicating that the 4-poi nt ordinal scale was strongly correlated with the time scores. Using the 5-point ordinal scale, this value dropped slightly to 0.71 but still reflected a strong correlation betw een the measures. The intention of the conversi on of the WMFT continuous ti me data to ordinal data was to preserve as much of the informati on contained in the continuous performance time data as possible with an optimal and effective rating scale. To that end, we compared two rating scales against quality control criteria. Bo th scales appeared to satisfy the majority of the criteria. However, the 4-point rating s cale was chosen since it satisfied all of the criteria especially with regard to criteria 1. Step 1: Dimensionality of the WMFT Table 15 presents the results from the WMFT PCA at T1 (left) and T2 (right). According to the Kaiser rule [157], the anal ysis retained two fact ors with eigenvalues greater than one. Each analysis showed the first eigenvalue to large (8.28 T1, 8.54 T2) indicating that a single factor explains the ma jority of the variance in the data, 55% and 59% respectively. The second eigenvalue is small (1.54 T1, 1.12 T2) meaning that a second factor contributes little (10% and 7%) explanation of the data. Table 16 presents the unrotated principal component factor pattern for each of the two retained factors at T1 (left) and T2 (ri ght). In an unrotated factor pattern matrix, factor loadings are similar to correlation coe fficients and are consider ed meaningful if the

PAGE 138

124 value is > 0.40 [157]. At each time point, a ll items correlate str ongly with one factor. First factor loadings range from 0.65 0.83 at T1 and 0.64 0.87 at T2. Table 17 presents Rasch derived Wolf Motor Function Test item measures, fit statistics, and point measure correlations. In fit statistics for all items at each time point are below the cutoff suggested by Linacre and Wright [247] (MnSq values are < 1.7 with associated ZSTD values < 2.0). The Cronbachs alpha values for the assessment were = 0.94 at T1, and = 0.95 at T2 suggesting the items have strong inter-correlations and the assessment is internally consistent. Item point measur e correlations (rPM) (table 17) are str ong; ranging from 0.63 0.78 at T1 and 0.64 0.80 at T2 suggesting that the items work together to measure a common trait. In summary, the dimensionality tests (PCA Rasch infit statistics, Cronbachs alpha and item point measure correlations) suggest that the WMFT is a unidimensional assessment. Table 18 presents the Wolf Motor Function Test item difficulty hierarchy with items arranged in T1 difficulty order so that easy items are at the bottom and more challenging items at the top. Item difficulties spread only 0.50 logits at T1 (0.25 logits to .25 logits) and 0.22 logits at T2 (0.09 log its to .13 logits). In general, items representing simple limb movement s without functiona l endpoint (e.g., hand to table and forearm to table) are easier than items representing functional activit ies requiring fine hand movements (e.g., turn key and lift pencil). Step 2: Longitudinal stability of the WMFT Table 19 presents the Wolf Motor Func tion Test Differential Item Function Analysis. From left to right, the table presen ts the T1 and T2 item difficulty measures and

PAGE 139

125 standard error of the measure followed by t-test results. The alpha value for this analysis was based on a Bonferroni correction for 15 comparisons, thus the level of significance was determined to be p<0.003. No items di splay statistically signi ficant differences in item difficulty, suggesting that the item difficulties remain consistent at each testing occasion. In summary, the dimensionality and diffe rential item functioning analyses suggest that the WMFT functions as a unidimensiona l, longitudinally stable assessment. The modified 30-item FMA-UE has been previous ly shown to have the similar measurement properties (see Chapter 3). Step 3: Linking and Concurrent Calib ration of the FMA-UE + WMFT The overall goal of this study was to develop a general UE motor ability scale by combing the FMA-UE with the WMFT. To accomplish this, we ran a co-calibrated Rasch analysis, with the items for both instruments Step 4: Dimensionality of th e FMA-UE + WMFT Item Pool A PCA was conducted on T1 and T2 meas ures derived from FMA-UE and WMFT items from the co-calibrated analysis (SPSS v. 11). Table 20 presents the eigenvalues of the FMA-UE + WMFT linked asse ssment PCA. Figure 5 presen ts the scree plots of the FMA-UE + WMFT linked assessment PCA. Usi ng the Kaiser rule [157] (retain factors with eigenvalues > 1) the PCA retained 8 f actors at T1 and 10 factors at T2. At each testing occasion the first eigenvalue wa s large (19.06 T1; 17.56 T2), the second eigenvalue was much smaller (3.21 T1; 3.29 T2), and subsequent eigenvalues were progressively smaller. The first factors accoun ted for 42% and 39% of the variance in the data at T1 and T2, respectively, while the second factors only cont ributed an additional 7% of explanatory variance at each time point.

PAGE 140

126 Because the Kaiser rule suggested retaini ng so many factors, Ca ttells scree test was employed [157, 158]. The scr ee plot (T1 to the left, T2 to the right of Figure 5) is a graph of all eigenvalues in their decreasing order. The plot looks like the side of a mountain, and "scree" refers to the debris fa llen from a mountain and lying at its base. According to the scree test, factors are retained if their data points oc cur prior to the point at which the mountain ends and the debris begins. Eigenvalues of scree are trivial because of measurement error, only those that markedly deviate from a linear trend are valuable to the analysis [157]. According to this rule, 2 or possibly 3 factors should be retained in this analysis (See Figure 5). Although we hypothesized that the item pool represents a single construct, UE motor ability, it is possible th at the item pool represents tw o constructs; one represented primarily by FMA-UE items another by WMFT items. The FMA-UE is comprised of a majority of items requiring the client to pe rform UE movements in free space, that is, achieve a limb configuration wi thout a specific target endpoint or interaction with object. In contrast, the WMFT items require the client to perform a specific task or move to a specific target. Furthermore the WMFT places a speed demand on the clients performance (e.g., move as fast as you can) whereas the FMA-UE places no such demand on the client. Therefore, to explore if the items split into assessment-specific dimensions, we retained two factors and performed a factor rotation (varimax) to facilitate factor interpretation. Table 21 presents the FMA-UE + WMFT item pool rotated factor matrix. The items from the FMA-UE and WMFT are listed in the left column, FMA-UE items above, and WMFT below. Factor load ings for two factors at T1 are presented first (middle

PAGE 141

127 columns), and factor loadings for two factor s at T2 are presente d next (right side columns). Strong factor loadings (values > 0. 40 [157]) are shaded in grey. In general, FMA-UE items tend to load on factor 1 while WMFT items tend to load on factor 2. Seventy-three percent (22/20) FMA-UE items solely load onto factor 1 at T1, and 86 percent (26/30) at T2. Fiftythree percent of the WMFT items load solely on factor 2 at T1 and 100 percent at T2. Two items in T1, scapular elevation, a nd movement without dysmetria, fail to load strongly onto any factor (items marked with a single asterisk). Twelve items at T1 and 4 items at T2 cross-load onto more than one factor (marked with a double asterisk). Cross-loaded items inter-correlate with the factors and are generally considered uninterpretable within the fram ework of one factor [255]. For example, the items shoulder flexion to 90 with elbow exte nded, and lift soda can load ont o factor 1 (0.66, 0.43) and factor 2 (0.45, 0.68) at T1 and the items s houlder flexion to 180 with elbow extended and movement with normal speed load onto factor 1 (0.60, 0.71) and fact or 2 (0.45, 0.42) at T2. In summary, while the majority of the PCA variance for the combined FMA-UE and WMFA is accounted by a si ngle factor, there are some indications that the two instruments may be measuring distinct UE mo tor ability constructs (see discussion of multidimensionality in the discussion secti on below). For purposes of this study, we chose to continue the analysis with the entire item pool. Table 22 presents the FMA-UE + WMFT ite m measure, fit stat istics, and point measure correlations. Only 1 item in th e item pool (FMA-UE movement without dysmetria) showed infit valu es that are beyond the acceptabl e ranges described by Wright

PAGE 142

128 and Linacre (1994) [169] at T1 and approach unacceptability at T2 (MnSq >1.7 with associated ZSTD values > 2.0, are shaded in grey). The results sugge st that all remaining items fit the unidimensional expectations of the Rasch model at each time point. Person separation was 5.02 T1 and 4.69 at T2 with person reliability at 0.96 at each time point, indicating that the samp le was split into 4 statistica lly distinct strata [254]. The internal consistency of the pooled ite ms within the overall UE motor ability scale was strong (Cronbachs = 0.96) at both time point s suggesting the items measure a similar underlying construct and the item pool is internally valid [249]. Table 22 presents correlations. The item point-measure item correlation values (rPM) at T1 ranged from 0.22 0.76 and at T2 ranged from 0.19 0. 79. Four items at T1 and 7 items at T2, all FMA-UE items, demonstrated poor rPM valu es (below 5.0). These values are shaded in grey in the columns labeled rPM in ta ble 22. While a subset of the FMA-UE items showed poor rPM values, a majority of the pool ed items (91 percent (41/45) at T1 and 84 percent (38/45) at T2) showed acceptable values suggesting that the majority of the items contribute to the measurement of a singl e overall construct of UE motor ability. Step 5: Longitudinal Stability of the FMA-UE + WMFT Item Pool Figure 6 presents the FMA-UE + WMFT differential item functioning scatter plot. The item difficulty parameters of the item pool calculated at T1 (x-axis) are crossplotted with the item difficul ties of the item pool calculated at T2 (y-axis). The solid lines represent the 95% confidence interval for thes e values. The items outside the error bands are labeled. Three of the 45 items fall outside the error interval (forearm supination, wrist stable with elbow extended, wrist stable with elbow bent). The items are below the error band indicating that the items we re unexpectedly more difficult for the sample at T1 than at T2. However, no T1 to T2 comparisons of item difficulty were statistically significant

PAGE 143

129 at a significance level corrected for multiple comparisons (p<0.0011) suggesting that the three items falling outside the error interval in figure 6 may do so because of chance (i.e., type I error, p<0.05). In summary, the dimensional and longitudinal stability analysis results show that while the FMA-UE + WMFT combined it em pool exhibits some degree of multidimensionality the results tend to suppor t viewing the item pool as a unidimensional set of items representing a single underlying trait. Both the traditional PCA and PCA of the standardized residuals suggest the WMFT items measure a some what different trait than the FMA-UE. However, a single factor expl ained the vast majority of the variance in the data in both of the PCA analyses. Although one FMA-UE item demonstrated poor infit statistics at one time point (movement w ithout dysmetria), this item did not misfit at T2, nor did it misfit in our previous Rasch anal yses of the FMA-UE with a larger dataset. The rPM statistics show that 11 FMA-UE items display poor item measures. The DIF statistics suggest that three FMA-UE items di splay instability of item difficulties at the two testing occasions, however, in a previous analysis with a larger sample only one of these items (wrist stable with elbow bent) not stable. Finally the Rasch analysis showed that all of the pooled items had acceptable inf it statistics with strong item point measure correlations). Step 6: Interpretation of Change Scores The FMA-UE + WMFT items were combin ed into an overall UE motor ability measure. The item difficulty parameters were concurrently calibrated so that range of content of the two assessments could be di rectly compared. Figure 7 displays the FMAUE + WMFT linked item map. FMA-UE items are located to the left of the diagram, WMFT are located to the right of the diag ram. The center scale represents UE motor

PAGE 144

130 ability, measured in logits, ordered from less ability (bottom of the diagram) to more ability (top of the diagram). The logit scale is an equal interv al scale. Note that each item is located along the UE motor ability me tric. Each items lo cation on the metric corresponds to the average UE motor ability required for an individual to have a 50% probability of successfully accomplishing th at motor behavior. For example, an individual has a 50% probability of performi ng the WMFT item, hand to lumbar spine if he/she has 0.5 logits of UE motor ability while 1.3 logits of UE motor ability is required for a person to have a 50% probabi lity of successfully performing the FMA-UE item, picking up a pencil. The FMA-UE + WMFT item map also illust rates the range of UE motor ability measured by each assessment. The shaded boxes surrounding each item set represent this range. The FMA-UE items span a wider range overall and are more evenly distributed along the measurement scale than the WMFT ite ms. The low end of the range (bottom of the colored box) is calculated as the mean UE motor ability score for the individuals who received the lowest rating ( for the FM A-UE and for the WMFT) on the easiest item on each assessment. The high end of the ra nge (top of the colored box) is calculated as the mean UE motor ability score of th e individuals who recei ved the highest rating ( for the FMA-UE and for the WMFT ) on the hardest item on each assessment. The FMA-UE covers a range of UE motor abilit y content 4.80 logits wide with the lowest average measure for the easiest FMA-UE item (elbow flexion) being .69 logits while the highest average measure for the hardes t item (shoulder flexion to 180 with elbow extended) being 2.11 logits. The WMFT covers a range of UE moto r ability content 3.22 logits wide with the lowest average measur e for the easiest WMFT item (hand to table)

PAGE 145

131 being .98 logits while the highest average measure for the hardest WMFT item (turn key) being 2.24 logits. The upper half of th e FMA-UE range overlaps with the full range of the WMFT items. Person measures have a direct relationship to the item difficulties of the combined FMA-UE WMFT instruments as illustrated by the person UE motor ability distribution curve that has been overlaid onto Figure 7. The peak of th e distribution curve corresponds to the mean person-ab ility measure of the sample at T1. Items at located at an individuals ability measure reflect items that the individual has a 50% probability of being able to achieve. For example, a person with an UE motor ability measure of -1.0 logits has a 50% probability of being able to pronate and supinate his forearm with elbow bent. He likely can accomplish shoulder abduction and scapul ar elevation (-1.0 logits), but not is not likely to be able to place ha nd to lumbar spine (-0.5 logits), achieve hook grasp (0.5 logits) or lift a pencil (1.3 logits). A person with 1.3 logits of UE motor ability is likely to be able to perform all of these motor behaviors. An individuals change score can also be directly connected to a descripti on of the UE motor beha viors that he can or cannot perform. For example, while a person with 0.0 logits of UE motor ability can perform palmar prehension (hold a pencil betwee n the tips of his thumb and index finger against slight resistance) (0.0 l ogits), he is unlikely to be able to lift the pencil (1.3 logits). If this individual responds to an UE rehabilitation intervention as evidenced by a 1 logit increase in his UE motor ability score (from 0. 0-1.0 logits), this he has likely gained the ability to pronate and supinate forearm with elbow extended (0.9 logits) and place hand to box (1.0 logits), but still ca nnot lift the pencil (1.3 logits). A further 0.05 logit increase

PAGE 146

132 in his/her UE motor ability would make it likel y that he/she can lift the pencil (1.3 logits) and turn a key (1.3 logits). Functional threshold We would expect that depending on the i ndividuals initial UE motor ability we should see different interpretations of FM A-UE change scores. We hypothesized that there exists a threshold zone of UE motor ability correspo nding to the point at which a FMA-UE score can be directly connected to the likelihood that the individual can successfully perform the WMFT functional item s (items 8 15). This threshold zone is defined as the point at which the functiona l items of the WMFT overlap the FMA-UE and is represented by the shaded rectangular area of the UE motor ability l ogit ruler in Figure 7. The lower boundary of this zone was determined by calculating the lowest category estimate of the easiest WMFT ite m. The easiest functional WMFT item was stack checkers. Its lowest category estimate (i .e., the mean person score of the individuals who received a on this item) was calculated by WINSTEPS as 0.45 logits. In essence, this overlapping area reflects the amount of UE motor ability (represented by the FMAUE items) needed to perform simple functional UE motor tasks (represented by the WMFT). Functional interpretation of a FMA-UE change score Figure 8 presents FMA-UE person ability measures overlaid onto the FMA-UE + WMFT item map to illustrate how the threshold zone aids in interpretation of FMA-UE change scores. Figure 8 presents the UE mo tor ability logit ruler running vertically through the center of the diagra m. The shaded box near the top of the UE motor ability logit ruler is the functional threshold (corre sponding to the shaded area of the ruler in

PAGE 147

133 Figure 7), that is, the range of UE motor ab ility measured by the WMFT functional items 8 15. Rasch derived FMA-UE motor ability meas ures (i.e., person ability measures in logits) for the 91-person sample, are overlaid onto this item map. T1 FMA-UE ability measures are displayed to the left of the ruler, and T2 FMA-UE ability measures are displayed to the right of the ruler. The T1 and T2 measures of two individuals are highlighted. The readers attention is drawn to in dividuals #49 and #29 whose identification codes are marked with a box. These individuals will be compared because while each individual demonstrat es similar FMA-UE change scores, the meaning of the FMA-UE change score differs dependi ng on the individual s T1 measure. At T1 individual #49T1 (boxed identifi cation code) demonstrated a FMA-UE ability measure of .39 logits. At T2 this individual demonstrated a FMA-UE ability measure of .55 logits (#49T 2). Individual #49 experienced a 1.84-logit gain in UE motor ability; however, his/her T2 ability m easure is not in the functional threshold zone (shaded area). Because his/her T2 ability measure does not matc h the difficulty of WMFT functional items, he/she is unlikely to be able to successfully perform these WMFT functional items. Therefore it is un likely that he/she e xperienced functional benefits from the rehabilitation intervention. In comparison, individual #29 (boxed iden tification code) experienced a similar FMA-UE change score but with a different functional interpretation. At T1 this individual #29 demonstrated a FMA-UE ab ility measure .64 logits, and at T2 a FMAUE ability measure of 1.16 logits. Similarly to individual #49, he/she demonstrated a 1.80-logit gain in UE motor ability, howe ver, by T2, individual #29 gained enough UE

PAGE 148

134 motor ability to move into the threshold z one. His/her T2 UE motor ability matches the difficulty of WMFT functiona l items making it likely that he/she can successfully perform these functional items. In contrast to individual #49, indi vidual #29 appears to have experienced functional benefits fr om the rehabilitation intervention. In summary, although individuals #49 a nd #29 experienced nearly identical FMAUE change scores, the individual beginning th e intervention with more UE motor ability appeared to have experienced greater f unctional benefits of the intervention. Discussion The purpose of this study was to lin k the FMA-UE and WMFT using Rasch analysis. The results suggest that the two inst ruments can be conceptualized as measuring a single trait UE motor ability. It was feasible to link the two assessments to a common measurement scale. While others [42] have related the FMA-UE score to external functional criterion, this st udy demonstrates a method to directly embed a functional criterion within the FMA-UE by linking th e FMA-UE to the WMFT. This method is advantageous because it affords a connec tion between the FMA-UE score an its functional interpretation. Linking an impairment-level motor assessment with a more functionally-oriented motor assessment o ffers a novel approach to exploring two important issues facing rehabilitation scientists; 1) interpreting impairment-level change scores in terms of more functional-level gains, and 2) understanding the relationship between stroke-related UE motor impairment and activity performance. Linking as a Method to Expl ore Significant Change The linked item map provides an opportunity to consider the interaction between changes in individual ability and functional responses to sp ecific items. The map shows that FMA-UE change scores for individuals at the lower end of the scale (those with less

PAGE 149

135 ability) will have different meaning than change scores for individuals in the middle or at the top of the scale (those with more ability ). Individuals with less ability will require larger increases in the FMA-UE score be fore functional gains are realized, and individuals with more ability will realize f unctional gains with smaller increases in the FMA-UE score. While traditional met hods for defining responsiveness (i.e., distribution-based approaches seek to define clinically significant change in terms of a single change score value [24, 138, 241, 242], the FMA-UE + WMFT item map demonstrates that a single value will not ad equately capture the significance of change for all individuals. Linking as a Method to Measure Body Function and Activity Performance The item map may be useful for testing specific hypotheses regarding measurement of impairment-level and/or activity-level moto r behaviors. For example, the creators of the WMFT hypothesized that th e assessment is most approp riate for individuals with mild and moderate stroke related UE moto r impairment [26, 200] Our results not only support this hypothesis but also po int to a specific quantity of UE motor ability that is necessary to be successful with WMFT items Viewing FMA-UE change scores (Raschderived person ability measures) with regards to the threshold marked by the UE motor ability necessary to successfully perfor m functional WMFT items illustrates that individuals with less post-stroke UE motor ability will require larger gains in motor ability before gaining functional motor skills when compared to those individuals with more post-stroke UE motor ability. Pang (2006) [42] found that high baseline FMA-UE scores experienced greater ability to perfor m WMFT items when comp ared to those with lower baseline FMA-UE scores. The FMA-UE + WMFT item map offers a direct way to

PAGE 150

136 examine the relationship between stroke-rela ted UE motor impairment and performance of specific functional activities. Linking as a Way to Conceptualize the Motor Rehabilitation Treatment Process The Rasch-derived item difficulty hierarchy may illustrate the pattern of post-stroke recovery. Specifically, the hierarchy illustrate s a progression of the reacquisition of single joint and multiple joint movements (FMA-UE it ems) to reacquisition of simple functional reaching, grasping and lifting tasks (WMFT ite ms). This hierarchy is consistent with contemporary motor control expectations th at arm movements are easier or more challenging based on the inherent task-spe cific neural, mechanical and contextual complexities of the movement [122, 126, 127, 256, 257] (see Chapter 2). Contemporary motor control research sugge sts that task-specific increases in neural, mechanical and/or contextual re quirements concomitantly increase the computational demands on the motor system [112]. This concept may explain why WMFT items are more difficult than FMA-UE items. First, the WMFT items require interaction with an object (e.g ., table, box, soda can etc.). Interacting with an object not only increases the mechanical load on the limb but also increases the perceptual processing requirements of the task, e.g., at tention to the characteristics of the object [126], feedforward adju stments of posture [127] and increased reliance on feedback to correct movement errors [70]. Second, WMFT items are considered serial rather than discrete movement tasks. A discrete move ment is a one step motion with definable beginning and end points, whereas a serial task is comprise d of an ordered series of discrete movements [66]. A serial task may be more difficult than a discrete task because the motor program must blend successive m ovement components to accomplish the goal [122]. Third, the WMFT items place a speed requirement on the individuals

PAGE 151

137 performance (do this task as fast as you can). An individual must make additional adaptations to his/her moto r plan to meet this externally imposed demand on the information processing system [258]. The map may assist in conceptualizing a progression of intervention activities and treatment goals for the UE motor rehabil itation process. For example, the item map reveals that in general, FMA-UE movement items are less difficult than WMFT functional items. This may imply that simp le UE movements wit hout functional endpoint may be appropriate treatment activities and/or treatment goals for those with more severe post-stroke UE motor impairment or for thos e earlier in the recovery process. These individuals may benefit from motor experien ces in which the motor control system is challenged to attain/maintain various arm pos tures. Individuals w ith less severe poststroke UE motor impairment (i.e., those with more UE motor ability) may benefit from more functionally-oriented interventions a nd treatment goals. These individuals may benefit from treatment aimed at challenging the motor system to interact with objects, perform serial tasks and m ove with varying speeds. Limitations of this Study: Conversion of Performance Time data to Ordinal Data It is possible that the precision of the WMFT performance time data was compromised during the conversi on to the ordinal rating scal e, although the correlations between the scores produced by timed performa nce data and ordinal scales was r = 0.77. Conversion of continuous data to ordinal/categorical da ta is not uncommon in the research literature because it is sometimes advantageous to group patients based on an interval measurement with cut-off scores. For example, Hands (2001) [243] desired to create an assessment of childrens motor deve lopment comprised of milestone behaviors. These behaviors (i.e., the items on the new assessment) included skipping, running, and

PAGE 152

138 hopping, however each item was scored with a unique unit of measurement (i.e., number of skips in a set time, time to run a set dist ance, number of hops for a set distance). To facilitate the comparison of the behaviors she created categories for each item, using percentile divisions of the item scores across pa rticipants, so that each childs item score (number of skips, running velocity etc.) fell into one of the categories. Each category was assigned an ordinal rank, which became the item score on the new assessment. Similarly, Williams (2005) [246] sought to create an assessment to measure high level mobility of individuals with brain inju ry comprised of behaviors measured as counts and continuous data (e.g., number of stairs ascended/descended in a period of time, gait velocity over uneven surfaces etc.). In a proced ure similar to Hands, he created categories for each item using percentile divisions of ite m scores across participants. The categories were assigned ordinal ranks, and Rasch analysis was applied to explore the psychometrics of the new assessment. We followed the examples of Hands and Williams in converting the WMFT time data to performance categories. To assure that our rating scale was effective at capturing client UE motor ability, we compared my results at each stage of the process against quality-control criteri a specified by Linacre (2002) [247] and Hands (1999) [244]. While this approach met these criteria, there is the possibility of at least some loss of the precision and psychometric prop erties of the timed scale wh en converting it to a rating scale. Limitations of this Study: Sample Characteristics The low categories (representing less UE motor ability) were not well represented in the sample that, by design, only included individua ls with mild and moderate stroke. This limitation was noticed when checking the number of observations

PAGE 153

139 per category in the low categories of the 5point scale. The pres ent study would benefit from re-analysis with a sample including pers ons of more severe UE motor impairment. It would also be interesting to determine wh ether similar findings would be obtained if I used the WMFT Functional Abilities Scale rati ng scale data rather than the converted time data. Unfortunately, the rating scale was not available the datasets used for this study. Limitations of this Study: Possible Mu ltidimensionality of the Item Pool It may at first appear contradictory that the PCA results suggested the FMA-UE + WMFT item pool exhibited multidimensionality; while the Rasch fit statistics suggested the item pool is unidimensional. IRT meas urement models assume unidimensionality, and thus the co-calibration pro cess is invalid if this assump tion is violated [144]. Muraki and colleagues (2000) [167] note the diffi culties meeting the unidimensionality assumption when IRT models are applied to performance assessments. Murakai argues that because performance assessments, by th eir very nature, require examinees to complete tasks that demonstrate their ability to apply their skills to a real life situation, the data gathered from the assessment is influenced by construct-irrelevant variance such as environmental effects, practice e ffects and examinee fatigue. According to the researchers, unidimensionality is violated to some extent in every dataset, and even more in data from performance assessments. Howe ver, according to Murakai, as long as the violations are not extreme IR T models appear to be r obust. McHorney (2002) [248] suggested that if greater than 20% of the variance in the da ta is accounted for by the first factor in a PCA, unidimensionality is estab lished. Finlayson and Mallinson (2005) [161] suggested that a factor must contribute a minimum of 10% of explanatory variance to be considered a separate factor in modeling the data. Our results (table 20) support the

PAGE 154

140 unidimensionality of the item pool according to these suggestions. The first factor explained 42% of the variance in the data at T1 and 39% of the variance in the data at T2. The second factor contributed 7% of additi onal explanatory varian ce at each testing occasion, and subsequent factors contribute d less than 7%. These findings appear to indicate noise in the data, but do not a ppear to support the pr esence of distinct multidimensional constructs. Because it is likely that performance assessment date is multidimensional to some extent, various multidimensional IRT models have been proposed but are not widely used [167]. The findings from this study suggest that these models should be considered for future studies. Limitations of this Study: Sample Size The use of a 91 person sample size for this study may be criticized because of the commonly referenced rule of thumb speci fying that a minimum of 200 subjects is necessary for Rasch analysis [203]. In Rasch me asurement, the researcher seeks to obtain stable person, item and rating scale estimates A stable parameter is one that has low standard error. In a Monte Carlo simula tion study, Wang (2005) [208] found unstable item parameters at either the higher or lower end of the item difficulty hierarchy with sample sizes of 100 2000 individuals and an assessment with 10 to 60 items. Item parameters were stable with 100 150 participants and l onger assessments with 30 40 items. Wangs study suggests that the pres ent sample size of 91 participants and assessment lengths provide stable item pa rameter estimation by limiting the error with which the parameters are associated. The fi ndings of this study should be empirically confirmed by conducting the study with larger sample sizes.

PAGE 155

141 Conclusion The use of IRT measurement methods to link existing motor function assessments is being increasingly applied to healthcare outcomes research [148, 191, 197, 198]. The conceptual basis of linking existing assessmen ts is that items, although from different assessments, share common underlying motor beha viors, and thus are appropriate to be measured on a common metric [148, 192]. In th is study we have shown that the FMA-UE and WMFT items require similar UE motor be haviors and represent a single UE motor ability construct. Linking assessments allows scores from one assessment to directly translate to a score on another assessment [197 ]. This method has been useful for tracking recovery across the continuum of care [148] and extendi ng an assessment that had previously exhibited a ceiling effect [198]. We have suggest ed that linking the FMA-UE to the WMFT improves the functional interpretability of the FMA-UE change score, and may be a useful platform to explore recove ry of impairment-level UE motor behaviors and functional performance. Specifically, the linked item map can show how changes in motor impairments, often the target of ther apy, can translate to improvements at the functional level. Calibrating the FMA-UE and WMFT together onto the same measurement scale may be one of example of, as Fisher (1997) stated, taking full advantage of scale-free measur ements possibilities [191].

PAGE 156

142 Table 13: Sample characteristics (n = 91) Table 13: Characteristics of the 91 person sample Years of Age (Mean SD) 69.4 10.3 Gender Female 43 Male 48 Race White 74 African-American 16 Other 1 Stroke Type Ischemic 82 Hemorrhagic 9 Stroke Location Right Hemisphere 44 Left Hemisphere 47 Days Since Stroke (Mean SD) 76.1 27.1 Stroke Severity Minor (Orpington Prognostic Score <3.2) 50 (54.9%) Moderate (3.2 <= Orpington <= 5.2) 36 (39.6%) Severe (Orpington > 5.2) 2 (2.2%) Orpington score missing 3

PAGE 157

143 Table 14: Example of rating scale statistics 4-point scale 5-point scale Item description Rating Scale category Number of observations per category Step Threshold Rating Scale category Number of observations per category Step Threshold 1 21 none 1 5 none 2 22 -0.44 2 20 -2.92 3 23 0.00 3 25 -0.03 4 20 0.44 4 25 1.11 Forearm to box 5 25 1.84 1 31 none 1 16 none 2 19 -0.97 2 10 -0.85 3 17 0.11 3 26 -1.07 4 19 0.86 4 24 0.65 Turn key 5 24 1.26

PAGE 158

144 Table 15: Principal components analysis eige nvalues of the Wolf Motor Function Test T1 T2 Factor Eigenvalue Proportion of Vari ance Eigenvalue Proportion of Variance 1 8.28 55% 8.54 59% 2 1.54 10% 1.12 07% Table 16: Principal com ponent factor loadings T1 T2 Item Description Factor 1 Factor 1 Forearm to table (side) 0.758 0.690 Forearm to box (side) 0.705 0.772 Extend elbow (side) 0.724 0.790 Extend elbow, push weight 0.650 0.701 Hand to table 0.821 0.867 Hand to box 0.816 0.816 Reach and retrieve 0.693 0.734 Lift soda can 0.831 0.831 Lift pencil 0.790 0.728 Lift paperclip 0.737 0.720 Stack checkers 0.755 0.750 Flip cards 0.684 0.735 Turn key 0.726 0.740 Fold towel 0.664 0.643 Lift basket 0.757 0.771

PAGE 159

145Table 17: Wolf Motor Function Test item measures, fit statistics, and point measure correlations. T1 T2 WMFT Item Description Difficulty measure (SE) Infit MnSq Infit ZSTD rPM Difficulty measure (SE) Infit MnSq Infit ZSTD rPM Forearm to table -0.17 (0.14) 0.97 -0.2 0.71 -0.16(0.14) 1.24 1.6 0.67 Forearm to box -0.09 (0.14) 1.14 1.0 0.69 -0.30(0.14) 0.94 -0.4 0.73 Extend elbow to side -0.11 (0.14) 1.07 0.5 0.70 -0.08 (0.14) 0.89 0.7 0.74 Extend elbow, push weight -0.11 (0.14) 1.32 2.1 0.63 -0.15 (0.14) 1.18 1.2 0.68 Hand to table -0.25 (0.14) 0.77 -1.6 0.78 -0.13 (0.14) 0.63 -2.8 0.80 Hand to box -0.06 (0.14) 0.79 -1.5 0.76 0.07 (0.14) 0.81 -1.4 0.76 Reach and retrieve -0.15 (0.14) 1.17 1.2 0.67 -0.16 (0.14) 1.09 0.6 0.68 Lift soda can 0.10 (0.14) 0.69 -2.3 0.76 -0.03 (0.14) 0.75 -1.8 0.77 Lift pencil 0.24 (0.13) 0.83 -1.2 0.73 0.10 (0.14) 1.09 0.7 0.70 Lift paperclip 0.06 (0.14) 1.01 0.2 0.69 0.10 (0.14) 1.11 0.8 0.69 Stack checkers 0.02 (0.14) 0.96 -0.2 0.71 0.07 (0.14) 1.01 0.1 0.69 Flip cards 0.13 (0.14) 1.18 1.2 0.66 0.03 (0.14) 1.06 0.5 0.71 Turn key 0.24 (0.13) 1.04 0.3 0.67 0.10 (0.14) 1.04 0.4 0.71 Fold towel 0.12 (0.14) 1.23 1.5 0.65 0.10 (0.14) 1.37 2.3 0.64 Lift basket 0.03 (0.14) 0.96 -0.2 0.70 0.19 (0.14) 0.93 -0.4 0.71

PAGE 160

146 Table 18: Wolf Motor Function Test item difficulty hierarchy T1 T2 Item Description Difficulty (logits) SE Difficulty (logits) SE hard Turn key 0.25 0.14 0.09 0.14 Lift pencil 0.23 0.14 0.09 0.14 Flip cards 0.10 0.14 0.03 0.14 Fold towel 0.12 0.14 0.09 0.14 Lift soda can 0.12 0.14 -0.03 0.14 Lift paperclip 0.06 0.14 0.09 0.14 Stack checkers 0.02 0.14 0.07 0.14 Lift basket 0.00 0.14 0.18 0.14 Hand to box -0.07 0.14 0.07 0.14 Forearm to box (side) -0.09 0.14 -0.03 0.14 Extend elbow (side) -0.11 0.14 -0.07 0.14 Extend elbow, push weight -0.11 0.14 -0.13 0.14 Reach and retrieve -0.13 0.14 -0.15 0.14 Forearm to table (side) -0.13 0.14 -0.15 0.14 easy Hand to table -0.25 0.14 -0.13 0.14 Table 19: Wolf Motor Function Test differential item function analysis T1 T2 t T1 item difficulty SE T2 item difficulty SE calculated t value df probability Forearm to table -0.13 0.14 -0.15 0.14 0.08 170 0.9385 Forearm to box -0.090.14-0.030.14-0.32 1700.7499 Extend elbow to side -0.11 0.14 -0.07 0.14 -0.22 170 0.8262 Extend elbow push weight -0.110.14-0.130.140.08 1700.9393 Hand to table -0.25 0.14 -0.13 0.14 -0.61 170 0.5426 Hand to box -0.070.140.070.14-0.71 1700.4763 Reach and retrieve -0.13 0.14 -0.15 0.14 0.08 170 0.9385 Lift soda can 0.120.14-0.030.140.76 1700.4505 Lift pencil 0.23 0.14 0.09 0.14 0.75 170 0.4516 Lift paperclip 0.060.140.090.14-0.13 1700.8987 Stack checkers 0.02 0.14 0.07 0.14 -0.22 170 0.8224 Flip cards 0.100.140.030.140.36 1700.7168 Turn key 0.25 0.14 0.09 0.14 0.85 170 0.3951 Fold towel 0.120.140.090.140.17 1700.8682 Lift basket 0.00 0.14 0.18 0.14 -0.91 170 0.3625

PAGE 161

147 Table 20: Eigenvalues of the FM A-UE + WMFT linked assessment T1 T2 Factor Eigenvalue Proportion of Variance Cumulative Eigenvalue Proportion of Variance Cumulative 1 19.06 42% 42% 17.56 39% 39% 2 3.21 07% 49% 3.29 07% 46% 3 2.29 05% 55% 2.41 05% 51% 4 1.52 03% 58% 1.72 04% 56% 5 1.37 03% 61% 1.61 04% 59% 6 1.35 09% 64% 1.42 03% 62% 7 1.26 03% 67% 1.40 03% 65% 8 1.12 02% 69% 1.26 03% 68% 9 na na na 1.17 03% 71% 10 na na na 1.03 02% 73% Figure 5: Scree plots of the FM A-UE + WMFT linked assessment

PAGE 162

148 Table 21: FMA-UE + WMFT item pool rotated factor matrix T1 Rotated Factor Matrix T2 Rotated Factor Matrix Factor 1 Factor 2 Factor 1 Factor 2 Scapular elevation *0.20 *0.35 0.71 0.26 Scapular retraction 0.12 0.44 0.54 0.11 Shoulder abduction 0.29 0.56 0.68 0.23 Shoulder external rotation 0.58 0.28 0.68 0.25 Elbow flexion 0.57 0.26 0.53 -0.02 Forearm supination 0.63 0.02 0.52 0.34 Shoulder adduction, int. rotation 0.59 0.23 0.57 0.16 Elbow extension 0.55 0.29 0.64 0.16 Forearm pronation 0.72 0.32 0.61 0.24 Hand to lumbar spine 0.49 0.37 0.63 0.15 Shoulder flex. to 90, elbow ext. **0.66 **0.45**0.72 **0.41 Pron-sup, elbow at 90 0.65 0.13 0.68 0.28 Shoulder abd. to 90, elbow ext **0.55 **0.44 0.72 0.29 Shoulder flex to 180, elbow ext **0.57 **0.40**0.60 **0.45 Pronation-supination, elbow ext 0.67 0.31 0.71 0.27 Wrist stable, elbow at 90 0.63 0.33 0.59 0.29 Wrist flex-ext, elbow at 90 0.62 0.29 0.66 0.21 Wrist stable, elbow ext 0.70 0.38 0.71 0.38 Wrist flex-ext, elbow ext **0.49 **0.41 0.53 0.17 Wrist circumduction 0.71 0.29 0.60 0.27 Finger mass flexion 0.76 0.15 0.64 0.36 Finger mass extension 0.77 0.18 0.64 0.31 Hook grasp 0.70 0.19**0.58 **0.43 Lateral prehension 0.76 0.14 0.60 0.32 Palmar prehension 0.70 0.17 0.56 0.37 Cylindrical grasp 0.61 0.27 0.49 0.23 Spherical grasp 0.74 0.05 0.54 0.23 Movement without tremor 0.48 0.26 0.36 0.14 Movement without dysmetria *0.20 *0.03 0.30 -0.07 Fugl-Meyer Upper Extremity Assessment Items Movement with normal speed 0.70 0.37**0.71 **0.42 Forearm to table 0.16 0.770.14 0.68 Forearm to box 0.11 0.790.15 0.77 Extend elbow side 0.13 0.760.25 0.75 Extend elbow push weight 0.09 0.720.10 0.71 Hand to table 0.31 0.790.32 0.81 Hand to box 0.28 0.800.24 0.78 Reach and retrieve 0.06 0.760.19 0.70 Lift soda can **0.43 **0.680.35 0.75 Lift pencil **0.55 **0.530.33 0.65 Lift paperclip **0.56 **0.480.25 0.67 Stack checkers **0.49 **0.530.27 0.69 Flip cards 0.62 0.380.34 0.66 Turn key **0.48 **0.490.22 0.71 Fold towel 0.33 0.520.19 0.62 Wolf Motor Function Test Items Lift basket **0.48 **0.580.23 0.74 grey shadings indicate strong factor loadings (value > 0.40), ** indicated cross-loadings, indicate a failure to load strongly on either factor, n = 91, varimax rotation

PAGE 163

149Table 22: FMA-UE + WMFT item measure, fit statistics, and point measure correlations T1 T2 Item Description Item Difficulty Error Infit MNSQ ZSTD rPM Item Difficulty Error Infit MNSQ ZSTD rPM FMA-UE shoulder flexion to 180, elbow extended 1.40 0.19 0.97 -0.1 0.67 1.29 0.18 0.85 -1.2 0.72 WMFT turn key 1.29 0.13 1.00 0.0 0.67 1.41 0.13 1.09 0.7 0.67 WMFT lift pencil 1.29 0.13 0.70 -2.3 0.75 1.41 0.13 1.02 0.2 0.70 WMFT flip cards 1.19 0.13 0.93 -0.5 0.71 1.36 0.13 0.98 -0.1 0.71 WMFT fold towel 1.18 0.13 1.29 2.0 0.61 1.41 0.13 1.36 2.3 0.60 WMFT lift soda can 1.16 0.13 0.70 -2.4 0.76 1.30 0.13 0.75 -1.9 0.77 WMFT lift paperclip 1.13 0.13 0.84 1.1 0.72 1.41 0.13 1.11 0.8 0.67 WMFT lift basket 1.10 0.13 0.87 -0.9 0.72 1.50 0.13 0.93 -0.5 0.69 WMFT stack checkers 1.09 0.13 0.90 -0.7 0.72 1.39 0.13 0.99 0.0 0.69 FMA-UE wrist circumduction 1.07 0.19 0.70 -2.4 0.71 1.08 0.19 0.74 -2.1 0.61 WMFT hand to box 1.02 0.13 0.88 -0.8 0.73 1.39 0.13 0.90 -0.7 0.74 WMFT forearm to box 1.00 0.13 1.28 1.9 0.62 1.30 0.13 1.07 0.5 0.67 WMFT extend elbow to side 0.98 0.13 1.28 1.9 0.62 1.26 0.13 0.94 -0.4 0.72 WMFT extend elbow, push weight 0.98 0.13 1.44 2.8 0.57 1.20 0.13 1.37 2.4 0.61 WMFT reach and retrieve 0.95 0.14 1.39 2.5 0.57 1.18 0.14 1.11 0.8 0.64 WMFT forearm to table 0.93 0.14 1.15 1.1 0.64 1.19 0.13 1.32 2.1 0.60 FMA-UE pronation-supination, elbow extended 0.89 0.19 0.98 -0.1 0.67 0.70 0.19 0.99 0.0 0.67 WMFT hand to table 0.85 0.13 0.87 -0.9 0.75 1.21 0.13 0.68 -2.5 0.79 FMA-UE wrist flexion-extension, elbow extended 0.82 0.19 0.72 -2.2 0.63 0.56 0.19 0.90 -0.8 0.49 FMA-UE hook grasp 0.53 0.19 1.41 2.8 0.64 0.24 0.19 1.30 2.1 0.69 FMA-UE spherical grasp 0.49 0.19 0.99 0.0 0.59 0.01 0.20 1.03 0.3 0.54 FMA-UE lateral prehension 0.31 0.19 1.08 0.6 0.65 0.74 0.19 0.78 -1.7 0.63 FMA-UE shoulder abduction to 90, elbow extended 0.24 0.19 0.97 -0.2 0.67 0.35 0.19 1.00 0.1 0.67 FMA-UE forearm supination 0.20 0.19 0.86 -1.0 0.47 -0.57 0.21 0.87 -0.8 0.60 FMA-UE movement with normal speed 0.17 0.19 0.87 -1.0 0.75 -0.03 0.20 0.90 -0.7 0.77 FMA-UE wrist stable, elbow extended 0.13 0.19 1.29 2.1 0.75 -0.70 0.22 1.19 1.2 0.74 FMA-UE palmar prehension -0.02 0.19 0.94 -0.4 0.62 -0.31 0.20 0.83 -1.3 0.64 FMA-UE shoulder external rotation -0.29 0.20 0.77 -1.9 0.63 -0.52 0.21 0.84 -1.1 0.64

PAGE 164

150Table 22 continued FMA-UE movement without dysmetria -0.29 0.20 1.75 4.7 0.22 -0.35 0.20 1.65 3.9 0.19 FMA-UE scapular retraction -0.41 0.20 0.98 -0.1 0.39 -0.39 0.21 1.06 0.5 0.45 FMA-UE hand to lumbar spine -0.45 0.20 0.84 -1.2 0.60 -0.19 0.20 0.89 -0.8 0.52 FMA-UE shoulder flexion to 90, elbow extended -0.57 0.20 1.16 1.2 0.77 -0.48 0.21 1.10 0.7 0.76 FMA-UE wrist flexion-extension, elbow at 90 -0.65 0.20 0.69 -2.5 0.64 -0.85 0.22 0.78 -1.4 0.58 FMA-UE wrist stable, elbow at 90 -0.82 0.21 1.42 2.7 0.69 -2.01 0.28 1.25 1.1 0.63 FMA-UE elbow extension -0.95 0.21 0.77 -1.7 0.57 -1.21 0.24 1.01 0.1 0.55 FMA-UE pronation-supination, elbow at 90 -1.00 0.21 0.87 -0.9 0.56 -1.21 0.24 0.83 -1.0 0.65 FMA-UE cylindrical grasp -1.14 0.22 0.91 -0.5 0.63 -1.78 0.27 1.08 0.4 0.47 FMA-UE movement without tremor -1.34 0.23 1.24 1.5 0.53 -1.10 0.23 1.33 1.9 0.38 FMA-UE finger mass flexion -1.44 0.23 0.77 -1.5 0.63 -1.05 0.23 0.63 -2.5 0.67 FMA-UE shoulder abduction -1.72 0.24 0.86 -0.8 0.59 -1.21 0.24 0.83 -1.0 0.62 FMA-UE scapular elevation -1.78 0.24 1.55 2.8 0.38 -1.10 0.23 0.89 -0.6 0.65 FMA-UE finger mass extension -1.84 0.25 0.72 -1.7 0.67 -1.93 0.28 0.69 -1.6 0.64 FMA-UE shoulder adduction with internal rotation -1.90 0.25 0.89 -0.6 0.56 -2.17 0.30 1.11 0.5 0.48 FMA-UE forearm pronation -2.16 0.27 0.62 -2.2 0.74 -2.36 0.31 0.79 -0.9 0.59 FMA-UE elbow flexion -3.66 0.39 0.51 -1.9 0.63 -3.39 0.42 0.96 0.0 0.35

PAGE 165

151 T1 T2 FMAUE WMFT Item Difficulty Measures -5 -4 -3 -2 -1 0 1 2 -5.00-4.00-3.00-2.00-1.000.001.002.00 FMA-UE Wrist stable, elbow at 90' FMA-UE Wrist stable, elbow extended' FMA-UE Forearm supination' Figure 6: FMA-UE + WMFT different ial item functioning scatter plot

PAGE 166

152 4 3 2 sh flex to 180, elbow ext flip cardslift pencilturn key lift basketlift papercliplift soda canfold towel wrist circumduction1forearm to boxhand to boxstack checkers pron/sup, elbow extwrist flex/ext, elbow ext forearm to tablereach and retrieveextend elbow to sideext elbow, push wt hook grasp hand to table spherical graspsh abd to 90, elbow ext scapular retractionlateral prehensionforearm supination movement with normal speedwrist stable, elbow extendedpalmar prehension0shoulder external rotationmovement without dysmetria scapular retractionhand to lumbar spine sh flex to 90, elbow extended wrist flexion/extension, elbow bentwrist stable, elbow at 90 elbow ext (with internal rotation)-1pron/sup, elbow bentcylindrical grasp movement without tremorfinger mass flexion shoulder abductionscapular elevationsh add with internal rotation-2finger mass extensionforearm pronation -3 elbow flexion -4 UE motor ability logit rulerFugl-Meyer Upper Extremity AssessmentWolf Motor Function Test Figure 7: FMA-UE + WMFT linked item map

PAGE 167

153 23T2 32T164T1438T2 55T2 27T1 17T2 44T2 54T2 64T2 78T2 23T150T1 54T1 06T2 11T2 22T2 25T2 37T2 50T2 63T2 70T2 79T2 44T13 34T2 57T2 71T2 89T2 17T122T125T155T1 18T2 32T2 43T2 60T2 72T2 74T2 01T103T134T174T1 03T2 04T2 09T2 13T2 16T2 19T2 28T2 61T2 76T2 16T189T1 46T2 58T2 84T2 87T2 05T128T1 43T1 46T1 58T1 60T1 63T1 70T1 79T187T1265T2 18T171T1 84T1 02T2 07T2 75T2 77T2 45T176T1 88T1 30T2 05T2 45T2 56T2 06T138T1 48T1 52T1 56T1 61T1 62T1 78T1 24T2 62T2 08T110T1 14T1 19T1 72T1 29T2 26T2 01T2 52T2 86T2 88T2 90T2 91T2 02T131T1 57T1 65T1 77T1112T2 14T2 20T2 31T2 04T113T1 15T1 15T2 81T2 69T1 27T2 82T2 12T137T1 86T1 90T1 91T1 67T2 17T185T1 47T1 30T1 48T2 07T120T1 24T1 67T1 81T1008T2 42T2 47T2 69T2 73T2 26T142T1 33T2 66T2 85T2 51T2 29T1 49T2 21T166T1 68T1 75T1 83T1 83T2 51T153T1-121T2 39T2 68T2 39T173T1 36T2 53T2 82T1 33T1 10T2 35T2 41T2 35T136T1 59T1-259T2 40T180T1 49T1 80T2 41T1 40T2 -309T1 -4 UE motor ability logit ruler Person Ability Scores T1Person Ability Scores T2 Figure 8: FMA-UE person ability measures overlaid onto the FMA-UE + WMFT item map

PAGE 168

154 CHAPTER 5 A COMPARISON OF THE RASCH MODEL AND A TWO PARAMERTER ITEM RESPONSE MODEL TO DERIVE ITEM AND PERSON PARAMETERS Item Response Theory (IRT) is a family of mathematical models each designed to describe the relationship between an indivi duals ability level a nd his/her pattern of responses to assessment items. All IRT models calculate one ability parameter for each individual, however the models vary with regards to the num ber of item parameters used to describe ability. The least complex, the one -parameter logistic models (1PL), express individual ability as a function of item difficulty. The Rasch model is typically considered a 1PL IRT model. The two-pa rameter logistic models (2PL) express individual ability as a function of item diffi culty and item discrimination, and the threeparameter logistic models (3PL) express indivi dual ability as a func tion of item difficulty, item discrimination and guessing [144]. The me rits and pitfalls of each model are debated, sometimes passionately, in the literature [160, 209, 211, 259]. The question of which IRT model to select for any analysis is complex because one must first carefully consider how the assump tions of the IRT model match ones a-priori hypotheses regarding the characteristics of th e given population (i.e., the data) [209, 212]. Models with additional parameters will more fl exibility fit a greater variety of data [260]. The Rasch model is unique among IRT models because it assumes all items are equally discriminating whereas the 2PL and 3PL m odels accommodate varying discriminations. Item discrimination describes how well a test item differentiates between test-takers of higher and lower ability [144] Items with high discrimina tion are thought to provide

PAGE 169

155 more information about a test-taker than items with low discrimination [261]. Traub [262] stated that the a-priori assumption that an assessments items all have equal discriminative ability flies in the face of common sense and a wealth of empirical evidence accumulated over the last 80 years. Indeed, studies have shown that item discriminations vary greatly among items m easuring activities of daily living [159], cognition [263], and self-esteem [261]. The evidence suggests that item discrimi nation values vary among items; therefore it is reasonable to assume that this is also true of FMAUE items. In choosing the Rasch model for this dissertation, we did not account for the likelihood that FMA-UE item discriminations vary and so it is possible th at we have inappropriately applied an IRT model to the dissertation da ta. Model misfit (i.e., choosi ng the wrong IRT model) is thought to negatively effect the estimati on of person and item parameters and may invalidate an analysis [209, 260, 264]. The purpose of the present study is to te st whether the choice of IRT model affected the estimation of FMA-UE item diff iculty and person ability parameters as accomplished in the previous chapters of this dissertation. We will address this purpose by testing three hypotheses: (1) the 2PL is a bett er fit to the data than the Rasch model, (2) item discriminations vary among FMA-UE items, and (3) the choice of IRT model affects estimates of item difficulty and person ability. The hypotheses will be tested by comparing 2PL-derived item and person para meters to Rasch-derived item and person parameters. We will compare the item and person parameters derived from two datasets: the 512-person dataset (D512) util ized in the first study of th e dissertation, and the 377 person dataset (D377) used as the post-test data in the second study of this dissertation.

PAGE 170

156 The reason that the two datasets were c hosen was to examine the person and item parameters influencing our conclusions rega rding the cross secti onal and longitudinal measurement properties of the FMA-UE. Methods Participants Two separate 2PL analyses were conducted. The first analysis was applied to the pooled dataset described in study #1 of this dissertation. The sample consisted of 512 individuals. Please refer to study #1 for a more complete description of the sample. For the purposes of the present study I wi ll refer to this dataset as D512. The second analysis was applied to the sample who were administered the FMAUE at the second time point (T2), 6 months post stroke, in the Kansas City Stroke study. This sample consisted of 377 individuals and is described in study #2 of this dissertation. For the purposes of the present study I will refer to this dataset as D377. Dimensionality Prior to conducting any IRT analysis a neces sary first step is to determine if the data meet the unidimensional assumptions co mmon to all IRT models. Because we have previously conducted analyses of unidimensi onality (e.g., principal components analysis) of the 30-item FMA-UE we will not repeat the analyses here. Preparing the data for 2PL analysis The software for 2PL analysis (Multil og [265]) does not reco gnize as an item score, instead treating the va lue as missing data. Therefore we recoded all FMA-UE data from both time points from a 02 rating scale to a 1-3 rating scale. The data was recoded in SPSS (v. 11).

PAGE 171

157 Model fit The fit of an IRT model to the data wa s examined with a likelihood ratio test, which assumes that data are better modeled with more mathematical parameters. We followed a procedure outlined by McHorney a nd Kirisci to perform the likelihood ratio test [213, 248]. If the like lihood for a model with fewer pa rameters is not significantly different than the likelihood for a model w ith more parameters, then the model with fewer parameters is chosen. The likelihood ratio is a chi-square statistic and the significance of the ratio te st was set at p<0.01. 2PL analysis FMA-UE data were fit to a 2-parameter graded response mode l using the Multilog for Windows software program [ 265] to calculate item diffic ulty and item discrimination parameters. The graded response model is us eful for analyzing responses in multiple ordered categories [266]. The data were subs equently used to cr eate maximum likelihood estimates (MLE) of person ability, also conducted with Multilog. Comparison of parameter estimates We compared three parameters estimated with each model. Item discrimination The Rasch model software calculates an item discrimination index [148]. However, in contrast to the 2-parame ter IRT model this index does not enter into person ability estimates. The index describes whether items are close to or depart from the ideal discrimination value of . In other words, the index reveals how consistent items are with the unidimensional expectations of the Rasch model. Low values indicate that an item fails to discriminate as well as other items. We compared item discrimination values as calculated by Rasch analysis to those calculated by the 2PL analysis.

PAGE 172

158 Person ability and item difficulty We compared estimates of person ability and item difficulty obtained with the 2PL analysis to the estimates of person ability and item difficulty that we had previously obtained with Rasch analysis. We utilized Pearson product moment correlations to determine the correlations between the estimates. Multilog calculates parameters in a unit of measurement termed probits while Winsteps (Rasch analysis) calculates paramete rs in logits. In ge neral, 1 logit = 1.7 probits [234]. To facilitate comparison of the item difficulty hierarchies obtained with each model, we converted logit values to probit values using the above conversion ratio. Results Model Fit Multilog was used to calculate the nega tive two times log likelihood values for the 1PL and 2PL logistic models. These va lues were 7645.6 and 9960.8 for the 1PL and 2PL models respectively. Th e likelihood ratio was 2315.2, the difference between the values. The 1PL analysis calculated 30 paramete rs (30 items in the assessment), the 2PL analysis calculated 60 parameters. The df fo r this analysis is equal to the difference between these parameters and therefore df = 30. The likelihood ratio is distributed as a chi square statistic, thus using the chi square table to calculate the critical value; the likelihood ratio of 2315.2 is signif icant at p<0.01 (the critical chi square value is equal to 50.89 at the 0.01 level). The result suggests th e 2PL model fit these data better than the 1PL model. Item Discrimination D512 The comparisons of Rasch-derived item di scrimination values and those obtained from the 2PL analysis in the first datase t, D512, are presente d in table 23. Item

PAGE 173

159 discriminations calculated by Multilog ranged from 2.26 to 4.92. Item discriminations calculated by Winsteps ranged from 0.35 to 1. 38. In general, the two models reveal the same information. That is, each item respons e model identified the same items as most and least discriminating. Specifically, wrist flexion-extension with the elbow at 90 degrees, shoulder flexion to 90 degrees with the elbow extended, wrist flexion-extension with elbow extended and pronation-supinati on with the elbow at 90 degrees are among the most discriminating items at in each da taset as identified by each model. Movement without dysmetria, hook grasp, and lateral pr ehension are among the least discriminating items. The item discrimination values calcu lated by each model at T1 were strongly correlated; r = 0.88. D377 Item Discrimination The comparisons of Rasch-derived item di scrimination values and those obtained from the 2PL analysis in the second datase t, D377, are presented in Table 24. Item discriminations calculated by Multilog ranged from 6.13 to 2.66. Item discriminations calculated by Winsteps ranged from 1.31 to 0. 50. As in the time 1 analysis above, the two models yield identical information. That is the items the 2PL model identified as the most discriminating were the same items Rasc h model identified as most discriminating. The items the 2PL identified as least discri minating were the same items the Rasch model identified as least discriminating. The ite m discrimination values calculated by each model were strongly correlated ; r = 0.89. The item disc rimination pattern for the 377subject analysis is similar to that ge nerated with the 512subject analysis.

PAGE 174

160 D512 Item Difficulty Item difficulty parameters calculated by the 2PL analysis in the first dataset, D512, are compared to those obtained by Rasch analysis in table 25. The 2PL results are presented on the left side of the table, the Rasch results on the right side of the table. Moving from left to right, item descriptions are in the first column, the next two columns present beta 1 and beta 2 values. These values are item difficulty estimates at each adjacent response category boundary. Beta 1 is the boundary between unable to perform and partially able to perform, beta 2 is the boundary between partially able to perform and near normal performa nce. The fifth and sixth columns from the left report the average of the betas values a nd the average standard er ror of the difficulty measures SEM). The betas were averaged so as to more easily compare 2PL item difficulty estimates with the Rasch item difficulty estimates that are reported as an average value. The two columns to the right report the Rasch-derived item difficulty values with its SEM. 2PL values are reporte d in probits, Rasch values are reported in logits. The item difficulty hierarchies are directly compared in figure 9. Item descriptions are located along the x-axis, item difficulty values (probits) are re ported on the y-axis. Each items difficulty value as calculated by th e 2PL analysis is shown as a circle, while its item difficulty value as calculated by Ra sch is shown as a square. Rasch estimates have been converted to probits by multiplying th e logit value (sixth column from the left in table 25) by 0.59. Error bars are reported as 2 SEM. Figur e 9 shows that, in general, the hierarchies are identical. The values for 22 of the 30 items are identical within error. Eight items show differe nt difficulty values. Linacre repor ts that the conversion ratio (1

PAGE 175

161 logit = 1.7 probits) is an appr oximate relationship and is not expected to yield identical values [165]. Figure 10 presents a scatter pl ot of the 2PL item difficulty estimates (x-axis) plotted against Rasch item difficulty estimates (y-a xis). The Pearson product moment correlation between the difficulty estima tes is strong; r = 0.99. D377 Item Difficulty Item difficulty parameters calculated by each model in the second dataset, D377, are compared in table 26. As described above the 2PL results are presented on the left side of the table, the Rasch results on the ri ght side of the table. The item difficulty hierarchies are directly compared in figure 14. Item descriptions are located along the xaxis, item difficulty values (probits) are re ported on the y-axis. Each items difficulty value as calculated by the 2PL analysis is shown as a circle, while its item difficulty value as calculated by Rasch is shown as a s quare. Rasch estimates have been converted to probits by multiplying the logit value (sixth column from the left in table 26) by 0.59. Error bars are reported as 2 SEM. Figure 11 is a comparison of item hierarch ies generated by Multilog and Winsteps in the D377 dataset. The figure shows that, in general, the hierarchies are identical. The values for 22 of the 30 (73%) it ems are identical within error. Figure 12 presents a scatter plot of the 2P L item difficulty estimates (x-axis) plotted against Rasch item difficulty estimates (y-a xis). The Pearson product moment correlation between the difficulty estima tes is strong; r = 0.98.

PAGE 176

162 D512 Person Ability Figure 13 presents a scatter plot of pers on-ability measures de rived with the 2PL analysis (x-axis) plotted against person-ability measures derived with the Rasch analysis (y-axis) for D512. The Pearson product mo ment correlation between the person-ability estimates is strong; r = 0.99. Figure 14 presents a cross plot of personability measures derived with the 2PL analysis (x-axis) plotted against person-ability measures derived with the Rasch analysis (y-axis) for D377. The Pearson product mo ment correlation between the person-ability estimates is strong; r = 0.99. Discussion The purpose of this study was to test whet her the choice of IRT model affected the estimation of FMA-UE item difficulty and pe rson ability parameters. The results show that, although the 2PL model fit the data bette r than a 1PL model and FMA-UE item discrimination values varied, item difficulty a nd person ability estimates generated by the two IRT models were essentially identical in each dataset. Which IRT Model, 2PL or Rasch, Best Fits the FMA-UE Data? The likelihood ratio was calculated to dete rmine which model, 2PL or 1PL, was the best fit to the data. This pr ocedure is often utilized as a fi rst step in IRT analysis [213, 248, 261, 267, 268]. The chi-square test reveal ed that the 2PL mode l better fits the FMAUE data than the 1PL model. This result is not surprising. Adding parameters to a mathematical model increases its flexibility for describing variations in assessment data [144]. From this result it is inferred that FMA-UE items differ in their discriminability and that the more complex 2PL was necessa ry to adequately describe FMA-UE item characteristics [261].

PAGE 177

163 Within Each Dataset, D o FMA-UE Item D iscriminations Vary? The results show that FMA-UE item di scrimination values vary within each dataset. D512 FMA-UE item discriminations ra nged from 4.92 to 2.26 probits and D377 item discriminations ranged from 6.13 to 2.66 probits, a 1.17% and 1.30% difference respectively (calculated as [m aximum-minimum]/minimum). These results are consistent with others who have shown activities of daily living, c ognitive, and personality items differ in their discriminability. McHorn ey and Cohen (2000) [159] showed a 10% variation in discrimination values among 71 ADL and IADL item (discriminations ranged from 0.31 to 3.46 probits). Teresi et al (1995) [263] showed an 8% variation in 50 items measuring cognition (discriminations rang ed from 0.51 5.00 to probits), and GrayLittle et al. (1997) [261] s howed a 1.32% variation in item discriminations among 30 items measuring self-esteem (discriminati ons ranged from 2.70 to 1.16 probits). The Rasch model is criticized because it assumes equal item discrimination, constraining discriminations to a value of when calculating item difficulty and person ability parameters [165]. Cook and co lleagues (2003) [160] ar gued that the Rasch assumption does not conform to reality. Ind eed, the results of the present study show that FMA-UE items are not equally discrimina ting. Item discrimination indices calculated by Winsteps support this findi ng, indicating that D512 item discriminations ranged from 0.35 to 1.38 and D377 discriminations range d from 0.50 to 1.31, a 2.94% and 1.62% variation respectively. Perhaps the important question is not do FMA-UE item discriminations vary? The results of both the 2PL and Rasch analyses show that they do. Instead, the important question is, given that FMAUE item discriminations vary does the choice of IRT

PAGE 178

164 model affect the measurement properties of the FMA-UE with regards to estimation of item difficulty and person ability? Does the Choice of IRT Model Affect Either Item or Person Parameters? The results show that within both datase ts, the item hierarchies estimated by the 2PL and Rasch models are consistent and the item difficulty values (measured in probits) are nearly identical. Moreover, person ability measures were strongly, nearly perfectly correlated. The results show that the item and person parameters estimated by the Rasch model were robust to unequal FMA-UE item discriminations. The results are consistent with Teresi et al. (1995) [263] w ho fit both a 1PL and 2PL model to her analysis of cognitive asse ssment among individuals of different ethnic groups and levels of education. Teresi showed an 8% variation in item discriminations, yet each model was robust with regards to item and person parameter estimation. Penfield (2004) [264] in simulation studies, found that wh en the data had small levels of variation in item discrimination about th e value of 1.0 (as calculated by Rasch analysis) the person and item parameters were not impacted by th e choice of Rasch partial credit model or 2PL graded response model. However, when th e variation was large for all items of the scale, the person and item parameters becam e increasingly biased, meaning that it was difficult to reliably estimate the parameters. It is possible that the FMA-UE items vary so little with regards to their discriminati on that the estimation of person and item parameters is not impacted. What Does Item Discrimination Mean? Item discrimination is conceptually rela ted to item bias [269]. A non-biased item measures the same underlying attribute acro ss subgroups. Item discrimination describes how well a test item differentiates between subgroups of test-takers: those with higher

PAGE 179

165 ability and those with lower ability [144]. Item discrimina tion is not an index of item bias, however items with different magnitude s of discrimination relate to the underlying attribute differently and therefore might meas ure different constructs in examinees of different ability levels [144, 263]. McHorney and Cohen (2000) [159] suggest that poorly discriminating items inefficiently measure and yiel d less information than other items because factors other than ability will be affecting the persons re sponse to the item. Gray-Little et al. (1997) [261] found that the poorly disc riminating self-esteem items were the same items that the factor analysis revealed to be less correlate d with the primary fact or in describing the unidimensional underlying construct. Masters (1988) [269] argued that in some cases items with the highest discrimination valu es were biased. He argued that item discrimination interacts with di mensionality of the assessment, i.e., highly discriminating items are biased in that they do not m easure the intended co nstruct underlying the assessment. Masters makes the point that a researcher must carefully inspect item discrimination values to learn what the values reveal about the data rather than just accepting discrimination indices at face value, as the basis for making a subsequent judgment about the items usefulness. Master s viewpoint touches upon the motivational differences that separate the Rasch from other IRT models. The underlying conflict between the Rasch model and other IRT models has to do with views of how data provide informati on [211]. In short: should a measurement model explain data, or should a measurement model disclose information perhaps hidden in data? Andrich (2004) [ 211] frames this controversy using the language of mathematician-philosopher Thomas Kuhn as a clash of incompatible paradigms.

PAGE 180

166 Kuhn stated that the purpose of measurement is to disclose anomalies in data that cannot be explained by the theory underlyi ng the measurement model with which the data is analyzed. With this approach one is able to ask questions as to why data does not fit a given theory thereby potentially lear ning new information that contributes to advancement of the theory. Kuhns view is in stark contrast to traditional measurement theory that asserts data s hould fit a model so as to s upport the theory underlying the model. This issue is notably evident in c ontroversies surrounding the Rasch model [211]. According to non-Rasch IRT models, data is best interpreted according to a mathematical model found to explain the larg est amount of variance in the data. Models are tried and abandoned depending on how well the model explains the data. The Rasch model has the inverse assumption. That is, data should be viewed with the assumption that it either fits or does not fit the model. If it does not f it, then, rather than abandoning the model in favor of another, the data is examined for anomalies [211]. According to Andrich [211], the process of examining misfitting data (either items or persons) uncovers information that would have been hidden otherwise. New information may lead to advances in theory. From the perspe ctive of the Rasch model, fitting a model to the data might mask important features in th e data that might provi de clues that inform researchers where to look so as to expand or revise an existing theory. The overarching purpose of this disserta tion was to examine the measurement properties of the FMA-UE. In the three stud ies preceding the present study we assumed an item-oriented viewpoint from which to investigate (1) the validity of the FMA-UE item content, (2) the reproducibility of its it em structure, and (3) interpretation of its score. Throughout the studies we have e xplored the FMA-UE item-level construct

PAGE 181

167 validity, which has served to focus attent ion on traditional and contemporary motor control theories. The intention of the dissert ation has been to chal lenge and enhance the measurement properties of the FMA-UE by testing and increasing its match to contemporary theory. The Rasch model has be en a valuable method for exploring the interdependence of measurement and theo ry. The present study demonstrates the robustness of the Rasch model for accomplis hing the purpose of this dissertation. Conclusion Choosing the correct IRT model for a given an alysis is said to be complex because the researcher must carefully consider how the assumptions of the IRT model describe the characteristics of the given population (i.e., the data) [209, 212]. Models with additional parameters, i.e., the 2PL or 3PL models, will more flexibility fit a greater variety of data [260]. The Rasch model assumes all assessment items are equally discriminating. In this study we have show n that the FMA-UE items are not equally discriminating. Yet, despite this violati on of Rasch model assumptions, Rasch-derived FMA-UE item and person parameters are robu st. The choice of the most parsimonious IRT model, the Rasch model, to conduct an alyses for this dissertation is supported.

PAGE 182

Table 23: Comparison of item discriminati on values calculated by Multilog to those calculated by Winsteps in the first dataset D512 Item Discrimination Item Description Multilog estimated Item Discrimination Winsteps estimated item discrimination Elbow extension 4.92 1.24 Sh flexion to 90, elbow extended 4.86 1.21 Pronation-supination, el bow extended 4.57 1.38 Pronation-supination, elbow at 90 4.56 1.29 Wrist flex-ext, elbow at 90 4.35 1.31 Forearm pronation 4.30 1.21 Sh abd to 90, elbow extended 4.30 1.23 Shoulder adduction with internal rotation 4.05 1.08 Wrist stable, elbow extended 3.93 0.99 Shoulder external rotation 3.90 1.11 Wrist stable, elbow at 90 3.88 0.96 Sh flex to 180, elbow extended 3.82 1.21 Hand to lumbar spine 3.78 1.10 Finger mass extension 3.76 1.12 Wrist flex-ext, elbow extended 3.73 1.25 Elbow flexion 3.69 1.00 Forearm supination 3.67 1.08 Finger mass flexion 3.49 0.92 Shoulder abduction 3.41 0.98 Cylindrical grasp 3.26 0.98 Wrist circumduction 3.21 1.11 Movement with normal speed 3.14 0.79 Scapular elevation 3.08 0.74 Scapular retraction 3.07 0.76 Movement without tremor 2.92 0.68 Palmar prehension 2.91 0.92 Spherical grasp 2.68 0.81 Hook grasp 2.61 0.56 Lateral prehension 2.56 0.77 Movement without dysmetria 2.26 0.35

PAGE 183

169 Table 24: Comparison of item discriminati on values calculated by Multilog to those calculated by Winsteps in the second dataset D377 Item Discrimination Item Description Multilog estimated Item Discrimination Winsteps estimated item discrimination Wrist flex-ext, elbow at 90 6.13 1.31 Sh flexion to 90, elbow extended 5.57 1.29 Wrist stable, elbow extended 5.29 1.14 Wrist flex-ext, elbow extended 5.13 1.31 Wrist stable, elbow at 90 5.13 1.16 Pronation-supination, elbow at 90 5.07 1.26 Sh abd to 90, elbow extended 5.01 1.27 Pronation-supination, elbow extended 4.90 1.30 Movement with normal speed 4.75 1.25 Shoulder external rotation 4.71 1.12 Finger mass flexion 4.45 1.11 Finger mass extension 4.36 1.19 Elbow extension 4.29 1.02 Cylindrical grasp 4.21 1.17 Sh flex to 180, elbow extended 4.01 1.10 Forearm supination 3.76 0.79 Wrist circumduction 3.69 1.06 Hand to lumbar spine 3.66 0.98 Shoulder adduction with internal rotation 3.52 0.96 Palmar prehension 3.47 0.92 Shoulder abduction 3.32 0.94 Movement without tremor 3.29 0.86 Forearm pronation 3.28 0.88 Elbow flexion 3.24 0.85 Scapular elevation 3.21 0.81 Spherical grasp 3.16 0.85 Scapular retraction 3.09 0.69 Movement without dysmetria 2.82 0.71 Hook grasp 2.69 0.30 Lateral prehension 2.66 0.50

PAGE 184

170 Table 25: D512 comparison of item parameters derived with the 2PL and Rasch models D512 2PL (probits) Rasch Model (logits) Item Description 1 2 Ave. Ave. SEM Item difficulty SEM Scapular elevation -0.89 0.15 -0.37 0.07 -1.40 0.12 Scapular retraction -0.71 0.76 0.03 0.08 0.03 0.10 Shoulder abduction -0.64 0.45 -0.10 0.05 -0.56 0.11 Shoulder external rotation -0.55 0.79 0.12 0.07 0.24 0.10 Elbow flexion -0.93 0.05 -0.44 0.08 -1.76 0.13 Forearm supination -0.49 0.98 0.25 0.07 0.65 0.10 Shoulder add with internal rotation -0.95 0.16 -0.40 0.07 -1.56 0.12 Elbow extension -0.74 0.48 -0.13 0.07 -0.64 0.11 Forearm pronation -0.61 0.32 -0.15 0.07 -0.87 0.11 Hand to lumbar spine -0.55 0.47 -0.04 0.06 -0.40 0.11 Sh flexion to 90, elbow extended -0.16 0.39 0.12 0.05 -0.21 0.11 Pronation-supination, elbow at 90 -0.46 0.56 0.05 0.05 -0.17 0.11 Sh abd to 90, elbow extended -0.19 0.64 0.23 0.05 0.28 0.10 Sh flex to 180, elbow extended 0.04 0.98 0.51 0.05 1.26 0.10 Pronation-supination, elbow ext -0.04 0.91 0.44 0.05 1.00 0.10 Wrist stable, elbow at 90 0.01 0.41 0.21 0.05 0.18 0.10 Wrist flex-ext, elbow at 90 -0.49 0.70 0.11 0.05 0.12 0.10 Wrist stable, elbow extended 0.21 0.67 0.44 0.05 0.95 0.10 Wrist flex-ext, elbow extended -0.13 0.66 0.27 0.06 1.06 0.10 Wrist circumduction 0.13 1.18 0.66 0.07 1.67 0.10 Finger mass flexion -1.00 0.20 -0.40 0.07 -1.44 0.12 Finger mass extension -0.77 0.20 -0.29 0.07 -1.25 0.12 Hook grasp 0.18 0.88 0.53 0.07 1.33 0.10 Lateral prehension -0.13 1.03 0.45 0.07 1.08 0.10 Palmar prehension -0.34 0.57 0.12 0.07 0.06 0.10 Cylindrical grasp -0.60 0.16 -0.22 0.07 -1.10 0.12 Spherical grasp -0.14 1.11 0.49 0.07 1.20 0.10 Movement without tremor -0.32 0.08 -0.12 0.08 -0.91 0.11 Movement without dysmetria -0.25 0.62 0.19 0.08 0.27 0.10 Movement with normal speed 0.10 0.73 0.42 0.06 0.89 0.10

PAGE 185

171 -1.50 -1.00 -0.50 0.00 0.50 1.00 1.50Wrist circumduction Hook grasp Sh flex to 180, elbow extended Spherical grasp Lateral prehension Wrist flex-ext, elbow extended Pronation-supination, elbow extended Wrist stable, elbow extended Movement with normal speed Forearm supination Sh abd to 90, elbow extended Movement without dysmetria Shoulder external rotation Wrist stable, elbow at 90 Wrist flex-ext, elbow at 90 Palmar prehension Scapular retraction Pronation-supination, elbow at 90 Sh flexion to 90, elbow extended Hand to lumbar spine Shoulder abduction Elbow extension Forearm pronation Movement without tremor Cylindrical grasp Finger mass extension Scapular elevation Finger mass flexion Shoulder adduction with internal Elbow flexion Multilog item difficulty (probits) Winsteps item difficulty (converted to probits) Figure 9: D512 comparison of item hierarchie s generated by Multil og (probits) to the item hierarchy generated by Winsteps (logits converted to probits by multiplying by 0.59). Error bars 2 SEM -2.00 -1.50 -1.00 -0.50 0.00 0.50 1.00 1.50 2.00 -0.60-0.40-0.200.000.200.400.600.80 2 PL item difficulty estimates (probits)Rasch item difficulty estimates (logits) Figure 10: D512 scatter plot of item difficulty estimates (n = 512, r = 0.99)

PAGE 186

172 Table 26: D377 comparison of item parameters derived with the 2PL and Rasch models D377 2PL (probits) Rasch Model (logits) Item Description 1 2 Ave. Ave. SE Item difficulty SE Scapular elevation -1.39 -0.18 -0.79 0.11 -1.90 0.17 Scapular retraction -0.90 0.46 -0.22 0.09 0.21 0.14 Shoulder abduction -0.97 0.08 -0.45 0.10 -0.88 0.15 Shoulder external rotation -0.79 0.47 -0.16 0.08 0.15 0.14 Elbow flexion -1.28 -0.21 -0.75 0.10 -1.90 0.17 Forearm supination -0.86 0.71 -0.08 0.09 0.77 0.13 Shoulder add with internal rotation -1.27 -0.16 -0.72 0.09 -1.81 0.17 Elbow extension -1.01 0.23 -0.39 0.08 -0.68 0.15 Forearm pronation -0.86 -0.02 -0.44 0.09 -1.02 0.16 Hand to lumbar spine -0.51 0.12 -0.20 0.08 -0.23 0.14 Sh flexion to 90, elbow extended -0.33 0.16 -0.09 0.06 -0.19 0.14 Pronation-supination, elbow at 90 -0.61 0.29 -0.16 0.07 -0.21 0.14 Shoulder abd to 90, elbow ext -0.28 0.35 0.04 0.06 0.37 0.13 Sh flexion to 180, elbow extended -0.19 0.83 0.32 0.06 1.65 0.12 Pronation-supination, elbow ext -0.23 0.71 0.24 0.06 1.29 0.12 Wrist stable, elbow at 90 -0.23 0.10 -0.07 0.08 -0.17 0.14 Wrist flex-ext, elbow at 90 -0.48 0.45 -0.02 0.07 0.33 0.13 Wrist stable, elbow extended 0.03 0.42 0.23 0.06 0.95 0.13 Wrist flex-ext, elbow extended -0.29 0.72 0.22 0.07 1.23 0.12 Wrist circumduction -0.05 0.92 0.44 0.07 2.02 0.12 Finger mass flexion -1.11 0.03 -0.54 0.08 -1.29 0.16 Finger mass extension -0.77 -0.04 -0.41 0.07 -1.02 0.16 Hook grasp 0.08 0.47 0.28 0.80 1.53 0.12 Lateral prehension -0.37 0.77 0.20 0.09 1.38 0.12 Palmar prehension -0.72 0.29 -0.22 0.08 -0.11 0.14 Cylindrical grasp -0.88 -0.27 -0.58 0.08 -1.64 0.17 Spherical grasp -0.39 0.71 0.16 0.08 1.20 0.12 Movement without tremor -0.43 -0.03 -0.23 0.09 -0.50 0.15 Movement without dysmetria -0.45 0.24 -0.11 0.11 0.08 0.14 Movement with normal speed -0.21 0.33 0.06 0.07 0.40 0.13

PAGE 187

173 -1.50 -1.00 -0.50 0.00 0.50 1.00 1.50Wrist circumduction Sh flex to 180, elbow extended Hook grasp Pronation-supination, elbow extended Wrist stable, elbow extended Wrist flex-ext, elbow extended Lateral prehension Spherical grasp Movement with normal speed Sh abd to 90, elbow extended Wrist flex-ext, elbow at 90 Wrist stable, elbow at 90 Forearm supination Sh flexion to 90, elbow extended Movement without dysmetria Shoulder external rotation Pronation-supination, elbow at 90 Hand to lumbar spine Scapular retraction Palmar prehension Movement without tremor Elbow extension Forearm pronation Finger mass extension Shoulder abduction Finger mass flexion Cylindrical grasp Shoulder adduction with internal rotation Elbow flexion Scapular elevation Multilog item difficulty (probits) Winsteps item difficulty (converted to probits) Figure 11: D377 comparison of item hierarchie s generated by Multil og (probits) to the item hierarchy generated by Winsteps (logits converted to probits by multiplying by 0.59). Error bars 2 SEM -2.50 -2.00 -1.50 -1.00 -0.50 0.00 0.50 1.00 1.50 2.00 2.50 -1.00-0.80-0.60-0.40-0.200.000.200.400.60 2PL-derived item difficulty (probits)Rasch-derived item difficulty (logits) Figure 12: D377 scatter plot of item difficulty estimates (n = 377, r = 0.98)

PAGE 188

174 -8.00 -6.00 -4.00 -2.00 0.00 2.00 4.00 6.00 8.00 -4.00-3.00-2.00-1.000.001.002.003.004.00 2P-derived person ability (probits)Rasch derived person abilit y (logits) Figure 13: D512 scatter plot of person ability estimates (n = 512, r = 0.99) -8.00 -6.00 -4.00 -2.00 0.00 2.00 4.00 6.00 8.00 -4.00-3.00-2.00-1.000.001.002.003.004.00 2PL-derived person ability estimatesRasch-derived person abilit y estimates Figure 14: D377 scatter plot of person ability estimates (n = 377, r = 0.99)

PAGE 189

175 CHAPTER 6 SUMMARY AND CONCLUSION The FMA-UE is considered the gold st andard and is the most widely used clinical assessment of post-stroke UE mo tor impairment. However its item-level measurement properties have not been well studied with contemporary measurement models, e.g., item response theory. A review of the literature identif ied several potential deficiencies in FMA-UE item level content and structure; (1) all items might not be consistent with current motor control theoreti cal expectations of pos t-stroke impairment and recovery, (2) the reproducib ility of the construct recovery illustrated by its item hierarchy had not been established across subj ects of various ability levels, nor within subjects over time, and (3) in its traditional format, the FMA-UE score had no clear functional meaning. These possible deficienci es were addressed in this project. The overall purpose of this dissertation was to examine, and potentially enhance, the measurement properties of the FMA-UE. Item Response Theory, i.e., the Rasch measurement model, was used to investigate (1) the validity of the FMA-UE item content, (2) the reproducibility of its item structure, and (3) interpretation of its score. The purpose of this dissertation wa s addressed in four studies. The aim of study 1 was to test the validity of the traditional post-stroke UE motor control theory that underlies the FMA-UE by examining its item-level dimensionality and item difficulty hierarchy. Principal Component s Analysis and Rasch analysis was applied to the data from 512 individuals 1-6 months post-stroke. The results indicated that the items of the FMA-UE contribute to the measurement of multiple constructs. Specifically,

PAGE 190

176 the three reflex items were empirically disconnected from the other items of the assessment. The reflex items were removed, creating a modified 30-item FMA-UE, a unidimensional assessment of post-stroke uppe r extremity motor ability. Rasch analysis revealed that the FMA-UE item difficulty order was not consistent with the item order proposed by Fugl-Meyer, but instead was c onsistent with contemporary motor control theoretical views that post-stroke upper ex tremity movements are easier or more challenging based on task-specific contex tual requirements of the movement. The Rasch-derived 30-item FMA-UE difficulty hierarchy remained consistent, independent of person ability, across 98% of the 512-person sample. The aim of study 2 was to test the longit udinal reproducibility of the 30-item FMAUE item structure. Rasch analysis was used to examine data from 377 individuals with stroke to whom the FMA-UE was administered at two points during th e recovery process, 1-3 months post-stroke, and 6 months post-stroke. Using Diff erential Item Functioning procedure, the item difficultly measures at each time point were compared. Ideally, item difficulties should be invariant each time an asse ssment is administered to assure that the score is reproducible and comparable. It was found that the item difficulty order varies slightly, but that this va riance has no practical conse quences for measuring upper extremity motor ability, suggesting that the 30-item FMA-UE item structure is reproducible in test-r etest situations. The aim of study 3 was to test a method for enhancing the interpretability of a FMA-UE change score by linking it to the Wolf Motor Function Test (WMFT) an upper extremity motor ability assessment with f unctionally oriented items. Using Rasch analysis it was determined that a single trait is measured by both assessments. It was

PAGE 191

177 feasible to link the two assessments to a common measurement scale. This method afforded an obvious connection between the FMA-UE score an its functional interpretation, thereby increasing the interpretability of the score. Because Rasch analysis was the primar y method used throughout these three studies, the aim of study 4 was to compare the results obtained with the Rasch model to those obtained with another item response th eory model to assure that the choice of statistical method was valid. The results show ed that the results obtained with a twoparameter item response theory mathematical model are virtually identical to those obtained by the Rasch analysis This finding suggests that the results obtained using Rasch analysis in the dissertation are valid. This project is significant for two reasons. First, the FMA-UE item content and structure represents traditional assumptions regarding the sequence of post-stroke upper extremity recovery. Rasch analysis provide d insights about the dimensionality and difficulties of FMA-UE items thereby challenging traditional assumptions and providing insight into the construct of recovery consis tent with contemporary motor control theory. Second, linking the FMA-UE to the WMFT has produced a item diffic ulty hierarchy that might illustrate the pattern of post-stroke recovery, progressing from reacquisition of single joint and multiple joint movements (F MA-UE items) to reacquisition of simple functional reaching, graspi ng and lifting tasks (WMFT items). The combined item difficulty hierarchy shows that FMA-UE change scores for individu als at the lower end of the scale (those with less upper extremity motor ability) will have different meaning than FMA-UE change scores for individuals in the middle or at the top of the scale (those with more upper extremity motor ability). Individu als with less ability will require larger

PAGE 192

178 increases in the FMA-UE score before functi onal gains are realized, and individuals with more ability will realize functional gains with smaller increases in the FMA-UE score. The combined item hierarchy also may be the basis for exploring hypotheses regarding post-stroke UE motor impairment and the func tional consequences of impairment and/or assisting rehabilitation prof essionals in designing the sequence for therapeutic interventions to reduce impairment and improve functional ability. The overarching purpose of this dissertati on and the aims of each project were consistent with that of an overall emphasis within the field of rehabilitation research to assure that clinical assessme nt tools are optimal for measuring targeted outcomes [58-60] and translate to functionally interpretable m easurements of change [61]. An emphasis on evidence-based practice in pos t-stroke motor rehabilitation has amplified the need for valid, reliable, and functionally relevant outco me measures to evaluate the effects of intervention. This dissertati on has examined and improved th e measurement properties of the FMA-UE, a long-standing and important post-stroke rehabil itation measure.

PAGE 193

179 APPENDIX A RASCH ANALYSIS CONTROL FILES Control File for Project 1 (Chapter 2) Title="Fugl-Meyer Upper Extremity Motor Assessment" ;512 person pooled dataset ;reflex items removed IDFILE=* 1 2 18 CODES=012 NAME1=1 NAMELEN=3 NI=33 ITEM1=5 TABLES=111111111111111111111 DISCRIM=YES &END 1 Biceps reflex 2 Triceps reflex 3 Scapular elevation 4 Scapular retraction 5 Shoulder abduction 6 Shoulder exte rnal rotation 7 Elbow flexion 8 Forearm supination 9 Shoulder adduction with internal rotation 10 Elbow extension 11 Forearm pronation 12 Hand to lumbar spine 13 Sh flexion to 90, elbow extended 14 Pronation-supination, elbow at 90 15 Sh abd to 90, elbow extended 16 Sh flex to 180, elbow extended 17 Pronation-supination, elbow extended 18 Normal reflex activity 19 Wrist stable, elbow at 90

PAGE 194

180 20 Wrist flex-ext, elbow at 90 21 Wrist stable, elbow extended 22 Wrist flex-ext, elbow extended 23 Wrist circumduction 24 Finger mass flexion 25 Finger mass extension 26 Hook grasp 27 Lateral prehension 28 Palmar prehension 29 Cylindrical grasp 30 Spherical grasp 31 Movement without tremor 32 Movement without dysmetria 33 Movement with normal speed END NAMES Control File for Project 2, T1 (Chapter 3) Title=T1 30-item Fugl-Meyer Upper Extremity Motor Assessment, n= 377" ;KC stroke study T1 data but only those with T2 data only n = 377 CODES=012 NAME1=1 NAMELEN=3 NI=33 ITEM1=5 IDFILE=* 1 2 18 TABLES=111111111111111111111 DISCRIM=YES &END 1 Biceps reflex 2 Triceps reflex 3 Scapular elevation 4 Scapular retraction 5 Shoulder abduction 6 Shoulder exte rnal rotation 7 Elbow flexion 8 Forearm supination 9 Shoulder adduction with internal rotation 10 Elbow extension 11 Forearm pronation

PAGE 195

181 12 Hand to lumbar spine 13 Sh flexion to 90, elbow extended 14 Pronation-supination, elbow at 90 15 Sh abd to 90, elbow extended 16 Sh flex to 180, elbow extended 17 Pronation-supination, elbow extended 18 Normal reflex activity 19 Wrist stable, elbow at 90 20 Wrist flex-ext, elbow at 90 21 Wrist stable, elbow extended 22 Wrist flex-ext, elbow extended 23 Wrist circumduction 24 Finger mass flexion 25 Finger mass extension 26 Hook grasp 27 Lateral prehension 28 Palmar prehension 29 Cylindrical grasp 30 Spherical grasp 31 Movement without tremor 32 Movement without dysmetria 33 Movement with normal speed END NAMES Control File for Project 2, T2 (Chapter 3) Title="T2 30-item Fugl-Meyer Upper Extremity Motor Assessment, n = 377" ;KC stroke study T2 data but only those with T1 data only, n = 377 CODES=012 NAME1=1 NAMELEN=3 NI=33 ITEM1=5 IDFILE=* 1 2 18 TABLES=111111111111111111111 DISCRIM=YES &END 1 Biceps reflex 2 Triceps reflex

PAGE 196

182 3 Scapular elevation 4 Scapular retraction 5 Shoulder abduction 6 Shoulder exte rnal rotation 7 Elbow flexion 8 Forearm supination 9 Shoulder adduction with internal rotation 10 Elbow extension 11 Forearm pronation 12 Hand to lumbar spine 13 Sh flexion to 90, elbow extended 14 Pronation-supination, elbow at 90 15 Sh abd to 90, elbow extended 16 Sh flex to 180, elbow extended 17 Pronation-supination, elbow extended 18 Normal reflex activity 19 Wrist stable, elbow at 90 20 Wrist flex-ext, elbow at 90 21 Wrist stable, elbow extended 22 Wrist flex-ext, elbow extended 23 Wrist circumduction 24 Finger mass flexion 25 Finger mass extension 26 Hook grasp 27 Lateral prehension 28 Palmar prehension 29 Cylindrical grasp 30 Spherical grasp 31 Movement without tremor 32 Movement without dysmetria 33 Movement with normal speed END NAMES WMFT Control File for Pr oject 3, T1 (Chapter 4) Title="T1 WMFT n= 91 quartiles" CODES=1234 ISGroups=0 NAME1=1 NAMELEN=4 NI=15 ITEM1=6 ;DIF=$S3W2

PAGE 197

183 ;DIF analysis with classification in column 34 with a width of 2 characters (T1 or T2) TABLES=111111111111111111111 &END 1 Forearm to table 2 Forearm to box 3 Extend elbow to side 4 Extend elbow push weight 5 Hand to table 6 Hand to box 7 Reach and retrieve 8 Lift soda can 9 Lift pencil 10 Lift paperclip 11 Stack checkers 12 Flip cards 13 Turn key 14 Fold towel 15 Lift basket END NAMES WMFT Control File for Pr oject 3, T2 (Chapter 4) Title="WMFT Pepper T2 n= 91 quartiles" ;Pepper study T2 n = 91, timed data made into ordinal with quartiles CODES=1234 ISGroups=0 NAME1=1 NAMELEN=4 NI=15 ITEM1=6 ;DIF=$S3W2 ;DIF analysis with classification in column 34 with a width of 2 charachters (T1 or T2) TABLES=111111111111111111111 &END 1 Forearm to table 2 Forearm to box 3 Extend elbow to side 4 Extend elbow push weight 5 Hand to table 6 Hand to box 7 Reach and retrieve

PAGE 198

184 8 Lift soda can 9 Lift pencil 10 Lift paperclip 11 Stack checkers 12 Flip cards 13 Turn key 14 Fold towel 15 Lift basket END NAMES FMA-UE + WMFT Control File for Project 3, T1 (Chapter 4) Title="FMA-UE + WMFT T1 Combined n= 91" ;Pepper study combined analysis of 91 FMA and WMFT data T1 CODES= 01234 NAME1=1 NAMELEN=4 NI=48 ITEM1=6 ISGROUPS=111111111111111111111111111111111000000000000000 ;DIF=$S3W2 ;DIF analysis with classification in column 34 with a width of 2 characters (T1 or T2) TABLES=111111111111111111111 IDFILE=* 1 2 18 &END 1 FMA-UE Biceps reflex 2 FMA-UE Triceps reflex 3 FMA-UE Scapular elevation 4 FMA-UE Scapular retraction 5 FMA-UE Shoulder abduction 6 FMA-UE Shoulder external rotation 7 FMA-UE Elbow flexion 8 FMA-UE Forearm supination 9 FMA-UE Shoulder adducti on with internal rotation 10 FMA-UE Elbow extension 11 FMA-UE Forearm pronation 12 FMA-UE Hand to lumbar spine 13 FMA-UE Sh flexion to 90, elbow extended

PAGE 199

185 14 FMA-UE Pronation-supi nation, elbow at 90 15 FMA-UE Sh abd to 90, elbow extended 16 FMA-UE Sh flex to 180, elbow extended 17 FMA-UE Pronation-supina tion, elbow extended 18 FMA-UE Normal reflex activity 19 FMA-UE Wrist stable, elbow at 90 20 FMA-UE Wrist flex -ext, elbow at 90 21 FMA-UE Wrist stable, elbow extended 22 FMA-UE Wrist flex-ext, elbow extended 23 FMA-UE Wrist circumduction 24 FMA-UE Finger mass flexion 25 FMA-UE Finger mass extension 26 FMA-UE Hook grasp 27 FMA-UE Lateral prehension 28 FMA-UE Palmar prehension 29 FMA-UE Cylindrical grasp 30 FMA-UE Spherical grasp 31 FMA-UE Movement without tremor 32 FMA-UE Movement without dysmetria 33 FMA-UE Movement with normal speed 34 WMFT Forearm to table 35 WMFT Forearm to box 36 WMFT Extend elbow to side 37 WMFT Extend elbow push weight 38 WMFT Hand to table 39 WMFT Hand to box 40 WMFT Reach and retrieve 41 WMFT Lift soda can 42 WMFT Lift pencil 43 WMFT Lift paperclip 44 WMFT Stack checkers 45 WMFT Flip cards 46 Turn key 47 WMFT Fold towel 48 WMFT Lift basket END NAMES FMA-UE + WMFT Control File for Project 3, T2 (Chapter 4) Title="FMA-UE + WMFT T2 Combined n= 91" ;Pepper study combined analysis of 91 FMA and WMFT data T2 CODES= 01234 NAME1=1

PAGE 200

186 NAMELEN=4 NI=48 ITEM1=6 ISGROUPS=111111111111111111111111111111111000000000000000 ;DIF=$S3W2 ;DIF analysis with classification in column 34 with a width of 2 characters (T1 or T2) TABLES=111111111111111111111 IDFILE=* 1 2 18 &END 1 FMA-UE Biceps reflex 2 FMA-UE Triceps reflex 3 FMA-UE Scapular elevation 4 FMA-UE Scapular retraction 5 FMA-UE Shoulder abduction 6 FMA-UE Shoulder external rotation 7 FMA-UE Elbow flexion 8 FMA-UE Forearm supination 9 FMA-UE Shoulder adducti on with internal rotation 10 FMA-UE Elbow extension 11 FMA-UE Forearm pronation 12 FMA-UE Hand to lumbar spine 13 FMA-UE Sh flexion to 90, elbow extended 14 FMA-UE Pronation-supi nation, elbow at 90 15 FMA-UE Sh abd to 90, elbow extended 16 FMA-UE Sh flex to 180, elbow extended 17 FMA-UE Pronation-supina tion, elbow extended 18 FMA-UE Normal reflex activity 19 FMA-UE Wrist stable, elbow at 90 20 FMA-UE Wrist flex -ext, elbow at 90 21 FMA-UE Wrist stable, elbow extended 22 FMA-UE Wrist flex-ext, elbow extended 23 FMA-UE Wrist circumduction 24 FMA-UE Finger mass flexion 25 FMA-UE Finger mass extension 26 FMA-UE Hook grasp 27 FMA-UE Lateral prehension 28 FMA-UE Palmar prehension 29 FMA-UE Cylindrical grasp 30 FMA-UE Spherical grasp

PAGE 201

187 31 FMA-UE Movement without tremor 32 FMA-UE Movement without dysmetria 33 FMA-UE Movement with normal speed 34 WMFT Forearm to table 35 WMFT Forearm to box 36 WMFT Extend elbow to side 37 WMFT Extend elbow push weight 38 WMFT Hand to table 39 WMFT Hand to box 40 WMFT Reach and retrieve 41 WMFT Lift soda can 42 WMFT Lift pencil 43 WMFT Lift paperclip 44 WMFT Stack checkers 45 WMFT Flip cards 46 WMFT Turn key 47 WMFT Fold towel 48 WMFT Lift basket END NAMES

PAGE 202

188 APPENDIX B SAS AND MULTILOG CONTROL FILES SAS PCA Control File for FMA-UE Data Project 1 (Chapter 2) data; input i1 5 i2 6 i3 7 i4 8 i5 9 i6 10 i7 11 i8 12 i9 13 i10 14 i11 15 i12 16 i13 17 i14 18 i15 19 i16 20 i17 21 i18 22 i19 23 i20 24 i21 25 i22 26 i23 27 i24 28 i25 29 i26 30 i27 31 i28 32 i29 33 i30 34 i31 35 i32 36 i33 37; label i1 = 'Biceps Reflex' i2 = 'Triceps Reflex' i3 = 'Scapular Elevation' i4 = 'Scapular Retraction' i5 = 'Shoulder Abduction' i6 = 'Shoulder External Rotation' i7 = 'Elbow Flexion' i8 = 'Forearm Supination' i9 = 'Shoulder Adduction with IR' i10 = 'Elbow Extension' i11 = 'Forearm Pronation' i12 = 'Hand to Lumbar Spine' i13 = 'Sh Flex to 90 Elbow Extended' i14 = 'Pronation Supination with Elbow 90' i15 = 'Sh Abd to 90 Elbow Extended' i16 = 'Sh Flex to 180 Elbow Extended' i17 = 'Pronation Supinati on with Elbow Extended' i18 = 'Normal Reflex Activity' i19 = 'Wrist Stable Elbow at 90' i20 = 'Wrist Flexion Extension Elbow 90' i21 = 'Wrist Stable Elbow Extended' i22 = 'Wrist Flexion Ex tension Elbow Extended' i23 = 'Wrist Circumduction' i24 = 'Finger Mass Flexion' i25 = 'Finger Mass Extension' i26 = 'Hook Grasp' i27 = 'Lateral Prehension' i28 = 'Palmar Prehension' i29 = 'Cylindrical Grasp' i30 = 'Spherical Grasp'

PAGE 203

189 i31 = 'Movement without tremor' i32 = 'Movement without dysmetria' i33 = 'Movement with normal speed' ; datalines; 001 222222222222212110212102100221021 002 220000000000000000000000000000000 003 222222222222222222222212201212222 004 220000000000000000000000000000000 005 221111111112111110000001100010111 (insert data here) proc factor rotate=obvarimax reorder; run; quit; SAS PCA Control File for WMFT Data Project 3 (Chapter 4) data; input i1 6 i2 7 i3 8 i4 9 i5 10 i6 11 i7 12 i8 13 i9 14 i10 15 i11 16 i12 17 i13 18 i14 19 i15 20; label i1 = 'Forearm to table' i2 = 'Forearm to box' i3 = 'Extend elbow' i4 = 'Ext elbow push weight' i5 = 'Hand to table' i6 = 'Hand to box' i7 = 'Reach and retrieve' i8 = 'Lift soda can' i9 = 'Lift pencil' i10 = 'Lift paperclip' i11 = 'Stack checkers' i12 = 'Flip cards' i13 = 'Turn key' i14 = 'Fold towel' i15 = 'Lift basket' ; datalines; 02T1 443333344342412 03T1 433443232222231 04T1 342433443431332 06T1 222331211232233 (insert data here)

PAGE 204

190 proc factor rotate=obvarimax reorder; run; quit; SAS PCA Control File for FMA-UE + WMFT Data Project 3 (Chapter 4) data; input i1 6 i2 7 i3 8 i4 9 i5 10 i6 11 i7 12 i8 13 i9 14 i10 15 i11 16 i12 17 i13 18 i14 19 i15 20 i16 21 i17 22 i18 23 i19 24 i 20 25 i21 26 i22 27 i23 28 i24 29 i25 30 i26 31 i27 32 i28 33 i29 34 i30 35 i31 36 i 32 37 i33 38 i34 39 i35 40 i36 41 i37 42 i38 43 i39 44 i40 45 i41 46 i42 47 i43 48 i44 49 i45 50; label i1 = 'FMA-UE Scapular elevation' i2 = 'FMA-UE Scapular retraction' i3 = 'FMA-UE Shoulder abduction' i4 = 'FMA-UE Shoulder external rotation' i5 = 'FMA-UE Elbow flexion' i6 = 'FMA-UE Forearm supination' i7 = 'FMA-UE Shoulder adducti on with internal rotation' i8 = 'FMA-UE Elbow extension' i9 = 'FMA-UE Forearm pronation' i10 = 'FMA-UE Hand to lumbar spine' i11 = 'FMA-UE Sh flexion to 90, elbow extended' i12 = 'FMA-UE Pronation-s upination, elbow at 90' i13 = 'FMA-UE Sh abd to 90, elbow extended' i14 = 'FMA-UE Sh flex to 180, elbow extended' i15 = 'FMA-UE Pronation-supi nation, elbow extended' i16 = 'FMA-UE Wrist st able, elbow at 90' i17 = 'FMA-UE Wrist flex-ext, elbow at 90' i18 = 'FMA-UE Wrist st able, elbow extended' i19 = 'FMA-UE Wrist flex -ext, elbow extended' i20 = 'FMA-UE Wrist circumduction' i21 = 'FMA-UE Finger mass flexion' i22 = 'FMA-UE Finger mass extension' i23 = 'FMA-UE Hook grasp' i24 = 'FMA-UE Lateral prehension' i25 = 'FMA-UE Palmar prehension' i26 = 'FMA-UE Cylindrical grasp' i27 = 'FMA-UE Spherical grasp' i28 = 'FMA-UE Movement without tremor' i29 = 'FMA-UE Movement without dysmetria'

PAGE 205

191 i30 = 'FMA-UE Movement with normal speed' i31 = 'WMFT Forearm to table' i32 = 'WMFT Forearm to box' i33 = 'WMFT Extend elbo' i34 = 'WMFT Ext elbow push weight' i35 = 'WMFT Hand to table' i36 = 'WMFT Hand to box' i37 = 'WMFT Reach and retrieve' i38 = 'WMFT Lift soda can' i39 = 'WMFT Lift pencil' i40 = 'WMFT Lift paperclip' i41 = 'WMFT Stack checkers' i42 = 'WMFT Flip cards' i43 = 'WMFT Turn key' i44 = 'WMFT Fold towel' i45 = 'WMFT Lift basket' ; datalines; 01T2 222111222111212110221211212221222444134244342313 02T2 222111212222222120212112221122222343444343232113 03T2 222222212221222222222212221121122332222113321312 (insert data here) proc factor nfact= 2 rotate=obvarimax reorder; run; quit; Multilog Control File for D512 (Chapter 5) MULTILOG for Windows 7.00.2327.2 Created on: 7 June 2006, 13:22:57 >PROBLEM RANDOM, INDIVIDUAL, DATA = '512FMAT1Recod.DAT', NITEMS = 30, NGROUPS = 1, NEXAMINEES = 512, NCHARS = 3; >TEST ALL, GRADED, NC = (3(0)30); >END ; 3 123

PAGE 206

192 111111111111111111111111111111 222222222222222222222222222222 333333333333333333333333333333 (33A1) Multilog Control File for D377 (Chapter 5) MULTILOG for Windows 7.00.2327.2 Created on: 10 June 2006, 13:37:43 >PROBLEM RANDOM, INDIVIDUAL, DATA = 'FMAN377T2.DAT', NITEMS = 30, NGROUPS = 1, NEXAMINEES = 377, NCHARS = 3; >TEST ALL, GRADED, NC = (3(0)30); >END ; 3 123 111111111111111111111111111111 222222222222222222222222222222 333333333333333333333333333333 (33A1)

PAGE 207

193 LIST OF REFERENCES 1. American-Heart-Association, 2005 Heart Disease and Stroke Statistics 2005 Update. 2005, American Heart A ssociation: Dallas, Texas. 2. Kwakkel, G, BJ Kollen, J van der Grond, and AJ Prevo, Probability of regaining dexterity in the flaccid upper limb: impact of severity of paresis and time since onset in acute stroke. Stroke, 2003. 34(9): p. 2181-6. 3. Duncan, PW, LB Goldstein, RD Horn er, PB Landsman, GP Samsa, and DB Matchar, Similar motor recovery of upper and lower extremities after stroke. Stroke, 1994. 25(6): p. 1181-8. 4. Broeks, JG, GJ Lankhorst, K Rumping, a nd AJ Prevo, The long-term outcome of arm function after stroke: results of a follow-up study. Disabil Rehabil, 1999. 21(8): p. 357-64. 5. Nakayama, H, HS Jorgensen, HO Raas chou, and TS Olsen, Recovery of upper extremity function in stroke patients: the Copenhagen Stroke Study. Arch Phys Med Rehabil, 1994. 75(4): p. 394-8. 6. Nudo, R, Adaptive plasticity in motor co rtex: implications fo r rehabilitation after brain injury. J Rehabil Med, 2003. 41: p. 7-10. 7. Lum, PS, CG Burgar, PC Shor, M Majmundar, and M Van der Loos, Robotassisted movement training compared with conventional therapy techniques for the rehabilitation of upper-limb motor function after stroke. Arch Phys Med Rehabil, 2002. 83(7): p. 952-9. 8. Winstein, CJ, JP Miller, S Blanton, E Taub, G Uswatte, D Morris, D Nichols, and S Wolf, Methods for a multisite randomized trial to investigate the effect of constraint-induced movement therapy in improving upper extremity function among adults recovering from a cerebrova scular stroke. Neur orehabil Neural Repair, 2003. 17(3): p. 137-52. 9. Michaelsen, SM, R Dannenbaum, and MF Levin, Task-specific training with trunk restraint on arm recovery in stroke: randomi zed control trial. Stroke, 2006. 37(1): p. 186-92. 10. Howle, J, Neuro-Developmental Treatm ent Approach: Theoretic al Foundations and Principles of Clinical Practice. 2002, Laguana Beach, CA: Neuro-Developmental Treatment Association.

PAGE 208

194 11. Maher, CG, C Sherrington, M Elkins, RD Herbert, and AM Moseley, Challenges for evidence-based physical therapy: accessing and interp reting high-quality evidence on therapy. Phys Ther, 2004. 84(7): p. 644-54. 12. Coster, W, International conference on evidence-based practice: A collaborative effort of the American Occupational Therapy Association, the American Occupational Therapy Foundation, and the Agency for Healthcare Research and Quality. Am J Occup Ther, 2005. 59(3): p. 356-8. 13. Whitall, J, Stroke rehabilitation research: time to answer more specific questions? Neurorehabil Neural Repa ir, 2004. 18(1): p. 3-8. 14. Salter, K, JW Jutai, R Teasell, NC Foley, and J Bitensky, Issues for selection of outcome measures in stroke rehabilitatio n: ICF Body Functions. Disabil Rehabil, 2005. 27(4): p. 191-207. 15. Fugl-Meyer, AR, L Jaasko, I Leyman, S Olsson, and S Steglind, The post-stroke hemiplegic patient I: a method for eval uation of physical performance. Scand J Rehabil Med, 1975. 7(1): p. 13-31. 16. Van Wijck, FM, AD Pandyan, GR Johnson, and MP Barnes, Assessing motor deficits in neurologi cal rehabilitation: patterns of instrument usage. Neurorehabil Neural Repair, 2001. 15(1): p. 23-30. 17. Duncan, PW, M Propst, and SG Nelson, Reliability of the Fugl-Meyer assessment of sensorimotor recovery following cereb rovascular accident. Phys Ther, 1983. 63(10): p. 1606-10. 18. Sanford, J, J Moreland, LR Swanson, PW Stratford, and C Gowland, Reliability of the Fugl-Meyer assessment for testing motor performance in patients following stroke. Phys Ther, 1993. 73(7): p. 447-454. 19. Wood-Dauphinee, SL, JI Williams, and SH Shapiro, Examining outcome measures in a clinical study of stroke. Stroke, 1990. 21(5): p. 731-9. 20. Malouin, F, L Pichard, C Bonneau, A Durand, and D Corriveau, Evaluating motor recovery early after stroke: comparison of the Fugl-Meyer Assessment and the Motor Assessment Scale. Arch Phys Med Rehabil, 1994. 75(11): p. 1206-12. 21. Filiatrault, J, AB Arsenault, E Du til, and D Bourbonnais, Motor function and activities of daily living assessments: a study of three tests for persons with hemiplegia. Am J Occup Ther, 1991. 45(9): p. 806-10. 22. Katz, RT, GP Rovai, C Brait, and WZ Rymer, Objective quantification of spastic hypertonia: correlation with cl inical findings. Arch Phys Med Rehabil, 1992. 73(4): p. 339-47.

PAGE 209

195 23. Gowland, C, P Stratford, M Ward, J Moreland, W Torresin, S Van Hullenaar, J Sanford, S Barreca, B Vanspall, and N Plew s, Measuring physical impairment and disability with the Chedoke-McMaster St roke Assessment. Stroke, 1993. 24(1): p. 58-63. 24. van der Lee, JH, H Beckerman, GJ La nkhorst, and LM Bouter, The responsiveness of the Action Research Arm Test and the Fugl-Meyer Assessment Scale in chronic stroke patients. J Rehabil Med, 2001. 33(3): p. 110-3. 25. Wang, CH, CL Hsieh, MH Dai, CH Chen and YF Lai, Inter-rater reliability and validity of the stroke rehabilitation assessment of movement (STREAM) instrument. J Rehabil Med, 2002. 34(1): p. 20-4. 26. Wolf, SL, PA Catlin, M Ellis, AL Arch er, B Morgan, and A Piacentino, Assessing the Wolf Motor Function Test as outcome measure for research in patients after stroke. Stroke, 2001. 32(7): p. 1635-9. 27. Duncan, PW, SM Lai, and J Keighley, De fining post-stroke recovery: implications for design and interpretati on of drug trials. Neuropharmacology, 2000. 39(5): p. 835-41. 28. Boissy, P, D Bourbonnais, C Kaegi, D Gr avel, and BA Arsenault, Characterization of global synkineses during hand grip in hemiparetic patients. Arch Phys Med Rehabil, 1997. 78(10): p. 1117-24. 29. Desrosiers, J, L Noreau, A Rochette D Bourbonnais, G Bravo, and A Bourget, Predictors of long-term participation afte r stroke. Disabil Rehabil, 2006. 28(4): p. 221-30. 30. Duncan, PW, LG Richards, D Wallace, J Stoker-Yates, P Pohl, C Luchies, A Ogle, and S Studenski, A randomized, controlled pilot study of a home-based exercise program for individuals with mild and moderate stroke. Stroke, 1998. 29(10): p. 2055-60. 31. Raghavan, P, E Petra, JW Krakauer, a nd AM Gordon, Patterns of impairment in digit independence after subcortical str oke. J Neurophysiol, 2006. 95(1): p. 369-78. 32. Harris-Love, ML, S McCombe Waller, and J Whitall, Exploiting interlimb coupling to improve paretic arm reachi ng performance in people with chronic stroke. Arch Phys Med Rehabil, 2005. 86(11): p. 2131-7. 33. Levin, MF, RW Selles, MH Verheul, a nd OG Meijer, Deficits in the coordination of agonist and antagonist muscles in stroke patients: implicati ons for normal motor control. Brain Res, 2000. 853(2): p. 352-69. 34. Michaelsen, SM, S Jacobs, A Roby-Brami, and MF Levin, Compensation for distal impairments of grasping in adults with hemiparesis. Exp Brain Res, 2004. 157(2): p. 162-73.

PAGE 210

196 35. Levin, MF, Interjoint coordination during pointing movements is disrupted in spastic hemiparesis. Brain, 1996. 119: p. 281-93. 36. Stinear, JW and WD Byblow, Rhythmic bilateral movement training modulates corticomotor excitability and enhances upper limb mo tricity poststroke: a pilot study. J Clin Neurophysiol, 2004. 21(2): p. 124-31. 37. Luft, AR, S McCombe-Waller, J Whitall, LW Forrester, R Macko, JD Sorkin, JB Schulz, AP Goldberg, and DF Hanley, Re petitive bilateral arm training and motor cortex activation in chro nic stroke: a randomized c ontrolled trial. JAMA, 2004. 292(15): p. 1853-61. 38. Chae, J, F Bethoux, T Bohine, L Dobos T Davis, and A Friedl, Neuromuscular stimulation for upper extremity motor and f unctional recovery in acute hemiplegia. Stroke, 1998. 29(5): p. 975-9. 39. Francisco, G, J Chae, H Chawla, S Ki rshblum, R Zorowitz, G Lewis, and S Pang, Electromyogram-triggered neuromuscula r stimulation for improving the arm function of acute stroke survivors: a randomized pilot study. Arch Phys Med Rehabil, 1998. 79(5): p. 570-5. 40. Page, SJ and P Levine, Back from the brink: electromyography-triggered stimulation combined with modified c onstraint-induced movement therapy in chronic stroke. Arch Phys Med Rehabil, 2006. 87(1): p. 27-31. 41. Viriyavejakul, A, R Vachalathiti, an d N Poungvarin, Botulinum treatment for poststroke spasticity: low dose regime. J Med Assoc Thai, 1998. 81(6): p. 413-22. 42. Pang, MY, JE Harris, and JJ Eng, A community-based upper-extremity group exercise program improves motor function and performance of functional activities in chronic stroke: a random ized controlled trial. Arch Phys Med Rehabil, 2006. 87(1): p. 1-9. 43. Krebs, HI, BT Volpe, ML Aisen, and N Hogan, Increasing pr oductivity and quality of care: robot-aided neuro-rehabilitati on. J Rehabil Res Dev, 2000. 37(6): p. 63952. 44. Finley, MA, SE Fasoli, L Dipietro, J Ohlhoff, L Macclel lan, C Meister, J Whitall, R Macko, CT Bever, Jr., HI Krebs, and N Hogan, Short-duration robotic therapy in stroke patients with seve re upper-limb motor impairment J Rehabil Res Dev, 2005. 42(5): p. 683-92. 45. Piron, L, F Cenni, P Tonin, and M Dam, Virtual Reality as an assessment tool for arm motor deficits after brain lesions. Stud Health Technol Inform, 2001. 81: p. 386-92.

PAGE 211

197 46. Page, S, Imagery improves upper extrem ity motor function in chronic stroke patients: a pilot study. The Occupational Therapy Journal of Research, 2000. 23(3): p. 200-215. 47. van der Lee, JH, RC Wagenaar, GJ La nkhorst, TW Vogelaar, WL Deville, and LM Bouter, Forced use of the upper extremity in chronic stroke patients: results from a single-blind randomized clinical tria l. Stroke, 1999. 30(11): p. 2369-75. 48. Page, SJ, SA Sisto, P Levine, MV Johns ton, and M Hughes, Modified constraint induced therapy: a randomized feasibility and efficacy study. J Rehabil Res Dev, 2001. 38(5): p. 583-90. 49. Page, SJ, S Sisto, MV Johnston, and P Levine, Modified cons traint-induced therapy after subacute stroke: a preliminary study. Neuroreh abil Neural Repair, 2002. 16(3): p. 290-5. 50. Page, SJ, S Sisto, MV Johnston, P Le vine, and M Hughes, M odified constraintinduced therapy in subacute stroke: a cas e report. Arch Phys Med Rehabil, 2002. 83(2): p. 286-90. 51. Bonifer, NM, KM Anderson, and DB Arciniegas, Constraint-induced movement therapy after stroke: efficacy for patien ts with minimal upper-extremity motor ability. Arch Phys Med Reha bil, 2005. 86(9): p. 1867-73. 52. Heart and Stroke Foundation of Ontario. Management of the Po st Stroke Arm and Hand: Treatment Recommendations of the 2001 Consensus Panel., http://209.5.25.171/ClientImages/1 /PostStrokeArmandHandFinal2002.pdf. Accessed May 3, 2006. 53. Hambleton, RK, H Swaminathan, and HJ Rogers, Fundamentals of Item Response Theory. 1991, Newbury Park, CA: Sage Publications. 54. De Weerdt, W and MA Harrison, Meas uring recovery of arm-hand function in stroke patients: a comparison of the Brunnstrom-Fugl-Meyer test and Action Research Arm test. Physiother Can, 1985. 37: p. 65-70. 55. Gladstone, DJ, CJ Danells and SE Black, The Fugl-M eyer Assessment of motor recovery after stroke: a cri tical review of its measurem ent properties. Neurorehabil Neural Repair, 2002. 16(3): p. 232-40. 56. Platz, T, C Pinkowski, F van Wijck, IH Kim, P di Bella, and G Johnson, Reliability and validity of arm function assessment w ith standardized guidelines for the FuglMeyer Test, Action Research Arm Test and Box and Block Test: a multicentre study. Clin Rehabil, 2005. 19(4): p. 404-11. 57. Berglund, K and AR Fugl-Meyer, Upper extremity function in hemiplegia. A crossvalidation study of two assessment methods Scand J Rehabil Med, 1986. 18(4): p. 155-7.

PAGE 212

198 58. Wolf, S and JA Kleim. Top 10 Take Home Messages. in The III Step Conference: Linking Movement Science and Interv ention. 2005. Salt Lake City, UT. 59. Winstein, C, Why is the functional in dependence measure used to identify some rehabilitation needs in stroke survivors wh en there are better tools? Physiother Res Int, 2004. 9(4): p. 182-4. 60. Research plan for the national center fo r medical rehabilitation research. 1993, National Institutes of Health. 61. Haley, SM and MA Fragala-Pinkham, In terpreting change sc ores of tests and measures used in physical therapy. Phys Ther, 2006. 86(5): p. 735-43. 62. International Classification of Functi oning (ICF), Disability and Health. 2001, Geneva: World Health Organization. 63. Jette, AM, Toward a common language fo r function, disability, and health. Phys Ther, 2006. 86(5): p. 726-734. 64. Salter, K, JW Jutai, R Teasell, NC Fo ley, J Bitensky, and M Bayley, Issues for selection of outcome measures in stroke rehabilitation: ICF Activity. Disabil Rehabil, 2005. 27(6): p. 315-40. 65. Salter, K, JW Jutai, R Teasell, NC Fo ley, J Bitensky, and M Bayley, Issues for selection of outcome measures in stroke rehabilitation: ICF Participation. Disabil Rehabil, 2005. 27(9): p. 507-28. 66. Shumway-Cook, A and MH Woollacott, Motor Control: Theo ry and Practical Applications. 2nd ed. 2001, Philadelphia: Lippincott Williams & Wilkins. 67. McDowell, I and C Newell, The Theoreti cal and Technical Foundations of Health Measurement, in Measuring Health: A Gu ide to Rating Scales and Questionnaires, I. McDowell and C. Newell, Editors. 1996, New York:University Press: p. 10-46. 68. Barreca, SR, PW Stratford, LM Masters, CL Lambert, and J Griffiths, Comparing two versions of the Chedoke Arm and Hand Activity Inventory with the Action Research Arm Test. Phys Ther, 2006. 86(2): p. 245-53. 69. Jackson, JH, Selected Writings. 1958, New York: Basic Books. 70. Schmidt, RA and TD Lee, Motor Cont rol and Learning: a Behavioral Emphasis. 3rd ed. 1999, Champaign, IL: Human Kinetics. 71. Clower, WT, Early contribu tions to the reflex chain hypothesis. J Hist Neurosci, 1998. 7(1): p. 32-42.

PAGE 213

199 72. Mathiowetz, V and JB Haugen, Motor behavior research: implications for therapeutic approaches to central nervous system dysfunction. Am J Occup Ther, 1994. 48(8): p. 733-45. 73. Bobath, B, Observations on adult hemiplegia and suggestions for treatment. Physiotherapy, 1960. 46: p. 5-14. 74. Kabat, H and M Knott, Proprioceptive facilitation techniques for treatment of paralysis. Phys Ther Rev, 1953. 33(2): p. 53-64. 75. Twitchell, TE, The restoration of moto r function following hemiplegia in man. Brain, 1951. 74: p. 443-480. 76. Brunnstrom, S, Motor testing procedur es in hemiplegia: based on sequential recovery stages. Phys Ther, 1966. 46(4): p. 357-75. 77. Brunnstrom, S, Movement Therapy in Hemiplegia: a Neurophysiological Approach. 1970, New York: Harper & Row. 78. Carr, JH and RB Shepherd, Movement Sc ience: Foundations for Physical Therapy in Rehabilitation. 2nd ed. 2000, Gaith ersburg, Md.: Aspen Publishers. 79. Krakauer, JW, Motor learning: its relevance to stroke recovery and neurorehabilitation. Curr Opin Neurol, 2006. 19(1): p. 84-90. 80. Zehr, EP, Neural cont rol of rhythmic human movement: the common core hypothesis. Exerc Sport Sci Rev, 2005. 33(1): p. 54-60. 81. Taub, E, Movement in nonhuman primates deprived of somatosensory feedback. Exerc Sport Sci Rev, 1976. 4: p. 335-74. 82. Levin, MF and AG Feldman, The role of stretch reflex threshold regulation in normal and impaired motor control. Brain Res, 1994. 657(1-2): p. 23-30. 83. Wolpaw, JR and AM Tennissen, Activity-dep endent spinal cord plasticity in health and disease. Annu Rev Neur osci, 2001. 24: p. 807-43. 84. Nakazawa, K, SI Yamamoto, T Ohtsuki, H Yano, and T Fukunaga, Neural control: novel evaluation of stretch reflex sensitivity. Acta Physiol Scand, 2001. 172(4): p. 257-68. 85. Ogiso, K, JM McBride, T Finni, and PV Komi, Stretch-reflex mechanical response to varying types of previous muscle ac tivities. Journal of Electromyography and Kinesiology, 2002. 12(1): p. 27-36. 86. Wolpaw, JR, Adaptive plasticity in the sp inal stretch reflex: an accessible substrate of memory? Cell Mol Neurobiol 1985. 5(1-2): p. 147-65.

PAGE 214

200 87. Wolf, SL and RL Segal, Reducing huma n biceps brachii spinal stretch reflex magnitude. J Neurophysiol, 1996. 75(4): p. 1637-1646. 88. Burne, JA, VL Carleton, and NJ O' Dwyer, The spasticity paradox: movement disorder or disorder of resting limbs? J Neurol Neurosurg Psychiatry, 2005. 76(1): p. 47-54. 89. Lindmark, B and E Hamrin, Evaluation of functional capacity after stroke as a basis for active intervention. Validation of a modified chart for motor capacity assessment. Scand J Rehabil Med, 1988. 20(3): p. 111-5. 90. Welmer, AK, LW Holmqvist, and DK Somm erfeld, Hemiplegic limb synergies in stroke patients. Am J Phys Med Rehabil, 2006. 85(2): p. 112-9. 91. Beer, RF, JD Given, and JP Dewald, Task-dependent weakness at the elbow in patients with hemiparesis. Arch Phys Med Rehabil, 1999. 80(7): p. 766-72. 92. Dewald, JP and RF Beer, Abnormal joint torque patterns in th e paretic upper limb of subjects with hemiparesis. Musc le Nerve, 2001. 24(2): p. 273-83. 93. Lum, PS, CG Burgar, and PC Shor, Evidence for strength imbalances as a significant contributo r to abnormal synergies in he miparetic subjects. Muscle Nerve, 2003. 27(2): p. 211-21. 94. Beer, RF, JP Dewald, ML Dawson, and WZ Rymer, Target-dependent differences between free and constrained arm moveme nts in chronic hemiparesis. Exp Brain Res, 2004. 156(4): p. 458-70. 95. Kamper, DG, AN McKenna-Cole, LE Kahn, and DJ Reinkensmeyer, Alterations in reaching after stroke and their relation to movement direction and impairment severity. Arch Phys Med Reha bil, 2002. 83(5): p. 702-7. 96. Trombly, CA, Deficits of reaching in subj ects with left hemiparesis: a pilot study. Am J Occup Ther, 1992. 46(10): p. 887-97. 97. Chae, J, I Labatia, and G Yang, Uppe r limb motor function in hemiparesis: concurrent validity of the Arm Motor Ab ility test. Am J Phys Med Rehabil, 2003. 82(1): p. 1-8. 98. Michaelsen, SM, A Luta, A Roby-Brami, and MF Levin, Effect of trunk restraint on the recovery of reaching movements in hemiparetic patients. Stroke, 2001. 32(8): p. 1875-83. 99. Latash, ML, Bernsteins traditions in movement studies, in Progress in Motor Control, M.L. Latash, Editor. 1998, Champaign:Human Kinetics.

PAGE 215

201 100. Wu, C, CA Trombly, K Li n, and L Tickle-Degnen, A kinematic study of contextual effects on reaching performance in persons with and without str oke: influences of object availability. Arch Phys Me d Rehabil, 2000. 81(1): p. 95-101. 101. Wu, C, CA Trombly, K Li n, and L Tickle-Degnen, Eff ects of object affordances on reaching performance in persons with and without cerebrovascular accident. Am J Occup Ther, 1998. 52(6): p. 447-56. 102. Wu, CY, CA Trombly, and KC Lin, Th e relationship between occupational form and occupational performance: a kinematic perspective. Am J Occup Ther, 1994. 48(8): p. 679-87; discussion 688. 103. Ma, HI and CA Trombly, The comparis on of motor performance between part and whole tasks in elderly persons. Am J Occup Ther, 2001. 55(1): p. 62-7. 104. Ma, HI, CA Trombly, and C RobinsonPodolski, The effect of context on skill acquisition and transfer. Am J O ccup Ther, 1999. 53(2): p. 138-44. 105. Jaric, S and ML Latash, Learning a moto r task involving obstacles by a multi-joint, redundant limb: two synergies within one movement. J Electromyogr Kinesiol, 1998. 8(3): p. 169-76. 106. Murray, IA and GR Johnson, A study of the external forces and moments at the shoulder and elbow while performing ever y day tasks. Clin Biomech, 2004. 19(6): p. 586-94. 107. Latash, ML and JJ Nicholas, Motor cont rol research in reha bilitation medicine. Disabil Rehabil, 1996. 18(6): p. 293-9. 108. Latash, ML and JG Anson, What are normal movements" in atypical populations? Behav Brain Sci, 1996. 19(1): p. 55-106. 109. Mercier, C, AM Bertrand, and D Bourbonnais, Comparison of strength measurements under single-joint and multi-joint conditions in hemiparetic individuals. Clin Rehabi l, 2005. 19(5): p. 523-30. 110. Zackowski, KM, AW Dromerick, SA Sa hrmann, WT Thach, and AJ Bastian, How do strength, sensation, spasticity and jo int individuation rela te to the reaching deficits of people with chr onic hemiparesis? Brain, 2004. 111. McCrea, PH, JJ Eng, and AJ Hodgson, Sa turated muscle activation contributes to compensatory reaching strategies afte r stroke. J Neurophysiol, 2005. 94(5): p. 2999-3008. 112. Scott, SH and KE Norman, Computationa l approaches to motor control and their potential role for interpreting motor dysf unction. Curr Opin Neurol, 2003. 16(6): p. 693-8.

PAGE 216

202 113. Shelton, FN and MJ Reding, Effect of lesion location on upper limb motor recovery after stroke. Stroke, 2001. 32(1): p. 107-12. 114. Gowland, C, H deBruin, JV Basmajia n, N Plews, and I Burcea, Agonist and antagonist activity during vol untary upper-limb movement in patients with stroke. Phys Ther, 1992. 72(9): p. 624-33. 115. Mercier, C, AM Bertra nd, and D Bourbonnais, Differen ces in the magnitude and direction of forces during a submaximal ma tching task in hemiparetic subjects. Exp Brain Res, 2004. 157(1): p. 32-42. 116. Cirstea, MC and MF Levin, Compensatory strategies for reaching in stroke. Brain, 2000. 123: p. 940-53. 117. Taub, E, G Uswatte, and T Elbert, New treatments in neurorehabilitation founded on basic research. Nat Rev Ne urosci, 2002. 3(3): p. 228-36. 118. Fisher, BE and KJ Sullivan, Activitydependent factors affecting poststroke functional outcomes. Top Stroke Rehabil, 2001. 8(3): p. 31-44. 119. Taub, E and G Uswatte, Constraint-induced movement therapy: bridging from the primate laboratory to the stroke rehabilitation labora tory. J Rehabil Med, 2003(41 Suppl): p. 34-40. 120. Davies, PM, Steps to follow: a guide to the treatment of adu lt hemiplegia based on the concept of K. and B. Bobath 1984, New York: Springer-Verlag. 121. Mulder, T, W Zijlstra, and A Geurts, A ssessment of motor recovery and decline. Gait Posture, 2002. 16(2): p. 198-210. 122. Rohrer, B, S Fasoli, HI Krebs, R Hughe s, B Volpe, WR Frontera, J Stein, and N Hogan, Movement smoothness changes duri ng stroke recovery. J. Neurosci., 2002. 22(18): p. 8297-8304. 123. Ellis, MD, BG Holubar, AM Acosta, RF Beer, and JP Dewald, Modifiability of abnormal isometric elbow and shoulder join t torque coupling af ter stroke. Muscle Nerve, 2005. 32(2): p. 170-8. 124. Michaelsen, SM and MF Levin, Short-te rm effects of practic e with trunk restraint on reaching movements in patients with chr onic stroke: a controlled trial. Stroke, 2004. 35(8): p. 1914-19. 125. Thielman, GT, CM Dean, and AM Gentile, Rehabilitation of reaching after stroke: Task-related training versus pr ogressive resistive exercise ,. Archives of Physical Medicine and Rehabilitation, 2004. 85(10): p. 1613-1618. 126. Gentilucci, M, E Daprati, M Gangitano, MC Saetti, and I Toni, On orienting the hand to reach and grasp an object. Neuroreport, 1996. 7(2): p. 589-92.

PAGE 217

203 127. Zattara, M and S Bouisset, Posturo-kineti c organisation during the early phase of voluntary upper limb move ment. 1. Normal subjects. J Neurol Neurosurg Psychiatry, 1988. 51(7): p. 956-65. 128. Roth, E, A Heinemann, L Lovell, R Ha rvey, J McGuire, and S Diaz, Impairment and disability: Their relati on during stroke rehabilitati on. Archives of Physical Medicine and Rehabilitation, 1998. 79(3): p. 329-335. 129. Mahoney, FI and DW Barthel, Functional evaluation: the Barthe l Index. Md State Med J, 1965. 14: p. 61-5. 130. Winstein, CJ, DK Rose, SM Tan, R Lewthwaite, HC Chui, and SP Azen, A randomized controlled comparison of upper-ext remity rehabilitation strategies in acute stroke: A pilot study of immediate and long-term outcomes. Arch Phys Med Rehabil, 2004. 85(4): p. 620-8. 131. Shelton, FD, BT Volpe, and M Reding, Motor impairment as a predictor of functional recovery and guide to rehabil itation treatment after stroke. Neurorehabil Neural Repair, 2001. 15(3): p. 229-37. 132. Mercier, C and D Bourbonnais, Relative s houlder flexor and handgrip strength is related to upper limb func tion after stroke. Clin Reha bil, 2004. 18(2): p. 215-21. 133. Lai, SM, PW Duncan, and J Keighley Prediction of func tional outcome after stroke: comparison of the Orpington Prognos tic Scale and the NIH Stroke Scale. Stroke, 1998. 29(9): p. 1838-42. 134. Wilson, DJ, LL Baker, and JA Craddock, Functional test for the hemiparetic upper extremity. Am J Occup Ther, 1984. 38(3): p. 159-64. 135. Feys, HM, WJ De Weerdt, BE Selz, GA Cox Steck, R Spichiger, LE Vereeck, KD Putman, and GA Van Hoydonck, Effect of a therapeutic intervention for the hemiplegic upper limb in the acute phase after stroke: a single-blind, randomized, controlled multicenter trial. St roke, 1998. 29(4): p. 785-92. 136. Wade, D, 'But is the diffe rence clinically significant?' Clin Rehabil, 2005. 19(4): p. 349-53. 137. Testa, MA, Interpretation of quality-of-lif e outcomes: issues that affect magnitude and meaning. Med Care, 2000. 38(9 Suppl): p. II166-74. 138. Terwee, CB, FW Dekker, WM Wiersi nga, MF Prummel, and PM Bossuyt, On assessing responsiveness of health-related qua lity of life instruments: guidelines for instrument evaluation. Qual Li fe Res, 2003. 12(4): p. 349-62. 139. McCombe Waller, S and J Whitall, Hand dominance and side of stroke affect rehabilitation in chronic stroke. Cl in Rehabil, 2005. 19(5): p. 544-51.

PAGE 218

204 140. Duff, MJ, LB Okun, and G Veneziano, Trialogue on the number of fundamental constants. Journal of High Energy Physics, 2002. 3(23). 141. Wright, B, Definition of objective measurement, http://www.rasch.org/define.htm Accessed May 27, 2005. 142. Bond, TG and CM Fox, Applying the Rasch Model: Fundamental Measurement in the Human Sciences. 2001, Mahwah, N.J.: Erlbaum. 143. Velozo, CA, G Kielhofner, and JS Lai, Th e use of Rasch analysis to produce scalefree measurement of functional ability. Am J Occup Ther, 1999. 53(1): p. 83-90. 144. Crocker, LM and J Algina, Introduction to Classical and Modern Test Theory. 1986, New York: Holt Rine hart and Winston. 145. White, LJ and CA Velozo, The use of Ra sch measurement to improve the Oswestry classification scheme. Arch Phys Me d Rehabil, 2002. 83(6): p. 822-31. 146. Arnould, C, M Penta, A Renders, an d JL Thonnard, ABILHAND-Kids: a measure of manual ability in childr en with cerebral palsy. Neurology, 2004. 63(6): p. 104552. 147. Sheehan, TJ, S DuBrava, J Fifield, S Reis ine, and L DeChello, Rate of change in functional limitations for patients with rheuma toid arthritis: effects of sex, age, and duration of illness. J Rheuma tol, 2004. 31(7): p. 1286-92. 148. Coster, WJ, SM Haley, PL Andres, LH Ludlow, TL Bond, and PS Ni, Refining the conceptual basis for rehabilitation ou tcome measurement: personal care and instrumental activities domain. Med Care, 2004. 42(1 Suppl): p. I62-72. 149. Avery, LM, DJ Russell, PS Raina, SD Walter, and PL Rosenbaum, Rasch analysis of the Gross Motor Function Measure: va lidating the assumptions of the Rasch model to create an interval -level measure. Arch Phys Med Rehabil, 2003. 84(5): p. 697-705. 150. Wright, BD, JM Linacre, RM Smith, AW Heinemann, and CV Granger, FIM measurement properties and Rasch mode l details. Scand J Rehabil Med, 1997. 29(4): p. 267-72. 151. Velozo, CA, LC Magalhaes, AW Pan, and P Leiter, Functional scale discrimination at admission and discharge: Rasch analysis of the Level of Reha bilitation Scale-III. Arch Phys Med Rehabil, 1995. 76(8): p. 705-12. 152. Fisher, AG, The assessment of IADL moto r skills: an applicat ion of many-faceted Rasch analysis. Am J Occup Ther, 1993. 47(4): p. 319-29.

PAGE 219

205 153. Smith, EV, Jr., Detecting and evaluati ng the impact of multidimensionality using item fit statistics and principal component analysis of residuals. J Appl Meas, 2002. 3(2): p. 205-31. 154. Linacre, JM, Detecting multidimensionality : which residual data-type works best? J Outcome Meas, 1998. 2(3): p. 266-83. 155. Hattie, J, Methodology review: assessing unidimensionality of tests and items. Applied Psychological Measurem ent, 1985. 9(2): p. 139-164. 156. Smith, EV, Jr. and BD Johnson, Attenti on Deficit Hyperactivit y Disorder: scaling and standard setting using Rasch measurem ent. J Appl Meas, 2000. 1(1): p. 3-24. 157. Lattin, J, JD Carroll, and P Green, An alyzing Multivariate Date. 2003, Pacific Grove, CA: Thomson Brooks-Cole. 158. Hong, S, SK Mitchell, and RA Harshm an, Bootstrap scree te sts: A Monte Carlo simulation and applications to published da ta. Br J Math Stat Psychol, 2006. 59(Pt 1): p. 35-57. 159. McHorney, CA and AS Cohen, Equating he alth status measures with item response theory: illustrations with f unctional status items. Med Care, 2000. 38(9 Suppl): p. II43-59. 160. Cook, KF, PO Monahan, and CA McHorn ey, Delicate balance between theory and practice: health status assessment and ite m response theory. Med Care, 2003. 41(5): p. 571-4. 161. Finlayson, M, T Mallinson, and VM Ba rbosa, Activities of daily living (ADL) and instrumental activities of daily living (I ADL) items were stable over time in a longitudinal study on aging. J Clin Epidemiol, 2005. 58(4): p. 338-49. 162. Haley, SM, WJ Coster, PL Andres, LH Ludlow, P Ni, TL Bond, SJ Sinclair, and AM Jette, Activity outcome measurement for postacute care. Med Care, 2004. 42(1 Suppl): p. I49-61. 163. Humphreys, LG and RG Montanelli, An investigation of the parallel analysis criterion for determining the number of co mmon factors. Multivariate Behav. Res., 1975. 10: p. 193. 164. Norman, GR and DL Streiner, Biostatistics: the Bare Essentials. 2000, Hamilton, Ont.: B.C. Deker Inc. 165. Linacre, JM, A Users Guide to Winste ps and Ministeps Ra sch-Model Computer Programs. 2006, Chicago: Winsteps.com. 166. Wright, B and GN Master s, Rating Scale Analysis. 1982, Chicago, IL: MESA press.

PAGE 220

206 167. Muraki, E, CM Hombo, and YW Le e, Equating and linking of performance assessments. Appl Psych Meas, 2000. 24(4): p. 325-337. 168. Hart, T, J Whyte, S Millis, RK Bode J Malec, RN Richardson, and F Hammond, Dimensions of Disordered Attention in Traumatic Brain Injury: Further Validation of the Moss Attention Rating Scale. Arch Phys Med Rehabil, 2006. 87(5): p. 647655. 169. Wright, BD and JM Linacre, Reasonable mean-squa re fit values, http://www.rasch.org/rmt/rmt83b.htm Accessed May 15, 2006. 170. Ryall, NH, SB Eyres, VC Neumann, BB Bhakta, and A Tennant, Is the Rivermead Mobility Index appropriate to measure mobility in lower limb amputees? Disabil Rehabil, 2003. 25(3): p. 143-53. 171. Linacre, JM, Test valid ity and Rasch measurement: construct, content etc. Rasch Measurement Transactions, 2004. 18(1): p. 970-971. 172. Wright, BD and MH Stone Best Test Design. 1979, Chicago: Mesa Press. 173. Kornetti, DL, SL Fritz, YP Chiu, KE Li ght, and CA Velozo, Rating scale analysis of the Berg Balance Scale. Arch Phys Med Rehabil, 2004. 85(7): p. 1128-35. 174. Russell, DJ, LM Avery, PL Rosenbaum, PS Raina, SD Walter, and RJ Palisano, Improved scaling of the gross motor functi on measure for children with cerebral palsy: evidence of reliabili ty and validity. Phys Ther, 2000. 80(9): p. 873-85. 175. Haley, SM, LH Ludlow, and WJ Cost er, Pediatric Evalua tion of Disability Inventory: clinical interp retation of summary scores using Rasch rating scale methodology. Phys Med Rehabil Clin N Am, 1993. 4: p. 529-40. 176. Penta, M, JL Thonnard, and L Tesi o, ABILHAND: a Rasc h-built measure of manual ability. Arch Phys Med Re habil, 1998. 79(9): p. 1038-42. 177. Sabari, JS, AL Lim, CA Velozo, L Le hman, O Kieran, and JS Lai, Assessing arm and hand function after stroke: a validity test of the hi erarchical scoring system used in the motor assessment scale for stroke. Arch Phys Med Rehabil, 2005. 86(8): p. 1609-15. 178. Carr, JH, RB Shepherd, L Nordholm, a nd D Lynne, Investigation of a new motor assessment scale for stroke patients Phys Ther, 1985. 65(2): p. 175-80. 179. Chiu, YP, SL Fritz, KE Light, and CA Velozo, Use of item response analysis to investigate measurement properties and clin ical validity of data for the dynamic gait index. Phys Ther, 2006. 86(6): p. 778-87. 180. Gluck, J and A Indurkhya, Assessing cha nges in the longitudina l salience of items within constructs. J Adolescent Res, 2001. 16(2): p. 169-187.

PAGE 221

207 181. Bingenheimer, JB, SW Raudenbush, T Leventhal, and J Brooks-Gunn, Measurement equivalence and differential item functioning in family psychology. J Fam Psychol, 2005. 19(3): p. 441-55. 182. Tesio, L, Measuring be haviours and perceptions: Rasc h analysis as a tool for rehabilitation research. Journal of Re habilitation Medicine 2003. 35(3): p. 105115. 183. Lai, JS, J Teresi, and R Gershon, Procedur es for the analysis of differential item functioning (DIF) for small sample sizes. Ev al Health Prof, 2005. 28(3): p. 283-94. 184. Teresi, JA, M Kleinman, and K Ocepek -Welikson, Modern psychometric methods for detection of differential item functi oning: application to cognitive assessment measures. Stat Med, 2000. 19(11-12): p. 1651-83. 185. Azocar, F, P Arean, J Miranda, and RF Munoz, Differential item functioning in a Spanish translation of the Beck Depressi on Inventory. J Clin Psychol, 2001. 57(3): p. 355-65. 186. Tennant, A, M Penta, L Tesio, G Grimby, JL Thonna rd, A Slade, G Lawton, A Simone, J Carter, A Lundgren-Nilsson, M Tr ipolski, H Ring, F Biering-Sorensen, C Marincek, H Burger, and S Phillips, Assessing and adjusting for cross-cultural validity of impairment a nd activity limitation scales through differential item functioning within the fram ework of the Rasch mode l: the PRO-ESOR project. Med Care, 2004. 42(1 Suppl): p. I37-48. 187. Roorda, LD, ME Roebroeck, T van Tilburg, GJ Lankhorst, and LM Bouter, Measuring activity limitations in climbing stairs: development of a hierarchical scale for patients with lower-extremity disorders living at ho me. Arch Phys Med Rehabil, 2004. 85(6): p. 967-71. 188. Dallmeijer, AJ, J Dekker, LD Roorda, DL Knol, B van Baalen, V de Groot, VP Schepers, and GJ Lankhorst, Differential item functioning of the Functional Independence Measure in higher performing neurological patients. J Rehabil Med, 2005. 37(6): p. 346-52. 189. Roznowski, M and J Reith, Examining the Measurement Quality of Tests Containing Differentially F unctioning Items: do biased items result in poor measurement? Educational and Psychol ogical Measurement, 1999. 59(2): p. 248269. 190. Linacre, JM, AW Heinemann, BD Wr ight, CV Granger, and BB Hamilton, The structure and stability of the Functional Independence Measure. Arch Phys Med Rehabil, 1994. 75(2): p. 127-32. 191. Fisher, WP, Jr., Physical disability construct convergence across instruments: towards a universal metric. J Outc ome Meas, 1997. 1(2): p. 87-113.

PAGE 222

208 192. Fisher, WP, RF Harvey, P Taylor, KM Kilgore, and CK Kelly, Rehabits a common language of functional assessment. Arch Phys Med Rehabil, 1995. 76(2): p. 113-122. 193. Wolfe, EW, Equating and item banking w ith the Rasch model. J Appl Meas, 2000. 1(4): p. 409-434. 194. Crane, PK, G van Belle, and EB Larson, Te st bias in a cognitive test: differential item functioning in the CASI. Stat Med, 2004. 23(2): p. 241-56. 195. Hanson, BA and AA Beguin, Obtaining a common scale for item response theory item parameters using separate versus c oncurrent estimation in the common-item equating design. Appl Psych Meas, 2002. 26(1): p. 3-24. 196. Kim, SH and AS Cohen, A comparison of linking and concu rrent calibration under item response theory. Appl Psyc h Meas, 1998. 22(2): p. 131-143. 197. Fisher, WP, Jr., RL Eubanks, and RL Marier, Equating the MOS SF36 and the LSU HSI Physical Functioning Scales. J Outcome Meas, 1997. 1(4): p. 329-62. 198. Segal, ME, AW Heinemann, RR Schall, a nd BD Wright, Rasch an alysis of a brief physical ability scale for long-term outcom es of stroke. Physical Medicine and Rehabilitation: State of the Ar t Review, 1997. 11(2): p. 385-96. 199. Latham, N and SM Haley, Measuring functional outcomes acr oss postacute care: current challenges and future directio n. Critical Reviews in Physical and Rehabilitation Medicine, 2003. 15(2): p. 83-98. 200. Wolf, SL and SA Binder-MacLeod, Electro myographic biofeedback applications to the hemiplegic patient. Changes in uppe r extremity neuromuscular and functional status. Phys Ther, 1983. 63(9): p. 1393-403. 201. Morris, DM, G Uswatte, JE Crago, EW Cook, 3rd, and E Taub, The reliability of the wolf motor function test for assess ing upper extremity function after stroke. Arch Phys Med Rehabil, 2001. 82(6): p. 750-5. 202. Linacre, JM, Sample size and item calibration stability. Rasch Measurement Transactions, 1994. 7(4): p. 328. 203. Hambleton, RK, Educational Measurement, in Principles and Selected Applications in Item Response Theory: 3rd Ed., R.L. Lynn, Editor. 1989, New York:American Council on Education/McMillan Publishing. 204. Embretson, S and SP Reise, Item Re sponse Theory for Psychologists. 2000, Mahwah, New Jersey: Lawrence Erlbaum Associates.

PAGE 223

209 205. Reeve, BB and P Fayers, Applying item response theory modeling for evaluating questionnaire item and scale properties, in Assessing Quality of Life in Clinical Trials Methods and Practice: 2nd Ed., P. Fayers and R.D. Hays, Editors. 2005, p. 55-73. 206. Lai, JS, D Cella, CH Chang, RK Bo de, and AW Heinemann, Item banking to improve, shorten and computer ize self-reported fatigue: an illustration of steps to create a core item bank from the FACIT-Fati gue Scale. Qual Life Res, 2003. 12(5): p. 485-501. 207. Holman, R, CA Glas, and RJ de Haan, Powe r analysis in randomized clinical trials based on item response theory. Control Clin Trials, 2003. 24(4): p. 390-410. 208. Wang, WC and CT Chen, Item parameter r ecovery, standard error estimates, and fit statistics of the Winsteps program for th e family of Rasch models. Educational and Psychological Measurement, 2005. 65(3): p. 376-404. 209. McHorney, CA and PO Monahan, Postscri pt: Applications of Rasch analysis in health care. Med Care, 2004. 42(1 Suppl): p. I73-8. 210. .Masters, GN, Undesirable item discri mination. Rasch Measurement Transactions, 1993. 7(2): p. 289. 211. Andrich, D, Controversy and the Rasch model: a characteristic of incompatible paradigms? Med Care, 2004. 42(1 Suppl): p. I7-16. 212. Reise, SP and MG Haviland, Item respons e theory and the measurement of clinical change. J Pers Assess, 2005. 84(3): p. 228-38. 213. Kirisci, L, RE Tarter, M Vanyukov, C Martin, A Mezzich, and S Brown, Application of item response theory to qua ntify substance use disorder severity. Addictive Behaviors, 2006. 31(6): p. 1035-1049. 214. Wright, BD and JM Linacre, Observa tions are always ordinal; measurements, however, must be interval. Arch Phys Med Rehabil, 1989. 70(12): p. 857-60. 215. Tate, DG, The state of rehabilitation research: art or science? Arch Phys Med Rehabil, 2006. 87(2): p. 160-6. 216. Murray, IA and GR Johnson, A study of the external forces and moments at the shoulder and elbow while performing ever y day tasks. Clin Biomech (Bristol, Avon), 2004. 19(6): p. 586-94. 217. Buchanan, TS, DP Almdale, JL Lewis, a nd WZ Rymer, Characteristics of synergic relations during isometric c ontractions of human elbow muscles. J Neurophysiol, 1986. 56(5): p. 1225-1241.

PAGE 224

210 218. Beer, RF, JP Dewald, and WZ Rymer, Deficits in the coordination of multijoint arm movements in patients with hemiparesi s: evidence for disturbed control of limb dynamics. Exp Brain Res, 2000. 131(3): p. 305-19. 219. Floyd, FJ and KF Widaman, Factor analysis in the de velopment and refinement of clinical assessment instruments. Psyc hol Assessment, 1995. 7(3): p. 286-299. 220. Duncan, PW, S Studenski, LG Richards, S Gollub, S-M Lai, D Reker, S Perera, J Yates, V Koch, S Rigler, and D Johnson, Ra ndomized clinical tr ial of therapeutic exercise in subacute stroke. Stroke, 2003. 34(9): p. 2173-2180. 221. Patel, AT, PW Duncan, SM Lai, and S Studenski, The relation between impairments and functional outcomes poststroke. Arch Phys Med Rehabil, 2000. 81(10): p. 1357-63. 222. Luft, AR, S Waller, L Forrester, GV Smith, J Whitall, RF Macko, JB Schulz, and DF Hanley, Lesion location alters brain activation in chronically impaired stroke survivors. Neuroimage, 2004. 21(3): p. 924-35. 223. Wright, BD and GN Masters, Rating scale analysis. 1982, Chicago: Mesa Press. xi, 206. 224. Gordon, J, ed. Assumptions underlying phys ical therapy intervention: theoretical and historical perspectives 2nd ed. Movement science: foundations for physical therapy in rehabilitation, ed J.H. Carr and R.B. Shepherd. 2000, Aspen Publishers: Gaithersburg, Md. xiii, 220. 225. Colebatch, JG, JC Rothwell, BL Day, PD Thompson, and CD Marsden, Cortical outflow to proximal arm muscles in man. Brain, 1990. 113 ( Pt 6): p. 1843-56. 226. Gordon, C, Assumptions underlying physic al therapy interventions, theoretical and historical perspectives, in Movement scie nce: Foundations for Physical Therapy in Rehabilitation, 2nd ed, J.H. Carr and R.B. Shepherd, Editors. 2000, Gaithersburg, Md:Aspen Publishers. 227. Warren, M, Strategies for sensory and neuromotor remediation, in Occupational Therapy: Overcoming Human Performance De ficits, C. Christiansen and C. Baum, Editors. 1991, Thorofare NJ:Slack: p. 633-664. 228. Pedretti, LW, Occupational Therapy: Practice Skills for Physical Dysfunction. 4th ed. 1996, St. Louis: Mosby. xv, 876. 229. Kamper, DG, RL Harvey, S Suresh, and WZ Rymer, Relative contributions of neural mechanisms versus muscle mechanic s in promoting finger extension deficits following stroke. Muscle Nerve, 2003. 28(3): p. 309-18.

PAGE 225

211 230. Flora, DB and PJ Curran, An empirical evaluation of alternative methods of estimation for confirmatory factor analys is with ordinal data. Psychol Methods, 2004. 9(4): p. 466-91. 231. Chang, WC and C Chan, Rasch an alysis for outcomes measures: some methodological considerations. Arch Phys Med Rehabil, 1995. 76(10): p. 934-9. 232. Duncan, P, LB Goldstein, RD Horner, PB Landsman, GP Samsa, and DB Matchar, Similar motor recovery of upper and lowe r extremities after stroke. Stroke, 1994. 25(6): p. 1181-8. 233. Feys, H, J Hetebrij, G Wilms, R Dom, and W De Weerdt, Predicting arm recovery following stroke: value of site of lesion. Acta Neurol Scand, 2000. 102(6): p. 3717. 234. Linacre, JM, WINSTEPS. 2004, Winsteps.com: Chicago. 235. Smith, RM, Polytomous M ean-Square Fit Statistics, http://www.rasch.org/rmt/rmt103a.htm Accessed October 10, 2005. 236. Portney, L and M Watkins, Foundations of Clinical Research, 2nd edition. 2000, Upper Saddle River, NJ.: Prentice Hall Health. 237. Kothari, DH, SM Haley, KM Gill-Bo dy, and HM Dumas, Measuring functional change in children with acquired brain in jury (ABI): compar ison of generic and ABI-specific scales using the Pediatric Ev aluation of Disability Inventory (PEDI). Phys Ther, 2003. 83(9): p. 776-85. 238. Crane, PK, DL Hart, LE Gibbons, and KF Cook, A 37-item shoulder functional status item pool had neglig ible differential item functio ning. Journal of Clinical Epidemiology, 2006. 59(5): p. 478-484. 239. Wright, BD, Time 1 to Time 2 Comp arison. Rasch Measurement Transactions, 1996. 10(1): p. 478-9. 240. Smith, RM, A comparison of methods fo r determining dimensionality in Rasch measurement. Structural Equati on Modeling, 1996. 3: p. 25-40. 241. Wallace, D, PW Duncan, and SM Lai, Comparison of the responsiveness of the Barthel Index and the motor component of the Functional Independence Measure in stroke: the impact of using different methods for measuring responsiveness. J Clin Epidemiol, 2002. 55(9): p. 922-8. 242. Rabadi, MH and FM Rabadi, Comparison of the Action Research Arm Test and the Fugl-Meyer Assessment as measures of upper-extremity motor weakness after stroke. Archives of Physi cal Medicine and Rehabilita tion, 2006. 87(7): p. 962-66.

PAGE 226

212 243. Hands, B and D Larkin, Using the Rasc h measurement model to investigate the construct of motor ability in young childre n. J Appl Meas, 2001. 2(2): p. 101-20. 244. Hands, B, B Sheridan, and D Larkin Creating performance categories from continuous motor skill data using a Rasc h measurement model. J Outcome Meas, 1999. 3(3): p. 216-32. 245. Reeves, TC and EW Ford, Strategic management and performance differences: nonprofit versus for-profit health organiza tions. Health Care Manage Rev, 2004. 29(4): p. 298-308. 246. Williams, GP, V Robertson, KM Green wood, PA Goldie, and ME Morris, The high-level mobility assessment tool (HiMAT) for traumatic brain injury. Part 2: content validity and discriminability Brain Inj, 2005. 19(10): p. 833-43. 247. Linacre, JM, Understanding Rasch m easurement: optimizing rating scale category effectiveness. Journal of Applied Measurement, 2002. 3(1): p. 85-106. 248. McHorney, CA, Use of item response theo ry to link 3 modules of functional status items from the Asset and Health Dynami cs Among the Oldest Old study. Arch Phys Med Rehabil, 2002. 83(3): p. 383-94. 249. Bland, JM and DG Altman, Cronbach's alpha. British Medical Journal, 1997. 314(7080): p. 572. 250. Nunnally, JC, Psychometric Theory. 2nd ed. 1978, New York: McGraw-Hill. 251. Jette, AM, SM Haley, and P Ni, Comparis on of functional status tools used in postacute care. Health Care Fina nc Rev, 2003. 24(3): p. 13-24. 252. Lai, S-M, PW Duncan, and J Keighley Prediction of Func tional Outcome After Stroke : Comparison of the Orpington Pr ognostic Scale and the NIH Stroke Scale. Stroke, 1998. 29(9): p. 1838-1842. 253. Stone, MH, Substantive scale constructi on. J Appl Meas, 2003. 4(3): p. 282-297. 254. Smith, EV, Jr., Evidence fo r the reliability of measur es and validity of measure interpretation: a Rasch measurement pers pective. J Appl Meas, 2001. 2(3): p. 281311. 255. Costello, AB and JW Osborne, Best practi ces in exploratory f actor analysis: four recommendations for getting the most from your analysis, http://pareonline.net/pdf/v10n7.pdf Accessed May 30, 2006. 256. Graham, KM, KD Moore, DW Cabel, PL Gribble, P Cisek, and SH Scott, Kinematics and kinetics of multijoin t reaching in nonhuman primates. J Neurophysiol, 2003. 89(5): p. 2667-77.

PAGE 227

213 257. Galloway, JC and GF Koshland, General coordination of shoulder, elbow and wrist dynamics during multijoint arm movements. Exp Brain Res, 2002. 142(2): p. 16380. 258. Fitts, PM, The information capacity of the human motor system in controlling the amplitude of movement. 1954. J Exp Psychol Gen, 1992. 121(3): p. 262-9. 259. Linacre, JM, What is item respons e theory? A tentative taxonomy. Rasch Measurement Transactions, 2004. 17(2): p. 926-927. 260. May, K and TS Jackson, IRT item parame ters and the reliabil ity and validity of pretest, posttest, and gain scores. Internat ional Journal of Testing, 2005. 5(1): p. 6373. 261. Gray-Little, B, VSL Williams, and TD Hancock, An item response theory analysis of the Rosenberg self-esteem scale. Pers Soc Psychol B, 1997. 23(5): p. 443-451. 262. Traub, RE, A priori considerations in choosing an item response model, in Applications of Item Response Theory, R.K. Hambleton, Editor. 1983, Vancouver, Canada:Educational and Research Institute of British Columbia. 263. Teresi, JA, RR Golden, P Cross, B Gurla nd, M Kleinman, and D Wilder, Item bias in cognitive screening measur es: comparisons of elderl y white, Afro-American, Hispanic and high and low education subgr oups. J Clin Epidemiol, 1995. 48(4): p. 473-83. 264. Penfield, RD, The impact of model misfit on partial credit model parameter estimates. J Appl Meas, 2004. 5(2): p. 115-28. 265. Thissen, D, Multilog for Windows. 2003, Scientific Software International: Lincolnwood, IL. 266. Samejima, F, Graded response model, in Handbook of modern item response theory, W. van der Linden and R. Hambleton, Editors. 1997, New York, NY:Springer: p. 85. 267. Marino, RJ and DE Graves Metric properties of the AS IA motor score: subscales improve correlation with functional activities Archives of Physical Medicine and Rehabilitation, 2004. 85(11): p. 1804-1810. 268. Krueger, RF and MS Finger, Using item response theory to understand comorbidity among anxiety and unipolar mood disorders. Psychol Assess, 2001. 13(1): p. 14051. 269. Masters, GN, Item discrimination: When more is worse. J Educ Meas, 1988. 25(1): p. 15-29.

PAGE 228

214 BIOGRAPHICAL SKETCH Michelle Woodbury graduated from Fre donia High school, Fredonia, NY, in 1982. From 1982 to 1986 she attended Elizabethtown College, Elizabethtown, PA, graduating with a Bachelor of Science degree in occupational therapy. She worked as an occupational therapist in both inpatient and outpatient neuror ehabilitation settings in New York, Florida and South Carolina full tim e from 1986 to 1998. From 1993 to 1995 she attended Columbia International University, Columbia, SC, graduating with a Masters of Arts degree in intercultural studies. She join ed the faculty of the Medical University of South Carolina in Charleston, SC, in 1998 and taught neurorehabilitation occupational therapy courses until 2002. From 2002 to 2006 she attended the Univer sity of Florida, Gainesville, FL, graduating with a PhD in re habilitation science a nd a concentration in movement science.


Permanent Link: http://ufdc.ufl.edu/UFE0015615/00001

Material Information

Title: Measuring Post-Stroke Arm Motor Ability: Measurement Properties of the Fugl-Meyer Upper Extremity Assessment Examined with an Item Response Theory Measurement Framework
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0015615:00001

Permanent Link: http://ufdc.ufl.edu/UFE0015615/00001

Material Information

Title: Measuring Post-Stroke Arm Motor Ability: Measurement Properties of the Fugl-Meyer Upper Extremity Assessment Examined with an Item Response Theory Measurement Framework
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0015615:00001


This item has the following downloads:


Full Text











MEASURING POST-STROKE ARM MOTOR ABILITY: MEASUREMENT
PROPERTIES OF THE FUGL-MEYER UPPER EXTREMITY ASSESSMENT
EXAMINED WITH AN ITEM RESPONSE THEORY MEASUREMENT
FRAMEWORK














By

MICHELLE LYNNE WOODBURY


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


2006




























Copyright 2006

by

Michelle Lynne Woodbury




































To Bernard Earl Watrous and Leah Michelle Smith
















ACKNOWLEDGMENTS

I must first thank my dissertation co-chairs for their help and guidance throughout

the process. Lorie Richards, a fellow Elizabethtown College alumnus, has been a steady

source of encouragement. When I have needed her, she is there with enthusiasm, empathy

and ideas. Craig Velozo is a model mentor. His integrity, clarity of thought, humility and

witty demeanor make it possible for him to expect, and get, excellence from his research

team. I also must thank the other members of my committee, Dena Howland and Kathye

Light. Dena, an occupational therapist turned neuroscientist, somehow is able to balance

her family life with a highly demanding research career. I am grateful that in the midst of

chaos she always had time for me, gave me her full attention, and was able to clearly

provide guidance and encouragement. Kathye is a highly intelligent wonderful teacher

who has shown me how to mix academic content with a healthy dose of "real world"

application and clinical intuition.

I am deeply grateful to Pam Duncan. This dissertation idea began with her; she

graciously allowed access to her databases, and she has been dangerously enthusiastic

about the proj ect from the beginning!

My family has been a constant source of support. Many thanks go to my sister and

brother-in-law, Melissa and Christopher Smith; my very cute niece, Leah Michelle

Smith; my mother and step-father, Bonnie and Red Beuglass; my dad and step-mother,

Gary and Robin Woodbury; my crazy cousin Rosemary Watrous; my wonderful "neena"

Ruth Watrous; and my clever and funny grandfather Bernard Watrous whom I miss.









I must also thank my "RSD family," the fellow students that are the source of 24/7

support and wisdom. Many thanks go to the Velozo research team (Leigh, Inga, Jia-Hwa,

Pey-Shan) for their assistance with Rasch and 2PL analyses. Thanks go also to Arlene,

Dennis (Steve), and Mike for "forcing" me to choose UF. I am grateful for the OT RSD

students who make me laugh: Megan, Jessica, Roxanna, Rick, Bhagwant, Eric, Kezia,

Sandy W., Sandy P., Patricia, Christine, and Mindy.

It is easy to loose perspective in the academic rehabilitation research context. Three

groups have helped me remember the truly important big picture. Prior to coming to

PhD-school I was privileged to "sit at the feet" of amazing teachers: Drs. Mullen,

Hulbert, Priest, Larkin, and Davidson (Columbia International University). They modeled

academic and personal excellence, becoming "real life" illustrations of wisdom and

knowledge. Waleed-Al Oboudi is an OTR with stunning talent whom I have been

privileged to call mentor and friend. His passionate desire to advance clinically relevant

stroke rehabilitation by training therapists to pursue excellence has been the goal to

which I have aimed both academically and clinically. Finally, I have very clear memories

of two former patients, Luke Harrell (14 years old) and Caitlin Hill (9 years old). I see

them smile and laugh despite devastating brain damage. Their memories remind me, with

crystal clarity, that improving outcomes for patients like Luke and Caitlin (and their

families) is unquestionably the only important reason for rehabilitation research.

This dissertation work was partially supported an NIH T-32 Neuromuscular

Plasticity Training Grant Fellowship (T32HDO43730). The Kansas City Stroke Study

was funded by the National Institute on Aging, Claude D. Pepper Center Older

Americans Center (5P60AGl463 5).



















TABLE OF CONTENTS


page

ACKNOWLEDGMENT S .............. .................... iv


LI ST OF T ABLE S ............_...... .__ ...............x....


LI ST OF FIGURE S .............. .................... xii


AB S TRAC T ......_ ................. ..........._..._ xiii..


CHAPTER


1 INTRODUCTION AND LITERATURE REVIEW ................. .........................1


The International Classification of Function and Disability ................. ................ ..5
M otor Control Theory................ .............. .. .........7
Traditional Theories of Motor Control and Recovery ................. ............... .....7
Challenges to the Traditional Theories of Motor Control and Recovery ............10
Role of reflexes ................. ... ........ ... ........... ............1

The pattern of upper extremity behaviors evidencing recovery ................... 12
Contemporary Theories of Motor Control and Recovery ................. ................18
UJE motor impairment. ...._ ......___._ ........... ............1
Post-stroke UE recovery ..............._ ...... ....._ ......... ...... ......... 2
Post-stroke recovery of functional UE motor abilities .............. ..............24
Clinical significance of the FMA-UE score ......____ ..... ... .._............27
Summary .................... ...............29.
The Rasch Measurement Model .............. ...............29....
M easurement .............. ...............29....
Item Response Theory .......................... ..............._ ........._ .........3
Potential Advantages for Using the Rasch Model to Explore the FMA-UE
Measurement Properties. .....__.....___ ..........__ ............31
Validity of the item content.. ....._ .....___ ........_ ...............3 1
Item difficulty hierarchy............... ...............3
Reproducibility of the item structure .............. ...............38....
Functional interpretation of the FMA-UE score ............_.. ......... .....43
Criticisms of the Rash Measurement Model ............ .....__ ..............47

Sam ple size............... ..... .. ..............4
Should data fit the model or visa versa? ............ ...............48.....
Conclusion ............ _...... __ .............50....












2 USING THE RASCH MEASUREMENT MODEL TO EXAMINE POST-
S TROKE UPPER EXTREMITY REC OVERY ......____ ........._ ...............54


M ethods ................. ...............57...
Dimensionality .............. ...............57....
Construct Validity .............. ...............60....
Re sults ......__................ .........__..........6
Participants .............. ...............61....
Fit Statistics ............... .... ...............62
Principal Components Analysis .............. ...............62....
Item Hierarchy ......__................. .........__..........6
K eyform s .............. ...............65....
Discussion ................. ...............67.................
Dimensionality .............. ...............68....
Item Hierarchy ................. ...............69.................
Limitations ................. ...............73.................
Conclusion ................ ...............74.................


3 LONGITUIDNAL STABILITY OF THE 30-ITEM FUGL-MEYER UPPER
EXTREMITY ASSES SMENT ITEM DIFFICULTY HIERARCHY ................... ....82


M ethods .............. ...............84....
Participants .............. ... ........... .. ...........8
The 30-item Fugl-Meyer Upper Extremity Assessment .............. ...................85
D ata Analysis..................... .............8
Step 1: Dimensionality ................. ........... ...............86. ....
Step 2: Equivalence of item difficulties ................ ................ ........ .87
Step 3: Differential item functioning .............. .. ...............87...
Step 4: Impact on assessment of arm motor function .............. ..................88
Re sults ................ ...............89.................
Participants .............. ........ ...............89
Step 1: Dimensionality ................. ........... ...............89. ....
Step 2: Equivalence of item difficulties ................ ................ ........ .89
Step 3: Differential item functioning T1 vs. T2 ................ ............... .....90
Step 4: Impact of DIF on person ability measurement............... ...............9
Discussion ................. ...............92.................

Study Limitations .............. ...............98....
Conclusion............... ...............9


4 LINKING THE FUGL-MEYER TO THE WOLF MOTOR FUNCTION TEST:
ASSESSING THE CONTINUM OF IMPAIRMENT TO FUNCTION ...............105


M ethods ................. ...............110......... ......
Participants ................. ...............110......... ......
Instrum ents ................. ...............110......... ......
Data Analysis ................. ...............112................
Step 1: Dimensionality ................. ...............112......... .....












Principal Components Analysis (PCA) ................. .......... ...............113
Rasch analysis ................. ...............114......... ......
Item correlations ................. ...............115................
Step 2: Longitudinal Stability.................. .......... ........ ............. 1
Step 3: Linking and Concurrent Calibration of the FMA-UE + WMFT Item
Pool ........._._... ..... .. .._._.. ......._._ .. ..........11

Step 4: Dimensionality of the Common Item Pool ........._._..... ...._._.........1 17
Step 5: Longitudinal Stability of the Common Item Pool ........._._... ...............117
Step 6: Interpretation of Change Scores ....._.__._ ..... ... .__. ......._........1 18
Functional threshold ........._._..............._... ... ...............118..
Functional interpretation of a FMA-UE change score............. ... ............1 19
Re sults........._..... ...__. ...............119...
Participants .........._.... .. ..... .... ...._._....... ...... ........... 1
Conversion of WMFT Performance Time Data to Performance Categories ....1 19
Step 1: Dimensionality of the WMFT ................. ...............123........... ..
Step 2: Longitudinal stability of the WMFT .......................... .. ..................2
Step 3: Linking and Concurrent Calibration of the FMA-UE + WMFT..........125
Step 4: Dimensionality of the FMA-UE + WMFT Item Pool ................... ........125
Step 5: Longitudinal Stability of the FMA-UE + WMFT Item Pool ...............128
Step 6: Interpretation of Change Scores ................. ............... ......... ...129
Functional threshold.........................................3
Functional interpretation of a FMA-UE change score .............. ..............132
Discussion ................... ........ ...._ ..... ........ .. ..... ........3
Linking as a Method to Explore "Significant" Change ................... .................1 34
Linking as a Method to Measure Body Function and Activity Performance....135
Linking as a Way to Conceptualize the Motor Rehabilitation Treatment
P process ....................... ..... ........ .... ............ 3
Limitations of this Study: Conversion of Performance Time data to Ordinal
Data .................. ......... .._ .. ......__ .. ...........13
Limitations of this Study: Sample Characteristics .............. ... .. .................13
Limitations of this Study: Possible Multidimensionality of the Item Pool .......139
Limitations of this Study: Sample Size .............. ...............140....
Conclusion ................ ...............141................

5 A COMPARISON OF THE RASCH MODEL AND A TWO PARAMERTER
ITEM RESPONSE MODEL TO DERIVE ITEM AND PERSON
PARAMETERS ............ ..... ._ ...............154...


M ethods .............. ...............156...
Participants ............ ..... ._ ...............156...
Dimensionality ................. ... ...............156
Preparing the data for 2PL analysis............... ...............15
Model fit............... ...............157..
2PL analysis .............. .... .... .............15
Comparison of parameter estimates ........__..........._. .........._._. ...157
Item discrimination .............. .... ...............157..
Person ability and item difficulty .........__......... .........___......15











Re sults................... ................ 158........ .....
M odel Fit ................. .............. ................. 158....
Item Discrimination D512 ................. ................. 158........ ...
D377 Item Discrimination............... ............ 159
D512 Item Difficulty ............ ......_ .............. 160.
D377 Item Difficulty ............ ......_ .............. 161..
D512 Person Ability ................. ................. 162........ ....
D discussion ............... .... ....... ....... ... ... .... ..... .. ...... .... ....... .... ...... 6
Which IRT Model, 2PL or Rasch, Best Fits the FMA-UE Data? ............. 162
Within Each Dataset, Do FMA-UE Item Discriminations Vary? ...............163
Does the Choice of IRT Model Affect Either Item or Person Parameters? .164
What Does Item Discrimination Mean? ............ ...............164.....
Conclusion............... .............. 167

6 SUMMARY AND CONCLUSION ........._..._......_._ ...._._ ...........17

APPENDIX

A RASCH ANALYSIS CONTROL FILES ................. .........._ ....... 179.........


Control File for Proj ect 1 (Chapter 2) .....__.___ ..... ....... ........._......179
Control File for Proj ect 2, T1 (Chapter 3) ................. ........___........._.180
Control File for Proj ect 2, T2 (Chapter 3) ................. ................ ...._.181
WMFT Control File for Proj ect 3, T1 (Chapter 4) ................. ................. 182
WMFT Control File for Proj ect 3, T2 (Chapter 4) ................. ... ............... 183
FMA-UE + WMFT Control File for Proj ect 3, T1 (Chapter 4) ................. 184
FMA-UE + WMFT Control File for Proj ect 3, T2 (Chapter 4) ................. 185

B SAS AND MULTILOG CONTROL FILES .........__ ...... ...__ .........__ ...188


SAS PCA Control File for FMA-UE Data Proj ect 1 (Chapter 2) .............188
SAS PCA Control File for WMFT Data Proj ect 3 (Chapter 4) ................. 189
SAS PCA Control File for FMA-UE + WMFT Data Proj ect 3 (Chapter
4) ................. ...... ...__ .......... ..._ ..............19
Multilog Control File for D512 (Chapter 5)............... ....................191
Multilog Control File for D377 (Chapter 5)............... ....................9

LIST OF REFERENCES ................. ...............193................

BIOGRAPHICAL SKETCH .............. ...............214....


















LIST OF TABLES


Table pg

1 The Fugl-Meyer Upper Extremity Assessment ................. ................. ........ 52

2 Descriptions of the traditional upper extremity limb synergies .............. ..... ........._.53

3 Characteristics of the 512-person sample............... ...............74.

4 Item measures, fit statistics and score correlations for the full 33-item FMA-UE
(n = 512) ................ ...............75.......... ....

5 Eigenvalues for the principal components analysis .............. .....................7

6 Item correlations with first principal component ........._.._........._._ .............77

7 Rotated factor loadings (oblique rotation) for 4 principle components ...................78

8 Item measures, fit statistics, and correlations for revised FMA-UE (n = 512, 30
Item s) ........... ..... .._ ...............79..

9 Item difficulty hierarchy revised 30-item FMA-UE .............. ....................8

10 Sample characteristics (n = 377) .....__.....___ ..........__ ..........10

11 Rasch derived infit statistics T1 and T2 (n = 377) ....._____ .... .. ..__ ............102

12 Differential item functioning analysis (n= 377, df = 594, ** p<0.0017, p<0.05)103

13 Sample characteristics (n = 91) ................ ...............142.............

14 Example of rating scale statistics ................. ......... ....___ ...........4

15 Principal components analysis eigenvalues of the Wolf Motor Function Test......144

16 Principal component factor loadings .......................... ....___ ............4

17 Wolf Motor Function Test item measures, fit statistics, and point measure
correlations. .............. ...............145....

18 Wolf Motor Function Test item difficulty hierarchy .........._.... ......_._..........146










19 Wolf Motor Function Test differential item function analysis .............. ..... ..........146

20 Eigenvalues of the FMA-UE + WMFT linked assessment ................. .................147

21 FMA-UE + WMFT item pool rotated factor matrix ...........__... ......__.........148

22 FMA-UE + WMFT item measure, fit statistics, and point measure correlations ..149

23 Comparison of item discrimination values calculated by Multilog to those
calculated by Win steps in the first dataset ................. ..............................16

24 Comparison of item discrimination values calculated by Multilog to those
calculated by Winsteps in the second dataset ................. .......... ................1 69

25 D512 comparison of item parameters derived with the 2PL and Rasch models....170

26 D377 comparison of item parameters derived with the 2PL and Rasch models....172

















LIST OF FIGURES


Figure pg

1 Keyform recovery maps for three individuals with stroke. ...........__... .................81

2 item difficulty measures of T1 (0 14 days post-stroke) and T2 (6 months post-
stroke) ........._ ...... ............... 103...

3 scatter plot of 3 0-item FMA-UE item difficulties at T1 versus T2.............._._......104

4 Comparison of mean person ability with and without DIF adjustment ...............104

5 Scree plots of the FMA-UE + WMFT linked assessment ...........__.................147

6 FMA-UE + WMFT differential item functioning scatter plot .............. ................151

7 FMA-UE + WMFT linked item map .............. ...............152....

8 FMA-UE person ability measures overlaid onto the FMA-UE + WMFT item
m ap ................ ...............153.................

9 D512 comparison of item hierarchies generated by Multilog (probits) to the item
hierarchy generated by Winsteps (logits converted to probits by multiplying by
0.59). Error bars a 2 SEM ........... __... ......... ...............171.

10 D512 scatter plot of item difficulty estimates (n = 512, r = 0.99) ........................171

11 D377 comparison of item hierarchies generated by Multilog (probits) to the item
hierarchy generated by Winsteps (logits converted to probits by multiplying by
0.59). Error bars a 2 SEM ........... __... ......... ...............173.

12 D377 scatter plot of item difficulty estimates (n = 377, r = 0.98).. ................... .....173

13 D512 scatter plot of person ability estimates (n = 512, r = 0.99) .........................174

14 D377 scatter plot of person ability estimates (n = 377, r = 0.99).............._._.........174
















Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

MEASURING POST-STROKE ARM MOTOR ABILITY: MEASUREMENT
PROPERTIES OF THE FUGL-MEYER UPPER EXTREMITY ASSESSMENT
EXAMINED WITH AN ITEM RESPONSE THEORY MEASUREMENT
FRAMEWORK

By

Michelle Lynne Woodbury

August 2006

Chair: Lorie Richards
Cochair: Craig Velozo.
Major Department: Occupational Therapy

There is a pressing need to ascertain whether assessment tools commonly used in

post-stroke upper extremity motor rehabilitation research and clinical practice are

accurately quantifying impairment, characterizing recovery and producing a score that

has clear functional interpretability. The Fugl-Meyer Upper Extremity Assessment

(FMA-UE) is the most widely used clinical assessment of post-stroke UE motor

impairment. Because of the importance of this assessment in post-stroke rehabilitation

research, it is critical to examine, and perhaps improve, the quality of its measurement

properties. The overall purpose of this dissertation was to examine the measurement

properties of the FMA-UE. An item-oriented measurement perspective was used to

investigate (1) the validity of the FMA-UE item content, (2) the reproducibility of its

item structure, and (3) interpretation of its score. Rasch analysis, a type of item response

theory statistical method, was applied to FMA-UE data collected from 512 individuals in









the Kansas City Stroke Registry and Kansas City Pepper Center exercise study. In the

four studies comprising this dissertation, it was determined that the items of the FMA-UE

contribute to the measurement of multiple constructs and demonstrate a difficulty order

that is not consistent with the item order proposed by Fugl-Meyer. The reflex items were

removed, resulting in a modified 30-item FMA-UE, a unidimensional measure of post-

stroke UE motor ability. It was found that the Rasch-derived item difficulty order of the

30-item FMA-UE remained consistent, independent of person ability, longitudinally

across two testing occasions. The modified 30-item FMA-UE was linked to the Wolf

Motor Function Test. Linking is a novel method to improve the interpretability of the

FMA-UE score and clearly illustrated how changes in motor impairment can translate to

improvements in activity performance. The item and person parameters estimated by

Rasch analysis were validated with the two-parameter item response theory graded

response model. This proj ect is significant in that it not only establishes the quality of the

30-item FMA-UE as an instrument to measure post-stroke UE motor impairment, but

also establishes the 30-item FMA-UE as an instrument with which to gather information

that will advance rehabilitation science motor control theory.















CHAPTER 1
INTRODUCTION AND LITERATURE REVIEW

Each year over 700,000 individuals in the United States have a stroke [1].

Approximately three-fourths of these individuals will experience immediate flaccid

paralysis of one arm and hand [2]. This flaccid period is followed by rapid recovery of

arm motor function over the first 3-months post-stroke [3]. However, full recovery of arm

and hand function is rare. At four years post-stroke 85% of stroke survivors report

residual arm or hand motor impairment [4]. For many individuals with severe stroke the

arm never becomes useful [5] and for many individuals with less severe stroke, residual

arm motor impairment interferes with performance of daily tasks [4]. Stroke is the

leading cause of long-term disability in the United States [1].

In recent years, progress in neuroscience and rehabilitation science has advanced

the understanding of the potential for neuronal recovery after stroke-related brain damage

[6]. In turn, this has led to the development of new rehabilitation interventions [7-9] and

revision of traditional rehabilitation interventions [10] to restore post-stroke upper

extremity (UE) motor function. The rehabilitation disciplines have adopted an evidence-

based practice paradigm [11, 12], necessitating accurate measurement of motor recovery.

There is a pressing need to ascertain whether assessment tools commonly used in post-

stroke UE motor rehabilitation research and clinical practice are accurately quantifying

impairment and characterizing recovery [13, 14].

The Fugl-Meyer Upper Extremity Assessment (FMA-UE) [15] is the most widely

used clinical assessment of post-stroke UE motor impairment [14, 16]. It has excellent









intra-rater reliability (r = 0.995 [17]), inter-rater reliability (r = 0.992 [17]), test-retest

reliability (Intraclass correlation coefficients 0.94 0.99 [18]), and internal consistency (r

= 0.97 [19]. It has been used as the standard from which to establish the construct validity

of other commonly used tests of post-stroke UE motor function such as the Wolf Motor

Function Test and Ashworth Scale [20-26]. Researchers consistently use the FMA-UE as

a descriptor of functional status following stroke [4, 20, 27]. To this end, FMA-UE scores

have been used to stratify research study participants into categories of stroke severity

[28], predict long-term functional participation [29], and to describe arm motor

impairment [30-32], arm movement quality [21], and residual arm motor function [33-

35]. Furthermore, the FMA-UE is the primary criterion for evaluating the success of

novel upper extremity interventions such as rhythmic bilateral movement training [36,

37], neuromuscular stimulation [38], electromyographic triggered electrical stimulation

[39, 40], Botox [41], home based exercise [30], community based exercise [42], robot

aided therapy [43, 44], virtual reality [45], imagery [46], forced use [47], and modified

constraint induced movement therapy [48-5 1]. Because of the importance of this

assessment in post-stroke rehabilitation research, it is critical to examine, and perhaps

improve, the quality of its measurement properties.

Table 1 presents the FMA-UE. The evaluator observes the client's performance of

33 items, scoring each item by applying a 3-point ordinal rating scale (0=the movement

cannot be performed at all, 1=the movement can be partially performed, and 2=the

movement can be faultlessly performed). Ratings are summed and the FMA-UE score is

reported out of the maximum 66 points. Higher scores indicate more UE motor ability.









Recently the Canadian Stroke Network identified 88 stroke-related UE motor

function assessments [52]. The obvious question is, which is the "best" assessment?

Salter and colleagues (2005) [14] argue that evidence based practice is advanced only

when researchers and clinicians choose assessments with "superior measurement

properties." The question then becomes, what are the essential elements of a high quality

assessment? The quality of an assessment is usually judged by its "essential"

measurement properties: reliability, validity, and responsiveness [14]. These criteria

assure that an assessment measures the skill that it is purported to measure (validity),

produces a reproducible score (reliability), and is sensitive to change over time

(responsiveness). Psychometric evaluation of an assessment' s measurement properties is

usually accomplished with classical test theory statistical methods such as correlating or

comparing assessment scores.

The item response theory (IRT) measurement framework, e.g., Rasch measurement,

offers another approach for judging the quality of an outcome measure. As the name

implies, IRT methods allow one to evaluate the measurement properties of items rather

than whole tests [53]. One can examine an assessment's item content (i.e., does a specific

item contribute to the measurement of an intended construct?), item reproducibility (i.e.,

do items measure the purported construct in the same way each time the assessment is

administered?), and item-level interpretation of the assessment' s score (i.e., with respect

to item content, what does an assessment's change score mean?). Although the

measurement properties of the FMA-UE have been extensively studied using classical

test theory [15, 17, 18, 20, 24, 54-57], presently there are no published studies applying

IRT approaches to the FMA-UE.









The overall purpose of this dissertation is to examine the measurement properties

of the FMA-UE. I will assume an item-oriented viewpoint from which to investigate (1)

the validity of the FMA-UE item content, (2) the reproducibility of its item structure, and

(3) interpretation of its score.

In the subsequent literature review I will develop the background to address the

study's purpose. First, I will review the World Health Organization' s International

Classification of Function and Disability (ICF). The ICF provides an operational

definition of the UE skills of interest to post-stroke UE rehabilitation researchers and

models a way to organize measurement of these skills. Next, I will review traditional and

contemporary concepts of UE motor control. This is important because exploring FMA-

UJE item-level construct validity will serve to focus our attention on motor control theory.

Third, I will review the Rasch measurement model because it permits testing the

theoretical claims about a construct. Finally, I will briefly review the critical differences

between the Rasch model and other IRT models as they pertain to this proj ect.

In the chapters that follow I addressed the purpose of this dissertation in four

studies. The aim of study 1 was to test the validity of the traditional post-stroke UE motor

control theory that underlies the FMA-UE by examining its item-level dimensionality and

item difficulty hierarchy in a cross-sectional analysis. The aim of study 2 was to test the

longitudinal reproducibility of the FMA-UE item structure. The aim of study 3 was to test

a method for enhancing the interpretability of a FMA-UE change score by linking to

another assessment with more functionally oriented items. Because Rasch analysis was

the primary method used throughout these three studies, the final proj ect was to compare

the results obtained with the Rasch model to those obtained with another IRT model to









assure that my choice of statistical method was valid. The overarching purpose of this

dissertation and the aims of each proj ect are consistent with that of an overall emphasis

within the Hield of rehabilitation research to assure that clinical assessment tools are

optimal for measuring targeted outcomes [58-60] and translate to functionally

interpretable measurements of change [61].

The International Classification of Function and Disability

The World Health Organization's International Classifieation of Function and

Disability (ICF) [62] offers a systematic approach to classify and describe post-stroke UE

motor function according to a theory of health and disability. The ICF provides a useful

taxonomy and standard language to define UE motor function and describe its

measurement. As a practical research application, the ICF describes sets of behaviors that

should be included in an instrument intended to measure given domain [63].

The purpose of the ICF is to describe components of health in which a person' s

functioning is conceived as a dynamic multi-dimensional interaction between individual

health attributes and contextual factors. The ICF identifies three primary health

dimensions of the individual: the functioning of the body or body parts, the functioning

of the individual in his/her daily activities, and the functioning of the individual in

relationship to his/her participation in the socio-cultural community. These health

dimensions can also be expressed in the negative to describe disability: body

function/structure impairment, activity limitation, and participation restriction.

The ICF models the concept that stroke-related UE motor deficits affect the

individual along a continuum of health dimensions; from impaired body functions to

limited performance of activities and restricted participation in the community.

Accordingly, post-stoke arm/hand motor function can be measured in each dimension.









Furthermore, the model suggests that altered arm motor function as measured in one

dimension (e.g., body function) conceivably translates to altered arm motor function

when measured in an adj acent dimension (e.g., activity performance). The purpose of this

dissertation centers on UE motor body function impairment and activity restriction; thus

the subsequent discussion is limited to these domains.

In series of papers [14, 64, 65] researchers from the Canadian Stroke Network

organized the most commonly used post-stroke assessments according to ICF domains

for psychometric review. The FMA-UE is classified as an assessment of body function

impairment because its item content is consistent with the ICF operational definition of

this domain [14]. For example, "neuro-musculoskeletal and movement related functions"

illustrate one aspect of body function. More specifically, the functions are described as

"control and coordination of single and multi-j oint motions" (p. 99-100). FMA-UE items

fit this description.

A valuable assessment for measuring upper extremity outcome after stroke is one

that provides information about the recovery spectrum, i.e., movement impairment and

the functional consequences of movement impairment. The ICF models a relationship

between body function impairment and activity performance. According to the ICF, it is

conceivable that body function impairment impacts behaviors in the adj acent activity

performance domain such as the ability to "carry, move and handle obj ects." More

specifically, "reaching," "grasping," and lifting," (p. 141-143). It is conceivable that an

UJE assessment intended to capture the functional aspects of reduced body function UE

impairment should be comprised of items with this variety of content.










Although the ICF offers a broad framework to define FMA-UE item content and

organize its structure, the ICF does not specify what UE motor behaviors are affected by

stroke, nor does it detail expectations regarding post-stroke recovery. For that, a

discussion of motor control theory is warranted.

Motor Control Theory

Motor control theory provides a conceptual framework of how the CNS organizes

and controls motor function by addressing potential "rules" governing movement

organization. Motor control theory models normal motor behaviors, defines injury-related

motor behaviors, and predicts a post-injury course of motor recovery [66]. One way in

which an abstract motor control theory becomes tangible to the rehabilitation professional

is in the design, structure and scoring of a clinical assessment [67]. The theory defines

"normal" and "abnormal" movement behaviors therefore, governing the content of

assessment items. Moreover, the theory describes behaviors that evidence "less" and

"more" motor control and, in doing so, suggests an appropriate array of items to measure

a range of motor skills. Barecca et al. (2006) [68] argues that stroke rehabilitation

knowledge is advanced only if the tools chosen to measure motor control are consistent

with contemporary motor control theoretical expectations of deficit and recovery. Recent

advances in movement science challenge the motor control theory upon which the FMA-

UJE was based. Prior to a review of these challenges, I will review the theoretical

foundation of the FMA-UE.

Traditional Theories of Motor Control and Recovery

The characterization of post stroke recovery has been the focus of theoreticians,

clinicians and researchers and represents one of the historic underpinnings of modern

rehabilitation. Many of the traditional stroke rehabilitation frames of reference emanate









from Jackson's late 19th century work with persons with epilepsy and other "nervous

disorders" [69]. Jackson, a neuroscientist influenced strongly by Darwin, applied the

"doctrine of evolution" to his observations and ultimately posited a theoretical order of

the CNS. According to his framework, the complex motor abilities seen in humans are

evidence that higher motor centers have evolved out of lower centers. In the healthy

nervous system these more advanced centers orchestrate and modify the reflex chains of

lower centers. Jackson's observations of individuals with epilepsy led him led him to

hypothesize that injury to the cortex resulted in "dissolution" of the CNS and reversal of

its evolutionary structure. That is, without input from the advanced motor centers,

movement control defaults to primitive lower centers evidenced in gross undifferentiated

reflexive motions.

In the early 20th century Charles Sherrington and his colleagues performed a series

of elegant experiments to show the existence of the reflex and describe the interaction of

afferent feedback with voluntary movement. In a classic primate experiment Sherrington

performed complete upper limb sensory denervation. This procedure appeared to abolish

all voluntary upper limb movement. The limb was held close to the body and not used for

daily tasks. Sherrington concluded that afferent information was a necessary building

block of voluntary movement [70]. The work of Sherrington and his colleagues formed

the experimental foundation of the classic reflex chain hypothesis [71]. The results of

Sherrington and Jackson's experiments and observations were drawn together and are

often referred to in the clinical literature as the reflex-hierarchical theory of motor control

[66, 72].









With the advances in medicine and improved post-stroke survival in the mid 20th

century, clinician-researchers applied the reflex-hierarchical theory to their clinical

observations [73-75]. In a classic paper Dr. Thomas Twitchell (1951) [75] described an

"orderly progression of phenomena" characterizing the course of post-stroke UE motor

recovery. Initial UE flaccidity was followed by emergence of gross flexor and extensor

voluntary movement synergies, gradual separation of synergy movement patterns and

Einally distal fine motor coordination. Twitchell attributed abnormal motor behaviors to

cortical dissolution and emergence of default reflexive behaviors as per Jackson's model

of the CNS. Twitchell suggested that recovery was a "re-evolution" of this hierarchy as

the cortex reassumed its natural inhibition of primitive proprioceptive reflexes.

Renowned physical therapist Signe Brunnstrom furthered Twitchell's work, and

made an important contribution to stroke rehabilitation by creating an assessment tool

that synthesized clinical observation with motor control theory [76, 77]. She mapped

post-stroke recovery into a six-stage sequence. Beginning with flaccidity (stage 1), motor

recovery was first evident as return of reflexive responses (stage 2). With further

recovery individuals gained voluntary motion within whole-limb flexion and/or extension

synergies (stage 3), movement requiring a combination of the synergy patterns (stage 4),

followed by movement independent of synergy patterns (stage 5), and Einally individual

joint motion and normal coordination (stage 6). The "Hemiplegia Classifieation and

Progress Record" [76] detailed specific motor behaviors typifying each stage. The

hierarchical ordering of movements documented in the Progress Record was useful to the

clinician for assessment and treatment planning. Using it, therapists could identify a

patient' s stage of recovery by observing the movements the patient could or could not










perform. Once the client's location along the recovery continuum was known, the

Progress Record enabled the clinician to map treatment to advance the client to the next

stage.

Noting the dearth of quantitative assessments available to stroke rehabilitation

clinicians and researchers, Fugl-Meyer et al. (1975) [15] created a rating scale assessment

based, at least in part, on Brunnstrom's Progress Record and Twitchell's observations.

Fugl-Meyer' s theoretical framework broadly reflects the assumptions of the reflex-

hierarchical motor control theory. Fugl-Meyer stated:

The form has been constructed following the hypothesis that the restoration of
motor function in hemiplegic patients follows a definable stepwise course. Thus
for a patient with hemiparalysis, recurrence of reflexes always precedes volitional
motor action. Thereafter through initial dependence on synergies, the active
motion will become successively less dependent upon the primitive reflexes and
reactions and finally complete voluntary motor function with normal muscle
reflexes may be regained. [15] p. 14

Fugl-Meyer chose items to exemplify the construct of motor recovery including

reflex items and voluntary movement items. Moreover, because the assessment was

intended to measure recovery, Fugl-Meyer arranged the items in a hierarchical order to

illustrate the recovery process. He expected motor recovery to proceed in a reflexive-to-

voluntary, and synergy-to-isolated progression as the cortex resumed its ability to

integrate reflexive movement components into coordinated limb motions.

Challenges to the Traditional Theories of Motor Control and Recovery

In recent years there have been many advances towards understanding how

movement is controlled, organized and recovered after injury. There has been a shift

away from a purely neurophysiologic explanation of motor control/recovery to a task-

oriented view of motor control/recovery [66]. Contemporary motor control science posits

that movement emerges from a dynamic, complex interaction of the individual's neural









and musculoskeletal apparatuses with specific tasks and environmental demands [78].

Recovery is viewed as re-acquisition of an individual's ability to successfully conquer

task-specific contextual demands rather than neurophysiologic re-evolution of the cortical

motor control hierarchy [79].

Recent advances in movement science challenge the "definable stepwise course" of

motor recovery upon which the FMA-UE is based in two ways: (1) the place of reflex

behaviors in an assessment of motor recovery, (2) the pattern of behaviors that evidence

the course of recovery. These challenges will now be expanded upon.

Role of reflexes

Contemporary views of the central nervous system indicate that reflexes and

voluntary movements are uniquely different types of motor behavior. A relatively simple

spinal neural circuit governs the tendon tap muscle contraction, while a complex

supraspinal and spinal neural network governs voluntary goal-directed movement [80].

Afferent input is not necessary for voluntary movements to occur [81], instead reflexes

(e.g., the stretch reflex) appear to be important adjuncts of voluntary motor control [80,

82, 83]. The reflex motor control system is complex as evidenced by research showing

that the sensitivity of the of the stretch reflex threshold is modulated according to

previous muscle activity [84, 85]. The research of Wolpaw (1985) and Wolf and Segal

(1986) has demonstrated that the magnitude of the stretch reflex is amendable to operant

training [86, 87]. Burne and colleagues (2005) [88] found that spasticity (measured as the

amplitude of the biceps stretch reflex) was different in a resting limb versus a moving

limb. The researchers suggest that reduced ability to dynamically modulate reflex activity

contributes more to an understanding of voluntary movement impairment than assessing

reflex activity in a static condition. The research evidence suggests that the FMA-UE









reflex items may be assessing a motor behavior that is uniquely different from other items

that assess voluntary movement.

The presence of the three reflex items in the FMA-UE has been long questioned.

Lindmark and Hamlin (1988) [89] removed the items, altered the rating scale, and added

items to form the Lindmark Functional Capacity Evaluation. Gladstone et al. (2002) [55]

stated that reflexes were "over represented" in the FMA-UE total score. Whitall et al.

(2004) [13] suggested that assessment of tendon-tap reflex behaviors does not contribute

to an understanding of post-stroke UE motor ability. It is reasonable to question whether

the presence of the three tendon tap FMA-UE reflex items confound the interpretation of

the FMA-UE score and threaten the assessment' s construct validity.

The pattern of upper extremity behaviors evidencing recovery

The extent to which post-stroke individuals exhibit abnormal UE flexor and/or

extensor synergy as per Fugl-Meyer' s descriptions [15] (adapted from Brunnstrom [76])

is debated in the literature. Table 2 reminds the reader of limb movements characterizing

traditional synergy patterns. Note that these movements are items 3 11 on the FMA-UE

(see table 1).

Recently, Welmer and colleagues (2006) [90] prospectively observed 64

consecutive patients with stroke at one week and three months post-stroke. The

researchers tested the hypothesis that stroke recovery is consistent with Brunnstrom's

sequence, specifically that UE flexor and extensor synergies emerge early in recovery

while non-synergistic isolated movements emerge later in recovery. Subj ects were

evaluated by trained, study-specific, occupational and physical therapists. One-week post

stroke 15% of the sample was unable to follow testing directions and 22% of the sample

exhibited flaccid hemiplegia. Among the subj ects who participated in voluntary UE









movement assessment, no participant exhibited only (i.e. "pure") flexor/extensor synergy

movements. Welmer found that 63% of the sample demonstrated at least one UE

movement inconsistent with synergy (e.g., was able to partially extend the elbow while

abducting the shoulder). Three months post stroke, 13% of the sample remained unable to

move. Among the others, 16% exhibited at least one UE movement typically considered

part of the flexor/extensor synergy pattern (e.g., tended to flex the elbow while flexing

the shoulder to or above 90 degrees) but 77% exhibited no evidence of synergy-related

movements. The results suggest that some flexion/extension synergy movements occur

early in recovery, but some isolated movements also occur early in recovery. The

researchers concluded that individuals with stroke rarely exhibit pure manifestations of

the flexor or extensor synergy patterns.

Welmer' s work does not discount longstanding clinical observations [75, 76] that

individuals with stroke exhibit impaired intra-limb control. Instead, Welmer's study

suggests that the abnormal pattern is not manifest as an "all or nothing" obligatory link

between all movements comprising the traditional definition of the synergy (table 2).

Indeed, contemporary experimental paradigms have elucidated the specific parts of the

abnormal pattern that are most likely to be present. In a series of studies Beer et al.

(1999) [91] and Dewald et al. (2001) [92] have shown that individuals with stroke were

unable to maintain maximal j oint torque of a one j oint in a single direction without

producing secondary torques of an adj acent j oint in an unintended direction during an

isometric task. In this study, eight individuals with chronic stroke and moderate spasticity

(Ashworth scores 3-5) were directed to attain and then maintain maximal shoulder

flexion or shoulder abduction force. Forces were measured with load cells on the









humerus and forearm. In both the shoulder flexion and abduction conditions, all subjects

concurrently produced unintended elbow flexion torque. In a similar experimental

paradigm, Lum et al. (2003) [93] showed that post-stroke subjects also produced

unintended humeral internal rotation torque. The results suggest some individuals with

stroke exhibited abnormally constrained intra-limb motor control evident by abnormally

linked shoulder flexion/abduction, elbow flexion, and internal rotation torques.

Abnormal linkages of elbow flexion with the shoulder flexor or abductor

musculature are also evident during supported and unsupported reach. Beer et al. (2004)

[94] studied eight persons with chronic stroke and moderate spasticity. In this study

subjects performed rapid movements in the horizontal plane from a central starting point

to five targets located at the outer limits of the reachable workspace. The targets were

placed to require different combinations of flexion and extension ranges of motion at the

elbow and shoulder. In one condition the hemiparetic limb was supported in a friction-

less arm trough, in another it was not. In both conditions subjects had difficulty generate

concurrent shoulder flexion and elbow extension torques necessary to reach the targets,

however the deficit was greater in the unsupported versus supported condition. Kamper et

al. (2002) [95] showed similar interj oint coordination deficits in an unsupported reaching

condition. In this study, sixteen participants with chronic stroke were directed to point to

toward a screen containing 75 targets displayed side to side across the arm reachable

workspace from waist to head. Kinematic analysis showed that participants exhibited

abnormally constrained elbow flexion and shoulder flexion evident in curved hand

traj ectories when attempting to reach to targets the farthest from the body. The deficit

was similar irrespective of target locations in the ipsilesional or contralesional workspace.









Dewald, Beer and Kamper' s results are consistent with those of Levin et al. (1996)

[3 5]. In this study, the researchers studied reaching traj ectories of 10 participants with

moderate post-stroke UE impairment (moderate spasticity measured on a scale designed

by the investigator and baseline FMA-UE scores 19 66). Targets were arranged on a

table so that subj ects were required to utilize muscle activation patterns within and

outside of the traditional synergy configurations. For example, to reach a target placed

contralateral to the affected limb, the participant was required to activate the extensor

synergy musculature; shoulder adductors, internal rotators, and elbow extensors. The

researchers reasoned if participant' s movements were strictly constrained into a flexor or

extensor synergy pattern, participants would be less able to reach targets outside this

pattern. That is, if extensor synergy were dominant, reach to a contralateral target would

be less impaired than reach to an ipsilateral target. If the flexor synergy were dominant,

reach to a near target would be less impaired than reach to a far target. The results

showed that this was not the case. Reaching traj ectories were less smooth and more

curved for individuals with stroke when compared to healthy individuals, however post-

stroke movements within a typical synergy pattern were no less impaired (i.e., were not

smoother and straighter) than movements to targets outside the synergy pattern.

In an earlier study Trombly (1992) [96] realized a similar phenomena. In this study,

5 subj ects with left hemiparesis were directed to reach to three targets placed so as to

require movement into an extensor synergy pattern or away from this pattern. Cinematic

and electromyographic (EMG) data were recorded. All clients demonstrated impaired

kinematics (less smooth traj ectories, more movement units, increased movement time,

and earlier time to peak velocity) and altered EMG (higher percentage of maximum









voluntary output) in the impaired UE as compared to the unimpaired UE. However, the

biceps-anterior deltoid coactivity indices were not significantly different between the

limbs indicating that an abnormal synergistic constraint was not hampering the

movement.

The work of Trombly, Levin, Kamper, Dewald, Lum and Beer provide evidence

that the post-stroke reach is impaired by an abnormally constrained inter-j oint

coordination pattern involving a tendency to flex the elbow and internally rotate the

humerus when attempting to lift the humerus against gravity. Tromby [96] suggested that

this observable pattern is not directly attributable to abnormal co-activity of the "flexor

synergy" musculature, but may evidence a motor control strategy employed to reach

targets despite muscle weakness. When put together with the work of Welmer et al. [90],

it is apparent that the observed post-stroke UE motor deficit is not fully explainable by

traditional definitions of synergy. Full flexor or extensor synergy patterns are not always

present, at least in the subj ects studied by the above labs. It is possible that the other

muscles typically considered part of the synergy patterns, specifically the scapular

retractors, scapular elevators, shoulder external rotators and forearm rotators have not

been studied due to limitations in collecting surface EMG from deep muscles and

limitations in UE kinematic and kinetic modeling. However, at least to date, the evidence

suggests pure synergies (as per the traditional definition) do not appear to be obligatory

patterns for all individuals with stroke.

The importance of this research lies in its potential impact on the interpretation of

the FMA-UE score. Fugl-Meyer designed the scale with an expected regularity of

recovery in mind [15]. The evidence only partially supports this expected synergistic









recovery pattern. Therefore, it is possible that the measurement properties of the FMA-

UE with regards to the interpretation of its score may be affected.

Researchers use the FMA-UE score to characterize post-stroke recovery. One

interpretation of the FMA-UE score is consistent with traditional expectations regarding

the pattern of recovery; specifically low scores represent synergy-dominated movement

abilities whereas higher scores represent more isolated movement abilities. For example,

Chae et al. (2003) [97] interpreted the FMA-UE score in this way "it (the FM4~r-UE)

measures evolving synegy patterns as ithr subjects 11 ithr vohmntary movements limited to

proximal joints and synelgy patterns exhibiting the lowest scores and a s ithr subjects

demonstrating isolated movements in distal joints, 11 ithr normal reflexes exhibiting the

highest scores (p. 7). Some researchers have used categories as a context from which to

interpret the FMA-UE score. For example, Michaelson et al. (2001) [98] stated that

FMA-UE scores of 0-19 points represented severe UE motor impairment, FMA-UE

scores of 20-64 points represented moderate UE motor impairment, and FMA-UE scores

of 65-66 represented normal UE motor ability. Pang [42] stated that FMA-UE scores of

0-27 points represented severe UE motor impairment, FMA-UE scores of 28-57 points

represented moderate UE motor impairment, and FMA-UE scores of 58-66 represented

mild UE motor impairment. Sometimes researchers designate a FMA-UE score to

indicate the boundary between synergy-dominated and non-synergy dominated

movements. In the original article, Fugl-Meyer provided evidence that a score

approximately equal to 60% of the total shoulder/arm score, a score of 13 out of 22

points, was the boundary between synergy and non-synergy movements [15]. Finley et al.

(2005) [44] utilized a maximum FMA-UE score of 15 as inclusion criteria for a study of









robotic training because "the score represented severe motor inspairnzent in the paretic

arm, as demonstrated by limited neenia, Il n~l ithrin synergy patterns and no vohmntary

wrist or hand function (p. 684). The ranges of FMA-UE scores and their relationship to

the categories are frequently used in clinical research as a reference point for describing

participants' motor behavior and interpret outcomes. However, whether the scores are

being interpreted correctly is not known.

Contemporary Theories of Motor Control and Recovery

To this point I have argued that the validity of the FMA-UE item content and the

interpretation of the FMA-UE score may be threatened by challenges to traditional motor

control theory upon which the assessment was founded. The purpose of this dissertation

includes enhancing the measurement properties of the FMA-UE if necessary and

possible. This necessitates further exploration of contemporary theory to provide a better

understanding of the UE motor recovery construct than was provided by traditional

theories.

UE motor impairment

Bernstein (1967) [99] observed that any given motor task could be achieved using

many different combinations of muscle activation patterns and limb configurations. The

motor system chooses from the vast number of "degrees of freedom" available to it

depending on the contextual demands of the movement task (e.g., influence of gravity,

inter-segmental joint torques). Bernstein argued that it is unlikely that the motor control

system exerts executive control over each movement possibility (i.e., one to one control

of each degree of freedom) because a vast amount of storage capacity would be required

of the cortex. Instead, Bernstein proposed "functional synergies" as the more probable

motor control strategy. "Functional synergies" are defined as "highly evolved ta~sk-










specific ensembles ofneuromuscular and skeletal components constrained to act as a

single unit so as to reduce the computational burden associated a ithr controlling

voluntary movements in multisegmented hinbs"\` [99] p. 47.

In the healthy CNS, functional synergies appear to be flexibly organized according

to specific task requirements and goals. Using kinematic assessment, researchers have

shown that distinctive motor patterns emerge in response to task and contextual demands

[100-105]. For example, Trombly's lab demonstrated that neurologically intact

individuals displayed unique kinematic patterns of movement when asked to scoop actual

coins from a table versus scooping imagined coins [100] and smoother reaching

traj ectories when reaching to a food chopper containing actual versus imagined food

[101]. Trombly suggested that the perceptual and cognitive environments in the imagined

versus real conditions were different 'sets' affording movement responses particularly

patterned to the respective context.

Functional synergies also appear to be dynamically organized according to certain

task-specific mechanical factors. For example, Murray and Johnson (2004) [106] found

that healthy individuals scaled UE j oint torques according to the number of joints

involved in a task. For example, healthy individuals generated less joint torque during

functional tasks requiring primarily a single joint movement such as feeding, when

compared to tasks requiring multiple joint movements such as simulated hair brushing.

This research suggests that healthy individuals dynamically, and flexibly organize

and execute motor programs in accordance with visuo-perceptual and mechanical task-

demands. Do individuals with stroke retain the capacity to flexibly organize and apply

"functional synergies" in task-specific ways?









Latash and Nicholas (1996) [107] suggested that an "abundant" motor control

system enables an individual to adapt to specific environmental conditions or states of

injury. He argued that the "abnormal" synergy is actually a "normal" attempt to

overcome muscle weakness and stroke-related neural damage in order to successfully

accomplish a task (i.e., lift the arm). Latash notes that the changed movement pattern

reflects the admirable adaptability of the human CNS [108].

Research supports the presence of post-stroke UE weakness as a possible

mechanism underlying the abnormal flexor synergy pattern. In a paradigm similar to that

of the Dewald and Beer lab, Lum et al. (2003) [93] found that individuals with stroke

producing unintended secondary torques in patterns consistent with traditional definitions

of flexor synergy. However, unlike the Dewald and Beer studies, Lum collected EMG

recordings of shoulder and elbow agoni sts/antagoni sts during the experiment. Subj ects

producing the largest secondary torques (i.e., less able to dissociate elbow flexion from

shoulder abduction) also exhibited the greatest force production deficits. According to

Lum, the results suggested that abnormal post-stroke synergy patterns are, at least in part,

as a consequence of stroke-related UE weakness.

Mercier et al. (2005) [109] examined single and multi-joint isometric muscle

activity requiring different combinations of shoulder and elbow flexion/extension.

Subj ects were included in the study if their UE movements were consistent with

Chedoke-McMaster Stage 3 [23] (similar to Brunnstrom's flexion synergy pattern;

participants demonstrated active UE movement, but mostly in flexion). Joint torques were

measured with force transducers placed at the forearm and humeral shaft. The results

showed that subj ects demonstrated overall decreases in force production at the shoulder.










However, the deficits were similar irrespective of the direction to which the force was

aimed.

In subjects with stroke, Zackowski et al. (2004) [110] found that a movement

requiring simultaneous shoulder flexion and elbow flexion was performed faster and

more accurately than an arm movement requiring shoulder flexion and elbow extension.

Zackowski suggested that activation of the biceps to flex the shoulder might be a strategy

to compensate for weakness of the anterior deltoid.

McCrea and colleagues (2005) [111] demonstrated that the anterior deltoid reached

maximal activation ("saturation") in persons with stroke much sooner than healthy

controls while abducting the shoulder. As saturation was achieved, subj ects flexed the

elbow and internal rotated the humerus to reach the target. McCrea found that the

impairments were worse in individuals with more severe stroke compared to less severe

stroke.

Taken together, these studies suggest that shoulder weakness contributes to the

observed "flexor synergy" movement pattern. Following stroke the motor control system

has reduced resources with which to successfully interact with (i.e., adapt to) the

environment [112]. Cortical stroke damages the corticospinal pathway and may unmask

spared secondary motor tracts (e.g., vestibulospinal and rubrospinal) [113]. Stroke

reduces the capacity to recruit motor units [114], thereby reducing strength [115] ,

altering patterns of agonist and antagonist recruitment [33] and reducing interj oint control

[116]. Latash and Anson (1996) [108] suggested that flexing the elbow (i.e., the

observed abnormal flexor synergy pattern) may afford a mechanical advantage by









reducing the length of the lever arm and allowing for the individual to adapt to the post-

stroke state yet still accomplish a task goal (i.e., flex the shoulder).

Extending this argument, Carr and Shepherd (2002) [78] propose that post-stroke

recovery can be viewed from a similar task-specific perspective. Individuals with stroke-

related motor control deficit are able to accomplish tasks if their motor resources are

sufficient to satisfy the task-specific demands placed on the motor system. Individuals

with less UE motor ability would be more likely to accomplish an "easy" versus a "more

challenging" task. As these individuals recover they gain the capacity to accomplish both

easy and more challenging motor tasks.

Post-stroke UE recovery

According to Carr and Shepherd, motor recovery following stroke is

"fundamentally" a process of relearning how to move [78]. Most contemporary

rehabilitation frameworks are predicated on the assumption that in order for a patient to

improve his/her motor skill (i.e., recover from stroke-related hemiparesis) the patient

must practice motor skills [79]. Principles of neural plasticity first researched in animal

models translated to humans suggest that healthy brain neuronal connections and cortical

maps are dynamically remodeled by experience [6, 117]. Post-stroke UE recovery may be

enhanced with intense task specific-practice that engages the system to actively solve

motor problems [118, 119].

Note the conceptual dissimilarity between this contemporary understanding of

recovery and its more traditional counterpart. Traditionally post-stroke recovery was

viewed as a passive (from the viewpoint of the patient) process. Rehabilitation

interventions were provided to patients who largely assumed the role of passive recipient

of the intervention, in order to facilitate (or inhibit) an internal neurophysiological









process [120]. In contrast current motor control researchers and theorists view post-stroke

recovery as an active process in which the individual, given his/her internal neural and

muscular deficits, re-learns how to interact with the external environment to satisfy the

demands of the activity [70, 121].

This may signal that recovery is characterized by enhanced ability to execute

movement patterns with the appropriate traj ectory, acceleration and relative muscle

forces necessary to meet the temporal and spatial requirements of a task [66]. Roher

(2002) [122] demonstrated that with one month of robotic training, individuals with

chronic stroke were able to reach to a target with a smoother hand traj ectory velocity

profile, suggesting increased ability to move independently of the elbow-shoulder flexor

"synergy". Ellis and Dewald (2005) [123] found that with practice, individuals with

chronic stroke were able to modify the abnormal coupling of isometric shoulder and

elbow j oint torques evidenced by increased ability to isolate j oint torque production at a

single joint during a multijoint task. Michaelson et al. (2006, 2004) [9, 124] and

Thielman et al. (2004) [125] have shown that with training, individuals with severe to

moderate stroke demonstrate improved shoulder-elbow inter-j oint coordination evidenced

by decreased trunk recruitment and shorter (straighter) hand path traj ectories during

reaching tasks in various parts of the reachable workspace.

Taken together, the above studies suggest that post-stroke recovery is characterized,

at least in part, by the patient relearning the "rules that govern motion" [107]. Individuals

with stroke become more able to control each j oint during a multij oint task. Patients

improve, using ICF terminology, the ability to "coordinate single and multiple joint

motions" [62].









Post-stroke recovery of functional UE motor abilities

The research presented to this point supports the view that post-stroke recovery is a

process in which individuals gain more UE motor ability with regards to overcoming task

demands. More functionally oriented activities requiring interaction with obj ects and at

increased speeds of movements. Task-specific increases in neural, mechanical and/or

contextual requirements concomitantly increase the computational demands on the motor

system [1 12]. Interacting with an obj ect not only increases the mechanical load on the

limb but also increases the perceptual processing requirements of the task, e.g., attention

to the characteristics of the obj ect [126], feedforward adjustments of posture [127] and

increased reliance on feedback to correct movement errors [70]. It is likely that these

tasks are more difficult and recover later than movements without functional endpoint.

This appears to be the expectation of Fugl-Meyer who found a strong correlation between

the FMA-UE score and ADL performance [15]. However, evidence suggests that the

progression from reduced body function impairment (i.e., improved intra-limb

coordination) to recovery of activity performance is neither linear nor clearly understood

[128].

The association between motor function impairment assessed by the FMA-UE and

functional assessments of activities of daily living is both supported and debated in the

literature. For example, both Filiatrault et al. (1991) [21] and Platz et al. (2005) [56] have

shown a low correlation (Spearman r = 0.04 and 0.08) between the FMA-UE score and

the score on the Barthel Index [129], an assessment of ADL performance. Francisco et al.

(1998) [39] found that four individuals with stroke, all of whom exhibited palpable wrist

extension at baseline, demonstrated concurrent gains in FMA-UE and Functional

Independence Measure (now known as the FIMTM) motor scores following inpatient









rehabilitation with usual care augmented by wrist EMG electrical stimulation. In contrast,

Winstein et al. (2004) [130] showed no relationship between gains on any impairment

measure (e.g., the FMA-UE) and changes in the FIMTM after a two week intervention.

Shelton (2001) [131] showed that admission motor impairment (measured by the FMA-

UJE) predicted discharge disability (measured by the FIMTM), however the researchers

note that much of the gain in FIMTM score came as a result of increased use of the less

affected limb to perform the tasks.

Shelton's observation is consistent with Winstein's argument [59, 130] that the

FIMTM like other global assessments of ADL such as the Barthel Index, are not sensitive

to impairment level changes in UE motor ability. This is because neither the FIMTM nor

Barthel directly assess the function of the hemiparetic arm. A client is able to use

compensatory one-handed strategies to accomplish tasks with the unaffected limb and

still receive a high score on these assessments. The association between the FMA-UE

score and scores of functionally oriented assessments where the assessment requires use

of the hemiparetic arm is stronger. The FMA-UE was highly correlated (Spearman r =

0.92) with scores on the Action Research Arm Test [56], the Wolf Motor Function Test

(Spearman r = 0.57 0.68) [26], the Arm Motor Ability Test (Spearman r = 0.92) [97]

and the Lindmark (1988) [89] scale. Each of these assessments requires the patient to use

the hemiparetic arm for functional reaching or grasping tasks.

Mercier et al. (2004) [132] notes that it is possible that the relationship between

body function impairment and functional may not be linear across different levels of

ability. For example, she suggests that a task requiring grip strength (e.g., lifting a pitcher

filled with water) will be performed at a lower stage of the recovery process than a task









requiring a precision grip (e.g., picking up a small obj ect from a table). Winstein (2004)

[130] found that participants in the less severe group (stratified by Orpington Prognostic

Scale score [133]) showed greater improvements in the Functional Test for the

Hemiparetic UE (an assessment comprised of 17 graded tasks assessing "integrated

function" of the UE) [134] than clients in the more severe group. Feyes et al. (1998)

[135] found that clients with more severe motor impairment demonstrated significant

gains on the FMA-UE but the results did not "generalize" to concurrent changes in the

Action Research Arm Test or Barthel Index.

It is possible that individuals with less UE motor ability do not have enough motor

ability to perform even the simplest functional tasks. Chae et al. (2003) [97] found that

individuals with higher scores on the FMA-UE hand subscale demonstrated a strong

correlation with the overall AMAT score. Chae suggests that these individuals most

likely had enough UE motor ability that enabled them to successfully complete AMAT

items (e.g., cutting meat, dialing a telephone). In contrast individuals with less UE motor

ability (a lower FMA-UE score) likely did not have enough motor capacity to accomplish

the more functional AMAT items. Pang et al. (2006) [42] found a similar relationship

between participant ability and functional performance. In this study, more severely

impaired clients showed significant pre to post intervention changes in FMA-UE scores

(2.5 f 2.3 points), but no change in WMFT score (-0.3 f 0.5 seconds). Moderately

impaired clients showed concurrent changes in both assessments (FMA-UE 7.4 f 3.8

points and WMFT -1.5 f 6.2 seconds). The least severely impaired clients demonstrated

little increase in FMA-UE score (7.4 f 3.8 points) but the greatest gains on the WMFT










test (-6.6 +1 6.4 seconds). In general, the evidence suggests that individuals with greater

ability will show changes in both domains, less able people will not.

Clinical significance of the FMA-UE score

The importance of the studies reviewed above is that the FMA-UE score is difficult

to interpret with regards to its functional relevance. That is, although individuals with

stroke may recover the ability to "coordinate single and multiple joint movements" as

evidence by improved FMA-UE score, the functional meaning of this improvement is not

always clear.

Issues of clinically significant or clinically meaningful changes in assessment

scores have become paramount in rehabilitation research [61, 136, 137]. The relationship

between "statistically significant" research results, and "clinically significant" client-

oriented functional improvement is not always apparent [136]. Bonnifer and colleagues

(2005) articulated this concern;

Although this study showed significant increases in FMA-UE (scores) after the
constraint induced movement therapy (CIMT) intervention, these findings do not
suggest that a 3-week CIMT program restores motor ability to pre-stroke levels. It
is not clear how increases in scores on any of these measures translate to real-world
functional abilities. For instance, a peak score of 33 on the FMA-UE score does not
suggest that a person can perform functional tasks using the involved upper
extremity. On the contrary, the arm is still quite impaired and most likely used as
an assist at most for some activities. [51]

Conceptually, there are two approaches for determining what constitutes

"significant" change: distribution and anchor (or criterion) based approaches [61, 137,

13 8]. "Di stribution-based" methods define an important change score as one that

exceeds a pre-defined level of error. These techniques used to calculate significant

change include minimal detectable change, the minimal important difference, effect size

and/or one of numerous responsiveness statistics. The drawback of all of the se










approaches is that the value obtained in the analysis is not a Eixed property of an

assessment, instead varying with the setting of the study and characteristics of the

sample. Different values of "significance" might be obtained in different groups of

patients based on their initial level of impairment. Ceiling or floor effects can mask

change and effect the calculation of a responsiveness or effect size value. Moreover,

these methods require a judgment of what is important change. But, who should be the

judge, client, therapist, physician, payer or test developer? Therefore it is unlikely that

there exists a single score to define significantt" change across all samples and contexts.

In contrast, "anchor-based" methods relate the change score to a functionally

relevant criterion pre-determined by client or clinician. This criterion "anchors" the

meaning (i.e., significance) of the change. Both approaches have been applied to

interpretation of FMA-UE change scores [61].

According to Van der Lee et. al. (1999) [47] the literature contains no criteria for

interpreting a FMA-UE change score with regards to its "minimal clinically important

difference (MCID)". Van der Lee et al. (2001) [24] determined that a 10% change in the

overall FMA-UE score made measurement error unlikely thus allowing for the

conclusion that real change occurred. Using Van der Lee' s criteria, McCombe-Waller

and Whitall (2005) [139] interpreted their 8% gain in FMA-UE score as "approaching

clinical relevance". Based on an effect size calculated in a pilot study, Feyes et al. (1998)

[135] determined that a 10% change in the overall FMA-UE score was "clinically

relevant" meaning that participants moved from a more impaired to a less impaired stage

of recovery. Pang et al. (2006) [42] explored the clinical significance of a FMA-UE

change score using external criterion. In this study the researchers administered a










satisfaction survey to participants in a community based exercise program. Clients in this

intervention showed a 0.71 0.81 effect size for pre to post test FMA-UE score change,

and concurrently reported increased ability to perform simple functional tasks such as

"now able to switch on/off lights". Pang notes that the intervention related changes

appear to be clinically significant from the patient's perspective.

Summary

In the section above I have identified three concerns with the FMA-UE. First, it is

possible that some items, specifically the reflex items, are not consistent with the

construct represented by the other items (i.e., voluntary UE motor ability). Second, many

assumptions have been proposed regarding the meaning of the FMA-UE score. These

assumptions have gone untested. Finally, the relationship between the FMA-UE score

and performance of functional "reaching, grasping, lifting" activities with the more

affected UE is not obvious.

Item Response Theory offers a method to test specific hypotheses addressing each

of these concerns. The following section will review Item Response Theory in general,

the Rasch measurement model specifically, to establish the background for using this

measurement framework to explore these areas.

The Rasch Measurement Model

Measurement

Measurement is defined as "determining the amount or quantity of something".

Mass, length, and time are considered fundamental quantities of measurement in the

physical sciences [140] and are measured with devices accepted as "objective"

measurement tools. An obj ective measurement remains constant and unchanging across










the persons measured and is described in a unit that maintains its size as the unit is

repeated [141].

Ideally the process of measuring a client' s UE motor function should be similar to

the process of measuring length, temperature or distance. That is, UE motor function

should be quantifiable using an assessment tool that is structured to measure UE motor

skill in the same way that a ruler is structured to measure length or that a thermometer is

structured to measure temperature. Bond and Fox argue that in order for a clinical

assessment to obj ectively measure a skill, e.g., UE motor function, the assessment tool

should display markings (e.g., assessment items) that clearly describe "less" and "more"

of the skill. The order of the markings should represent the acquisition of the skill in a

pattern consistent with its recovery or development [142]. Also, an "obj ective"

measurement of UE motor function should connect the numbers produced by an

assessment with its content [143].

Item Response Theory

The increasing need for meaningful and psychometrically sound rehabilitation

assessments has led many to apply modern measurement theoretical statistical methods

(IRT) to the construction, re-design, and scoring of clinical assessments. IRT

measurement models are comprised of a family of mathematical models [53]. Each

model estimates an item difficulty parameter. The two-parameter logistic models (2PL)

also estimate an item discrimination parameter, and the three-parameter logistic models

(3PL) estimate item discrimination and guessing parameters [144]. The Rasch model is

usually considered a one-parameter IRT model, although the technicalities of this are the

subj ect of ongoing debates in the literature.









IRT, as its name implies allows for an item-oriented rather than test-oriented

analysis of an assessment' s measurement properties. The appeal of IRT is that person-

ability and item-difficulty are calculated on the same measurement scale which means

that a score can be directly related to the probability of item responses [53].

Potential Advantages for Using the Rasch Model to Explore the FMA-UE
Measurement Properties

Rasch analysis has been widely utilized to construct and/or refine clinical health

care assessment tools [for example [145-152]]. The model is a powerful framework with

which to address the measurement properties of the FMA-UE: specifically to test (1) the

validity of the FMA-UE item content, (2) the reproducibility of its item structure, and (3)

functional interpretation of its score.

Validity of the item content

Using Rasch analysis an item-by-item analysis is conducted to test whether each

item contributes to the measurement of UE motor ability. This is accomplished by

examining the dimensionality of item set and the item difficulty hierarchy.

In order to create a legitimate measure, all items must contribute to the same

construct, that is, the assessment must be unidimensional [143]. Unidimensionality

means that a single trait or skill accounts for performance of assessment items [153, 154].

Unidimensionality should remain invariant each time the assessment is used because

without unidimensionality it is not clear what construct the items define or what the score

means [153]. All IRT models assume that the data undergoing analysis are

unidimensional [53].

There is no single satisfactory method to assess the unidimensionality of an item set

and so researchers often use a variety of techniques to assess it [154-156]. Principal










Components Analysis (PCA), a classical test theory statistical method, is often partnered

with Rasch analysis for this purpose. The intent of PCA is to reorganize multivariate data

into a limited number of components (factors or dimensions) so that each component

captures a substantial amount of the overall variance within the dataset [157]. PCA can

be applied to either the original raw score data or the Rasch-derived residuals [154].

PCA produces an index of variance statistic for each factor, an eigenvalue, to

describe how well items measure a single construct. Techniques for interpreting the

eigenvalue(s) vary. The Kaiser rule [157] is typically employed as the criteria for

determining the number of factors to retain. According to the Kaiser rule, factors with

eigenvalues > 1.0 are retained, and each is thought to represent a distinct construct.

Cattell's scree test is also used to interpret eigenvalues [157, 158]. The scree plot is a

graph of all eigenvalues in their decreasing order. The plot looks like the side of a

mountain, and "scree" refers to the debris fallen from a mountain and lying at its base.

According to the scree test, factors are retained if their data points occur prior to the point

at which the "mountain" ends and the "debris" begins. Eigenvalues of "scree" are

considered trivial because they are thought to detect "noise" (e.g., measurement error)

rather than actual multidimensionality. Only points that markedly deviate from a linear

trend are valuable to the analysis [157]. McHorney and Cohen (2000) [159] suggest that

if greater than 20% of the variance is accounted for by the first factor in a PCA,

unidimensionality is established. However, Cook et al. (2003) [160] argued that this often

cited criterion is a "perpetuation of an incorrect interpretation" of an earlier work

concerning the stability of item parameter estimates in multidimensional tests rather than

suggesting a criterion for establishing unidimensionality. Finlayson and Mallinson (2005)









[161] suggest that a factor must contribute a minimum of 10% of explanatory variance to

be considered a unique factor in modeling the data, while Haley and colleagues [162]

suggest that meaningful factors are those that each account for 5% of the variance.

Parallel analysis provides an alternative, empirically derived criterion [163]. In parallel

analysis dimensionality is determined by retaining only the number of eigenvalues from

the real data that exceed the eigenvalues estimated from a parallel data set of the same

size that is randomly simulated from a Rasch model [162].

Once the number of factors has been identified the meaning of the factors must be

determined. PCA typically involves a second step, factor rotation. Factor rotation is a

mathematical technique intended to re-express factors in a more simple structure to

facilitate their interpretation. A factor is interpreted (i.e., subj ectively labeled as per the

construct it represents) by determining what trait is shared among the items that load

strongly onto the factor [164].

An application of PCA to the Rasch-derived standardized residuals is also intended

to elucidate potential multidimensionality in the data. Residuals are the data not

explained by the Rasch model. If the data are unidimensional there should be no

dimensionality remaining in the residuals once the Rasch model has extracted the

dominant factor [154]. The presence of distinct factors within the data is determined by

visually inspecting the factor structure plot produced by the analysis [165]. This diagram

cross plots the standardized residual loading on each factor against the item difficulty

calibration. If the data is unidimensional the graph should show a random distribution of

the items. Multidimensionality is evident when items cluster together in "obvious"

groupings [166].









Smith (2002) [153] acknowledges that defining unidimensionality as the existence

of one (and only one) skill accounting for performance on assessment items is likely too

strict of a definition. Smith recognizes the likelihood of many factors (e.g., cognitive,

psychological or physical) influencing test-takers' responses during assessment. He states

that "unidimensionality will hold" as long as the same factors affect the performance on

each item consistently throughout an assessment. Muraki and colleagues (2000) [167]

agree, and state that it is difficult to satisfy unidimensionality when IRT models are

applied to performance assessments, because performance assessments, by their very

nature, require examinees to complete tasks that demonstrate ability to apply skills to a

"real life" situations. These situations are complex, thus it is likely that the data is

influenced by "construct-irrelevant" variance such as environmental effects, practice

effects and examinee fatigue. Murakai argues that unidimensionality is violated to some

extent in every dataset, and even more in data from performance assessments. The

authors suggest that as long as the violations are not "extreme" IRT analysis appears to

be robust.

One way to interpret the apparent presence of more than one factor in data was

exemplified in a recent study by Hart and colleagues (2006) [168]. In this study the

researchers used factor analysis to examine the dimensionality of the Moss Attention

Rating Scale (MARS). The analysis suggested the presence of 3 factors in the data. The

authors interpret the results as meaning the instrument measures a single overarching

construct, attention, but is comprised of 3 inter-correlated sub-dimensions; initiation;

restlessness/di stractibility; and sustained/consi stent attention. The authors argue that the









sub-dimensions are not unique constructs, but reflect different ways of expressing the

overarching construct.

Rasch analysis takes the analysis of unidimensionality one-step farther in the sense

that it tests the extent to which persons have responded to assessment items in an

expected manner. Using Rasch analysis, the unidimensionality of an item is evaluated

with infit and outfit statistics. Under the Rasch model, each person with a given level of

UJE motor ability is expected to do well (score higher) on relatively easy tasks, and have

more difficulty (progressively lower scores) on tasks of increasing difficulty [142]. The

model expectations are assessed through item goodness-of-fit statistics. Fit statistics are

reported as the mean square (MnSq) of the item residuals, where a residual is the

difference between a person's actual response to an assessment item and the response

expected by the Rasch model. Consequently the MnSq statistic has an ideal value of 1.0

with departures from this value indicating potential departures from unidimensionality.

Fit statistics are also reported in a standardized form (ZSTD), which is the transformation

of the mean square of the residuals to a statistic with a z-distribution [153]. Two types of

fit statistics are reported. The infit statistic is sensitive to items that depart from model

expectations when the item difficulty is of a similar level as person ability. The outfit

statistic is sensitive to items that depart from model expectations when the item is much

harder or easier than a person's ability level (i.e., outliers). The acceptable criteria for

unidimensionality depends on the intended purpose of the measure and the degree of

rigor desired. For clinical observations, Wright and Linacre [169] suggest reasonable

ranges of MnSq fit values between 0.5 and 1.7 associated with standardized z-values

(ZSTD) less than 2.0 for clinical observations. High values indicate that scores are









variant or erratic, suggesting that an item belongs to a construct that is different from that

represented by the other items of the instrument. High values could also indicate that the

item is being inaccurately scored. A low MnSq value suggests that an item is failing to

discriminate individuals with different levels of ability (i.e., with different amounts of

motor ability) or that an item is redundant (i.e., other items represent similar motor

challenge). Items with high MnSq values represent a greater threat to construct validity.

Item difficulty hierarchy

An assessment intends to measure a skill. Rasch analysis paints a picture of the

skill by arranging the assessment items along a hierarchical continuum (an "item map")

from "least difficult" to "most difficult". The item difficulty hierarchy is therefore a

visual representation of the abstract skill the assessment is intended to measure. Ryall et

al. (2003) [170] argues that examining the agreement between the a-priori predicted

hierarchical ordering of items and the empirical item order is one way to test the validity

of the theoretical construct underlying the assessment. The empirical disordering of one

or more items may indicate that the item(s) contains) flaws, or that the theoretical

construct governing item order is "deficient" [171].

Rasch analysis places person ability and item difficulty on the same linear

continuum measured in "logits". A logit is the natural logarithm of the odds of a person

being able to perform a particular task to the probability of being able to perform the task

[172]. Logits of greater magnitude represent increasing item difficulty (and person

ability). Because items and people are measured on the same scale, person ability scores

reflect what a person can or cannot do on the basis of item content.

The item hierarchy affords a method to clearly interpret the assessment score which

is beneficial for measuring person ability and informing theory [173]. Because the items









are ordered according to their difficulty level, the expected "next step" of the client' s

progress is clearly visible. The Gross Motor Function Test (a test of motor development)

item map presented in Russell et al. (2000) [174] provides an example of how the item

map connects a score to a description of behavior. In this example, a GMFM score of

"80" is located towards the higher end of the developmental continuum. Because

developmental behaviors (i.e., assessment items) are also positioned on the continuum,

the score of "80" marks the location at which the child is able to accomplish the

behaviors below that level (i.e., easier items) but is not yet capable of accomplishing the

behaviors above that level (i.e., harder items). Similar item maps (also termed

"keyforms") are used for scoring the Pediatric Evaluation and Disability Inventory [175]

and the ABLEHAND [176] questionnaire.

The item hierarchy is useful to ascertain whether an assessment score is interpreted

in the way expected. White and Velozo (2002) [145] applied Rasch analysis to the

Owestry Low Back Pain Disability Questionnaire to determine if the often used Owestry

classification scheme was empirically supported. By matching assessment scores to the

item hierarchy, the researchers learned that the Owestry classification categories did not

describe actual client behaviors in the way intended by the classification taxonomy. The

researchers suggest re-categorization of the response scale to improve the meaningfulness

of the score.

Sabari and colleagues (2005) [177] applied Rasch analysis to the Motor

Assessment Scale [178]. The researchers found that the Rasch-derived item hierarchy

both supported and challenged the underlying conceptual foundations of the assessment.

For example, the item hierarchy revealed that overhead reach was less difficult in supine










than while standing, a finding consistent with contemporary motor control principles that

movement difficulty is governed by external mechanical factors, e.g., gravitational

torque. The item hierarchy also showed discrete, proximal hand movements were of

similar difficulty as some functional activities requiring simultaneous control over

multiple movements at a variety of joints, a finding that challenged theoretical

expectations. Chiu and colleagues (2006) [179] applied Rasch analysis to the Dynamic

Gait Index. The hierarchical ordering of locomotor behaviors was consistent with clinical

expectations. Penta et al. (1998) [176] utilized Rasch analysis to develop the

ABILHAND questionnaire. The item hierarchy "defined" manual ability, a trait

previously not well explored in motor control theory. For example, the most difficult

items elucidated characteristic behaviors (strength, mental effort, dexterity) that added to

an understanding of the construct.

For the construct underlying the FMA-UE, we might expect flexor synergy items to

represent easy items along the continuum of UE motor ability, while "hand" items which

involve more intricate prehension patterns and isolated j oint movements to represent

more challenging items across the continuum. Comparing the Rasch-derived item

hierarchy to Fugl-Meyer' s expected course of recovery will provide an opportunity to test

traditional theoretical expectations of recovery.

Reproducibility of the item structure

Rehabilitation researchers are often interested in whether an intervention is

effective. To this end, researchers choose an appropriate assessment to score an outcome

of interest and compare scores between and/or within groups. Regardless of the statistic

used to assess effectiveness (e.g., comparison of group means, correlation test, etc.) it is

implicitly assumed that the measurement properties of the assessment are identical each









time the assessment is used. Gluck (2001) [180] points out that this often untested

assumption may negatively impact the reliability of an assessment.

Bingenheimer et al. (2005) [181] argues that measurement equivalence is necessary

for valid interpretation of assessment scores. Specifically, an instrument should display

item and scalar equivalence meaning that an assessment's items should assess the same

trait each time it is given (item equivalence), and identical scores should always have the

same interpretation (scalar equivalence). The items of an assessment are the operational

definition of the trait the assessment is intended to measure, and the score quantifies how

much of the trait the test-taker possesses. If an item varies with regards to its meaning or

the score varies with regards to its interpretation, the assessment is not an accurate

measure of the trait. Bingenheimer argues that without convincing evidence of item and

scalar equivalence, there is no basis for comparing scores on an assessment between or

within groups.

Differential Item Functioning (DIF) is a statistical procedure in which the items of

a test are examined, one at a time, to ascertain whether test-item characteristics are

consistent irrespective of test-taker characteristics. The presence of DIF means that a test

item measures a unique feature (trait) of one group of people when compared to another

group of people, or that the test item was not administered correctly [182].

There are various approaches for examining DIF (see [183] for review), the most

commonly used is the Mantel-Haenszel (MH) procedure. MH begins with the assumption

that the ratio of correct to incorrect responses is the same between the groups for each

item. To test this hypothesis, test-takers are split into two groups (e.g., time 1 and time 2)

stratified by ability level (test score). Item difficulties between identical strata of each










group are compared, item-by-item, using the odds ratio statistic. An odds ratio value of 1

means that the item displays no DIF. Based on the magnitude of the value's deviation

from 1, items are rated as having negligible, intermediate or large DIF. Proponents of the

IVH procedure support excluding items showing large DIF, while leaving to clinical

judgment decisions regarding the inclusion or exclusion of items with intermediate DIF.

Rasch analysis employs a relatively straightforward DIF procedure in which the

item difficulty parameters between two groups (or two time points) are directly

compared. This procedure is similar to the above IVH procedure, but is independent of

sample-dependent person ability (i.e., individual raw scores are not used in the

mathematical calculation [184]). The Rasch procedure has reliably detected DIF with

small sample sizes (N < 200) [183].

In the healthcare literature DIF analyses have been useful to study item equivalence

(i.e., to detect potential test item bias) among persons of different cultures and diagnostic

groups. For example, using DIF analyses, Azocar and colleagues (2003) [185] learned

that Latino-Americans were more likely to receive a higher score on some items of the

Beck Depression Inventory than non Latino-Americans, potentially flagging a distinct

Latino cultural interpretation of the items. Azocar recommended a reconsideration of the

assessment' s Spanish translation in order to facilitate cross-cultural comparisons of

assessment scores. Tennant and colleagues (2004) [186] found that certain Functional

Independence Measure self-care items displayed DIF. The researchers learned that

culturally unique ways of bathing and grooming influenced the way the items were

administered among European countries. To facilitate cross-country comparisons of

scores, the assessment scoring procedure was adjusted for those items. Roorda and









colleagues (2004) [187] found that the hierarchical item ordering of a stair climbing

assessment was different for amputees versus non-amputees. The researchers recommend

using a shortened assessment, with DIF items removed, if the intention is to compare

scores between the two diagnostic groups. Dallmeij er and colleagues (2005) detected DIF

in the Functional Independence Measure [188]. When comparing Functional

Independence Measure scores among individuals with various neurological diagnoses

(stroke, multiple Sclerosis, and TBI) DIF was detected in 7of 11 motor items and 4 of 5

cognitive items. For example, persons with MS found the item "dressing lower body"

more difficult than did persons with stroke, and persons with stroke found the item

"problem solving" much more difficult than did persons with MS. The authors caution

that FIM scores cannot be compared across these patient groups.

Instead of testing the measurement equivalence of an assessment between cultures

or diagnostic groups, it is also useful to employ DIF analyses to compare the

measurement equivalence of an assessment between testing occasions within a single

group. The use of DIF analyses for this purpose has precedence in the literature. Tesio et

al. (2003) [182] unexpectedly detected DIF among several FIM mobility items (transfer

to a chair, locomotion, and stairs) when admission scores were compared to discharge

scores. The items were "easy" for a group of orthopedic patients with mobility

restrictions at admission (e.g., no lower extremity weightbearing due to orthopedic

precautions). However, the items became "more difficult" for this group at discharge

after the mobility restrictions were removed. This counterintuitive result led the staff at

this facility to examine the manner in which the assessment was scored. They learned that

the admission score was based on a prediction of the patient' s expected outcome, while









the discharge score was based on actual observation of patient performance. In essence,

the assessment was measuring a different construct at patient admission (e.g., ability of

the staff to predict patient status) versus patient discharge.

Finlayson and Mallinson (2005) [161] applied DIF analysis to the primary

functional outcome measure used in a large longitudinal study of aging and ADL

performance in various healthcare settings. Rasch analysis of a 24-item ADL assessment

was conducted at three time points over the 13-year study. Although DIF was detected

among 4 items, the researchers were unable to discover a specific reason for it. Because

statistically significantly differences in item difficulty do not always translate into

practical differences in person ability scores [189], the researchers measured ADL ability

using an assessment in which the 4 unstable items were removed and compared the

results to measurement using the full assessment. The analysis suggested that DIF had

only minimal impact on the Rasch-ability measures over time or across settings, leading

the researchers to continue using the unadjusted entire assessment.

The presence of DIF can have significant impact on the validity of an assessment

for measuring person ability. Linacre' s (1994) [190] study provided a classic example of

the impact of item instability on the measurement of patient change. Using Rasch

analyses, Linacre found that when the 13 motor and 5 cognitive items of the Functional

Independence Measure were combined into a single 18-item assessment, the admission

and discharge item difficulty calibrations of the items were not equivalent. He suggests

that the presence of variant item difficulty measures indicated that a single pattern of

disability could not be identified when comparing patient admission scores to discharge

scores. That is, the meaning of the score was dependent on the time point (admission









versus discharge) at which the assessment was administered. Linacre demonstrated that

the statistical validity of the scale was improved if the motor and cognitive items were

separated into sub-scales.

Functional interpretation of the FMA-UE score

As previously noted, one of the primary problems facing uses of the FMA-UE is

the functional interpretation of its score. IRT provides the theory and method for linking

the FMA-UE to a more functionally oriented assessment [143, 191, 192]. Linking is a

method that re-calibrates separate assessments onto a common measurement scale. The

theory behind this process has been termed "scale free measurement" [143]. Scale free

measurement means that what one measures is independent of the instrument used to

make the measurement. Applied to post-stroke UE rehabilitation, scale-free measurement

theorizes that an abstract trait (e.g., UE motor ability) can be measured by an "infinite"

number of items representing the trait. Any assessment used to measure this trait is

simply a subsample of this item pool.

Item pools (termed an "item bank"[193]) are the foundation of Computerized

Adaptive Testing. This testing procedure is considered "adaptive" because different test-

takers answer different sets of questions depending on their level of skill with regards to

the content (construct) being tested. The computer selects question sets from the item

pool, which are items that have been previously validated and calibrated. A person's test

score is derived from responses to the items administered meaning that test takers are not

required to complete the same items to receive comparable scores. Item banking and

computerized adaptive testing is widely used in educational research (the Graduate

Record Exam is a computerized adaptive test) and has been increasingly applied to health

care outcomes research [159, 194].









Items within an item pool are calibrated to a common metric using one of two

processes. The first involves combining the items together for a single analysis that

concurrently calibrates difficulty parameters. A variation of this method is to estimate

difficulty parameters for one set of items first, then use these difficulty estimates to fix

(anchor) this set of items in a combined analysis with the remaining items [193]. Hanson

et al. (2002) [195] suggests that, at least for 2PL analyses, concurrent calibration realizes

less error in item parameter calculation than separate calibrations. Kim and Cohen et al.

(1998) [196] suggest that the two methods yield nearly identical results in situations

where a large percentage of items are common to the participant sample.

Linking two existing assessments is another way to create a common item pool. If

two assessments measure the same underlying trait, then the items of each assessment are

derived from the common item pool and therefore the assessments can be mathematically

linked. Linking establishes the relationship between different assessments, therefore

allowing direct comparison of items and scores [167]. Fisher (1995) [192] linked two

assessments of physical function to a common scale with Rasch analysis. Thirteen motor

FIMTM items and Patient Evaluation and Conference System (PECS) items were co-

calibrated to single scale that allowed a patient' s score on one assessment to be directly

translated to a score on the other assessment. Fisher suggested that the assessment scores

could be reported in "rehabits", a unit of measurement representing the common

measurement scale to which the assessments were linked. Costner and colleagues (2004)

[148] linked five existing widely used ADL and IADL assessments (FIMTM, Minimum

Data Set (MDS), MDS-post acute care, OASIS and Physical Functioning -10) for the

purpose of tracking recovery across the continuum of care. The conceptual basis









underlying the link was that the assessments each contain daily living items sharing a

common requirement for upper limb and hand skills. The researchers concurrently

calibrated the items from the existing assessments, retaining 62 items in a unidimensional

item pool.

Fisher and Eubanks (1997) [197] concurrently calibrated the physical functioning

subscales of the Medical Outcomes Study Short Form 36 physical functioning subscale

(PF-10) and the Louisiana State University Health Status Instruments Physical

Functioning Scale (PFS) to a common metric using Rasch analysis, demonstrating that

the method was feasible. Segal and colleagues (1997) [198] used Rasch methodology to

link the PF-10 and Functional Independence Measure motor items. Six PF-10 and 4

FIMTM items were co-calibrated to form a new assessment for post-stroke outpatients. As

separate assessments floor effects in this population hampered the PF-10 and ceiling

effects hampered the FIMTM. However, the linked assessment demonstrated excellent

measurement properties for measuring long term physical functioning in this population.

The ICF provides a conceptual framework that identifies UE motor behaviors

relevant to a single domain (i.e., body function and/or activity performance) and specifies

a relationship between domains. This operationalizes health care dimensions and

provides a framework for rehabilitation outcome measurement within and across

dimensions [199]. Linking a post-stroke UE motor assessment from the body function

domain to an assessment from the activity domain sharing similar UE motor behaviors is

a way to test whether the domains are distinct or related constructs in post-stroke

recovery. Moreover, linking is a way to directly investigate the functional relevance of









reduced body function impairment because linking connects the assessments to a

common metric.

The Wolf Motor Function Test (WMFT) [26, 200] is a logical choice for linking to

the FMA-UE. Both the FMA-UE and the WMFT measure UE motor ability: the FMA-

UJE is a measure of UE body function impairment, while the WMFT is a measure of UE

activity performance restriction. The WMFT is an assessment of post-stroke UE motor

function [201]. It is intended to measure a "wide range of functional UE tasks" in

individuals with mild to moderate stroke-related UE motor impairment [201]. Its items

are arranged in a sequential order according to j points involved (shoulder to hand) and

level of difficulty (gross movements to fine motor movements) [26]. The first seven

items involve simple limb movements without functional endpoint (e.g., extend elbow to

the side on a table top, lift hand to table), and the remaining eight items involve

performance of common tasks (e.g., lift soda can, flip cards). The WMFT has been

shown to have high interrater reliability (intraclass correlation coefficients, > 0.88),

internal consistency (Cronbach a > 0.86), and test-retest reliability (r > 0.90) [26, 201].

Placing the FMA-UE and WMFT items on the same "ruler" as client ability allows

the individual's Rasch ability score to be directly related to his/her probability of

performing a particular set of movements or tasks. This permits one to describe a

person's motor ability in terms of specific movements or tasks that he can/cannot

perform. Furthermore, changes in score can be directly linked to acquisition of more

difficult movements or tasks. By linking the FMA-UE and WMFT in this manner one

should be able to convert improvement scores into performance of specific UE

movements and tasks (e.g., moving from elbow extension to grasping a soda can) thereby









identifying a critical threshold at which impairment reduction translates to functional

ability. Translating improvements in UE ability from impairment to function is the

foundation of documenting the clinical effectiveness of restorative rehabilitation.

Criticisms of the Rash Measurement Model

Sample size

The purpose of a sample size justification is to predict how many subj ects are

needed to satisfactorily answer the research obj ective. In traditional statistics, a power

analysis guides the researcher in determining the appropriate sample size for the

obj ectives of the study. With IRT analyses sample size guidelines are not as

straightforward. Linacre (1994) [202] suggests that as few as 50 subj ects may be an

appropriate sample size for a valid analysis using the one parameter Rasch model,

however Hambleton (1989) [203] suggests that at least 200 subj ects are necessary for

Rasch analyses. Embretson and Reise (2000) [204] indicate that the two-parameter model

may require 250 to 500 participants while Hambleton indicates that a minimum of 500

subj ects are necessary. Hambleton also suggests a minimum sample size of 1000 subj ects

if the researcher intends to use a three-parameter IRT model.

The purpose of the study affects the sample size recommendation. According to

Reeve and Fayers (2005) [205], if the aim of a study is to examine the properties of an

existing assessment, one does not need a large sample size to describe it clearly.

However if the study purpose is to calibrate items for test development then larger (over

500 subjects) sample sizes are needed. Lai et al. (2003) [206] recommends that test

equating requires a sample size of 400 for the Rasch model and 1,500 for a three-

parameter model.









There are very few empirical studies influencing IRT sample size

recommendations. Holman's (2003) [207] simulation study results suggests 450, 90, and

40 participants are needed in each arm of a randomized trial in order to detect an effect

size of 0.2, 0.5, or 0.8 respectively with a significance of 0.05 and a power of 80% using

a two parameter IRT analysis. In comparison, a t-test powered to detect the same effects

would require 394, 64, or 26 patients per arm of the study. Holman also indicates that the

number of "informative" items in the assessment affects the sample size. That is, more

items with more precise information require lower sample size.

Obviously these sample size guidelines may make using IRT methods prohibitive

for many health care researchers. In Rasch measurement, the researcher seeks to obtain

stable person, item and rating scale estimates. A stable parameter is one that has low

standard error. Linacre indicated that + 0.3 logits is the best that can be expected for most

variables and that the sample size needed to have 99% confidence that no item calibration

is more than 1 logit away from its stable value is 50 [202]. In a Monte Carlo simulation

study, Wang and Chen (2005) [208] found that with sample sizes of 5 20 individuals,

item parameters were unstable at either the higher or lower end of the item difficulty

hierarchy in short tests (10 to 20 items). However, the parameters were stable with longer

tests and the same sample size. The researchers conclude that longer assessments contain

more information and thus stabilize item parameter generation in a small sample size.

Should data fit the model or visa versa?

The Rasch model is criticized because it assumes equal item discrimination. Item

discrimination refers to an item's ability to differentiate between test takers of different

levels of ability. As stated, the two-parameter IRT model contains an item discrimination

parameter while the one parameter Rasch model does not. The Rasch model constrains









the item discrimination value at "1". According to McHorney [209] it is critically

important to examine the item discrimination capabilities of assessment items. It is

possible that equally valid items will discriminate in various ways.

To this end, the Rasch model software calculates an item discrimination index

[148]. However, in contrast to the 2-parameter IRT model, this index does not enter into

the original person scoring. The index describes whether items are close to or depart from

the ideal discrimination value of "1". In other words, the index reveals how consistent

items are with the unidimensional expectations of the Rasch model. Low values indicate

that an item fails to discriminate as well as other items; high values may indicate that the

item is idiosyncratic in some way. Misfitting items are reexamined for redesign if

necessary. The Rasch model uncovers possible anomalies in item design thereby enabling

further investigation of and modifications to the test [210].

The underlying conflict between the Rasch model and other IRT models has to do

with views of how data provide information [211]. In short: should a measurement

model explain data, or should a measurement model disclose information perhaps hidden

in data? Andrich (2004) [211] frames this controversy using the language of

mathematician-philosopher Thomas Kuhn as a clash of "incompatible paradigms".

Kuhn stated that the purpose of measurement is to disclose anomalies in data that

cannot be explained by the theory underlying the measurement model with which the

data is analyzed. With this approach one is able to ask questions as to why data does not

fit a given theory thereby potentially learning new information that contributes to

advancement of the theory. Kuhn' s view is in stark contrast to traditional measurement









theory that asserts data should fit a model so as to support the theory underlying the

model. This issue is notably evident in controversies surrounding the Rasch model [211].

According to CTT and non-Rasch IRT models, data is best interpreted according to

a mathematical model found to explain the largest amount of variance in the data. Models

are tried and abandoned depending on how well the model explains the data. For

example, a recent article eloquently advocates IRT methods to measure clinical cognitive

change [212]. The researcher describes how various models were tried and discarded

until he settled on a 2-parameter IRT model because it provided a "better fit" than the

other models. Kirsci et al. (2006) [213] employed significance testing of chi-square

statistics as a method for comparing one and two parameter IRT models as best fitting the

data. The point is that the a-priori assumption in the above paradigms was that a

mathematical model should fit the data.

The Rasch model has the inverse assumption. That is, data should be viewed with

the assumption that it either fits or does not fit the model. If it does not fit, then, rather

than abandoning the model in favor of another, the data is examined for "anomalies"

[21 1]. According to Andrich [21 1], the process of examining misfitting data (either items

or persons) "uncovers" information that would have been hidden otherwise. New

information may lead to advances in theory. As Wright states: "Misfitting items can be

redesigned. Misfitting populations can be reassessed" [214].

Conclusion

There exists an interdependent link between theory and measurement. The theory

guides the design of the measurement tool. In turn, the information gathered from the

measurement tool advances the theory. This concept was recently articulated by Dennis

Tate in the John Stanley Memorial Lecture to the American Congress of Rehabilitation










Medicine [215]. He stated "the process by which we measure rehabilitation domains and

test our theories about reality shapes what constitutes rehabilitation knowledge."

As recent advances in neuroscience and rehabilitation science have increased

knowledge about the potential for upper limb motor recovery after stroke, recognition of

the need to adequately measure impairment and the functional implications of

impairment has also increased. Stroke rehabilitation professionals face an urgent need to

assure that assessments used to establish the efficacy and effectiveness of translational

intervention are of the highest quality for measuring client ability, change with

intervention, informing the clinical decision-making process and providing valuable

quantitative information about impairment and recovery.

The FMA-UE is considered the "gold standard" clinical assessment of post-stroke

UE motor impairment, however its item-level measurement properties have not been well

studied with contemporary measurement models, e.g., item response theory. Several

deficiencies potentially exist in FMA-UE item content and structure that could impair its

use as an outcome measure; (1) some items may not be consistent with current motor

control theoretical expectations of post-stroke impairment and recovery, (2) the

reproducibility of the construct "recovery" illustrated by its item hierarchy has not been

established across subj ects of various ability levels, nor within subj ects over time, and (3)

in its current format, the FMA-UE score has no clear functional meaning. The aims of the

subsequent studies address these deficiencies and serve the purpose of exploring and

potentially enhancing the measurement properties of this important and long-standing

stroke rehabilitation measurement tool.










Table 1: The Fugl-Meyer Upper Extremity Assessment
FMA-UE Item
NumberFMA-UE Item Description
Shoul der/Elb ow/F orearm
Item 1 Biceps reflex elicited
Item 2 Triceps reflex elicited
Item 3 Scapular elevation
Item 4 Scapular retraction
Item 5 Shoulder abduction
Item 6 Shoulder external rotation
Item 7 Elbow flexion
Item 8 Forearm supination
Item 9 Shoulder adduction with internal rotation
Item 10 Elbow extension
Item 11 Forearm pronation
Item 12 Hand to lumbar spine
Item 13 Shoulder flexion to 900 degrees with elbow extended
Item 14 Pronation-supination of forearm with elbow at 900
Item 15 Shoulder abduction to 900 with elbow extended
Item 16 Shoulder flexion to 900-1800 with elbow extended
Item 17 Pronation-supination of forearm with elbow extended
Item 18 Normal reflex activity
Wri st
Item 19 Wrist stable with elbow at 900
Item 20 Wrist flexion-extension with elbow at 900
Item 21 Wrist stable with elbow extended and shoulder at 300
Item 22 Wrist flexion-extension with elbow extended and shoulder at 300
Item 23 Wrist circumduction
Hand
Item 24 Finger mass flexion
Item 25 Finger mass extension
Item 26 Hook grasp (MPs extended, PIPs and DIPs flexed)
Item 27 Lateral prehension (thumb adduction to hold paper)
Item 28 Palmar pinch (thumb to index finger to hold pencil)
Item 29 Cylindrical grasp (hold small can)
Item 30 Spherical grasp (hold tennis ball at fingertips)
Coordination/Speed
Item 31 Movement without tremor
Item 32 Movement without dysmetria
Item 33 Movement with normal speed







53


Table 2: Descriptions of the traditional upper extremity limb synergies
Flexor Synergy Extensor Synergy
Shoulder retraction Shoulder adduction with internal rotation
Shoulder elevation Elbow extension
Shoulder abduction to 90 degrees Forearm pronation
Shoulder external rotation
Elbow flexion
Forearm supination















CHAPTER 2
USINTG THE RASCH MEASUREMENT MODEL TO EXAMINE POST-STROKE
UPPER EXTREMITY RECOVERY

The characterization of post-stroke upper extremity (UE) motor recovery has long

been of the focus of rehabilitation clinicians and researchers. An assessment tool that

quantifies UE motor impairment can identify the amount of UE motor recovery, i.e.,

locate a patient along the motor recovery progression. Accurate measurement of UE

motor impairment is important to capture the effects of translational interventions,

explore optimal training parameters of existing interventions, and to predict future UE

motor function. There is a pressing need to ascertain whether assessment tools commonly

used in post-stroke UE motor recovery research are accurately quantifying impairment

and characterizing recovery.

The Fugl-Meyer Upper Extremity Assessment (FMA-UE) [15] is the most widely

used clinical assessment of post-stroke upper extremity motor impairment [16]. It has

excellent inter-rater reliability (Pearson Product-Moment Correlations r = 0.97-0.99 [17])

and test-retest reliability (Intraclass correlation coefficients 0.94 0.99 [18]). It has been

used as the standard from which to establish the validity of other commonly used tests of

upper extremity motor function such as the Wolf Motor Function Test and Ashworth

Scale [20-26]. Researchers consistently use the FMA-UE as a descriptor of functional

status following stroke [4, 20, 27]. To this end, the FMA-UE scores have been used to

stratify research study participants into categories of stroke severity [28], predict long-

term functional participation [29], and to describe arm motor impairment [30-32], arm










movement quality [21], and residual arm motor function [33-35]. Furthermore, the FMA-

UE is the primary criterion for evaluating the success of novel upper extremity

interventions such as rhythmic bilateral movement training [36, 37], neuromuscular

stimulation [38], electromyographic triggered electrical stimulation [39, 40], Botox [41],

home based exercise [30], community based exercise [42], robot aided therapy [43, 44],

virtual reality [45], imagery [46], forced use [47], and modified constraint induced

movement therapy [48-51i]. Because of the importance of this assessment in post-stroke

rehabilitation research, it is critical to examine, and perhaps improve, the quality of its

measurement properties.

The measurement properties of an assessment tool are usually studied with

traditional psychometric methods such as reliability, validity, and responsiveness. The

item response theory (IRT) measurement framework offers another way to examine an

assessment' s measurement properties. Stroke rehabilitation researchers have increasingly

used IRT methods, e.g. Rasch analysis*, to develop new assessments [176] or redesign

existing assessments [177]. Rasch analysis offers the advantage of examining an

instrument at the item level rather than as a whole. With this approach one can explore

item content (i.e., does a specific item contribute to the measurement of an intended

construct?), and item structure (i.e., does the item difficulty hierarchy progress from "less

of to "more of the intended trait?). Although the measurement properties of the FMA-





SSome authors use the terms Item Response Theory (IRT) and Rasch analysis interchangeably. However,
the measurement models are different. The Rasch model requires that items have equal discrimination,
whereas IRT models include a parameter for items to have different discrimination. The merits of each
model are the subject of ongoing debates. For the purposes of this article the simpler of the models, the
Rasch model, is used.









UJE have been extensively studied using traditional methods [15, 17, 18, 20, 24, 54-56],

presently there are no published studies applying IRT approaches to the FMA-UE.

McDowell and Newell (1996) suggest that health care instruments be founded on a

specific theoretical framework [67]. Fugl-Meyer et al.'s (1975) [15] theoretical

framework broadly reflects the observations of Twitchell (1951) [75] and Brunnstrom

(1966) [76] regarding the stages of post-stroke UE sensorimotor recovery. Fugl-Meyer

stated:

The form has been constructed following the hypothesis that the restoration of
motor function in hemiplegic patients follows a definable stepwise course. Thus
for a patient with hemiparalysis, recurrence of reflexes always precedes volitional
motor action. Thereafter through initial dependence on synergies, the active
motion will become successively less dependent upon the primitive reflexes and
reactions and finally complete voluntary motor function with normal muscle
reflexes may be regained (p. 14).

Fugl-Meyer chose items to exemplify the construct of motor recovery including reflex

items and voluntary movement items. Moreover, because the assessment is intended to

measure recovery, Fugl-Meyer arranged the items, from easy to hard to map the recovery

process. It is clear that Fugl-Meyer expected motor recovery to proceed in a reflexive-to-

voluntary, and synergy-to-isolated progression. We may therefore expect the FMA-UE

items to be ordered according to this hypothesis.

Recent advances in movement science challenge the "definable stepwise course of

recovery" upon which the FMA-UE is based. Contemporary views of the central nervous

system suggest that the reflex items measure a different behavior than other items. A

tendon tap reveals the integrity of a relatively simple spinal neural control circuit, while a

goal directed voluntary movement reveals the integrity of a more complex supraspinal

neural network [80]. Furthermore, recent studies of UE motor control suggest that

coordinated arm motion and patterns of muscle recruitment are influenced by task-










specific mechanics [98, 216-218] and environmental demands [100]. For example,

Michaelson et al. (2001) [98] suggested that trunk restraint "normalized" post-stroke

reaching kinematics. Wu and Trombly (2000) [100] found that characteristics of the

task-specific target influenced reaching kinematics in both healthy individuals and

individuals with stroke. While Fugl-Meyer' s recovery model may provide a

neurophysiological description of post-stroke UE motor impairment, task-specific

interactions of neural, biomechanical and contextual factors may also influence the

course of post-stroke UE motor recovery.

The purpose of this study is to investigate the dimensionality and construct validity

of the FMA-UE using Rasch analysis. The aims of this study are to (1) determine if all

items of the FMA-UE contribute to the measurement of a single construct

(dimensionality), and (2) determine if the items are ordered according to Fugl-Meyer' s

expected "stepwise" sequence (construct validity).

Methods

Dimensionality

In order to create a legitimate measure, all items must contribute to the same

construct [143]. Using Rasch analysis, the extent to which items contribute to a

unidimensional construct is evaluated employing infit and outfit statistics. The infit

statistic is most sensitive to ratings on items that are closely matched to subj ects' ability,

the outfit statistic is most sensitive to ratings on items that are much easier or much

harder than subjects' ability [142]. Fit statistics are reported as mean square standardized

residuals (MnSq) produced for each item of the instrument. MnSq represents observed

variance divided by expected variance [172]. Consequently, the desired value of MnSq

for an item is 1.0. The acceptable criteria for unidimensionality depends on the intended










purpose of the measure and the degree of rigor desired. For surveys using ordinal rating

scales, Wright and Linacre [169] suggest reasonable ranges of MnSq fit values between

0.5 and 1.7 associated with standardized Z values (ZSTD) less than 2.0. High values

indicate that scores are variant or erratic, suggesting that an item belongs to a construct

that is different from that represented by the other items of the instrument. High values

could also indicate that the item is being inaccurately scored. A low MnSq value suggests

that an item is failing to discriminate individuals with different levels of ability (i.e., with

different amounts of motor ability) or that an item is redundant (i.e., other items represent

similar motor challenge). In this analysis we focused on items with high MnSq values,

because they represent a greater threat to construct validity.

Fit statistics alone are inadequate to determine unidimensionality [154-156]. A

principal components analysis (PCA) was performed using the SAS Factor procedure

(SAS v 8.2) with oblique rotation. The intent of this PCA was to reorganize the

multivariate data (participants' FMA-UE item ratings) into a limited number of

components (factors) so that each component captures a substantial amount of the overall

variance within the dataset. That is, the purpose of the PCA was to detect "obvious"

factors within the FMA-UE. The PCA was carried out in two steps: an initial analysis to

examine the dimensionality of the assessment, and an oblique rotation to examine if

traditional subscale divisions were empirically supported.

We hypothesized that the assessment items illustrated a single factor, "UE motor

ability". To test our hypothesis we examined PCA eigenvalues and factor loading

statistics. We employed the Kaiser rule [157] as the criteria for retaining principal

components, i.e., we examined only the components with an eigenvalue > 1.0. We










expected one eigenvalue would explain the maj ority of the variance in the data.

Furthermore we expected that the maj ority of FMA-UE assessment items would correlate

with a single principal component as evidenced by factor loadings on the first component

> 0.40 [219].

Fugl-Meyer suggested dividing the assessment into four subscales: arm (items 1 -

18), wrist (items 19 23), hand (items 24 30), and coordination speed (items 31-33)

[15]. Each of Fugl-Meyer' s subscales may represent a unique motor control construct. To

examine if there was empirical evidence for this subscale division, we executed a second

step in the PCA analysis. We performed an oblique rotation (obvarimax) of four principal

components. We chose an oblique rotation because we anticipated inter-factor

correlations, i.e., movements of the hand were expected to correlate with movements of

the arm. Factor rotation is a mathematical technique in which the multi-dimensional axes

of initial PCA factor loadings (the number of axes matches the number of retained

factors) are rotated to model the data. The rotation re-expresses the factors as a less

complex structure to facilitate interpretation [164]. This procedure produces a factor

matrix with factor loading values (equivalent to regression coefficients) indicating the

"weight" given to the item when interpreting the factor [157]. We considered factor

loadings meaningful if the absolute value exceeded 0.40 [219]. We interpreted the

factors (i.e., we labeled the factor as per the construct it represents) by subj ectively

determining what trait was shared among the items loading strongly onto that factor.

Because the purpose of our factor rotation was to explore whether Fugl-Meyer' s

suggested subscale pattern was present in our data, we examined the groups of items

comprising a factor to determine if they matched Fugl-Meyer' s suggested item groupings.









Construct Validity

The foundation of "obj ective" measurement is to connect the numbers produced by

an instrument with its content [141]. Through Rasch measurement, items represent

"difficulty" markings (calibrated in log-equivalent units called "logits") along the

continuum of a construct. For example, for the construct underlying the FMA-UE, we

may expect flexor synergy items to represent easy items along the continuum of motor

ability, while hand items which involve more intricate prehension patterns to represent

more challenging items across the continuum. For purposes of this study, we postulated

that the item hierarchy would illustrate the trait "UE motor ability". Analysis of data from

a cross section of individuals with different abilities post-stroke may provide an initial

description of the hierarchy.

By placing item difficulties and person abilities on the same continuum, Rasch

analysis can be used to match the difficulty of the assessment items to the ability of the

sample tested. In general, the analysis will tell us what items are easy enough for

individuals with "poor" arm motor skills (e.g. individuals who have little arm movement)

and what items are challenging enough for individuals with "good" arm motor skill (e.g.

individuals who have fine-motor coordination). The placement of item difficulty and

person ability on the same continuum also provides information on how specific

individuals are expected to respond to particular items. For example, if a person receives

a logit measure of 6.0 and the item shoulder flexion to 90 degrees nI ithr elbow extended

also receives a logit measure of 6.0, then it would be expected that that person would

have a 50% probability of being able to perform this movement successfully. This

individual would be expected to have a greater than 50% probability of being able to

accomplish easier items such as those representing the "flexor synergy" items.









Furthermore, he/she would be expected have less than 50% probability of being able to

accomplish more difficult types of movement, such as palmer prehension (grasping a

pencil with the pads of the thumb and index finger).

Results

Participants

A secondary analysis was performed on data pooled from two studies. The first

dataset consisted of 100 persons enrolled in a randomized clinical trial of therapeutic

exercise [220]. FMA-UE data was collected prior to the exercise intervention by

therapists trained in standard administration of the assessment. Persons with stroke were

included in this study if they met the following inclusion criteria; "(1) Stroke within 30

to 50 days, (2) ability to ambulate 25 feet independently, (3) mild to moderate stroke

deficits defined by a total Fugl-Meyer score of 27-90 (upper and lower extremities) and

Orpington Prognostic Scale score of 2.0 to 5.2, (4) palpable wrist extension on the

involved side, and (5) Folstein Mini-Mental Status exam score of greater than 16" [220].

The second dataset consisted of 459 individuals enrolled in the Kansas City Stroke Study

[221]. Participants were included if they met the following criteria; (1) stroke onset

within 0 14 days prior to enrollment, (2) 18 years of age or older, (3) ischemic stroke as

diagnosed by physician, (4) lived in the community prior to stroke onset, and (5) able to

participate in baseline testing. Study staff trained in the administration of the FMA-UE

evaluated patients. Pooling of these data resulted in a 559-person dataset. However,

because lesion location may effect neural reorganization and clinically observed patterns

of UE recovery [222], subjects with non-cortical (e.g., brainstem or cerebellum) stroke

were not selected from the pooled dataset. This resulted in a Einal dataset of 512

participants. Characteristics of the 512-person sample are presented in Table 3.









Fit Statistics

Item Fit Statistics from the initial analysis of the entire 33-item FMA-UE are

presented in Table 4. Two items (biceps reflex and triceps reflex), shaded in gray, show

infit values that are beyond the acceptable ranges described by Wright and Linacre

(1994) [169]. These items have infit statistics that are above 1.7 with a standard

deviation greater than 2.0. These items also showed abnormally high outfit statistics. In

addition, the two items showed relatively low score correlations (0.36 and 0.26). Infit and

outfit statistics are within acceptable ranges for all other items.

Principal Components Analysis

The PCA retained 4 factors having eigenvalues > 1.0 (Kaiser rule [157]). See

Table 5. A single component explained 68% of the variance in the data. Three other

factors explained an additional 5%, 4%, and 3% of the variance. Table 6 displays the

factor loadings onto the first principal component. Thirty of 33 items loaded strongly

onto the first component (factor loading values ranging from 0.75 0.91). Three items

(biceps reflex, triceps reflex, normal reflex activityi) had a poor to moderate loadings (r =

0. 14 0.53) with this factor.

Rotated factor loadings for the 4 components with eigenvalues > 1 are presented in

Table 7. Primary loadings (factor loadings with an absolute value > 0.40) for each

component are shaded in gray. The values are sorted into descending order of magnitude

and grouped into columns to facilitate interpretation of the item groupings. Several items

do not load strongly onto any factor. For example, pronation-supination 11 ithr elbow at

90, and ilovement~'ll n without dysmetria do not display any factor loading greater than 0.40.

Five items "cross-load" onto more than one factor. For example, wrist flexion-extension

11 ithr elbow extended and wrist stable 11 ithr elbow extended each load onto factor 2 (0.41









and 0.46) and factor 3 (0.54 and 0.52). Items with no factor loadings and those that cross-

load are grouped near the bottom of the chart and marked with a double asterisk.

Examination of the shared traits among items that load onto each factor may

illustrate the motor behavior represented by each factor. For example the items, elbow

flexion, shoulder abduction, scapular elevation, scapular retraction, shoulder external

rotation, and forearm supination load onto the first factor. In addition, this factor contains

shoulder adduction 11 ithr internal rotation and forearm pronation. These items are

consistent with the traditional description of the post-stroke UE flexor and extensor

synergies as described by Twitchell and Brunnstrom. Therefore this factor may pick up a

"synergy trait" underlying the FMA-UE. The grasp and prehension items load onto the

second factor. Therefore this factor may detect a "hand trait" evident in the data. This

item grouping roughly matches the hand subscale division suggested by Fugl-Meyer.

Items requiring movements away from the body using elbow extension (e.g., shoulder

flexion to 180 in ithr elbow extended, pronation-supination al ithr elbow extended) load

strongly onto the third factor. However, it is difficult to interpret this factor because it

contains five cross-loaded items. Finally, triceps reflex and biceps reflex load onto the

fourth factor, perhaps detecting a "reflex" trait. Although the items located in factors 1

and 2 are roughly similar to Fugl-Meyer's arm and hand subscales, the other item

groupings do not match Fugl-Meyer' s FMA-UE sub scale suggestions.

Although the PCA suggests that four motor control constructs (four eigenvalues >

1) may be present, the maj ority of the variance in the data is explained by a single

principal component. However, poor factor loadings onto the first principal component,

and abnormal fit statistics suggest that the reflex items do not fit with the intended









measurement construct of the assessment; therefore these items were removed from

subsequent Rasch analysis. Fit statistics, the large first eigenvalue and strong principal

component factor loadings support the unidimensionality of the remaining items. Because

the intent of our study is to look at the relationship of items across the entire upper limb

we chose not to subdivide the assessment.

Item fit statistics the revised 30-item FMA-UE (reflex items removed) are

presented in Table 8. All 30 items fit the unidimensional assumptions of the Rasch

model. The revised 30-item FMA-UE showed good internal consistency. The person

reliability index, analogous to coefficient alpha, was 0.96. The instrument divided the

sample into 7 statistically significant strata (Number of Strata = [4(Gp = 4.67) + 1]/3,

where GP= person separation [223]) indicating a good person to item match.

Item Hierarchy

Table 9 presents items in order of decreasing challenge. FMA-UE items at the

bottom of the center column indicate the least challenging items; those at the top

represent most challenging items. Elbow flexion (measure = -1. 76 f 0.1~3 logits) and

shoulder adduction 0I ithr internal rotation (measure = -1.56 f 0.1~2 logits) were the

easiest items for this sample to perform, and wrist circumduction (measure = 1. 67 f0.10O

logits), and hook grasp (measure = 1. 33 f 0.10O logits) were the most difficult items for

this sample to perform. Five of the nine "flexor synergy" and extensorr synergy" items

(items 3 through 1 1) group towards the easy end of the hierarchy, while some flexor and

extensor synergy items span the item difficulty hierarchy. For example, elbow flexion is

the easiest item in the hierarchy (-1.76 + 0.13 logits), scapular elevation is slightly more









difficult (-1.40 f 0.12 logits), and forearm supination is moderately difficult (0.65 f 0.10

logits) .

Movements requiring combinations of shoulder flexion and elbow extension are

much more difficult than movements demanding no shoulder motion with the elbow

positioned at 90 degrees. For example, pronation-supination 11 ithr the elbow at 90

degrees, wrist flexion-extension 11 ithr the elbow at 90 degrees, and wrist stable 11 ithr the

elbow at 90 degrees are among the moderately difficult items. .\lrlA Allr flexion to 180 O

11 ithr the elbow extended, wrist flexion-extension 11 ithr the elbow extended, and pronation-

supination 11 ithr the elbow extended are among the most difficult items.

In addition, although the maj ority of the "hand" items (items 24 through 30) group

at the "difficult" end of the hierarchy, some of these items also span the item difficulty

hierarchy. For example, hook grasp (1.33 f 0. 10 logits), spherical grasp (1.25 f 0. 10

logits), and lateral prehension (1.08 f 0.10 logits) are among the most difficult items,

while finger mass flexion (-1.44 f 0. 12 logits) is among the easiest items and palmar

prehension (0.06 f 0. 10 logits) is calibrated near the middle of the scale. Surprisingly,

finger mass extension (-1.25 f 0.12 logits) calibrates as an easy item for this sample to

perform (see discussion).

Keyforms

While the above results show the overall hierarchical pattern for the sample under

study, a critical question is the consistency of this pattern across individual subjects.

Similar to the fit statistics that are produced for items are fit statistics for persons. Ninety-

eight percent of the sample showed acceptable infit statistics (MnSq <1.7 and ZStd <2.0),

suggesting that individual subj ects in the sample are responding similarly to the item









difficulty hierarchy. Figure 1 demonstrates the item difficulty hierarchy relative to the

scoring pattern of three clients representing different UE abilities. Items are listed in

terms of increasing difficulty level (i.e., elbow flexion being the easiest item at the bottom

of each panel, wrist circumduction representing the hardest item at the top of each panel).

The 3-point rating scales for each item are presented to the left of each panel. As the item

difficulty increases, the rating scale stair steps to the right. The bottom scale represents

item difficulty. For example, for the item elbow flexion, a rating of "1" is at a difficulty

of approximately -2.0 logits. This bottom scale also reflects person ability (i.e., the solid

vertical line crossing the scale represents the individual's person ability measure, the

dotted vertical lines represent the 95% confidence interval around this ability measure).

Circled numbers represent the actual ratings for the individual. The triangle shapes

represent ratings that are statistically erratic (p=0.05).

A person of low ability (-1.25 logits) is represented on the top left panel (Person

A), of moderate ability (0.20 logits) is represented on the top right panel (Person B), and

of high ability (3.88 logits) (Person C) is represented on the bottom left panel. The

individual of low ability (Person A) has a tendency to receive ratings of "2" (faultless

performance) and "1" (partial performance) on the easy items (elbow flexion, shoulder

adduction al ithr internal rotation, finger mass flexion), and receive ratings of "O" (unable

to perform) on the most difficult items (shoulder flexion to 90 degrees tI ithr elbow

extended, hook grasp, and wrist circumduction). The individual of moderate ability

(Person B) has a tendency to receive ratings of "2" (faultless performance) on the easy

items, and receive ratings of "1" (partial performance) on the moderately difficult items

(scapular retraction, shoulder external rotation), and receives ratings of "1" or "O"










(unable to perform) on the most difficult items. The individual of high ability (Person C)

has a tendency to receive ratings of "2" (faultless performance) on the easy items, and the

same the moderately difficult items. However this person receives ratings of "1" (partial

performance) on the most difficult items (shoulder flexion to 180 degrees nI ithr elbow

extended, wrist circumduction). In general, while these individuals are of different

abilities their scoring pattern is the same; all three score higher on easier items and lower

on harder items. The pattern retains its structure when measuring a low ability, moderate

ability, and high ability person.

There are exceptions to this pattern as evidenced by the ratings denoted with a

triangle. Person A with lower ability receives a higher than expected rating of "2" on a

moderately challenging item (scapular retraction). Person C of high ability receives a

lower than expected rating of "1" on a moderately difficult item (lateral prehension).

Overall, 12 out of 512 people (2%) showed erratic scores based on infit statistics. Among

these 12 individuals, only 2-5 items were statistically erratic.

Discussion

The purpose of this study was to investigate the dimensionality and construct

validity of the FMA-UE. Using item response theory statistical techniques we determined

that the items of the FMA-UE contribute to the measurement of multiple constructs and

demonstrate a difficulty order that does not reflect with the item order proposed by Fugl-

Meyer. Furthermore, we determined that this item difficulty order remains consistent,

independent of person ability. The above results will be discussed relative to traditional

and contemporary motor control theories.









Dimensionality

The initial finding from both the Rasch Analysis and the more traditional PCA

revealed that the reflex items empirically appear to be disconnected from the remainder

of the FMA-UE. While, according to Fugl-Meyer' s suggested "stepwise" course of

recovery, it would be expected that persons of little to no UE motor ability would be the

only individuals without reflex behavior, the high infit and outfit statistics indicate that at

least some individuals of high ability also show this behavioral response. Multiple

statistical findings (i.e. low PCA weights, low correlations and high fit statistics) suggest

that the reflex items are contributing little explanatory variance for the FMA-UE and

support our decision to remove these items from further analysis.

The three reflex items are exceptional to the assessment in two ways. First, the

items evaluate a passive response rather than active motion. Specifically, the behavior of

the phasic stretch reflex is assessed through tendon taps of the biceps (items 1 and 18),

triceps (items 2 and 18) or finger flexors (item 18). Secondly, the rating scales for the

reflex items are different than the rest of the assessment. For example, items 1 and 2

utilize a two-point scale (0 = No reflex activity elicited, and 2 = Reflex activity elicited),

and item 18 utilizes a uniquely defined three-point scale (0 = At least two of three pha~sic

reflexes are hyperactive, or 1 = One reflex is hyperactive or two reflexes are lively or 2 =

No more than one reflex is lively and none are hyperactive). These procedural and

scoring differences from that of the maj ority of the assessment may cause the items to

misfit and correlate as a separate factor.

While the above statistical findings may be a product of administration procedures

(passive vs. active) or the rating scale structure, they also may indicate that the FMA-UE

reflex items reflect a neurological mechanism that is different than that of the other









assessment items. A relatively simple spinal neural circuit governs the tendon tap muscle

contraction, while a complex supraspinal and spinal neural network governs voluntary

goal-directed movement [80]. Rather than a "hardwired" precursor to voluntary

movement (as suggested by traditional motor control theory), stretch reflexes are

important adjuncts of voluntary motor control [80, 82, 83] and the magnitude of the

stretch reflex is amendable to operant training [86, 87]. Given that the FMA-UE is an

assessment of voluntary movement ability, our results suggest the three reflex FMA-UE

items diminish the construct validity of the instrument. As suggested by Gladstone et al.

(2002), the reflex items may confound interpretation of the FMA-UE total score [55].

Item Hierarchy

Fugl-Meyer attributed the assessment structure and item choice, at least in part, to

the earlier works of Twitchell (1951) [75] and Brunnstrom (1966) [76]. Twitchell

described an "orderly progression of phenomena" characterizing the course of post-stroke

UE motor recovery. Initial UE flaccidity was followed by emergence of gross flexor and

extensor voluntary movement synergies, gradual separation of synergy movement

patterns and finally distal fine motor coordination. Twitchell attributed this recovery

sequence to neurophysiological mechanisms, i.e., the cortex reassuming its natural

inhibition of primitive proprioceptive reflexes [75]. Brunnstrom, furthering Twitchell's

work, documented specific movements that typified each stage of motor recovery thereby

mapping the recovery progression. Fugl-Meyer appears to have, at least in part, applied a

numerical rating scale to Brunnstrom's "Hemiplegia Classification and Progress Record"

[76] to create the FMA-UE [15].

The Rasch item difficulty hierarchy may illustrate the pattern of post-stroke UE

recovery. The keyform provides evidence that this hierarchy is replicated across different









individuals in this sample. This hierarchy however, does not follow the expected

"stepwise" item arrangement of the FMA-UE. Contemporary motor control theoretical

models suggest that UE movement is dynamically organized in response to the demands

of a task rather than in a strict reflexive sequence [224]. Following stroke, UE muscle

activation patterns and reaching kinematics for a task depend on contextual and

mechanical factors such as target location [217, 218], characteristics of the environment

[100] and external constraint [98]. Murray and Johnson (2004) found that healthy

individuals scaled arm j oint torques according to the number of joints involved in a task

[106]. In addition, post-stroke UE motor control is also influenced by descending inputs

from secondary non-cortical motor areas to axial musculature (likely spared from infarct)

and inputs directly from primary cortical motor areas to forearm and hand muscles (likely

damaged with infarct) [225]. Taken together, these studies suggest that for individuals

with and without stroke, UE motor control is dependent on a dynamic interaction of

neural, mechanical, and contextual factors. Our results suggest that the FMA-UE

assessment items are arranged in a Rasch-derived difficulty order compatible with the

task-specific complexity of motion. That is, arm movements are "easier "or "more

challenging" based on the inherent demands of the movement. Elbow flexion may be

"easier" than wrist circumduction because the movement is inherently less complex and

the movement may be orchestrated via spared secondary motor areas. .\bouA Allr flexion to

180 degrees nI ithr the elbow extended may be "difficult" because of the influence of

gravity on a long lever arm.

Our task-specific interpretation of the item hierarchy may affect the interpretation

of the FMA-UE score. Our results indicate that the FMA-UE aggregate score should not









be interpreted in a strict "proximal-to-di stal" or synergi stic-to-i isolated" manner. Thi s i s

particularly important when the FMA-UE score is used to describe patient movement

characteristics. Chae and colleagues (2003) [97] suggest that a low FMA-UE aggregate

score indicated recovery of only proximal UE motor function, while higher FMA-UE

aggregate score indicated recovery of both proximal and distal nE motor function.

Chae' s interpretation is consistent with Fugl-Meyer' s proposed "stepwise" item order in

which movements assumed to recover early in the process appear first on the assessment,

while those to recover later in the process appear further along the assessment. Our

results indicate that an individual with a low FMA-UE score is likely to exhibit some

proximal arm motions characteristic of an abnormal synergy pattern (e.g., the "easy"

items; elbow flexion, shoulder adduction 0I ithr internal rotation, scapular elevation), but

is unlikely to exhibit other proximal synergy-based movements (e.g., the "more difficult"

items; scapular retraction, shoulder external rotation, and forearm supination). An

individual with a low FMA-UE score is likely to exhibit some "easy" distal, non-

synergistic, hand movements (e.g., finger mass flexion, finger mass extension, and

cylindrical grasp). Participant groups formed using FMA-UE scores may be less

homogenous than desired. Similarly, our item hierarchy, if assumed to illustrate the

pattern of post-stroke UE recovery, shows that recovery does not proceed in a strict

proximal-to-distal sequence as was traditionally endorsed by some therapists [226] and

suggested by some therapy textbooks [227, 228].

Our results do not entirely contradict the original clinical observations of Twitchell

and Brunnstrom. Indeed, the results provide at least partial support for their classic

works. For example, both clinicians observed "flexion synergy" and "extension synergy"










movements to be among the first motions to return after stroke. Our results show that

three of the six "flexor synergy" movements and two of the three "extension synergy"

movements are among the least challenging items of the FMA-UE (elbow flexion,

shoulder abduction and scapular elevation; and forearm pronation and shoulder

adduction 0I ithr internal rotation). Furthermore both clinicians described "movements

deviating from synergy" occurring later in recovery. The Rasch hierarchy also partially

supports this observation. Items requiring wrist motion are easier when the shoulder is

neutral and the elbow is bent, than when the shoulder is flexed and the elbow is extended.

For example forearm pronation-supination, elbow at 90 and wrist flexion-extension,

elbow at 90 are moderately challenging items, while pronation-supination, elbow

extended and wrist flexion-extension, elbow extended are among the hardest items.

In addition, Brunnstrom proposed separate hand and proximal limb evaluation

scales. Several researchers [31, 34] have separated the hand items to focus their

assessment on an area interest. It is feasible that different motor control constructs

underlie arm versus hand movements given the different supraspinal areas likely

influencing arm versus hand motorneurons [225]. Although the PCA rotated factor

pattern provided some statistical support for this division, other results (e.g., Rasch fit

statistics and PCA item correlations) support keeping the arm and hand voluntary

movement items together as a unidimensional scale.

One surprising finding in the present study was that the finger extension item

calibrated as an "easy" item. Studies have shown Einger extension to be highly impaired

after stroke [3 1, 229]. Our results may be explained in two ways. First, the vast maj ority

of participants in our sample (90%) sustained mild or moderate stroke as defined by the










Orpington score. It is possible that our sample was not severely impaired and therefore

did not exhibit impaired finger extension. Secondly, the FMA-UE defines finger mass

extension as gross release of the mass flexion grasp [15]. Finger extension, as defined by

the FMA-UE may actually reflect the natural tenodesis mechanism of grossly

"extending" the fingers by flexing the wrist. It should be noted that some researchers

have more rigorously evaluated finger extension by controlling for wrist position and

defining finger extension as return to metacarpophalangeal joint neutral [31, 229]. One

weakness of the FMA-UE is that there exist no standard administration guidelines [18,

56]. It is feasible that if the assessment were administered such that the wrist

configuration during this item was controlled, the item would have a different difficulty

level .

Limitations

There are a number of statistical and demographic limitations to this study that

could limit the generalizability of our results. Traditional PCA was designed for

continuous rather than ordinal data. Flora and Curran (2004) [230] suggest using factor

analysis approaches designed specifically for multivariate categorical data when

analyzing ordinal data (e.g., FMA-UE ratings). While these approaches are not yet

common in the health care literature they should be considered for future studies.

Additionally the generalizability of the findings is confined to sample characteristics of

the study. Although our participants represented a range of stroke severity, the maj ority,

90%, of the sample experienced mild or moderate stroke. In order to generalize these

findings across the broader range of stroke severity the study needs to be replicated with

individuals with more severe motor impairment. Finally, the most critical limitation of,










the present study is that we used cross-sectional data to challenge Fugl-Meyer' s

"recovery" paradigm. We are currently replicating this study using longitudinal data.

Conclusion

In summary the Rasch analysis of the FMA-UE challenges the use of reflex items

to measure UE movement and challenges the strict "definable stepwise" structure implied

by the instr-ument. A better understanding of the post-stroke UE recovery progression and

possible revision of the theoretical mechanisms underlying this recovery may be critical

for clinical intervention. Instruments clearly reflecting the recovery pattern may more

accurately identify the client' s stage of recovery and inform the clinician of the

appropriate treatment progression.

Table 3: Characteristics of the 5 12-person sample
Table 1: Characteristics of the 512 person sample
Years of Age (Mean f 69.8 f 11.12
SD)

Gender Males 242
Females 270

Race White 411
African-American 85
Other 16

Stroke Type Ischemic 474
Hemorrhagic 38

Stroke Location Right Hemisphere 250
Left Hemisphere 262

Days Since Stroke 16.88 f 31.23
(Mean f SD)
Range of Days Since Minimum 0
Stroke
Maximum 145
Stroke Severity Minor (Orpington Score <3.2) 191 (37.3%)

Moderate (3.2 <= Orpington <= 5.2) 269 (52.5%)

Severe (Orpington > 5.2) 52 (10.2%)














Table 4: Item measures, fit statistics and score correlations for the full 33-item FMA-UE (n


512)


Init
Measure Error
MnSq
3.40 0.11 1.34
1.55 0.09 0.79
1.26 0.09 1.29
1.20 0.09 0.70
1.15 0.09 0.96
1.05 0.09 0.96
1.03 0.09 0.65
0.98 0.09 0.57
0.93 0.09 0.96
0.88 0.09 1.10
0.68 0.09 0.70
0.09 0.72
0.37


Init
ZSTD
3.8
-3.2
3.9
-5.0
-0.6
-0.5
-5.8
-7.4
-0.5
1.4
-4.8
-4.3

4.7
-5.0
0.1
-6.9
-0.7
-1.5
-6.5
-3.2
-4.0
-2.9
-6.6
-5.9
2.3
-1.5


Outfit
MnSq
1.10
0.92
1.08
0.57
1.02
1.21
0.57
0.44
0.69
0.93
0.73
0.53

1.21
0.63
0.71
0.52
0.83
0.96
0.48
0.52
0.62
0.77
0.53
0.64
0.98
0.76


Outit
ZSTD
0.4
-0.3
0.5
-2.7
0.2
1.2
-2.8
-4.1
-2.0
-0.4
-1.9
-4.0

1.4
-3.0
-2.3
-4.3
-1.2
-0.3
-4.8
-4.4
-3.3
-1.8
-4.2
-2.9
-0.1
-1.7


Score
Correlation
.57
.77
.72
.80
.77
.76
.82
.83
.78
.76
.84
.83

.76
.85
.80
.86
.81
.82
.86
.84
.85
.84
.87
.86
.80
.84


FMA-UE Item

Normal reflex activity
Wrist circumduction
Hook grasp
Shoulder flexion to 1800, elbow extended
Spherical Grasp
Lateral prehension
Wrist flexion-extension, elbow extended
Pronation-supination, elbow extended
Wrist stable, elbow extended
Movement with normal speed
Forearm supination
Shoulder abduction to 900, elbow
extended
Movement without dysmetria
Shoulder external rotation
Wrist stable, elbow at 900
Wrist flexion-extension, elbow at 900
Palmar prehension
Scapular retraction
Pronation-supination, elbow at 900
Shoulder flexion to 900, elbow extended
Hand to lumbar spine
Shoulder abduction
Elbow extension
Forearm pronation
Movement without tremor
Cylindrical grasp


0.36
0.34
0.29
0.24
0.19
0.16
0.00
-0.03
-0.19
-0.32
-0.39
-0.57
-0.60
-0.74


0.09
0.09
0.09
0.09
0.09
0.09
0.10
0.10
0.10
0.10
0.10
0.10
0.10
0.10


1.37
0.68
1.01
0.58
0.95
0.89
0.59
0.78
0.72
0.79
0.57
0.60
1.19
0.88













Table 4 continued
Finger mass extension -0.86 0.10 0.70 -4.1 0.58 -3.3 .86
Scapular elevation -0.97 0.11 0.93 -0.8 0.87 -0.8 .83
Finger mass flexion -1.00 0.11 0.79 -2.7 0.70 -2.0 .85
Shoulder adduction with internal rotation -1.09 0.11 0.65 -4.8 0.49 -3.8 .87
Elbow flexion -1.23 0.11 0.67 -4.4 0.81 -1.1 .85
Triceps reflex -3.66 0.13 5.20 9.9 9.90 9.9 .26
Biceps reflex -4.40 0.14 3.51 9.9 9.90 8.2 .36










Table 5: Eigenvalues for the principal components analysis
Table 3: Eigenvalues for the Principle Components Analysis
~Proportion of.
Factor Eigenvalue .Cumulative Variance
Variance
1 22.48 0.68 0.68
2 1.60 0.05 0.73
3 1.31 0.04 0.77
4 1.04 0.03 0.80


Table 6: Item correlations with first principal component
Item Description Com ponent 1
Elbow Extension 0.91
Forearm Pronation 0.91
Pronation Supination with Elbow 90 0.91
Wrist Flexion Extension Elbow 90 0.90
Sh Flex to 90 Elbow Extended 0.90
Shoulder Adduction with IR 0.90
Finger Mass Extension 0.90
Hand to Lumbar Spine 0.89
Sh Abd to 90 Elbow Extended 0.88
Finger Mass Flexion 0.88
Elbow Flexion 0.88
Cylindrical Grasp 0.88
Shoulder Abduction 0.88
Shoulder External Rotation 0.88
Pronation Supination with Elbow Extended 0.87
Forearm Supination 0.86
Wrist Stable Elbow at 90 0.86
Scapular Elevation 0.85
Palmar Prehension 0.85
Wrist Flexion Extension Elbow Extended 0.85
Movement without tremor 0.84
Scapular Retraction 0.84
Sh Flex to 180 Elbow Extended 0.83
Wrist Stable Elbow Extended 0.83
Movement with normal speed 0.81
Wrist Circumduction 0.80
Lateral Prehension 0.80
Spherical Grasp 0.80
Movement without dysmetria 0.78
Hook Grasp 0.75
Biceps Reflex 0.24
Triceps Reflex 0.14
Normal Reflex Activity 0.53










Table 7: Rotated factor loading (oblique rotation) for 4 p inciple components
Item Description Factor 1 Factor 2 Fatr3 Factor 4
Seuar Blevation 0.7 0.@ 0.10 0. 15
Blbowmlexion 0.7 0. 19 -0.0 0. 15
Scapular Retraction 0.7 -0.0 0.2 0.10
Shoulder Alxkiction 0.7 0.@ 0.18 0.11
Shoulder Adduction with IR 0.67 0.2 0.10 0.14
Shoulder Ex~temal Rotation 0.61 0.0 0.3 0. 13
BlbowEx~tension 0.61 0.21 0.19 0. 13
ForearmPronation 0.59 0.30 0.0 0.16
Mobvemeut without tremor 0.4 0.3 0.11 0.10
~Hnd to lunbar Spine 0.6 0.2 0.23 0.17
Forearm Sulpination 0.0 0.31 0.23 0. 15
Lateral Prehension -0.@ 0.7 0.15 0.09
Palmer Prehension 0.14 0.73 0.10 0.05
Spherical Grasp 0.0 0.71 0.10 0.0
Hok~d Grasp 0.01 0.69 0.18 0.01
Cylindrical Grasp 0.3 0.63 -0.03 0.0
Finger Ma~ss Ex~tension 0.37 0.61 -0.01 0.14
Finger Ma~ss Flexion 0.3 0.55 0.@ 0. 13
Wrist Stable Ebow at 90 0. 13 0.51 0.31 0.14
Movement with normal speed 0.21 0.8 0.2 0.M
Wrist Flexion Ex~tension Blbow 90 0.21 0.6 0.33 0. 13
Nomial Reflex Activity -0. 13 -0.17 0.99 0.0
Sh Fex to 180 BlbowEx~tended 0.2 0.0 0.7 0.M
Pronation uinton with Ebow~xtended 0.17 0.2 0.@ 0.11
Triceps Reflex -0.23 -0.10 -0.0 0.96
Biceps Reflex 0.0 -0.16 -0.11 0.92
Pronation Sulpination with Blbow 90 ** 0.37 0.35 0.2 0. 13
I~ovemeut without dysmedtria ** 0.32 0.37 0.18 0.09
Wrist Flexion Ex~tension Blbow Extended 0.0 0.41* 0.54* 0.1
Wrist Stable~lbowEx~tended -0.@ 0.46* 0.52 0.11
Wrist Gircumrduction -0.0 0.47 0.50 0.0
Sh Abd to 90 Blbow Etended 0.45* 0.0 0.50 0.05
Sh Flex to 90 BlbowEx~tended 0.44* 0.18 0.42 0.09










79







Table 8: Item measures, fit statistics, and correlations for revised FMA-UE (n = 512, 30

Items)

FMA-UE Item Measure Error q S nI oeai

VI rist circum duction 1 67 0 10 0 90 -1 5 0 96 -0 1 77
Hook grasp 1 33 0 10 1 43 5 6 1 28 1 7 73
Shoulder flexion to 1800, elbow extended 1 26 0 10 0 84 -2 5 0 80 -1 4 80
Spherical Grasp 1 20 0 10 1 08 1 1 1 17 1 1 78
Lateral prehension 1 08 0 10 1 10 1 5 1 30 2 0 78
W rist flexion-extension, elbow extended 1 06 0 10 0 77 -3 7 0 78 -1 6 82
Pronation-supination, elbow extended 1 00 0 10 0 70 -4 8 0 63 -3 1 83
W rist stable, elbow extended 0 95 0 10 1 13 1 9 0 96 -0 2 78
M ovem ent w ith norm al speed 0 89 0 10 1 25 3 4 1 13 1 0 77
Forearm supination 0 65 0 10 0 82 -2 6 0 88 -1 0 85
Shoulder abduction to 900, elbow extended 0 28 0 10 0 86 -2 1 0 71 -2 7 85
M ovem ent w ith out dysm etria 0 27 0 10 1 62 7 4 152 3 9 77
Shoulder external rotation 0 24 0 10 0 83 -2 5 0 82 -1 6 86
W rist stable, elbow at 900 0 18 0 10 1 20 2 7 0 96 -0 3 82
W rist flexion-extension, elbow at 900 0 12 0 10 0 69 -4 8 0 66 -3 2 88
P almar prehension 0 06 0 10 1 10 1 3 1 01 0 1 83
Scapularretraction 0 03 0 10 1 08 1 2 1 29 2 4 84
Pronation -sup nation, elb ow at 900 -0 1 7 0 11 0 72 -4 2 0 67 -3 1 88
Shoulder flexion to 900, elbow extended -0 21 0 11 0 94 -0 8 0 71 -2 7 86
Hand to lumbar spine -0 40 0 11 0 90 -1 4 0 85 -1 2 87
Shoulder abduction -0 56 0 11 0 99 -0 1 1 02 0 2 87
Elbow extension -0 64 0 11 0 70 -4 2 0 70 -2 4 90
Forearm pronation -0 87 0 11 0 76 -3 1 0 81 -1 4 89
M ovem entw without trem or -0 91 0 11 1 51 5 4 1 36 2 2 83
Cylindrical grasp -1 10 0 12 1 09 1 1 0 94 -0 3 87
Finger mass extension -1 25 0 12 0 89 -1 3 0 76 -1 4 89
Scapular elevation -1 40 0 12 1 25 2 7 1 31 1 6 86
Finger mass flexion -1 44 0 12 1 04 0 4 1 03 0 2 89
Shoulder adduction with Internal rotation -1 56 0 12 0 88 -1 4 0 74 -1 4 90
Elbow flexion -1 76 0 13 0 92 -0 8 1 18 0 8 89

























































-1.76 0.13


Measure Error


Table 9: Item difficulty hierarchy revised 30-item FMA-UE
FMA-UE
item umberFMA-UE item description


23
26
16


Wrist circumduction
Hook grasp
Shoulder flexion to 1800, elbow
extended
Spherical grasp
Lateral prehension
Wrist flexion-extension, elbow
extended
Pronation-supination, elbow extended
Wrist stable, elbow extended
Movement with normal speed
Forearm supination
Shoulder abduction to 900, elbow
extended
Movement without dysmetria
Shoulder external rotation
Wrist stable, elbow at 900
Wrist flexion-extension, elbow at 900
Palmar prehension
Scapular retraction
Pronation-supination, elbow at 900
Shoulder flexion to 900, elbow
extended
Hand to lumbar spine
Shoulder abduction
Elbow extension
Forearm pronation
Movement without tremor
Cylindrical grasp
Finger mass extension
Scapular elevation
Finger mass flexion
Shoulder adduction with internal
rotation
Elbow flexion


1.67
1.33
1.26


0.10
0.10
0.10

0.10
0.10
0.10

0.10
0.10
0.10
0.10
0.10

0.10
0.10
0.10
0.10
0.10
0.10
0.11
0.11

0.11
0.11
0.11
0.11
0.11
0.12
0.12
0.12
0.12
0.12


1.20
1.08
1.06

1.00
0.95
0.89
0.65
0.28

0.27
0.24
0.18
0.12
0.06
0.03
-0.17
-0.21

-0.40
-0.56
-0.64
-0.87
-0.91
-1.10
-1.25
-1.40
-1.44
-1.56
























Individual with Moderate Upper Extrem ity Motor Impairm ent
FMA-UE Score = 32/60, Rasch Ability Measure = 0 20 (SE 0 32) loqlts
I ting Scale Item Description


Individual with Severe Upper Extremity Motor Impairment
FMA-UE Score = 18/60, Rasch Ability Measure= -1 25 (SE 0 34) loqlts
~ting Scale Item Description


-6 -2 2 4 6 Measure (Logits;

Individual with Mild Upper Extrem ity Motor Im pairm ent
FMA-UE Score = 57/60, Rasch Ability Measure = 3 88 (SE 0 62) logits
Rating ItemDecito
0 0 2 Worst Ilrc mduction
Shoulder flexion to 1800, elbow extended
Spherical grasp
Lateral prehension
Wrlst flexion-extension, elbow extended
Pronation-supination, elbow extended
Worst stable, elbow extended
Movement with normalsrpeed
Forearm supination
Shoulder abduction to 900, elbow extended
Moverneentx ethoutrdyasrnetria

Worst stable, elbow at 90.
Wrlst flexion-extension, elbow at 90.
Palmer prehension
Scapular retraction
Pronation-supination, elbow at 90.
Shoulder flexion to 900, elbow extended
Hn Ito lumbar spine
Elbow extension
Forearm pronation
Movement without tremor
Cylindrical grasp
2 Finger mass extension
Scapular elevation
Finger mass flexion
Shoulder adduction with Internal rotation
Flbnw finalnn


A)














0
On



0 *
0
0 1

1
11


s6 4 -2 o 2 4 6 Measure (Logits)


Worst circumduction
Hook grasp
Shoulder flexion to 1800, elbow extended
Spherical grasp
Lateral prehension
Wrlst flexion-extension, elbow extended
Pronation-supination, elbow extended
Worst stable, elbow extended
Movement with normal speed
Forearm supination
Shoulder abduction to 900, elbow extended
Movemeentxeithoutrdtyasmetria

Worst stable, elbow at 90.
Wrlst flexion-extension, elbow at 90.
Palmer prehension
Scapular retraction
Pronation-supination, elbow at 90.
Shoulder flexion to 900, elbow extended


Elbow extension
Forearm prowntountrmr

Cylindrical grasp
Finger mass extension
Scapular elevation
Finger massflexion
Shoulder adduction with Internal rotation
Elbow flexion


Wrlst circumduction
Hook grasp
Shoulder flexion to 1800, elbow extended
Spherical grasp
Lateral prehension
Wrlst flexion-extension, elbow extended
Pronation-supination, elbow extended
Wrlst stable, elbow extended
Movement with normal speed
Forearm supination
Shoulder abduction to 900, elbow extended
Mhoo Irent without strnetria
Wrlst stable, elbow at 90.
Wrlst flexion-extension, elbow at 90.
Palmer prehension
Scapular retraction
Pronation-supination, elbow at 90.
Shoulder flexion to 900, elbow extended


Elbow extension
Forearm pron tounter

Cylindrical grasp
Finger mass extension
Scapular elevation
Finger massflexion
Shoulder adduction with Internal rotation
Elbow flexion


-6 -4 -2


6 Measure (Logits)


0




0


2





1 2




2


2


2




2


Figure 1: Keyform recovery maps for three individuals with stroke.














CHAPTER 3
LONGITUIDNAL STABILITY OF THE 30-ITEM FUGL-MEYER UPPER
EXTREMITY AS SES SMENT ITEM DIFFICULTY HIERARCHY

The primary goal of many post-stroke upper extremity (UE) motor rehabilitation

providers is to observe change in UE motor function attributable to the antecedent

intervention. Assessing change requires repeated measurement of function before and

after intervention. Furthermore many rehabilitation researchers and clinicians desire to

understand the time course of post-stroke UE motor recovery. This also necessitates

repeated measurement of function longitudinally over time. An assessment used to

analyze clinical change and/or describe recovery must measure in a valid and consistent

manner each time the assessment is administered [231].

The Fugl-Meyer Upper Extremity Assessment (FMA-UE) [15] is the most widely

used clinical assessment of post-stroke UE motor function impairment [16]. It is often

used to assess changes in UE motor function with rehabilitation. For example, the FMA-

UE was the primary criterion for evaluating response to interventions such as rhythmic

bilateral movement training [36, 37], neuromuscular stimulation [38], electromyographic

triggered electrical stimulation [39, 40], Botox [41], home based exercise [30],

community based exercise [42], robot aided therapy [43, 44], virtual reality [45], imagery

[46], forced use [47], and modified constraint induced movement therapy [48-51].

Furthermore, the FMA-UE is often used in longitudinal characterizations of the post-

stroke UE motor recovery process [2, 29, 232, 233]. When studied as a whole

assessment, the FMA-UE has shown excellent psychometric properties in test-retest










situations [17, 18, 56]. However, there are no published analyses examining how

individual FMA-UE items perform in test-retest situations.

Because the FMA-UE is often used as a longitudinal measure of UE motor

function, it is critical to test whether the items forming this construct remain

longitudinally stable. Item Response Theory, e.g., the Rasch measurement model, offers a

method to examine assessment stability at the item level. Rasch analysis has been

increasingly applied to health-care outcomes research to examine the stability of

assessment items across testing occasions [161, 162, 182, 190].

In a previous cross-sectional study we used Rasch analysis and Principal

Component Analysis to test the dimensionality and construct validity of the FMA-UE in

a participant sample of 512 individuals 2 weeks to 3 months post-stroke. We found that

three tendon-tap reflex items (biceps reflex, triceps reflex, normal reflex activity) were

empirically disconnected from the other voluntary UE movement items. We removed the

reflex items to form a modified 30-item FMA-UE.

The 30-item FMA-UE items were arranged along a hierarchical continuum from

"least difficult" to "most difficult". This item hierarchy is a visual representation or map

of the post-stroke UE recovery progression. This item difficulty hierarchy was consistent

with contemporary motor control science suggesting that UE movements are more or less

difficult depending on the task-specific complexities of the movement. The Rasch-

derived item hierarchy was replicated across 98% of our sample with mild to severe UE

motor impairment.

Ideally, the assessment should have the same item hierarchy each time it is

administered. A stable item order assures that individuals with equivalent UE motor









ability will have equal probability of receiving a score on a given FMA-UE item

irrespective of the occasion at which it was assessed. While we would expect that

individual UE motor abilities would change over time, we would not expect the item

hierarchy (i.e., the post-stroke UE recovery map) to change over time. An unstable item

order may threaten the assessment' s construct validity making it difficult to compare

FMA-UE scores from one testing occasion to another testing occasion. It is possible that

an assessment' s item difficulty order may unexpectedly reorganize with a repeated test

even if the overall consistency of the assessment as a whole remains high [180].

The purpose of the current study is to test whether the item difficulty hierarchy of

the modified 30-item FMA-UE is equivalent at subsequent testing occasions.

Specifically, we explored whether FMA-UE items display the same statistical

characteristics, specifically item difficulties, in one sample of post-stroke individuals to

whom the FMA-UE was administered at 2-weeks (T1) and 6 months (T2) post-stroke.

Testing the longitudinal stability of the item hierarchy addresses the important issue of

whether it is valid to make comparisons of repeated-measure FMA-UE scores.

Methods

Participants

A secondary analysis of FMA-UE scores from 459 individuals enrolled in the

Kansas City Stroke Study [221] was conducted. Participants included in the Stroke Study

met the following inclusion criteria; (1) stroke onset within 0 14 days prior to

enrollment, (2) 18 years of age or older, (3) ischemic stroke as diagnosed by physician,

(4) lived in the community prior to stroke onset, and (5) able to participate in baseline

testing. The FMA-UE was administered to participants at admission to the study, i.e., O -

14 days post stroke (T1) and 6 months post stroke (T2). Patients were evaluated by staff,









all of who were health care professionals (e.g., nurses, physical therapists) and had

undergone at least 2-weeks of training in the administration of the FMA-UE. For the

purposes of our study we included only participants with FMA-UE scores at both T1 and

T2. This resulted in exclusion of 82 participants from the larger database for a final

sample size of 377 participants.

The 30-item Fugl-Meyer Upper Extremity Assessment

The modified 30-item FMA-UE, produced in a previous cross-sectional study, is

identical to the traditional assessment except that the 3 tendon-tap reflex items (biceps

reflex, triceps reflex, and normal reflex activity) present in the traditional 33-item FMA-

UE [15] have been removed. Each of the 30-items are rated on a 3-point ordinal scale

("O" = unable to perform, "1" = partial performance, and "2" faultless performance). The

items of the modified 30-item FMA-UE are arranged in a difficulty order consistent with

contemporary motor control framework in which post-stroke UE recovery is viewed as a

process of re-acquiring the ability to meet task-specific UE movement demands.

Movements that were easier (i.e., less complex) for most people with stroke (shoulder

adduction with internal rotation, elbow flexion, mass finger flexion) are located towards

the easier end of the item hierarchy, and movements that were more challenging (i.e.,

more complex) for most people with stroke (wrist circumduction, lateral prehension,

spherical grasp) are located towards the more difficult end of the item hierarchy. Ideally,

item locations should be reproducible on different testing occasions.

Data Analysis

The analysis proceeded in four steps.










Step 1: Dimensionality

In order to create a legitimate measure, all items must contribute to the same

construct, that is, the assessment must be unidimensional [143]. We ascertained that all

items of the 30-item FMA-UE illustrated a single underlying trait using the WINTSTEPS

[234] software program. Rasch analysis produces infit and outfit statistics to determine

the extent to which items contribute to a unidimensional construct. The infit statistic is

most sensitive to ratings on items that are closely matched to participants' ability, the

outfit statistic is most sensitive to ratings on items that are much easier or much harder

than participants' ability [142]. Fit statistics are reported as mean square standardized

residuals (MnSq) produced for each item of the instrument. MnSq represents observed

variance divided by expected variance [172]. Consequently, the desired value of MnSq

for an item is 1.0. We defined the acceptable criteria for unidimensionality as MnSq infit

values < 1.7 associated with standardized Z values (ZSTD) < 2.0 as suggested by Wright

and Linacre for assessments employing clinical observation [169]. High values (i.e., infit

MnSq > 1.7, ZSTD > 2.0) indicate that scores are variant or erratic, suggesting that an

item belongs to a construct that is different from that represented by the other items of the

instrument, or that the item is being inaccurately scored. Low values indicate that the

item is too predictable in its measurement. In this analysis we focused on items with high

infit MnSq values, since this misfit represents a threat to construct validity.

Examining the point-measure correlation values of each item further

dimensionality. This value is the correlation of an individual item with the other items of

the assessment, omitting that item. The value indicates the extent to which the item

contributes to measurement of the overall construct [23 5].