<%BANNER%>

Time Lag Model for Quality of Life Assessment in HIV-Infected Patients with Highly Active Antiretroviral Therapy

University of Florida Institutional Repository
Permanent Link: http://ufdc.ufl.edu/UFE0021744/00001

Material Information

Title: Time Lag Model for Quality of Life Assessment in HIV-Infected Patients with Highly Active Antiretroviral Therapy
Physical Description: 1 online resource (94 p.)
Language: english
Creator: Watcharathanakij, Sawaeng
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2007

Subjects

Subjects / Keywords: assessment, haart, hiv, lag, model, quality, time
Pharmaceutical Outcomes and Policy -- Dissertations, Academic -- UF
Genre: Pharmaceutical Sciences thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Quality of life (QOL) assessment plays a pivotal role in determining the longitudinal effect of highly active antiretroviral therapy (HAART) in HIV-disease. This research addresses how the timing of clinical lab tests, such as CD4+ count and viral load, affect patients' self-reported quality of life. Objectives of this study are to (1) verify whether a lag time between clinical lab test results (CD4+ cell count and viral load) and QOL exists by comparing two statistical models, time-lag and non-time lag model, and (2) determine how well change in CD4+ cell count and change in viral load over time can predict change in QOL over time. Subjects treated a drug regimen called HAART were selected from a secondary database called the Multicenter AIDS Cohort study (MACS). The MACS is a prospective observational cohort study of the natural and treated histories of HIV-1 infection in homosexual and bisexual men. Data were analyzed with both time lag and non time-lag random coefficient models because each patient had a unique CD4+ cell count, viral load and QOL trajectory. The effect of CD4+ cell count and viral load on two dimensions of QOL--physical health component (PHC) and mental health component (MHC)--was examined by comparing the time lag and non time-lag random coefficient models with the model fit statistics, Akaike information criterion (AIC). PHC and MHC in HIV-infected patients who were on HAART slightly decreased over time. The change in viral load over time significantly predicts change in PHC and MHC over time, whereas the change in CD4+ cell count significantly predicts PHC over time only. CD4+ cell count has a positive longitudinal relationship with PHC, whereas viral load has a negative longitudinal relationship with both PHC and MHC. Overall, time-lag models were not different from non time-lag models in terms of the model fit statistics and regression coefficients.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Sawaeng Watcharathanakij.
Thesis: Thesis (Ph.D.)--University of Florida, 2007.
Local: Adviser: Segal, Richard.
Local: Co-adviser: Kimberlin, Carole L.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2007
System ID: UFE0021744:00001

Permanent Link: http://ufdc.ufl.edu/UFE0021744/00001

Material Information

Title: Time Lag Model for Quality of Life Assessment in HIV-Infected Patients with Highly Active Antiretroviral Therapy
Physical Description: 1 online resource (94 p.)
Language: english
Creator: Watcharathanakij, Sawaeng
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2007

Subjects

Subjects / Keywords: assessment, haart, hiv, lag, model, quality, time
Pharmaceutical Outcomes and Policy -- Dissertations, Academic -- UF
Genre: Pharmaceutical Sciences thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Quality of life (QOL) assessment plays a pivotal role in determining the longitudinal effect of highly active antiretroviral therapy (HAART) in HIV-disease. This research addresses how the timing of clinical lab tests, such as CD4+ count and viral load, affect patients' self-reported quality of life. Objectives of this study are to (1) verify whether a lag time between clinical lab test results (CD4+ cell count and viral load) and QOL exists by comparing two statistical models, time-lag and non-time lag model, and (2) determine how well change in CD4+ cell count and change in viral load over time can predict change in QOL over time. Subjects treated a drug regimen called HAART were selected from a secondary database called the Multicenter AIDS Cohort study (MACS). The MACS is a prospective observational cohort study of the natural and treated histories of HIV-1 infection in homosexual and bisexual men. Data were analyzed with both time lag and non time-lag random coefficient models because each patient had a unique CD4+ cell count, viral load and QOL trajectory. The effect of CD4+ cell count and viral load on two dimensions of QOL--physical health component (PHC) and mental health component (MHC)--was examined by comparing the time lag and non time-lag random coefficient models with the model fit statistics, Akaike information criterion (AIC). PHC and MHC in HIV-infected patients who were on HAART slightly decreased over time. The change in viral load over time significantly predicts change in PHC and MHC over time, whereas the change in CD4+ cell count significantly predicts PHC over time only. CD4+ cell count has a positive longitudinal relationship with PHC, whereas viral load has a negative longitudinal relationship with both PHC and MHC. Overall, time-lag models were not different from non time-lag models in terms of the model fit statistics and regression coefficients.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Sawaeng Watcharathanakij.
Thesis: Thesis (Ph.D.)--University of Florida, 2007.
Local: Adviser: Segal, Richard.
Local: Co-adviser: Kimberlin, Carole L.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2007
System ID: UFE0021744:00001


This item has the following downloads:


Full Text
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 E20101108_AAAACB INGEST_TIME 2010-11-08T22:10:32Z PACKAGE UFE0021744_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES
FILE SIZE 857 DFID F20101108_AABSSU ORIGIN DEPOSITOR PATH watcharathanakij_s_Page_48.txt GLOBAL false PRESERVATION BIT MESSAGE_DIGEST ALGORITHM MD5
13919b03f7b5f9d3cce06c167a592fa7
SHA-1
6ea544b12b77b81bb3865e1b47d2a8a99aafa6c2
32307 F20101108_AABSTJ watcharathanakij_s_Page_60.jp2
1c63b7b040ff8dc3fa850de8270e0c86
5ec2e6e240e10b39f232d7b19399abb26aa35a92
19821 F20101108_AABSSV watcharathanakij_s_Page_34.jp2
5aba5886ae5cf42da3b57d191a680415
bb954af2abc5912581267e62909cf9c6e54ce78a
1053954 F20101108_AABSTK watcharathanakij_s_Page_83.tif
7837aa90989caebe39bb1651869328ac
740ec3657ea809a392830068c6fb6d5fae161e65
3647 F20101108_AABSSW watcharathanakij_s_Page_71thm.jpg
e954907a59ceaae7680218fdc148a02d
a5ccb6cc94b70968f3f7977357e257718546f0b5
113908 F20101108_AABSTL UFE0021744_00001.mets FULL
c4dec710dabed923805c3e9d1364e942
560128c0ef16469d665da4a265048c9b8d0b2710
33921 F20101108_AABSSX watcharathanakij_s_Page_60.jpg
f283b810061975f88482f86e36e318d3
7bb8c52303e40224be8fb5ab78ddf3730798de3e
114673 F20101108_AABSUA watcharathanakij_s_Page_16.jpg
85e7c33d5509eeaba6e692cad0854adc
c1e3db9e246d4fc9d3a2f5d3ea0b1bab64eb27db
21535 F20101108_AABSSY watcharathanakij_s_Page_35.jpg
6a5788c1fb82aea0e68e2544fd55389a
697b2272d34913e29f3ad54c0ce8f41faa0f3e52
93366 F20101108_AABSUB watcharathanakij_s_Page_17.jpg
7c1e59e3cfde905f211ab28cf12cbf3d
858a9a09dd6ccd3a6a45be3a98e8953d70a10a03
33902 F20101108_AABSSZ watcharathanakij_s_Page_38.QC.jpg
7a610f2992933be060897c07b091ad08
89522220d7b66628b220ddfeb2cffe15c92c181f
82355 F20101108_AABSUC watcharathanakij_s_Page_18.jpg
5e81d12c291082afd9f159f606b6c0dc
61f901970d87849c97ef0689ea50218a6b502b5d
82134 F20101108_AABSUD watcharathanakij_s_Page_19.jpg
80e059f9747be8ccf90c1939324f22bd
000fdddddfbcc34253bd872f371596d888a0faca
28963 F20101108_AABSTO watcharathanakij_s_Page_01.jpg
64d911d34f6ffaa5cb038a45ef38b1d5
4693d3e44814831e9fbdd310eda3dd7b39cac435
104173 F20101108_AABSUE watcharathanakij_s_Page_20.jpg
ed84073ff490531f1cd76bbb4ae24f18
4b5966595a34fd06f6f9b087326586302abce805
4626 F20101108_AABSTP watcharathanakij_s_Page_02.jpg
bc2059fc016a3badbc4c529b952d817b
94ca2c73b8d14fd060b351835c7c6b72ec81fcf0
27697 F20101108_AABSUF watcharathanakij_s_Page_21.jpg
495d0574d51708fc731f4b8513b3c092
33be9a7b60142c467b4f286ea5035f756cea50e0
4495 F20101108_AABSTQ watcharathanakij_s_Page_03.jpg
cb22e9f2aa30d2106fe1b1cec21933fe
9ac52a95db21033bbe7b095f68596d6dde57580d
98939 F20101108_AABSUG watcharathanakij_s_Page_22.jpg
81dbf8321a1351abab204f9db0cd1cf3
e543fd67e34d1c332d9331d5b21fc20fef186d5c
31760 F20101108_AABSTR watcharathanakij_s_Page_04.jpg
e02ab24ee1f436402deb0838011e7791
a40873435a6aafab64adb05914ff102344588842
25271604 F20101108_AABTAA watcharathanakij_s_Page_08.tif
70edc17777647a7d94c29bf971178b20
89a49e40588780a5b5ace0f183cd472727545f8b
98361 F20101108_AABSUH watcharathanakij_s_Page_23.jpg
5b0b2c7ef84df8e18d6d318ff45b2a96
41e20a320b7242d892778643cdfa169256fb81e6
123111 F20101108_AABSTS watcharathanakij_s_Page_05.jpg
526931da1e03dd8a7fff9818700878fd
434d72bec71d583565a99e12c68976f811921fb8
F20101108_AABTAB watcharathanakij_s_Page_09.tif
80579fcc897cb0e3fa1fff9689c0c92f
2a46cafca3efb21dd36b0d44f97dcf6309655086
103554 F20101108_AABSUI watcharathanakij_s_Page_24.jpg
47d106cb7c5e79e27c0b3abb8c505d0a
6d8cfcf9d48ee4dfeec42396ca82718a63373f14
119671 F20101108_AABSTT watcharathanakij_s_Page_06.jpg
6dca4d8fc9736ab83f631296e8b6d6e7
a71f007b5948e87133d9fa8d0364224a6cda6d74
F20101108_AABTAC watcharathanakij_s_Page_10.tif
c6c2eb4ee103fb7c247746f216ddb271
61245dc329223f0ff8b252e4c5df5d47e6eac83a
106533 F20101108_AABSUJ watcharathanakij_s_Page_25.jpg
0c0f66c547b5c0b787f8226a6ba1258a
d78670f490e84b79b926c0bcf6b4973e33d8cecc
59262 F20101108_AABSTU watcharathanakij_s_Page_07.jpg
2ee71fbda7055cdcf37de609712944b9
b2595d336cda7df772d013911bee0d0c19d300df
F20101108_AABTAD watcharathanakij_s_Page_11.tif
8df1d2971336cbf33ae15a53b40ab68c
0292bfc1352d822991de9f5324c44ee78fe846cb
106415 F20101108_AABSUK watcharathanakij_s_Page_26.jpg
1b101be6b6d59ef4ab1dc09a66bd8444
aec16d3702c864c20966c6b9fdde22e112b89757
115954 F20101108_AABSTV watcharathanakij_s_Page_08.jpg
5f73c83bde959770aaa037b3f6d41a16
473828ef42dee10d39ff20af4743466089dc0b05
F20101108_AABTAE watcharathanakij_s_Page_13.tif
8816b6aceaa97cb030a2038084ba2bfa
682ea203156e06504f670336f2dfed43daf6e9cf
112572 F20101108_AABSUL watcharathanakij_s_Page_27.jpg
aa9354f9d7ac9cd475717c21c1b5cd98
443dd41672843c2ac5f58857bc080188b5694cbe
7555 F20101108_AABSTW watcharathanakij_s_Page_09.jpg
2741cadcaf3daa6e931e84011b9cb4cb
89e58ec1764f68b60cad152f01678ddca9a28e83
F20101108_AABTAF watcharathanakij_s_Page_14.tif
8a1d356073a02c38e8a0d67bfbe6a081
bca5ab87ed1dddab269a60e5493afd89aa4a1fab
53995 F20101108_AABSVA watcharathanakij_s_Page_48.jpg
9c3a5d89ea8217224b9e5025fdbd754f
fa5c3de23ad153bc91a21bc65ee9c63e861015b3
109490 F20101108_AABSUM watcharathanakij_s_Page_28.jpg
ebea0559f66c7cdb8894e23c52e45837
03c2ae2e49e4f933c688629768fa0f4163d03aa6
86024 F20101108_AABSTX watcharathanakij_s_Page_10.jpg
fbf1bcfff0e4cd00d2dbaf1da49fcb6b
bd223f7e91d8925447896a991d712e2bb0d767cf
F20101108_AABTAG watcharathanakij_s_Page_15.tif
b85df9c0ed647a96cbf48cf0ec84fb7f
d8970faff795522a957dddf5abfb753dda215165
75593 F20101108_AABSVB watcharathanakij_s_Page_49.jpg
218a6c813330e40957f5f50af3262f2f
04c589e79512c508050b213fc75833085dc137a5
107222 F20101108_AABSUN watcharathanakij_s_Page_29.jpg
0fff7cbbe1821b6c76d1422776433ddc
017832242757cf1e22a93c6364593aa4c375f58b
102548 F20101108_AABSTY watcharathanakij_s_Page_13.jpg
2b43bb778c319db9a49c3f0db30eb500
812bb99c3c6df090c8efa2d325e422b4c317be9e
F20101108_AABTAH watcharathanakij_s_Page_16.tif
3e7197ef9a4ed5e6efbed48ca7a99716
089e3f9a49b31456e01a5fc6a483cf3a390f0e1d
100923 F20101108_AABSVC watcharathanakij_s_Page_50.jpg
d6f13d98dbb195835c0f225bfe55f01a
7c48c109d7ef25826620b5ea41329f5245088abd
107234 F20101108_AABSUO watcharathanakij_s_Page_30.jpg
d0c0cc058c6fe83cacd504a1ef9ed159
25dc2125a9022c58e3a6630699b9708e00ed42ee
106334 F20101108_AABSTZ watcharathanakij_s_Page_14.jpg
4b086674c52fb199875c47b6564de1a0
bb8d5eda046c9ca0022597bf7ec656f85823e4ae
F20101108_AABTAI watcharathanakij_s_Page_18.tif
1d9c2ab1b59deb7775cd8d14f04a9bfc
13c0b2d2e02ea8afee1dd6b3258834ff30604298
F20101108_AABTAJ watcharathanakij_s_Page_19.tif
66bcd549cd0d0f51bb510b9460d1ec04
ecafdbefe7e6b1efd9c0e8280919424b40a3b7a5
101095 F20101108_AABSVD watcharathanakij_s_Page_52.jpg
3d53993fc33053ca972f3734e1f6a507
e13920384eac5fc439fd4ea77e366c0a76eef68f
40328 F20101108_AABSUP watcharathanakij_s_Page_32.jpg
19c2c8e5aef3461022faa0f273148005
ad35c58b5b2f4d106cb1fff0e62724d0b11318ee
F20101108_AABTAK watcharathanakij_s_Page_20.tif
d5117bc4bf662b080d0d078997338411
532d0acb02045936bc230fedbbdbfae5497689e7
102820 F20101108_AABSVE watcharathanakij_s_Page_53.jpg
6050471c2d2f36b6750c47bd6a9598ce
0902c4c46d2c4e06c05218a237882f7e20be4296
19548 F20101108_AABSUQ watcharathanakij_s_Page_34.jpg
86e601f3fd8fbe070afc73e0c6446bd5
2809d0174e66ba7b4cd966b25b0250bb30d9d89f
F20101108_AABTAL watcharathanakij_s_Page_22.tif
f033d6a634f42411664685309aa6d53b
74b9845d2a704be2239b676985c5be8f52649a9f
103514 F20101108_AABSVF watcharathanakij_s_Page_55.jpg
78e3d8cffbf9b4138d6adfc4ef8bf97f
5649301b481e0ff1c0f01ed7dcfcc5273b2cc5ad
15391 F20101108_AABSUR watcharathanakij_s_Page_36.jpg
230b77c55988227155e762704872587a
bf3ea458d1fc8874c806505c88af1c6223c66dff
F20101108_AABTBA watcharathanakij_s_Page_41.tif
8391f473145dea856f69dfc1b07162c2
bbce60c88f746d0f653a7716e1674c369bb70ea9
F20101108_AABTAM watcharathanakij_s_Page_23.tif
6539f054c4716a96e82fa33fe0ad34a7
7fe3f9077f79e3e688d5392f2726d64711711395
105586 F20101108_AABSVG watcharathanakij_s_Page_56.jpg
b993d5b1a857a47e67784d12f0d117da
cf39fdbd03f29ec4e6eb07cc97f800142453d6c7
101879 F20101108_AABSUS watcharathanakij_s_Page_37.jpg
4c0a1b3146cf38f3027f454aff46862e
81e5a753139dc0943d62919fdab8234d11a57b34
F20101108_AABTBB watcharathanakij_s_Page_42.tif
d44ff3d94aa24f92fa39783b4d134cc8
cb9dc9922efdf9db0a814de5a54b49b22cfb9139
F20101108_AABTAN watcharathanakij_s_Page_24.tif
a475a6eb575d790fb99bfe0480cc4d4b
5fa93c86990d77357799f3d5c214002cc244caa0
105092 F20101108_AABSVH watcharathanakij_s_Page_57.jpg
8673d0f8830687a01c815ccba768a658
b627f37a04dab37ee449e0122596b013d6be6863
96798 F20101108_AABSUT watcharathanakij_s_Page_39.jpg
35cbe81767ed69f020fb62ab58b83e24
f52998105e00d7de583ac65247bf3464548232e7
F20101108_AABTBC watcharathanakij_s_Page_43.tif
9f46905e975d8a4e0684d856a851ca63
8986b16503960aa471feb04c37cca7df0dcf1d7b
F20101108_AABTAO watcharathanakij_s_Page_25.tif
10fc63912b02c2aff1cd497c96acde5d
b17af9d974b76f77616a3cbe5fa4c260d78d40fe
102222 F20101108_AABSVI watcharathanakij_s_Page_58.jpg
9d6791b0e257d5c072bd97fa4d3ee9a0
93be95e1d1da95ed472da14ea33c665d753fe775
98373 F20101108_AABSUU watcharathanakij_s_Page_40.jpg
7de4f51b0d1b64f0d2a182b1ff16f9c9
fdfdd72c98490c26ae0189d606cc7cf8883959f4
F20101108_AABTBD watcharathanakij_s_Page_44.tif
11ce0c3f8601e63fb842fa7d264ea9cf
42222ae8eb6ebe771e07fd9957f1059435d229b1
F20101108_AABTAP watcharathanakij_s_Page_26.tif
a35347de4ad07410d8823615571b505c
4578d5e816314bc3a04cb546da2b37c92420fbab
48873 F20101108_AABSVJ watcharathanakij_s_Page_59.jpg
cfb568b6a4db926556586c368d7c89a6
f36d3caaccb711adce500b0e54c7f607ab10de42
88472 F20101108_AABSUV watcharathanakij_s_Page_42.jpg
cc9d4ad1b73798f82b1617fac043cbe1
0fd1dbd9b536d1ee1dbbed91ce02f8ba92cab579
F20101108_AABTBE watcharathanakij_s_Page_45.tif
3db95169994207b7cd07ba36d20f20e2
b5eaa646e570ecc4ab660245240f0b3ddef93114
F20101108_AABTAQ watcharathanakij_s_Page_27.tif
e5e8afac4958a89f287cc06f40a5a5b6
db4559c2952212d2f02b50d7a2ffa91313df23f2
68152 F20101108_AABSVK watcharathanakij_s_Page_61.jpg
528038e5d1c724c0744ed17870716a54
4796f80a7a01bc472784166d02010321d15b2f62
73291 F20101108_AABSUW watcharathanakij_s_Page_43.jpg
8c91f011f22356a533e894fe296c7f4e
e62570e51407ec2974647c75aa2286c7cb49cca3
F20101108_AABTBF watcharathanakij_s_Page_46.tif
ea48abae092d359616909f309523e033
bff062c501b1002724840c4ee773263e6e24c364
F20101108_AABTAR watcharathanakij_s_Page_29.tif
07737e7128bf75fad9996f4706bf3f49
6eb055dfb05a0647ddc73028bd2e48f6efc9e487
59119 F20101108_AABSVL watcharathanakij_s_Page_62.jpg
43a2561ac0ef336597128ecb3b70b905
172ec6e208cfd28b672721694178a10320bd1196
36126 F20101108_AABSUX watcharathanakij_s_Page_45.jpg
5dae690f87a99e8a64c25f78d09f558a
3ea2ae1a52cade1296935e9a427bfea0f7716627
F20101108_AABTBG watcharathanakij_s_Page_47.tif
1dc370b922628502d03c02f9346782fb
7c3d70ffca275ddd1dad40b5b28a362bb7bacd56
91013 F20101108_AABSWA watcharathanakij_s_Page_80.jpg
1bf62a0e3f7a5dd726109602e3b0bf7a
d4dde0c5cbefa8b4b905defa68ea1c3e4d0bddc6
F20101108_AABTAS watcharathanakij_s_Page_31.tif
5fec73e2e4d76f6399ce0a2f431013f0
c8d7a234f076f0c3ad7c8d53bb72e8bb0e59b0dd
20581 F20101108_AABSVM watcharathanakij_s_Page_64.jpg
84f5c6486311779ac12bf84b76f00e7f
4c53ad231fc5fff64838e8a4fcfe0e115dd21bd5
51013 F20101108_AABSUY watcharathanakij_s_Page_46.jpg
29186828b92ac6524ee9d2b211c2682a
654f0e9a88cdc8da4d3d5b9870c6f4f96f7b4d97
F20101108_AABTBH watcharathanakij_s_Page_48.tif
ebb291926dc7c990e88bdce6fbfd3e70
0f64542257fb6109cf654550e7aac2b6f6462b23
110212 F20101108_AABSWB watcharathanakij_s_Page_81.jpg
c2772502e652419e33bb7ca70f600019
da8914022eacb53e744eca5f2b90833c78ebee06
F20101108_AABTAT watcharathanakij_s_Page_33.tif
d9652eead8c0c6a1a70f49047af866aa
086836f75336a348654505913f2d278f9d2068c1
40741 F20101108_AABSVN watcharathanakij_s_Page_66.jpg
62ae5c536f4a6c9577d1264d790ea3c8
c7a2de11272348329101141b7ea71c3195bc2cb7
39555 F20101108_AABSUZ watcharathanakij_s_Page_47.jpg
4cb2bce31a370ec29f0688e80a1d174d
6d7935334dd3b0d1b463dce9d489aaaa492a8123
F20101108_AABTBI watcharathanakij_s_Page_49.tif
408e5814bceb73e47520997adcf8a048
3ed38bdf1c9d593216910cc892812ae7bff3c296
109176 F20101108_AABSWC watcharathanakij_s_Page_82.jpg
2f256c648c4119ad37a82d427e8cab33
1646756ac07c536cacf22c6a8ad12c365e3baf21
F20101108_AABTAU watcharathanakij_s_Page_34.tif
d2f86795a0b7457dde1e429c52dab305
234603f37c1348b0064bb4dc4bfa4f2070d13f70
20682 F20101108_AABSVO watcharathanakij_s_Page_67.jpg
e9e750007b26fcea19a9a3bc6edc776c
b26da51331c609c63f923f2fc99679e959f0bd76
F20101108_AABTBJ watcharathanakij_s_Page_50.tif
c69f666df7305dfdcd62ec3dbca91934
98e649a921117bd59ef0ad17b61c1f6ffdfca9ce
106330 F20101108_AABSWD watcharathanakij_s_Page_83.jpg
c180cef123cc59e26ff01568fc1f6b4e
93679fba823f7b25efce4fb6b3d0877993ca1c54
1054428 F20101108_AABTAV watcharathanakij_s_Page_36.tif
7431a8bbd39e2710c163f54a7f737a92
535160fcc4a90ec338ca5d7f0cc5c8360ebde1f9
37359 F20101108_AABSVP watcharathanakij_s_Page_68.jpg
278220ed4dc23dcbd09690db66fdc905
c0586050a49ded61b7841ec78db767352d648c24
F20101108_AABTBK watcharathanakij_s_Page_51.tif
709e5d3c53e40b537d828b241be4e1be
632504dc75d118f525dac15f9bcd3ab59ab4cbfc
101976 F20101108_AABSWE watcharathanakij_s_Page_84.jpg
c1bb70dffb63ae103dfe3b12802e143d
544f1ac71ab7a0007c038bc3f83cf1adde63ffd7
F20101108_AABTAW watcharathanakij_s_Page_37.tif
45be977617b3ab75aebe3dc450e05679
93187159de821bd21184f668d4ef96a5e9bb4b1d
F20101108_AABTBL watcharathanakij_s_Page_52.tif
7fc1dc35c55fcfdb4cf0c80ceba4986d
ecaf662686225bea247e6761421eef5b5429a357
88060 F20101108_AABSWF watcharathanakij_s_Page_85.jpg
e0195a2db1eac2ab69c613ab2fd0edf8
84874971a463390d1603a9735346a711f5f10540
F20101108_AABTAX watcharathanakij_s_Page_38.tif
2aaedfecc362fcee895543dcec57b641
89b4a6124d3d7af43af7dc9ae5befd2eadb2ea59
40992 F20101108_AABSVQ watcharathanakij_s_Page_69.jpg
0ca5943ec9168837340b3d1ddbeddfde
7b51b4a370f58a88ab0d081e641e49464c66fd58
F20101108_AABTBM watcharathanakij_s_Page_53.tif
fde451a29ad2f2025b36e513f579d867
ef90dbca95ca7aac06fa8424c02c95a0e005fb50
80994 F20101108_AABSWG watcharathanakij_s_Page_86.jpg
e20edc9be3e80182837d6cadf437554c
aede5f804da69690bc1a71476597ae840d1aca82
F20101108_AABTAY watcharathanakij_s_Page_39.tif
b5547482703c6d8f2c7b861d7de15592
b5d7642af07b4a45b90b7e1a352d9d23e0445af7
36208 F20101108_AABSVR watcharathanakij_s_Page_71.jpg
32e367d50ecebc13f5239c3fa8955506
051ba108d1e5d4dada1debefdf573e05c726d6ff
F20101108_AABTCA watcharathanakij_s_Page_69.tif
a14613a5cbac2f5af642796503704bbb
1113b7333fa4bd15069032a2f8b0f0ae75d667f0
F20101108_AABTBN watcharathanakij_s_Page_54.tif
344c3542882e4a7552b71e046e3441cf
80ab3fc47e951d6c8da743ee39931e794afa949c
67501 F20101108_AABSWH watcharathanakij_s_Page_87.jpg
255f64fbf3bb4e89fbc520b6c00ccadc
368e078ca708616f77bfd84a2067efc523481a93
F20101108_AABTAZ watcharathanakij_s_Page_40.tif
858cd538c0f4f0d545a309274ea65ac9
a2a5386719ce35017f94cc85cdf6b084b78347d6
24491 F20101108_AABSVS watcharathanakij_s_Page_72.jpg
9d6473734ce0103c1a6282d6f06ff896
bbd1888ca7a67d8d65fd05d9f284b41183f582bf
F20101108_AABTCB watcharathanakij_s_Page_70.tif
b8f460f083c23c893e323bebd567ada8
73417e540876529cbe40aff1794f70787aabe1bc
F20101108_AABTBO watcharathanakij_s_Page_55.tif
c5b040cd5ee84ac4da4cfa30535b00e5
52b7c9819dc905e9da9e0c839e69d97743aa432a
49293 F20101108_AABSWI watcharathanakij_s_Page_88.jpg
a23339b532d0a6c231b4d662f330cf8e
238ab1f879d8ce906acaf55b7029ad70daadaf66
25953 F20101108_AABSVT watcharathanakij_s_Page_73.jpg
61765aa4e02247de3e7bf75750d65f44
22430675ad37e45d2fd1fc337c7859258707c98d
F20101108_AABTCC watcharathanakij_s_Page_71.tif
4de505b8f5f0e4c17d4a1ae108c50cec
d5531514d68b33b8a958ca7c86c2ddc929adab7d
F20101108_AABTBP watcharathanakij_s_Page_56.tif
0bb9954c5aa251e325e61450ec67c363
d6789cca4e54a8d41d1753317a0e595e88950480
114366 F20101108_AABSWJ watcharathanakij_s_Page_89.jpg
de50c3152f92f2d5f934f1396bb58716
4f6d17e970989888ff9c8bb34a1f36d3a43d5332
36619 F20101108_AABSVU watcharathanakij_s_Page_74.jpg
0c1860dfa3b889f33d619faf3473b00c
dea4e77a3957da009ac2fcd311feab912536620b
F20101108_AABTCD watcharathanakij_s_Page_72.tif
aa1089b27ca9993b56de6f309564dce7
80aa96c854f9aaa60cb3de89e31df56b89dc060f
F20101108_AABTBQ watcharathanakij_s_Page_57.tif
70831eb06feac1b6e709d34f79ffa5ac
27e995d795b17ce92c485680e49a4a08aca03062
124421 F20101108_AABSWK watcharathanakij_s_Page_90.jpg
26460ea313d746cfa1ae7fbb00ae4de9
04a6a02bc1a8f65b44f298a232e42e5ce49a35a3
23179 F20101108_AABSVV watcharathanakij_s_Page_75.jpg
aa6a9a8a1a0f039e5076c688fe8d1a37
874d168ceaa7db637f2908a4549a3a37e6038899
F20101108_AABTCE watcharathanakij_s_Page_74.tif
1d69f3ce9f1bdc70ac7c885d8314bf4a
5be78d0b8b4100e51f56458161439c7d3552c7b7
F20101108_AABTBR watcharathanakij_s_Page_58.tif
35e71eae17fa86e1dc0b68f22b74e1ab
f23cf609344d9dc9b6bce75f821cccc0940b9a40
128514 F20101108_AABSWL watcharathanakij_s_Page_91.jpg
a13d62cf4f89ad866829d81e2a61bf9d
daa2d906fb0b8cd9ab185b8940f61faca6092083
35069 F20101108_AABSVW watcharathanakij_s_Page_76.jpg
4c88e5df792c17757328881d2904831c
42989b10dd82e69135c0297cfc85bdd75b77783c
F20101108_AABTCF watcharathanakij_s_Page_75.tif
7be873ed45f7460b48f8ad99d0f61185
03898b65b1a101d17e4d44a502f6e8e2e6c75518
106923 F20101108_AABSXA watcharathanakij_s_Page_13.jp2
54ca8bbbc9678bb795d08a28d58a0ea1
2d6844b5624697a61e69dba2ce23973ca44cf1a5
F20101108_AABTBS watcharathanakij_s_Page_59.tif
93d38c96e457a5b45de0c9728fd0cad7
aff61777f56d9f2666205aa6202456f76f5cfde2
95798 F20101108_AABSWM watcharathanakij_s_Page_93.jpg
fcbbe0a079e0cc2c2d61de72aa096439
4c1df4a1799af28c5172625d92b11d2c6b2cc1c3
23756 F20101108_AABSVX watcharathanakij_s_Page_77.jpg
15f4c38b8568266b190eb7644034e52f
e4df37c1aa293ad0ef67e7cd03a2f09b2c4a297d
F20101108_AABTCG watcharathanakij_s_Page_76.tif
e3d2a2261295611c28bfc5b3387b1e9b
aa3bfdb80b3e01622e68d6a84a900ecc9a7e6c0a
112392 F20101108_AABSXB watcharathanakij_s_Page_14.jp2
00037684b7469df8a74996453b7f972d
a1fc14acc120baf779f623a9614b9e7761d42638
F20101108_AABTBT watcharathanakij_s_Page_60.tif
1516d9e15f190e2e9097fb45b431d98a
4f10ef36451c91534ff668a4736a689d75e40626
38042 F20101108_AABSWN watcharathanakij_s_Page_94.jpg
1c971ea58ccb846c8178fd61c730e548
a1f7223be5a67df9ddeb3395e93d00a81c57d179
20188 F20101108_AABSVY watcharathanakij_s_Page_78.jpg
528c408f40149e531e71a9e59a77b360
4852fc1f7ee43df3c59c60dc5ce6fe60758043f6
F20101108_AABTCH watcharathanakij_s_Page_77.tif
bbdbdd0fde58f55f930c774c2b3f49b5
6c6251ee64cdcb340383e05b77b1f78f7bec626c
117202 F20101108_AABSXC watcharathanakij_s_Page_15.jp2
f7f816ac26fe6fc04bdd53a28c1a843b
d69bd54f453da77c12c2ae59f77eebf28e7b6760
F20101108_AABTBU watcharathanakij_s_Page_61.tif
94d12ffff67ed118c092749de38b7438
4ba167e42eaede446b2676c1b2ee97f0fe4adf2c
27171 F20101108_AABSWO watcharathanakij_s_Page_01.jp2
c77bd6ef0bfa0db0f95119446592bb27
03615f02242b45d63a8ef014798cf728630ec36f
20194 F20101108_AABSVZ watcharathanakij_s_Page_79.jpg
1044276a7bdd3a6b88c895ca70c1f246
97edd4a7785408c79cbe684e971d5fe58c76a30a
F20101108_AABTCI watcharathanakij_s_Page_80.tif
5ca5f6163c0da7bb181e4ff85af0180a
67381920a52056c07c5a65f6600b294602b9a2bc
119652 F20101108_AABSXD watcharathanakij_s_Page_16.jp2
7995a49f2a5ebf7159cb9eb11b85c12e
6bc0d04a319657afa3a5f169225709494826159d
F20101108_AABTBV watcharathanakij_s_Page_62.tif
57df4692a70757dfd10d50268925d20a
35aca0ea7daadb623b41fd37d0d762865ed25112
6181 F20101108_AABSWP watcharathanakij_s_Page_02.jp2
e09c81fc037893b2844b74e12d13e428
31894e444a59a60799e74c6591460c7479c2a1c2
F20101108_AABTCJ watcharathanakij_s_Page_81.tif
c9492d72f942d422d29c139e315857b8
36162dcadfe21f0e9ca889fa9b0707057263f548
100997 F20101108_AABSXE watcharathanakij_s_Page_17.jp2
b6f4f14a6846860d7feaf31656adb585
5e131f6019d7b49029a44546610f2ef9a420268a
F20101108_AABTBW watcharathanakij_s_Page_63.tif
a661cd9536c2fefa5e586a765f2c1f88
cd9f983d3fd0961e0b6241cec370e57bf1c61734
5821 F20101108_AABSWQ watcharathanakij_s_Page_03.jp2
269f16315f47ba15f1c73c0f9c99dbfe
78c7ef02dc3bdb6414d03e6db04ac73d2046b87a
F20101108_AABTCK watcharathanakij_s_Page_82.tif
4ded72bfa8a80d3defd3e9680b32582e
43e652378a71cf7fa1dadb379e346c1cdd650032
85845 F20101108_AABSXF watcharathanakij_s_Page_19.jp2
f39160d92c81fc12f11ffcc8d4aaa80a
6f4f7251bd20204abed0a09976a02af1a6f7f0d1
F20101108_AABTBX watcharathanakij_s_Page_65.tif
f7f2ab90d62186ce4b092d1fc3ed6759
b10587512e893de7da8f55acfc400de6fc8c2edf
F20101108_AABTCL watcharathanakij_s_Page_84.tif
f5e38f036629fd1ca5737e6e8a4443ff
6af5eee966f76969adddf67c1759be93b8fe7ca3
108772 F20101108_AABSXG watcharathanakij_s_Page_20.jp2
8e2573cb38c236ec59627f5722d0a17e
2b381a1bf96afbe5aeb2c4cc3ca04f56446c45a4
F20101108_AABTBY watcharathanakij_s_Page_66.tif
a004e317d9366eeb7281617ad5e9b426
b4ec5a8f2e8206cb88bdcd2f1bc1b1ee0fa4a1d6
32191 F20101108_AABSWR watcharathanakij_s_Page_04.jp2
7c6e7a55e3a6df4c35efcf1168aff9c3
cbd4f2acbf0de7015c03276eda18172b7fa281c3
58255 F20101108_AABTDA watcharathanakij_s_Page_08.pro
87fa86f2d303bfd799f6ed5ca6943a32
cbf2c4bceefc32b5a462a72c1d78edf4ae3f2697
F20101108_AABTCM watcharathanakij_s_Page_85.tif
aeb3d523f67c9ff6b6f4b0682b58df2f
30b3cd254efab0da97889c3e23c44f68f471cbd0
39245 F20101108_AABSXH watcharathanakij_s_Page_21.jp2
53bdcbe2546c5bb6f6e8fc1f02fde8b8
9187a7a7109d01410ac52858f0b2f4e5034391cf
F20101108_AABTBZ watcharathanakij_s_Page_67.tif
ced0b9973858a2ce7301e72efd65449d
f172dca8c7a41cde6c32b96a7885dfe31fc56559
1051961 F20101108_AABSWS watcharathanakij_s_Page_05.jp2
96a0725fa2f68f81681e223977de786c
0ce2f4b9711fcca5d2a7aafd531632b9a7cba1b7
2702 F20101108_AABTDB watcharathanakij_s_Page_09.pro
f36085227336ec9c9bb46866714679b3
085df366aee92d2ecc9300be884f937d6da77d3c
F20101108_AABTCN watcharathanakij_s_Page_86.tif
7007e83279b0efa337453f869a484316
ca680b7b8750c8ac3e50b9f77cb16e3e21942d99
104734 F20101108_AABSXI watcharathanakij_s_Page_22.jp2
a7a64d1d15dc8d3dbe8e0f2ac796e6fd
1b74f32f32e31618c9a31a17609487ad7a9f3a06
1051934 F20101108_AABSWT watcharathanakij_s_Page_06.jp2
82c95ec68301a6c5f2c7266e7d2aa096
41e784e8f7dd966e311e05b2e609c2c8f247f573
41977 F20101108_AABTDC watcharathanakij_s_Page_10.pro
57feb099175297f5c74da65c10846300
1683fecbc65e656dd616054244dc443d5319e253
F20101108_AABTCO watcharathanakij_s_Page_87.tif
a0d6a89ff8451ab64be3a500e030c760
6ebe89f57d3ae3e1f566c7943eae95addf30ef65
104184 F20101108_AABSXJ watcharathanakij_s_Page_23.jp2
3d55cb508e7b45aa15f3b801c112cc55
ea573dfa384ee2c7e6605e8f82e1e807cb9266c8
1051964 F20101108_AABSWU watcharathanakij_s_Page_07.jp2
596cdd9933a952dda395932b9e05d5ea
3ef939d40ef5f7cac4ca35a5d05a9146a86b64f2
42950 F20101108_AABTDD watcharathanakij_s_Page_11.pro
7d09a1f8dc61e2fee960c7db0b0ed68e
5730d8dead4f809fdbd0cba1f1cbfb65f77f489c
F20101108_AABTCP watcharathanakij_s_Page_88.tif
583d146ffe023a56ec97e84774f404a8
b82466e4cb528d626574480e84bb8ce0bf1224fa
109095 F20101108_AABSXK watcharathanakij_s_Page_24.jp2
466d012e71eaa371a82f46de9037accc
27a4b1e858e305dffa4c01dcacaae5368b60a218
1051984 F20101108_AABSWV watcharathanakij_s_Page_08.jp2
a74907e56d26c985c76ed53add610c2e
b21f1aaff3ef94f653fcdcdf63705ccc90853332
19299 F20101108_AABTDE watcharathanakij_s_Page_12.pro
38a5c4f3c7a88ec672ca12cc4753adb1
81cceb90fd9aaac453d2b745e1b1cf7cfbb33118
F20101108_AABTCQ watcharathanakij_s_Page_89.tif
2af49954e13fe3f25eba098b8c34540f
8481dd3d22d28fda779d3b9001787cb6160908ef
112445 F20101108_AABSXL watcharathanakij_s_Page_25.jp2
c9cf6901408efe53338f8806e27504fe
34a81ac8ebc97ad4e02933acc54da736a2563a70
82589 F20101108_AABSWW watcharathanakij_s_Page_09.jp2
5e7d97bc318b4bb3a87bd54842690a59
14843f61dc6184db416d37b7659f859115f32d3f
49979 F20101108_AABTDF watcharathanakij_s_Page_13.pro
284c16db5831620cc69006229ec9a98b
a93db04926cc440dee29bdd0712a8f3ab6457e79
F20101108_AABTCR watcharathanakij_s_Page_90.tif
8235a81fd12c9f247007a57416a86369
6582e6aa18ebf1b281a33804786dba9cb70199ea
112100 F20101108_AABSXM watcharathanakij_s_Page_26.jp2
9ae40264462ea25e7c788ce419bc387c
6e013a20f569ce1b2bb168c13eafa8e7a565c2a6
1051986 F20101108_AABSWX watcharathanakij_s_Page_10.jp2
bf6331b4820e6ff90f932bf24385c2d3
65ddb8e0809bf1f1901b74bcc2f386c4cd2677a5
51819 F20101108_AABTDG watcharathanakij_s_Page_14.pro
fbe99c121722f53c4c1f10f3c5663f86
39511b97962fa8ca6fa526b7508416728f445a85
78396 F20101108_AABSYA watcharathanakij_s_Page_44.jp2
c4c5dbb7432ab5dc767015b0c803642e
6f39368f223e8689fc018e3154a19fdd17a14431
F20101108_AABTCS watcharathanakij_s_Page_91.tif
c7a4e20a22ea9d58a78947e966d260e1
39711e4ad8fcf6650bc6c7351f72af1096b068ea
118955 F20101108_AABSXN watcharathanakij_s_Page_27.jp2
4a9227185f96539728c5bdd3ff60d49d
27cd32ec06925e67e718c7113a08fd68a089c7f0
94789 F20101108_AABSWY watcharathanakij_s_Page_11.jp2
bc1ad0863096df7bd4617f1b873280ff
0cbcaf1a11abeb2d2faa443cdecab2f780cd9daf
54596 F20101108_AABTDH watcharathanakij_s_Page_15.pro
df65fa5752145d81f7f1fe3e56efffd8
ddc23d04111a392186024b16ce81fff1755931a0
32796 F20101108_AABSYB watcharathanakij_s_Page_45.jp2
9dda9d990f5254c88784e1371dfbca57
75104bfb5b35ae7022324c25acbc963303151699
F20101108_AABTCT watcharathanakij_s_Page_92.tif
daf56b73dcca4c608ec5c9466af19cac
d7642ac9221b0cd14e6f7e0254fc08bc2658c480
114560 F20101108_AABSXO watcharathanakij_s_Page_28.jp2
a0524b14b6f0b9b720867dd01f132604
ed36d3763a6f09e7b9a494eee8986d4825d02a5c
448160 F20101108_AABSWZ watcharathanakij_s_Page_12.jp2
7a60808b1d7d864253922223394628f2
be951b215524ba00e4c8d0dfcf8428eedd6d8e4a
56246 F20101108_AABTDI watcharathanakij_s_Page_16.pro
f62c659334ba9ac6ef9f073561f832ad
5f7d45afa6284805e7504ccf38a212db7099eaa7
541776 F20101108_AABSYC watcharathanakij_s_Page_46.jp2
df0dc381846beca5573780ced2115237
4ab3934623aadb4630d77c926ae6db41908b10ad
F20101108_AABTCU watcharathanakij_s_Page_94.tif
e15b0ec9e24490bf7efff6538c0d0b7a
e7edc86304a868c41237a60811d7e0e910709473
112800 F20101108_AABSXP watcharathanakij_s_Page_29.jp2
6fc92937f9699e8d3968bd1344b734d6
3d97eebde77d4b9f15de0e96c3995ac64737e8fc
46086 F20101108_AABTDJ watcharathanakij_s_Page_17.pro
bfd04dd3763cbca3bad39cdbb9d99081
f55951d7cfd5d2dc8dfdc4d8b683a3796ad6dfe1
292762 F20101108_AABSYD watcharathanakij_s_Page_47.jp2
61663390216708597c6d0f01408637a5
2dc43e31fc1ed3530d60783f4910d485e499a50d
9269 F20101108_AABTCV watcharathanakij_s_Page_01.pro
fc4f1ff44075731cfc0c16611ec8c997
48e32f1d07d0691230e71846e916f0699cf49d28
112730 F20101108_AABSXQ watcharathanakij_s_Page_30.jp2
e5471eb795869980927081a8c720a5b0
7b2417bdeb403f38d1c5120c2ab8e203414be53f
39734 F20101108_AABTDK watcharathanakij_s_Page_18.pro
3fb5dce593ec7ade42ff1b555b5b7d90
31f855b2cd22838418c3631da100f8dd84480e85
50515 F20101108_AABSYE watcharathanakij_s_Page_48.jp2
47cb78474c2903fec9395709594222d0
263f6ce88d338437d5bf0785557a28239acbb2d1
1125 F20101108_AABTCW watcharathanakij_s_Page_03.pro
d0d33d1cd89a8d2065e308eaf1f418af
de2771a6172ffe185b397533850e97db2cc2df57
43575 F20101108_AABSXR watcharathanakij_s_Page_31.jp2
ff15749300d6ba29fe7e3cb7fa28f1a1
cc1753a5fb016aa3092e6236712073a210f0b54f
39509 F20101108_AABTDL watcharathanakij_s_Page_19.pro
1ac56c6c72fa6319e4a149c33c71c86c
ae1f9fb2b4c1194be14904c6e92efb3e24f3affe
106254 F20101108_AABSYF watcharathanakij_s_Page_52.jp2
eb11d1714eaf04d6c601f8526e8a33f4
4ff8d79db3b082a2b737e140f4aee0bb115854d0
13302 F20101108_AABTCX watcharathanakij_s_Page_04.pro
69f13f95250df58aa08ca1f7c80f52fc
62737eb5c91d1f02a7e8ec4a908eaccc0956c490
2490 F20101108_AABTEA watcharathanakij_s_Page_36.pro
baa314a032dd38468d72c48e0b0f286c
9dc279a8a0877a5eb7a8f20000a8cbcef9f42349
51425 F20101108_AABTDM watcharathanakij_s_Page_20.pro
ad48b2761b0b064e745b2f13964fd50b
0ce8d5c34174dad4cbede9fbd30d5f08f0e53735
110093 F20101108_AABSYG watcharathanakij_s_Page_53.jp2
22ab594be8906e81fea1a1bdcd742af3
5702ca1cf61477aa6e21350be15a2a55b01dd66f
84200 F20101108_AABTCY watcharathanakij_s_Page_05.pro
51449610b26318301bda7585f3f8c2c7
2e6332ff431f7e5aa9caab5d50f3df881da195b8
61080 F20101108_AABSXS watcharathanakij_s_Page_32.jp2
f70ab225945ddb61c543a99a41760332
d2a7426d330fa3357b89ad9e9179efa4ac4584f5
49585 F20101108_AABTEB watcharathanakij_s_Page_37.pro
e73efb285f49d5a4d68d7acbc18a1de8
c934cdb51b961345d83e947e2cd57f00ff738f74
6877 F20101108_AABTDN watcharathanakij_s_Page_21.pro
0340c6da51c1364f2943cabdbf5e3381
be677171e170261096125dafac9b2139b2b4116a
105485 F20101108_AABSYH watcharathanakij_s_Page_54.jp2
f6cc62d33505b24006dbdf20233a0665
bc46f445c1a7fb66f1b1486c72c9358b553f689a
33813 F20101108_AABTCZ watcharathanakij_s_Page_07.pro
ced62763b2fd6f19bca91268a0bc2652
a39d79318d5e0ffabb952d13d285c9879a591167
104867 F20101108_AABSXT watcharathanakij_s_Page_33.jp2
0e333fd5cf5e4189501669bab3d83374
44c9263fe40f332d16843eb5857076c0433732be
50184 F20101108_AABTEC watcharathanakij_s_Page_38.pro
d665911a6d421aa32ac769ab6c594753
5c8a16b42e2f4fd0074a24ea20f47fbc14e3c325
46755 F20101108_AABTDO watcharathanakij_s_Page_22.pro
3df8b790f49af937453f03d5e26e8f35
5f88e8a9f9e8d72fc19a03054b5098b0eba7dc76
109986 F20101108_AABSYI watcharathanakij_s_Page_55.jp2
d4ddf4f8823492365dc2c873e25a7137
9ad769d7c3b21079b20d6b2b5fc6dc151c9d732b
106961 F20101108_AABSXU watcharathanakij_s_Page_37.jp2
c2e9b9dec363a0698a2b2a03b8dbb3e1
b5b5c385a1ab0adfe07f71cd3d64f218f64d29bb
45252 F20101108_AABTED watcharathanakij_s_Page_39.pro
a9e514198ca925258b6f58e1d2e79211
897a9b8467163466bd260f036aa402ade6ff5511
48002 F20101108_AABTDP watcharathanakij_s_Page_23.pro
80742e98a86e1294106aa7c46af8ffc4
7bcc061d47ae2f9cf3aa1ffa43a16a22a5fe475e
110051 F20101108_AABSYJ watcharathanakij_s_Page_56.jp2
4f3833c5f2d78e88fa30c14054b28c62
1ab73f4252e729bb5ba707c2ff74849e18a4ecec
109692 F20101108_AABSXV watcharathanakij_s_Page_38.jp2
e793df0de74127018e8440a6393b79ab
8cc8fd96a3d938c108ac303123a2d576214b7687
46558 F20101108_AABTEE watcharathanakij_s_Page_40.pro
e758843b68c26fcaba0728c896cdec32
303096e576ecce6218976af11dce7f1959162be0
50494 F20101108_AABTDQ watcharathanakij_s_Page_24.pro
37d50902a1e58f8dbe14b64385c32045
08bf47cd8965d8a8fbbb80e0cd4842b2ce146d5e
110259 F20101108_AABSYK watcharathanakij_s_Page_57.jp2
02b21612495975cadb4b18a0c09e36b0
c9d12a8405cf906f7dcb907a7600601ddd03131c
102251 F20101108_AABSXW watcharathanakij_s_Page_39.jp2
5df3d6a633b9f31c549b0bdebbcbc565
9c3cf3dfb2ea2cf436af064764b057a522ae33e2
33064 F20101108_AABTEF watcharathanakij_s_Page_43.pro
e044710bdc6f1552a35417b06717259a
59efbb8c290ef4893470804d86e96fc5d7f03352
51809 F20101108_AABTDR watcharathanakij_s_Page_26.pro
5d9131f8a13b36a348aa72ca5fd91b78
02334aa6299821ad60ca225ad1e4f572166bde1b
106741 F20101108_AABSYL watcharathanakij_s_Page_58.jp2
d6089e7e38cff53bb0d038d312851bfc
98543d414910c87b7210310ce7a5100d4a69d76c
99010 F20101108_AABSXX watcharathanakij_s_Page_41.jp2
8627dc3426396ef21dc8bc7b14e8541d
b00d8897e8657a7a8ac2d8bb7c1f99904c3cc110
34913 F20101108_AABTEG watcharathanakij_s_Page_44.pro
0edad6c0ede709f8758c27c0784b0264
9b6fe1afd594ab0166acbb2cc26f6fcb2cb32e5f
203589 F20101108_AABSZA watcharathanakij_s_Page_75.jp2
7dce7dd4725a59f23113de011f113569
e6faa6810600fe1cf1dc6c2d25526c2dc13a78fd
51875 F20101108_AABTDS watcharathanakij_s_Page_28.pro
570c398cf56e62c82cce20297ed5dc1f
85d3b5a35b4da567fb529d4425b29adf6c834789
46449 F20101108_AABSYM watcharathanakij_s_Page_59.jp2
876968ecea84c5579e08615957cba74c
7aae27e42213c09f2b74d0f0e46177d0363330e7
94673 F20101108_AABSXY watcharathanakij_s_Page_42.jp2
8e7ab37aeea39a97b5c2d93efb309a25
3a3424a0110ea82e390f550bc42980468711ed3d
5897 F20101108_AABTEH watcharathanakij_s_Page_45.pro
967e846efea26d3defbfed5e96a778b2
57e763260622387bb51fa67b98d261f1930038c5
395378 F20101108_AABSZB watcharathanakij_s_Page_76.jp2
6f29a0dbdd0712e0d3e419baeeed2442
9d4ab381da709f22c9779779b09a233a06a18e75
50747 F20101108_AABTDT watcharathanakij_s_Page_29.pro
8ed2e9a8f07916be9a783b4edef01c30
35c3b11249ae158f5b2d37eb40d5aa62e4383b51
67371 F20101108_AABSYN watcharathanakij_s_Page_61.jp2
94dad16d3c5ada192acc53c5d43e21c0
d1820427f8f4174c58a808e66a407d85ae85d462
76615 F20101108_AABSXZ watcharathanakij_s_Page_43.jp2
00a5a2d8b435f6eb3398199c60b178e3
8811730a3d7aa7bb2bae4a78a16abf2593b3a86c
21788 F20101108_AABTEI watcharathanakij_s_Page_46.pro
fc9dd055187e2e3df4cc4fa8f05088e9
bfa5c02e9954e0c0d6e837c5c4c0ac4f19711002
144983 F20101108_AABSZC watcharathanakij_s_Page_78.jp2
565df576ea53efd953877485b2b2ebd4
98192bf2ab3afa822965d72f9d33264912c9052b
52037 F20101108_AABTDU watcharathanakij_s_Page_30.pro
b348e0951f3074f28fa81c2b07925b0f
ab41e33e67466d783e9a57ac9ccaf1f09960821b
60733 F20101108_AABSYO watcharathanakij_s_Page_62.jp2
8755db52bbe95000aa55b16e86eb8e2c
cd4bdc09bcad33e712f8c50d9ec96377cfcab130
7943 F20101108_AABTEJ watcharathanakij_s_Page_47.pro
2b27ef61ba9d98e6f8486b54791a4be0
45730efc427c417462d35bc8e589bf1b6adbf1b4
144771 F20101108_AABSZD watcharathanakij_s_Page_79.jp2
8f92cb1d65fc1f4e63729f877eaea649
93dc255347fb9a98bfc705ffb7ad4aa31b04a8d9
18260 F20101108_AABTDV watcharathanakij_s_Page_31.pro
d04f43bfc09854f34ed5879fededc489
40f1f3940246f9ed21f8ca5b0ddcb46f7bcb9f80
81135 F20101108_AABSYP watcharathanakij_s_Page_63.jp2
102da1dc22a0768db01bbe9d2f661b82
9293c05742b77bf15c0e348017551fc3479c83b7
18174 F20101108_AABTEK watcharathanakij_s_Page_48.pro
15b81b50a1e8fd0085d7e19eb49318d4
bc79b15ad76169aa6d289899949cf2464e634e43
95558 F20101108_AABSZE watcharathanakij_s_Page_80.jp2
0dbd8c6686fb16b893ff88ad8770b45e
31a19f4358f2aaa1e1fe230aae58efa1c8bafe9d
26602 F20101108_AABTDW watcharathanakij_s_Page_32.pro
0233514f1184f98d88897db3af295745
a509bdad77af580659afd15b0c980a8277709751
32869 F20101108_AABSYQ watcharathanakij_s_Page_64.jp2
f2f452cbf5ed6cb6a2a5f0e045168fec
85abbaebfb818f1af547dbe3c1b30395f5ad3264
36719 F20101108_AABTEL watcharathanakij_s_Page_49.pro
23432b49ee3f61399ce3db1d00f22406
c76755e9a0efa78e6d5ea8d575d1b908dfe4a7f4
115389 F20101108_AABSZF watcharathanakij_s_Page_81.jp2
9c62358b2350c8eff491e37786a17083
349b08024829525acb6f816efb375689337645ff
54981 F20101108_AABTDX watcharathanakij_s_Page_33.pro
3670aa54d216c1a89679ab45096206bf
0935f2523c566ae041644801034c72d55bc7987a
63448 F20101108_AABSYR watcharathanakij_s_Page_65.jp2
56ac54fb3fd7d8dcab505e5d28432d30
16cc641df31e78d083fde489ee148abc6b2803f9
48114 F20101108_AABTEM watcharathanakij_s_Page_50.pro
3b4a03ddc19e77fa6ced0821c6e85866
4b22af6ec3ba411f73279532f001662f1f5c266c
113533 F20101108_AABSZG watcharathanakij_s_Page_82.jp2
63ac52f895b7c7f92bc25c1616b05125
1fb4d6013b4bcd8dbd7ee8fd44d79c2a8e531b30
7288 F20101108_AABTDY watcharathanakij_s_Page_34.pro
eb6b4129565efcf1e6e225e23d8d9d75
dda4df0fb9646bdc630dc985a13d06938924484d
33273 F20101108_AABSYS watcharathanakij_s_Page_67.jp2
ca6f072caafe6a8885ab787f76a5739c
8c58f208b3ac020f92dbb1e5fb1f0d057e4cb3bb
29917 F20101108_AABTFA watcharathanakij_s_Page_65.pro
5195c3006e617417b7b25b301e3fcb0e
5603bf0ef1efb5013d451197fd0ab5d6b7dfd73f
49715 F20101108_AABTEN watcharathanakij_s_Page_51.pro
73a83bb5127398b89413dbc55f54a072
89029e17e2c1f65429578665a7741c51152c13d9
111875 F20101108_AABSZH watcharathanakij_s_Page_83.jp2
83ded3f7444e557fef7bb57948e9033a
01295a80754dd2e029c3d42600ee941c600df05e
2750 F20101108_AABTDZ watcharathanakij_s_Page_35.pro
c3d584039ec91286b62a88a9fbd2b03a
e857eb5e9217dbbb3b5370d600c4bacbd79bfe92
31233 F20101108_AABTFB watcharathanakij_s_Page_66.pro
eac0b28c8e198ff72f48d0d756e8d807
76367995ff6dae71b564a8134461b2dcc369025c
49545 F20101108_AABTEO watcharathanakij_s_Page_52.pro
0c6d7ef354619060b119cb32df29611c
4086feb7cbfd0da848020ff3c47505ac69a0c798
107219 F20101108_AABSZI watcharathanakij_s_Page_84.jp2
dac531189ab9eaf65e9662955950d695
795dde383f1c6b55cc733c5e3d0cd52e20480329
59959 F20101108_AABSYT watcharathanakij_s_Page_68.jp2
3a638dc1de4ded48ed16a74ce1b6f483
bca042b8e2967ba66dedb81b5c62454904557342
15253 F20101108_AABTFC watcharathanakij_s_Page_67.pro
e603ae409c805c245de51a5ff77d9214
8e44cdd5c925f426e8dd7d8f7f9b5575fd92787f
50725 F20101108_AABTEP watcharathanakij_s_Page_53.pro
e009d2c76750db6e08e1dd318ed7213c
a6df0f390c9ac9ea09e4fe9b87b5a63eafbfc7fb
92734 F20101108_AABSZJ watcharathanakij_s_Page_85.jp2
ef6c8b51d5ca5f273609f1b4a3cc0084
384b8394c8954a5254264181df90fcce10252d55
64458 F20101108_AABSYU watcharathanakij_s_Page_69.jp2
28e8ddd1b43de1c0d7083bb7e217c69f
eb85941d15aded086389c91b6b269294606ff9ec
27682 F20101108_AABTFD watcharathanakij_s_Page_68.pro
447fd8dbf73da80bcb300ce880c8b15a
91887c0bed765d99c5580de70d9512cd71e38a3f
48548 F20101108_AABTEQ watcharathanakij_s_Page_54.pro
85d35c8613cb1f4ac39579acd44fe86c
ecd14b7268a8bee529ccbca254a160477e0211b0
74362 F20101108_AABSZK watcharathanakij_s_Page_86.jp2
0ea3aaaa5655d7de4d2956062297f1c2
f5fd424e908133d2b45518209e2e1070ef70f105
406175 F20101108_AABSYV watcharathanakij_s_Page_70.jp2
7feb293611aa86bc9dad19d9f43efacc
3ad95b59fc2a08e6a2da9ad13dc7fcd877ec285f
31314 F20101108_AABTFE watcharathanakij_s_Page_69.pro
0a2d0873c81fc2012764e9711607b959
cd6ebbac2c0a942d58b3f743a56e5a0ffbe48bbb
51159 F20101108_AABTER watcharathanakij_s_Page_55.pro
c84f89a078b36d57132459aa724f13d6
a3c986bff5c1ae5f10814e892e9787980e09f864
62710 F20101108_AABSZL watcharathanakij_s_Page_87.jp2
cfb9142f32c78bf6d19d7c8fc34c6d35
d23cdd504a563b1afb5acf3770ded4176f305f40
428503 F20101108_AABSYW watcharathanakij_s_Page_71.jp2
31ca78c60d471b51f48517f8b6f1e0d2
4be182e06ddcdababf323ca0fff549962d99d0ef
2290 F20101108_AABTFF watcharathanakij_s_Page_70.pro
0747bf02d591257e70efd150ca73ea94
caffb795538014ede818a779655ceb5ab79bf493
51683 F20101108_AABTES watcharathanakij_s_Page_56.pro
4b394094207d59ee5b0277f1b1feae2b
6855067b8f43e5782291ae1df463987353f2d2f7
45049 F20101108_AABSZM watcharathanakij_s_Page_88.jp2
c6ca5d4384371e1533727cec2f2605f0
d361fcc4ab5a137f2defed4158bca621c8d21bf6
262845 F20101108_AABSYX watcharathanakij_s_Page_72.jp2
e3eefbb4b6e63c4a5ab80091aee47c02
5b6f0a875c7f0c04d2db5249453e971b621fb5e7
2428 F20101108_AABTFG watcharathanakij_s_Page_72.pro
0a162597b98f28d1b16a3010e591f04c
1310ea1027f81f94d01cc6c26100f75ed675f717
51022 F20101108_AABTET watcharathanakij_s_Page_57.pro
49b3f9864bd4851d8454b5d1db735a39
faaaa1a2c9fc09e1a6cdb1be4462115571cc819f
117503 F20101108_AABSZN watcharathanakij_s_Page_89.jp2
fa3c9735ef86e84769112628499ec68b
c00011824ad02606ddda895175770b398d9ce6e2
283138 F20101108_AABSYY watcharathanakij_s_Page_73.jp2
3fca8097651aec15bc986d3b6260cb3b
74e54382a7baa42de65f711a34588f5e38c974bd
1998 F20101108_AABTFH watcharathanakij_s_Page_73.pro
a5caa64c5f5a430ec57f24e544bc56c1
3832942c9be5cb03a001c19cc5957cb00e476f5b
49276 F20101108_AABTEU watcharathanakij_s_Page_58.pro
b031f837292a8e7f7cc1990cc6152308
a615966af071a67b32a2f7a1db3e27a5727132e0
127556 F20101108_AABSZO watcharathanakij_s_Page_90.jp2
70a4e4719ef0a88540f0d0564022f5c9
2badf2808195093bb1ff60d7d61a43f24bdefaa8
431438 F20101108_AABSYZ watcharathanakij_s_Page_74.jp2
f092ec0e8cdb3b442f6e2b27969bd033
2f5821bcec9ac1b464f7127fd3e52177a0c0d86c
2049 F20101108_AABTFI watcharathanakij_s_Page_74.pro
021d75a91bdb66940a3729e590ff2a1b
0e3e65d561391cd5d41cfdcf08513c90b73fb507
12408 F20101108_AABTEV watcharathanakij_s_Page_60.pro
2cc3187048e1149cc70f5d490b774c32
2a7d67db400cb5d42cb34bdf923b10923dd70e81
133587 F20101108_AABSZP watcharathanakij_s_Page_91.jp2
c88d58f061a417084b6813bab118e276
451c74d649d521214e2ff9a991844e16e543aee3
1874 F20101108_AABTFJ watcharathanakij_s_Page_75.pro
5950eafb6b3336f5d0830794b1c74cb2
f8482f1d3e366524b2da62d0721977d6ca72d6dd
28003 F20101108_AABTEW watcharathanakij_s_Page_61.pro
c1ab6b719f77e96d8abae7942f167ce1
6dfc4eb7b7931ace03bc1f95ae40177ae1aa7ba2
133016 F20101108_AABSZQ watcharathanakij_s_Page_92.jp2
43d8f11b41ca182b500d73aa6479030d
50f56374f6ef6ee0f07339a426792eef55d9aec4
2833 F20101108_AABTFK watcharathanakij_s_Page_76.pro
a605e3b557966daf48a4d78d56c1a6e2
0c53a65a1f06ab71af44969bad63f9ab5473ce7d
25557 F20101108_AABTEX watcharathanakij_s_Page_62.pro
bb5af0a11e5bf5f73f43203e1f555ebf
b78b696c882ccb61908883008387f1c10105e31f
98535 F20101108_AABSZR watcharathanakij_s_Page_93.jp2
9630e6c04df9216c89704be561c4d871
dc052d6293ae9aa4bfb6af7b806c6db00a990719
1795 F20101108_AABTFL watcharathanakij_s_Page_77.pro
0d41a3f444de44598f8af77a63cd95ba
9bf477edaa8bd32a9603d2a67aaa5c92f6d17891
33684 F20101108_AABTEY watcharathanakij_s_Page_63.pro
875b92fab7cf1f70f682d7d57283d231
b6c20481f5a4b95b41c9ae3696335c9649fe099f
39665 F20101108_AABSZS watcharathanakij_s_Page_94.jp2
48f262ba901d51818ad30eaa55ce671d
2acc6dc1f7586c45fe486ffdbb5250c7445a2fa5
15764 F20101108_AABTGA watcharathanakij_s_Page_94.pro
b7b56f0b237d202ff3008e3bf129a45c
d5c60ae627b21cee23f7e8deacd4af00b46d58a3
2311 F20101108_AABTFM watcharathanakij_s_Page_78.pro
b54f6e2e2e6215cf0a251bb49b1d631d
c619515dfb58bb8b10eaf366193d12905f79ed48
F20101108_AABSZT watcharathanakij_s_Page_01.tif
51d5c31d912b23491149fc1d5476ef1a
2fceaacf68c0786eb1cbc3bd58c7b00a2ade44e6
96 F20101108_AABTGB watcharathanakij_s_Page_02.txt
7bd989cdbefba20adae0d8e61ff6cce4
0cd6652a4477cf3f201620f507d32bf7eb94fae4
2321 F20101108_AABTFN watcharathanakij_s_Page_79.pro
1162a64184b3b21624d58e9b9199fcd1
4becd6cb8bbfccde8a345f9ce6f1725b0b797c98
15365 F20101108_AABTEZ watcharathanakij_s_Page_64.pro
0549fd0c8df2bdc90e1446b99774e7e3
f1fbec55759fcbf3145e3c98bdd3bed4aeef535b
97 F20101108_AABTGC watcharathanakij_s_Page_03.txt
a76b42bbe19af96df00696c4fd9d7b58
0464a1d866c4149d724425b77cbd5d116e923e97
43244 F20101108_AABTFO watcharathanakij_s_Page_80.pro
7f0b891f1ce9dc71d16aefdaf80e920d
0b6826847c4ebfc4af8e9b2b095a7058d1e85e4c
F20101108_AABSZU watcharathanakij_s_Page_02.tif
588d0cdce2303859131aa07a401fb698
465023023c1ba73ab7c92c3e1508acd57adb07cf
569 F20101108_AABTGD watcharathanakij_s_Page_04.txt
5520b6651a4089f12502dc3834fef09a
149668a98e8ce7058cad2379d205c710f010f302
53497 F20101108_AABTFP watcharathanakij_s_Page_81.pro
a3b7644e703812413cd06b12fec66da6
c13eafa8c40e11fc5a339ce67e35f7a6efadb186
F20101108_AABSZV watcharathanakij_s_Page_03.tif
864284220ad42f2f76746a73620d320e
5f37446daa4a61a341e2944a2da4eb2bc244fc05
3694 F20101108_AABTGE watcharathanakij_s_Page_05.txt
fc02237d1ab2e5bba65476942d2e7468
3d3c17673a32fee27e954a98a00d0c743fc7e3b4
53223 F20101108_AABTFQ watcharathanakij_s_Page_82.pro
4ecbb43782938e35674c8f2b096a85e7
b03a3687150eafc6924ea85f84457bed3f47deaa
F20101108_AABSZW watcharathanakij_s_Page_04.tif
abfd31327f57c1fc9be882ab01d8a3ad
65aa53d9bc0c68b47da942c452f4e304274234c2
3073 F20101108_AABTGF watcharathanakij_s_Page_06.txt
79e43edac0ef868424bfe26b2686359b
6f2cfbf9abf905999fd608776ee5550e6b02095b
52905 F20101108_AABTFR watcharathanakij_s_Page_83.pro
7b99beedde704ba0d8ae571a84d52747
a661ae64e6d142b2ae7e738d429bc3520a6ba569
F20101108_AABSZX watcharathanakij_s_Page_05.tif
1f258168d6c01a3553c3a10227297ee4
71aa3c1870b59c150f1fd58a4f3d020f4837e83d
1485 F20101108_AABTGG watcharathanakij_s_Page_07.txt
f3280d26c72e5047bf4679703360f27e
ec10c3bdc7749ef37d25f77f03edcb13c93feb63
49396 F20101108_AABTFS watcharathanakij_s_Page_84.pro
a800f256d945200c08cb0b7ba386314b
e3a3cccc094963dcc209f1c49f74702e10a16f80
F20101108_AABSZY watcharathanakij_s_Page_06.tif
5c6a421a0a3830f252634e9e69c8e70b
6c6680460a1ce22cdfc025b0108a60029a0d77ee
2351 F20101108_AABTGH watcharathanakij_s_Page_08.txt
a64bc21cba77b9b0f80bca704ef8cfbb
754fd00c9306dd51434581b454326172879ea257
29290 F20101108_AABTFT watcharathanakij_s_Page_86.pro
b7fdbe85c3ca82409ef4b90763c253c3
488e166bdbacd057171362dd6bd22de6281fe680
F20101108_AABSZZ watcharathanakij_s_Page_07.tif
65bfe5d30a7b9846c07e1d7488e37a14
c2e598003d4c8654940581b9496f3652a8f2dc51
1696 F20101108_AABTGI watcharathanakij_s_Page_10.txt
e93305def28032748eb9505cef1d821c
186014048be721e9213fa6fa299c3675c3ede9ef
23889 F20101108_AABTFU watcharathanakij_s_Page_87.pro
6c229565709aa631e1e4419ba2614067
c66109badf38277d2673b3736e5cce90b0fabcc3
1922 F20101108_AABTGJ watcharathanakij_s_Page_11.txt
20d612fc9b8183b208422b4e1d31873f
8d94bcb7ddf97773000f22595c60cd6e4ff96284
16890 F20101108_AABTFV watcharathanakij_s_Page_88.pro
6a23ba027b70117c69d1d25e9e75d569
fa57da41788e440c49d1490916aed276ccaf00c1
771 F20101108_AABTGK watcharathanakij_s_Page_12.txt
2b92aa35105a32619fe07825471aafee
12c8406644a368d41c0942457659af8dfb9ff2e4
54271 F20101108_AABTFW watcharathanakij_s_Page_89.pro
55ebeb7d8e15d71bd14f3fbbbca13948
996eb5410429c088e5654559b65bf29cabc57b45
2076 F20101108_AABTGL watcharathanakij_s_Page_14.txt
b37c2abba10340ed816f5e90fefe7a8a
916768126e13e2532507abc212249a2dafb0505d
63809 F20101108_AABTFX watcharathanakij_s_Page_91.pro
7cf1f4073102faa819396460a5934ae4
442bf2a156d29d1f2cc77c7a665f970560f51283
F20101108_AABTHA watcharathanakij_s_Page_30.txt
835d603bd47466b7cb50ea479b9a5ec1
289245219a0b1e3f61193bfcd4d40088c24c7697
2139 F20101108_AABTGM watcharathanakij_s_Page_15.txt
82448b4870ac47041273fe5dcba1c42f
4a54bdc176a7e00caa32ef56a3e4c9bb3019f966
62865 F20101108_AABTFY watcharathanakij_s_Page_92.pro
3bbc45bbea2431aa223a0b066bb06857
e70160fc4825b9f2730c0ff001e009ee5f9641a9
1866 F20101108_AABTGN watcharathanakij_s_Page_17.txt
9f8e0b1b66a95b5b8605a91a2e57c685
48db7933cbd6af7e611b8b69839c6595867a2191
45841 F20101108_AABTFZ watcharathanakij_s_Page_93.pro
8f88cadf40e2f42a7cef614caf5e4692
dc0b0083bafa6fee8663e5b3b91cfdd76cab0094
1222 F20101108_AABTHB watcharathanakij_s_Page_32.txt
0603d9be1ac951f3f94c872744eec0c4
983a4761a0dd9efb65a842fb79ce13fba0618e98
1611 F20101108_AABTGO watcharathanakij_s_Page_18.txt
a51700c5e78e6b66b1e85154a07fcbc3
7cf1d4993fa57b172a06f41e2269278c862a3f33
2299 F20101108_AABTHC watcharathanakij_s_Page_33.txt
1e9426ab65c929dec49ca4b46a18e992
4c2b2491dcfff2586298a633fed0dde850e21608
1601 F20101108_AABTGP watcharathanakij_s_Page_19.txt
568ce8bbe629a5794d79ea8d9d6def93
afd7da9eb4d10021f6009ff5ad2a11e3e5f6cac6
369 F20101108_AABTHD watcharathanakij_s_Page_34.txt
5520d5647f15a5db7484d4b9085a1204
5fa73c18548bd2ba996054759cd16c402090cc3e
2063 F20101108_AABTGQ watcharathanakij_s_Page_20.txt
6eb816201d6dc20ba49df18b3af59286
81c2ab972afe83d8960170ef144b72844ca6a882
164 F20101108_AABTHE watcharathanakij_s_Page_35.txt
d858e6105d8da4e43289293a135033d1
95bd2e0897529e0876b2f9b12908805644c7c3f2
327 F20101108_AABTGR watcharathanakij_s_Page_21.txt
952a7e04ea2b6bed38e87294680324d8
2bd969f3935953ede35cd0febe4b75de22cec054
152 F20101108_AABTHF watcharathanakij_s_Page_36.txt
15f0452df2ef0f4de20ec5e2fef8b171
984744737f657304ca1988f21807f7bbeeec347f
1985 F20101108_AABTGS watcharathanakij_s_Page_22.txt
8ffa971fe07e73af400165b68be0cbe7
ee84149c09307bbc96084e1a61af1b8bbea4a60e
2068 F20101108_AABTHG watcharathanakij_s_Page_37.txt
9de1ac4198a3ef84294b2963e6197c1b
715662725f8bc1e173337a98fb2f7a8235bcc6a0
1961 F20101108_AABTGT watcharathanakij_s_Page_23.txt
6ce348c6b4ac3bd812116a0d48964d24
f401bec04cf3ede3071ab3a3c3e0d88b5c4c69f5
2021 F20101108_AABTHH watcharathanakij_s_Page_38.txt
6f2942d915568706ebbe2628d0e6c2b3
826f16799a46f8f85ecc2759b1dedb222d20e0d6
F20101108_AABTGU watcharathanakij_s_Page_24.txt
eb648a6efb55ebf1ec2c5c87d86563f2
d99c604a9767fa34bad580bb39364de31e4a7a77
1796 F20101108_AABTHI watcharathanakij_s_Page_39.txt
d60bb81e6adcd75b52dc5ea4f9debcda
364abb38f27f41c00de891660b49d9ecd5242001
F20101108_AABTGV watcharathanakij_s_Page_25.txt
572249b3b7c1acb6e13e0136931890b8
43b6cd4855ed95555e5b6241d58cccc54c7fdec7
1913 F20101108_AABTHJ watcharathanakij_s_Page_40.txt
6b2122a0874a8943693600dd1948fb9a
e236cbc0e063f942ca8e393f414852b8feb94643
2065 F20101108_AABTGW watcharathanakij_s_Page_26.txt
601d81f1fe7935b65b1a2e12a6b380b6
c94c170285c51525c6541786eb479f475bdb967f
F20101108_AABTHK watcharathanakij_s_Page_41.txt
a7028f7c865c79921d956a5a732e3a68
49fe8b5c4fadfa108c87ef8f7966f3c9f73b15ec
2115 F20101108_AABTGX watcharathanakij_s_Page_27.txt
b0f79a3de5f291c76f50b100ab03e262
d97942dc8c5b0f8fc15efff2d59e39f6c1111ec5
1276 F20101108_AABTIA watcharathanakij_s_Page_61.txt
4e1c1560192a09798a16acedcebdf6c9
1c57fa5b81337e883bc98e2e0f9451943b835b53
1728 F20101108_AABTHL watcharathanakij_s_Page_42.txt
37432710c15108f6284bf8372a816126
b76d3f24e94023b6de99e10de8ea14ffc6b0b531
2041 F20101108_AABTGY watcharathanakij_s_Page_28.txt
c61d5a7cf9af12925ab84407e6300d64
e6034de1534ba2ef8b0f37b73406322e96762850
1140 F20101108_AABTIB watcharathanakij_s_Page_62.txt
27d0dfdb933b93ebfc65c93c95edac8f
e648b424bb1ab644cb72bc631381d80659a8bc3b
1368 F20101108_AABTHM watcharathanakij_s_Page_43.txt
7868b31a4128e3a26b70424356546ca6
3d0dc31056b1d27dbc0b2c1339adf4594d6f4211
2003 F20101108_AABTGZ watcharathanakij_s_Page_29.txt
5c0816d53de08d0714e6b33a331043e8
8cf1abb33bbb4877f06e707f756acfd6b8801b74
1473 F20101108_AABTHN watcharathanakij_s_Page_44.txt
8b43a6521ec8493a877235f50ab59c92
c34b7ba649067696be957e0112858cbafb577be3
1418 F20101108_AABTIC watcharathanakij_s_Page_63.txt
2f82c350f8023d0b191023bb6262feec
45da95e5ddd8c336fcfaf1b63c51133200c1ddd5
324 F20101108_AABTHO watcharathanakij_s_Page_45.txt
1811eed4ef7a04963a0258b505aee87f
038a3146807e7c4799bcf1af5a8c1f8282f859cd
816 F20101108_AABTID watcharathanakij_s_Page_64.txt
f5d50575e23778a6115c6a0356ab1315
06fd3075bf21689d070e4e01156c723746718f1f
965 F20101108_AABTHP watcharathanakij_s_Page_46.txt
a9213de1f16025a850c1e54a669ab2c7
3d8014c11c5d6bb5cc71e3024e2acbab6eb7c256
1585 F20101108_AABTIE watcharathanakij_s_Page_65.txt
1c7d27cc7b089c30dc7ba0fe1947d6cd
53f093e49e762f7ef6ecf1bd1b71519f84ecfb06
451 F20101108_AABTHQ watcharathanakij_s_Page_47.txt
3b868f97d6c26a45ea8e840a243f7884
73a7070318bd55de880eda7afbf71d9980426c18
1794 F20101108_AABTIF watcharathanakij_s_Page_66.txt
7492d637d038ca8fcca52e2b4c0afaab
477f1338097ab5735306e2033fe7d92abab31acd
1595 F20101108_AABTHR watcharathanakij_s_Page_49.txt
f0d57254eb0c72feda28789f41788dad
c31eb84fe01240351f568726316e42625009912f
840 F20101108_AABTIG watcharathanakij_s_Page_67.txt
dc16549a0a0874d40be622ca238e6078
ccb9899bda7fb299fab2faae0a63548467cdae02
1962 F20101108_AABTHS watcharathanakij_s_Page_51.txt
ecae37c8e94cc140597eb7dcb79fa364
059346b1130e8e86167821814eca9ea767a8a251
1498 F20101108_AABTIH watcharathanakij_s_Page_68.txt
d75f9f27a3bc97c66535d632b6b31f9a
52ea8caf8371a11a3ee646b9a9a4aca5e4c4dc6e
1988 F20101108_AABTHT watcharathanakij_s_Page_52.txt
22640417c8a51aed146cd9275fedf2ea
4f304a22d9ddb9e8d74d2380957c0be1d07e7e2a
1878 F20101108_AABTII watcharathanakij_s_Page_69.txt
295091845a391f0cfb14a0192266e6b7
e3a6f9ee669805987201a38d358b8706ff854380
1920 F20101108_AABTHU watcharathanakij_s_Page_54.txt
2eefb9cbaec148e84ffe86cee15e38b2
be83f29e0f6393c2a627325ae36cb99ddad836d7
109 F20101108_AABTIJ watcharathanakij_s_Page_70.txt
5cb433c58b90ff7fd992d3b413e6128c
3355c673498f0469a08757cfb88a3db7e7fc60d8
2024 F20101108_AABTHV watcharathanakij_s_Page_55.txt
0ee50d0da18e9fb91a5290c48a0b0d0b
cfce4295ca640b678954a53a52a3c4bbb89f7181
116 F20101108_AABTIK watcharathanakij_s_Page_71.txt
01bef79fa0abb71643448e8cc2f6e179
254d5931041dc230a058e674e9233911ade8920e
2043 F20101108_AABTHW watcharathanakij_s_Page_56.txt
ef860ba5455c0f67f40e0893c10373e4
06c76d1ab6d932ba4c607752d7dee228107586b5
2509 F20101108_AABTJA watcharathanakij_s_Page_90.txt
7d6d2ab18f2cc5909b26db8fa2978f4f
c1419daf41d169315fcc83abbf2732e03b5fec59
153 F20101108_AABTIL watcharathanakij_s_Page_73.txt
828d7cb7d819446cce58b11d043c7d70
1bffa2d62382819de38aa1bf494d16e049974184
2022 F20101108_AABTHX watcharathanakij_s_Page_57.txt
a0ba500415bcfd5eebe603b5bb66716f
640952ec9672988340eeac3402a631ed63cfecde
2604 F20101108_AABTJB watcharathanakij_s_Page_91.txt
4dfad3a240fbb4728cb97e19de5e204b
28373c4c9844563dca89108fde1f4caa3b8f753f
144 F20101108_AABTIM watcharathanakij_s_Page_74.txt
658e094f732f8289c8867ae9f627afb0
5039f7cf7f6e7736c204792ef6b1d429ad6a51e6
1953 F20101108_AABTHY watcharathanakij_s_Page_58.txt
99d6e22d40faf0b3690ae337c7f9b799
87399d641e4e8133690301a1789a81fe39a71766
2587 F20101108_AABTJC watcharathanakij_s_Page_92.txt
5bb0564eb6510b40621f7737c7bc51a0
130b28e33d63d70635fdba71f65afdbf5e8b1bc7
132 F20101108_AABTIN watcharathanakij_s_Page_75.txt
bb3fc4521397ddc474769f58f98c46ff
b65da00a5e68364aef3b49964c9191f6882e182c
1118 F20101108_AABTHZ watcharathanakij_s_Page_59.txt
e13da8e62ab8565000b719ae48703ab1
21fa2901f7fe25eedeca0ec7852a28dbb5f0d778
221 F20101108_AABTIO watcharathanakij_s_Page_76.txt
9a319b578394c253677175aa77ba16ba
be7c6f5a5526f96d7272959c08da38751e1ea0e1
1911 F20101108_AABTJD watcharathanakij_s_Page_93.txt
aa0211e25ad642b2adc7341eec2f18dd
f581e786c347468cc50fcb632ce951278a753da6
131 F20101108_AABTIP watcharathanakij_s_Page_77.txt
3afffb2a80224bd008e0fac1c26836bb
44326285f5d079781265431e01e9f417d94044c6
664 F20101108_AABTJE watcharathanakij_s_Page_94.txt
93cf3c91528bf3f4cf67ddbf22a14b84
dab4a69e757a0b7bafaa575849f95baf36ad28a8
204 F20101108_AABTIQ watcharathanakij_s_Page_78.txt
e59842c83657dfa847dddb5d16900772
775aaf5240904677b7e4bee7c6764bdf7558a32d
18326 F20101108_AABTJF watcharathanakij_s_Page_48.QC.jpg
cdbe666cffc364fddebad5d23d845fa7
c6aa590cd8fcfac20b39a6d09b4ff1e0345b6acd
170 F20101108_AABTIR watcharathanakij_s_Page_79.txt
3cea0498508983f0ff69e7a3483305dc
ccb2a55a0179408ea244845773931ae8a3e491e6
8592 F20101108_AABTJG watcharathanakij_s_Page_56thm.jpg
f28eadce1621fa6dbd56381cd06a70e1
ec92175949c8c18cbd06119860223e686655527d
2145 F20101108_AABTIS watcharathanakij_s_Page_82.txt
eb1109ec46f373adea090629ab58a119
0d5d1b5a9a22b101a7392b87d369ce1e11ae8631
24980 F20101108_AABTJH watcharathanakij_s_Page_33.QC.jpg
581c3ddfdfa93368d57eacf2db3cdc89
01c2e7512a9eb6a25d487816482bcb50f4a9db20
2086 F20101108_AABTIT watcharathanakij_s_Page_83.txt
5aa79e0a6b2ac60bd14c6334cc9c5e99
aa6ea4368df4d8d5b00e85f0075d614baefc39e9
12664 F20101108_AABTJI watcharathanakij_s_Page_45.QC.jpg
1aa86de359bbd3be8f1b20a0335855eb
8feb4ae1478b96921d6478050e75802ee753096a
1952 F20101108_AABTIU watcharathanakij_s_Page_84.txt
5ba6f6c3fa3fa1758a6ed71e0c526e9c
b0f21af3616551a4cc27c7f8ead40d77a609b100
23082 F20101108_AABTJJ watcharathanakij_s_Page_86.QC.jpg
a81445662c6d1601d7f5a468d7d167da
48cf19e4380caa998c087717c1a54e708415a8bb
1764 F20101108_AABTIV watcharathanakij_s_Page_85.txt
0d7874571e1cd3c5971a287127026990
880d94aa8ac1a0e0f8140712fa7273ecc0116861
33770 F20101108_AABTJK watcharathanakij_s_Page_24.QC.jpg
055813c9dad8ce832d2aacd2c3f87433
f8a31b7685a6b1c9c6a3662412f1a9ebfa646139
1195 F20101108_AABTIW watcharathanakij_s_Page_86.txt
4af2e84a0a878914b954544e21a4fa56
5f378d44857adc51627a5763ab0da9c39d0359d7
8388 F20101108_AABTJL watcharathanakij_s_Page_58thm.jpg
86d068775a644651cdb5e6fa81e56075
292f916d36ce6fcff24b875694e8e41673ae73ff
980 F20101108_AABTIX watcharathanakij_s_Page_87.txt
1722429409c2767f5f96d0132078c073
65e2a742f2d9acae32d46ade45a1b631256409e5
4734 F20101108_AABTKA watcharathanakij_s_Page_62thm.jpg
933ae0558869ab2683d5b0995e475385
9edbbf411d5160ed0571d44cd18e513b18503255
8022 F20101108_AABTJM watcharathanakij_s_Page_22thm.jpg
514fb97feee5c9867418ba3aadcb674d
7afa983924b2f0b9a7fcee41871e3feddc7fe162
696 F20101108_AABTIY watcharathanakij_s_Page_88.txt
4c8e22f8bff30b1926c72ea6d68abb72
48c6b3f51255dfe88f81ba6904e6651f009e004f
3890 F20101108_AABTKB watcharathanakij_s_Page_69thm.jpg
e1c4d1a12ac272fe12c915aada565547
ad3318fc27796703d68966be879c5e0a84eec322
8882 F20101108_AABTJN watcharathanakij_s_Page_83thm.jpg
df7b7084821570f8d6d9dd3b0372b3e2
01e62b2fff3ffd9512440f1571317e9bbd30e4e5
2210 F20101108_AABTIZ watcharathanakij_s_Page_89.txt
56e69186d0cc1ab01abc8dfef40f340d
de19d4fb0e413863e8655f468332d2c85e0333b7
3910 F20101108_AABTKC watcharathanakij_s_Page_66thm.jpg
e593183fbf25637a495e3574b6c407a2
9441c1cf3c7dfbb9511ef8cd2e7b6bee3418716b
2141 F20101108_AABTJO watcharathanakij_s_Page_67thm.jpg
eccc81746e9fdd5a0030d9deb8eadb0e
59178e2605b96916e027e8cdd70ad494e0427453
8858 F20101108_AABTKD watcharathanakij_s_Page_01.QC.jpg
fdd78e9a73bb78a3f904da2cf61b8d2c
8f8a690abaf1c029f58996fb3c147f574600c7a8
14445 F20101108_AABTJP watcharathanakij_s_Page_12.QC.jpg
da22051f0cd8b090dee88599a8b6637b
c0ed283b6ccd5118695404c5d676c7015e166ca2
5728 F20101108_AABTJQ watcharathanakij_s_Page_86thm.jpg
cc89b3ee3d305e7d54daa0b37af1c93c
3590301a1ebfb908ba02fd93bba8b450dd66248f
6512 F20101108_AABTKE watcharathanakij_s_Page_33thm.jpg
1f2e5d0afb382114e86cd0e9c24bb884
ae8cfc783a0098f814875e983c3cf2d1a26f6048
8878 F20101108_AABTJR watcharathanakij_s_Page_89thm.jpg
94757a5b9f8435e82b9d6721003fac4c
363e6c96e9afb15fbcea8b092a0d22833c91b809
32560 F20101108_AABTKF watcharathanakij_s_Page_23.QC.jpg
bedc16a2bc2943932bdf791279cdfdf7
0e2d75f2627abe298967455f6e0d7d816790d893
8289 F20101108_AABTJS watcharathanakij_s_Page_20thm.jpg
bf1e118661750b8bfa3236509240f073
4e5d9a56edd86d707923551e60e00a1fe06c4ea0
16093 F20101108_AABTKG watcharathanakij_s_Page_88.QC.jpg
1010fd8fec145ce3bb3c06f9620e393f
1a59924699c25f8cb9c724465fc514ce365a91d0
24809 F20101108_AABTJT watcharathanakij_s_Page_06.QC.jpg
cedcc1fafc37cd9aafab888fa79133c0
882eff63a4c5f1e93c684bd74f0b914b23c6a744
3708 F20101108_AABTKH watcharathanakij_s_Page_74thm.jpg
57548264cb398458a6278f60a9a4e575
dc8bc99818a0528fe763153dc751e1e972c5abc5
3000 F20101108_AABTJU watcharathanakij_s_Page_72thm.jpg
48606336820cc69693a9c064c3115acb
5c9501b3fc198bb87ef9cbf5e0d26131c349a2ea
6942 F20101108_AABTKI watcharathanakij_s_Page_93thm.jpg
f56e59e8cb451583021831c6d4c3bfd0
a9f94bcb9862b63160fef03211ac4b09cd58a6b0
2611 F20101108_AABTJV watcharathanakij_s_Page_79thm.jpg
a7691fa1c7de0aa6b540f2a9545d2140
d6977993e8becfef57912e37bf83fcafc9a030ad
33883 F20101108_AABTKJ watcharathanakij_s_Page_52.QC.jpg
c6bb7874aeda03d4bafd6695e2f6bd5e
b475a4c711c18f14af95406ecbf3c71cd85ae764
7874 F20101108_AABTKK watcharathanakij_s_Page_17thm.jpg
ae479dd0048545317002cc41718b4740
fc4c914f614738a8846750befed56e6bba30a6f9
15625 F20101108_AABTJW watcharathanakij_s_Page_63.QC.jpg
237c04b3c4c5a0f65f0a2762dc7ac0f6
261e188e40f821e00077fb6cd056119f2ef8f5d9
9037 F20101108_AABTLA watcharathanakij_s_Page_77.QC.jpg
c2b785c53bada9363cdc15890850fbcd
5eb1ee9cb98fc5f4017f894d9efc5c2cb81baaea
7234 F20101108_AABTKL watcharathanakij_s_Page_11thm.jpg
04367645f5fa89344a6b8edaf24090ac
0153904f02f9437367629eea5e75fd2a02638c56
35699 F20101108_AABTJX watcharathanakij_s_Page_83.QC.jpg
44b802b3e7bb161ba274878ba5a19a53
94a041b0d7d8db2676df401a50887905a91ff5f5
7086 F20101108_AABTLB watcharathanakij_s_Page_18thm.jpg
0d7e4d498cabc0ad643bf82c122d2f8c
4ef70b89ffa38ec6e860816518cfef9c5676a97f
16680 F20101108_AABTKM watcharathanakij_s_Page_46.QC.jpg
98d4c31c4d08a842e0fb7373f7a39a8a
2e31c062b73be2421d0d542f265653ba5c270d61
30919 F20101108_AABTJY watcharathanakij_s_Page_39.QC.jpg
d8e2c488af1e39acfd468f849e860cd5
a219e23b41940631314fa4f4626f7bb29fb3445d
2849 F20101108_AABTLC watcharathanakij_s_Page_21thm.jpg
4942251c441c2ae9d95a1bf4fad1ed58
edb69aad23cbcce792f0b8adc8fd4e9e4731e9e6
8067 F20101108_AABTKN watcharathanakij_s_Page_13thm.jpg
3ccc5e3204ac53c5889282dbf53a62ff
4e1654798d06831a699aec7a8edde02546dbde00
554 F20101108_AABTJZ watcharathanakij_s_Page_03thm.jpg
e255a8c8990753cd74aa911c3b189f21
0b31773a4d2bab303aa4b17005e8543aaca9ba4f
5561 F20101108_AABTLD watcharathanakij_s_Page_36.QC.jpg
018890277252dc3452c519377ab5a327
1023d6fbe2b10730816828a0ff14064ada60c405
5103 F20101108_AABTKO watcharathanakij_s_Page_47thm.jpg
b9633dc94dabe52562c285213ed71419
ec45788174e348ff724c8bb1472af529dbf75d99
8646 F20101108_AABTLE watcharathanakij_s_Page_57thm.jpg
abe87751308cbf91271396688dbcdcf7
72eb815f2df7f589beec511f20c05f77234c7f30
7859 F20101108_AABTKP watcharathanakij_s_Page_79.QC.jpg
7fb3c2f029b25e4d914531bf1c6bc16a
dca429ed6c1f04a15d96de1aaa2d86e1d131e475
13993 F20101108_AABTKQ watcharathanakij_s_Page_31.QC.jpg
217f241a38ac9aba6dcf5e7dd2df8bef
f382631eae3945b27ba874d38e43a05ae1c2440d
36409 F20101108_AABTLF watcharathanakij_s_Page_15.QC.jpg
2f57185a1c4f12c7cfb61f2c03d4372a
90029f16a39d2b26bc54ae77c2598d3d1c7eb45d
3117 F20101108_AABTKR watcharathanakij_s_Page_94thm.jpg
66fa2fb92c77c38c89f3db81889dbed7
47e02e168d4b2141a3f0e3c5c9faa8aba0ea6ab3
6504 F20101108_AABTLG watcharathanakij_s_Page_43thm.jpg
fb2b13fd5ff7fd10f31428a9b9f961fd
d58d02511a6dbd3f51b0e75769b96eaa1a2d97f4
37774 F20101108_AABTKS watcharathanakij_s_Page_91.QC.jpg
787007b5f54827e7ac2c1d35b1502e55
f6d5cd6ce977cee4523227b76a110b9c460ff80c
36083 F20101108_AABTLH watcharathanakij_s_Page_28.QC.jpg
02f9f2e3f4654d71222613e4ebc7f71c
6a32a1416cf2e2b1e47b6040bd51459c8467ed7c
8263 F20101108_AABTKT watcharathanakij_s_Page_84thm.jpg
412cfadfffc470ecd70de289ee1cf5dd
48d940546222ac029ef672f778019f2c293173e1
8060 F20101108_AABTLI watcharathanakij_s_Page_80thm.jpg
086c5404e24a0d8c968a5cdf8183f3e9
2f0c0caf2fe0d507c6df35f564eae5799d20ee58
32999 F20101108_AABTKU watcharathanakij_s_Page_54.QC.jpg
e6b63e9557ffffc61e13e1aecd8ee65a
1fc140df2b7827508c8546a9221b64ec803739be
7851 F20101108_AABTLJ watcharathanakij_s_Page_40thm.jpg
c5677c5890453e296e3685ce26f691a5
4041d9a5d822cffa02e4be412606f8f0ba14558d
9477 F20101108_AABTKV watcharathanakij_s_Page_91thm.jpg
be7295b59541602680cbfdd25aaf1297
7fe78f557eb8cc98ec015badbc523df53fbcac95
12231 F20101108_AABTLK watcharathanakij_s_Page_65.QC.jpg
05f4e2be92fb18ff811e2b99e5fee642
444039b3bbd12f803201da45f7536ec09072fb30
3616 F20101108_AABTKW watcharathanakij_s_Page_76thm.jpg
b56a9e71ad23682fb3dac9e624598a4b
3ed2ccbc24ed87f5d32c47e17be4757605538920
3627 F20101108_AABTLL watcharathanakij_s_Page_68thm.jpg
3ac058d662eee0ab06f5536154e058cd
562ed6857cb415323e765df12ed98fefcdd03cf5
3481 F20101108_AABTKX watcharathanakij_s_Page_45thm.jpg
289a92695b447f895765d37441abe40d
1acfa79967598341089c79fb70c44dc2ccbbedb8
10323 F20101108_AABTMA watcharathanakij_s_Page_04.QC.jpg
cf8834e1a11eedcaad152b5b32c2fea5
fdfca1dcf1ed5c5fc2a6b375a0ffb8fd268b30c8
1268 F20101108_AABTLM watcharathanakij_s_Page_03.QC.jpg
daf78e03c1ca6a67f79ab24844609f6f
9b8ff52a52f7873bb33532dcfd7e2012cae6cac5
31753 F20101108_AABTKY watcharathanakij_s_Page_08.QC.jpg
dc30e6fd53b9f9e5e281107b1c76cf79
5e53c675635998ccf438a293fbb81b8dfb5c7e84
2507 F20101108_AABTMB watcharathanakij_s_Page_04thm.jpg
2b8c059ea659d421354b1dec18f08e8d
67df99aa839cb3df377911e919b419549d72fc1e
32821 F20101108_AABTLN watcharathanakij_s_Page_37.QC.jpg
25e5c851d585cb49cdb09633cb8385c6
e5715c7f8e16654c4f56f034939b866e5a64af36
12223 F20101108_AABTKZ watcharathanakij_s_Page_70.QC.jpg
98a02a4e0c2ee3478d31fcbb11085dfa
39f040c46a2e81306dc8a78d48df6ab4c5461437
26190 F20101108_AABTMC watcharathanakij_s_Page_05.QC.jpg
6fbb2df9830162ae435246d74f0810c2
8a0a0dd2844ead4c0fcf3a6455a3313b8393099d
33690 F20101108_AABTLO watcharathanakij_s_Page_89.QC.jpg
f94f4992410abcfabaab6d3d8ec81198
4700ccb141621267f692a417f756c376a28fffa6
6022 F20101108_AABTMD watcharathanakij_s_Page_05thm.jpg
c2702651dac1d7ba6a3695dabb44c5f4
0c156ea39d7a3d21d7225c7c4e814843c6563dfb
9216 F20101108_AABTLP watcharathanakij_s_Page_16thm.jpg
9e5c1e7e5ff74730934340e713d9e18d
9e33316fbba9b069f1960d10c2bd507319a5db94
13118 F20101108_AABTME watcharathanakij_s_Page_07.QC.jpg
bfd12070408f130d003f1a82865627e6
36de07d904911685697090d78b255f500c89f18a
8899 F20101108_AABTLQ watcharathanakij_s_Page_25thm.jpg
05b2029371a333aa56cff882b72e1e12
fc6a9ca0b817d7445bb36eda07223d0f545d84b5
3461 F20101108_AABTMF watcharathanakij_s_Page_07thm.jpg
565a0df9195c73df008be342411849d8
8603c2244d7b10f72e0a5aeda8627679835da48d
35740 F20101108_AABTLR watcharathanakij_s_Page_90.QC.jpg
da47748f7f469fe6b9c6f89ee41159ba
b1b02a01dd36f72fce678216d58a4ba12ea8eb6e
37559 F20101108_AABTLS watcharathanakij_s_Page_27.QC.jpg
fe1ea78fa1c903b728ffeb569f3b5028
aaf5a6d3aeee6e728663c309e4c2373a6bacea91
7825 F20101108_AABTMG watcharathanakij_s_Page_08thm.jpg
775e534aae8d5644521070d976f45af3
49c54be021ee3aa61b79180641fb7f97eb6922b4
31264 F20101108_AABTLT watcharathanakij_s_Page_41.QC.jpg
f22b4e7b3e0934397b660daf45ef7780
e45807238b1e3941ef16704d96da475134874bb9
2297 F20101108_AABTMH watcharathanakij_s_Page_09.QC.jpg
7c18a150a0ff674bd8239c1f8539e5fe
e695d87d53490b3b71e0a60fcc85f292a01d6784
7629 F20101108_AABTLU watcharathanakij_s_Page_42thm.jpg
4f738ea05385ef92ee761b3a430a8b81
83fa7e927b751deb8b1d4ca8b258dc93b913e742
856 F20101108_AABTMI watcharathanakij_s_Page_09thm.jpg
6b80b997c0876f8a92277f27a10977ff
857a69f9745dd6e21e9d5f38e633cf47b277ef58
33363 F20101108_AABTLV watcharathanakij_s_Page_53.QC.jpg
39b9860fdc24df1cbc32feb1db1cdd07
c6226b88b2b5b81a60b7fc0326ddc952a57e7ca6
24450 F20101108_AABTMJ watcharathanakij_s_Page_10.QC.jpg
47f0be319c2f3af1a5afe7575f51d539
7582ebcf3ea88b4a193759ad45c11fdc414b046c
147960 F20101108_AABTLW UFE0021744_00001.xml
1cfad229b67769dd83a3fb5577763512
e4d9c6d5ebdf6c9c5555f68f994db548ed6eb714
5960 F20101108_AABTMK watcharathanakij_s_Page_10thm.jpg
bec888ff30c41e7629bd24e11cfb0a84
0ee386349b172153f2f66d31f4c3b4422bbf0668
2219 F20101108_AABTLX watcharathanakij_s_Page_01thm.jpg
9da9b70778429f0af65d30746480e387
415501cf3b18a1f6d66b37437beb4a0509dd4b52
35111 F20101108_AABTNA watcharathanakij_s_Page_26.QC.jpg
d7ec36ec0b923831db0ea031ce154249
3d7bd037caa2903e5cd6df92d6c1891cef477009
28682 F20101108_AABTML watcharathanakij_s_Page_11.QC.jpg
5c5488bb66f3e48fb9ef4cd8777f108c
90e38f3a121f3e43ddba69a72fba266765f69854
1309 F20101108_AABTLY watcharathanakij_s_Page_02.QC.jpg
5e5a663b3d63db0b8c39297b7e9e1352
569eee721ea8ea023631244bf7de6fc97b28b123
8379 F20101108_AABTNB watcharathanakij_s_Page_26thm.jpg
18e4cd294d8500b219af7907fde43a92
08e4a46670b1ec19825c2ebaa447721b7ce6ed82
3584 F20101108_AABTMM watcharathanakij_s_Page_12thm.jpg
dbe78d4650f41064dd35b3b59bf15533
08d075efb1dd28b7d6c799f566d6f060946eb21f
553 F20101108_AABTLZ watcharathanakij_s_Page_02thm.jpg
1b5557cd2e5ac97d7a132ee7d13c467b
bb9d1b7a677fe1aa1c1ea6a7e9e96f2e7baf9721
8841 F20101108_AABTNC watcharathanakij_s_Page_27thm.jpg
48fdd3a25a4bdcec189c1f925e201df6
eb48c9bd1415c1186b2e18490694f23ca75799b8
33129 F20101108_AABTMN watcharathanakij_s_Page_13.QC.jpg
2dc6f67ca0db8cb78a914e57f1930107
088020b9a85faa15098f0635bc1e2fad5f212ce8
8591 F20101108_AABTND watcharathanakij_s_Page_28thm.jpg
67a51323491240a05032c0141db939b8
ee77d4f555abc92ec85ca7f943583393407756d2
35033 F20101108_AABTMO watcharathanakij_s_Page_14.QC.jpg
9938b2c1548987be268b99bf4a7a0753
06bb700da912a2cfd65bd870cefafb1ee471148f
34886 F20101108_AABTNE watcharathanakij_s_Page_29.QC.jpg
7037c6e320d3b5b7fe81c4fc202f44a6
aebd59af3aa03b0bbedf889117893b57da9893e8
8726 F20101108_AABTMP watcharathanakij_s_Page_14thm.jpg
8dad726d3fc64ee82d7969d6a1a15634
2c551a429334cb8375dba8b43bdd6fcdf4df0d91
8512 F20101108_AABTNF watcharathanakij_s_Page_29thm.jpg
0de1db87911b6dd8efbe097be7ab4b4c
591e2fb0474e1d23eae240612fb010f465465c5f
8768 F20101108_AABTMQ watcharathanakij_s_Page_15thm.jpg
4451f4e1d74585f2d9be8734ad04f338
e552cd01281ddf424590f2bba8e4276f2b81608e
3468 F20101108_AABTNG watcharathanakij_s_Page_31thm.jpg
36bb2d6bd24ab37124c819c6abaa5abf
e387e7ab99ddc3a062a0c18c48c4d55e648232b8
37438 F20101108_AABTMR watcharathanakij_s_Page_16.QC.jpg
a362d84b1e4f872878f7ad6def86373c
1961b16408263aef49f205d210d093b38d4e5744
30782 F20101108_AABTMS watcharathanakij_s_Page_17.QC.jpg
cbae0016dc512d9ad0955b5930aa38cc
e1d7172517028e5284ed8655e719fa1e22babda3
13141 F20101108_AABTNH watcharathanakij_s_Page_32.QC.jpg
4639ab0e18e58a32f254a5e409994d1b
b0f8c289c8c64198c69159329ab9a558eb989d53
24899 F20101108_AABTMT watcharathanakij_s_Page_18.QC.jpg
e343ac452fb8f51016156ca0b2b27dac
cd07ffe7ca458f4be12877f5a93ceaab6e334c0c
3763 F20101108_AABTNI watcharathanakij_s_Page_32thm.jpg
554dd0a939beaca6036bdeb81df79fc0
2365d656f73e30e90a15d1494653c19718f89259
24907 F20101108_AABTMU watcharathanakij_s_Page_19.QC.jpg
ec6869575ad0e5424e5a0e6ce4edf735
b1a124d6f05899c279e0cbed8f5205fd2b137933
6544 F20101108_AABTNJ watcharathanakij_s_Page_34.QC.jpg
94d1dbc4712a2f3328219629b11be3c4
1513305d14570b1326926a8c769bb8c620c0b4cb
7048 F20101108_AABTMV watcharathanakij_s_Page_19thm.jpg
8e50f27b99163d2a3b473ff9bc647a96
7c941dc5b3fb96c3d0b2e7e6fb512a93badc1a15
F20101108_AABTNK watcharathanakij_s_Page_34thm.jpg
a289d51ffe8f5334e1b69ad31ebd357c
9007f477f495f123a9fb72443ea134b1c3f59227
33117 F20101108_AABTMW watcharathanakij_s_Page_20.QC.jpg
2a6bc652bd1494ee81d1a6daa3a7b912
ccc71e9b634a915604e29f98a567176a91471149
34624 F20101108_AABTOA watcharathanakij_s_Page_51.QC.jpg
839cacd45933182a9bda2fc0a6580c90
2bd122a0df02195f0b6bb0f594f262b4be6ec8a6
7583 F20101108_AABTNL watcharathanakij_s_Page_35.QC.jpg
456dcb6bce9725efb3e9b70da9a3ad0b
f1aebe8ef8ca2951acea472c36557da3b5959d6f
9246 F20101108_AABTMX watcharathanakij_s_Page_21.QC.jpg
c221b18f9d19a4a8e4792ecc0e83f4a8
a24c4caa214d30bce9eabdf4d3f01379422fbf3a
8604 F20101108_AABTOB watcharathanakij_s_Page_52thm.jpg
49a182a02641166b5067ccca5e390a22
ed65e15b6f676e3a9b918fc9dc216e2ea11efe12
2072 F20101108_AABTNM watcharathanakij_s_Page_36thm.jpg
c7a74bf1f1739eef2d5b4a57f152c30a
e86ff29a2dc08617880328fed97a2be85c88cf54
8243 F20101108_AABTMY watcharathanakij_s_Page_24thm.jpg
cf025fae91489762a11c9dc54ebd54a0
d4b2bcca1884b9df9cdb691eef50a8de94cc7f76
8467 F20101108_AABTOC watcharathanakij_s_Page_53thm.jpg
2f4bb678175d682dacf2065c8ef79e43
186dec022dfe17006c85b361e4cbe51c7003a339
8064 F20101108_AABTNN watcharathanakij_s_Page_37thm.jpg
a4e1b16eee00d16d74eaf1a74ecd5c36
49c6d72f95ac642513af06b749db0bbad31d7ccd
35730 F20101108_AABTMZ watcharathanakij_s_Page_25.QC.jpg
b6eb5945f392ac97996875c3c6ef122a
11901b9749b8b975d33adc750dbd086b6a369a91
7750 F20101108_AABTOD watcharathanakij_s_Page_54thm.jpg
bd7103d06dd99c37e55706f1aefd2b65
653f15478a0495a2a2aaab450aca0bb8a6543b93
8250 F20101108_AABTNO watcharathanakij_s_Page_38thm.jpg
27b3af4cb8f5ee5c1462a997c537304e
39921a09cd5d1fe2684cd5e26566c27521aeebd3
8520 F20101108_AABTOE watcharathanakij_s_Page_55thm.jpg
8b30a81a5c8a3d6ce4c84612b6fc66c2
455889784ab7b77bcdc93337b40ac52ffe73d801
7782 F20101108_AABTNP watcharathanakij_s_Page_39thm.jpg
d514a9bf471b5f6a4cc87cdeb4e72029
6f415b1f83b5da275dccfef1ffafd8ab2dcf7b12
34395 F20101108_AABTOF watcharathanakij_s_Page_56.QC.jpg
1c1f7ab98b11eeac0fd2c49e8f25d452
3768a070942e674c5b2547139834b52f44eabf32
32217 F20101108_AABTNQ watcharathanakij_s_Page_40.QC.jpg
cc462ae8d89dc54aeee4028691ea4a5a
dc5428c361e8467a6fceb937647df56cf3e7f730
34335 F20101108_AABTOG watcharathanakij_s_Page_58.QC.jpg
b4894e44a86b9dbb14c93dd813221dde
5dcdfeb266491670d1d881d062ec0447ede66662
24979 F20101108_AABTNR watcharathanakij_s_Page_44.QC.jpg
4c6e96bd76e70cfc3f7222101d76744c
1c1c8c8bdd99aff09304c742717c32141c3f520e
16080 F20101108_AABTOH watcharathanakij_s_Page_59.QC.jpg
4003b618f9abdcd5aa826fd4a01a97e2
2f09f27e73be9920ef0c02d755542de6b62efdb1
6363 F20101108_AABTNS watcharathanakij_s_Page_44thm.jpg
d5e5a06eea7f576312f3b96bf40854ac
9a006e00c542a334237528b46d3395a2a55a8100
4931 F20101108_AABTNT watcharathanakij_s_Page_46thm.jpg
71098cb9bec0fc6f3fa37655c30856be
a19c12135122065dea86dc0bb67a9c3dedae9f39
4521 F20101108_AABTOI watcharathanakij_s_Page_59thm.jpg
f8ba2ec67c5ff8b698a8a018578e2a00
5ec2e03bad342da66284a20fe4d83cdf3d9f52cc
15572 F20101108_AABTNU watcharathanakij_s_Page_47.QC.jpg
ba9f5e29f331ffc1ac3fb5bfe28cd25c
2df8e4a52cee3cba289df54e599a5814bbe65002
11154 F20101108_AABTOJ watcharathanakij_s_Page_60.QC.jpg
48b44671d6ce980f246e5ec449c22e10
3a8f0dddc94b7c763a00e50edfcc9aef189174ce
5931 F20101108_AABTNV watcharathanakij_s_Page_48thm.jpg
3d6c80bd4c4a46f95c9603682cc3f636
faffa9f9aa0430e7c6fcf9d9fba64c588a4ef143
3012 F20101108_AABTOK watcharathanakij_s_Page_60thm.jpg
8f89aff0964e2c24a55f50b741e280a4
6fb688dc053af431ca6fc4854e2ac75b67f5d7a8
25475 F20101108_AABTNW watcharathanakij_s_Page_49.QC.jpg
6ea57d80ed9cb37bb21645a735549cb6
24256a806c7469952cad43e4cedba7a8dcdb08da
21093 F20101108_AABTOL watcharathanakij_s_Page_61.QC.jpg
fd071f858e80bd742fce2283ea3595ac
1bcc6896be4866884224954f3c7833a2381b38ea
6744 F20101108_AABTNX watcharathanakij_s_Page_49thm.jpg
e2fe15f349f4fe0af6b3b492cafe5352
fe72949edfdfd795981a2a0117eb9e0bf70bbd74
12281 F20101108_AABTPA watcharathanakij_s_Page_76.QC.jpg
34da806ec9403a7c884d0cda2ed9f4ac
79a5cb2552147273ab92905a1d77c9acb45be93c
5442 F20101108_AABTOM watcharathanakij_s_Page_61thm.jpg
be44232722934548b2cda50dea45e4f5
520fb2a85c6ed2a6fddd588ea47cdcd90ab08892
32953 F20101108_AABTNY watcharathanakij_s_Page_50.QC.jpg
660ee2e31719642f4a6f41b209f8e7af
13b92b7f2fb59020b6fb609e16c8edeaf42f3001
3014 F20101108_AABTPB watcharathanakij_s_Page_77thm.jpg
aa7bc564ef98bce01c71ec33c5dab627
6d70e82a918a727cb9d4811d49bf0b97db31235a
17778 F20101108_AABTON watcharathanakij_s_Page_62.QC.jpg
58211295cad1ed61598e8d52e6a9f7a3
ef1722162ff1679a6870788b8dc9229e0f1a678f
8432 F20101108_AABTNZ watcharathanakij_s_Page_50thm.jpg
6d4d9ee9a14168a75252b1ed4013e64e
367ccf2fd0b3ac0f9459bbc920da2f496c04a217
29563 F20101108_AABTPC watcharathanakij_s_Page_80.QC.jpg
011a151b37ee3d4a9523c678b2a092b1
34d13906732eeb254db8423525cf357d1a291a91
4057 F20101108_AABTOO watcharathanakij_s_Page_63thm.jpg
edce7476ae5bfd025e27bdca2da95966
2873b90b0f08e38c0bc8bab3b91b6bf2e90291a7
36025 F20101108_AABTPD watcharathanakij_s_Page_81.QC.jpg
1c87df55423b12779252b18aa92a9863
8fe8f2d1e6b50f264c518395a30f8fcee2d991ef
7114 F20101108_AABTOP watcharathanakij_s_Page_64.QC.jpg
1518ab3546c15a8352194315829c92b9
fd92774cac75af6f3cb13c075e15eda66049c6a8
8920 F20101108_AABTPE watcharathanakij_s_Page_81thm.jpg
82e6767534d00c96a87f58a345651373
a642e02d5a4a0de0cb4ee19596b926656b4f831c
2167 F20101108_AABTOQ watcharathanakij_s_Page_64thm.jpg
ddf8dce882e7efb158ac0f0929287be2
ee13e615d6da259ca667b8f41a62b307bb4349e0
32744 F20101108_AABTPF watcharathanakij_s_Page_84.QC.jpg
ab439529339255b8fbb651f1c72d244a
3698d863075061ce027e92896f42df8435a1b7a2
3731 F20101108_AABTOR watcharathanakij_s_Page_65thm.jpg
61a7d3f852b7dd36f9c3afea66ca4432
9c716bf2b2036f81cd7807a7ce44e399215f8470
28571 F20101108_AABTPG watcharathanakij_s_Page_85.QC.jpg
f96203b821f6a7b7eeea624adec485d0
c11fb57ff6f9b53abb914f8a254c62ac619edec8
12518 F20101108_AABTOS watcharathanakij_s_Page_66.QC.jpg
ff6ba37db5075285498b61b188e54c9b
e1c48996564fae00553d981e5b55b7b236acf0cd
19848 F20101108_AABTPH watcharathanakij_s_Page_87.QC.jpg
206400374e3e46d1a933e185f85c45b8
0667c7d9849ec88f2986902610e2b1b007cbdc44
5288 F20101108_AABTPI watcharathanakij_s_Page_87thm.jpg
e7d94997b78aab78d17e4a0d2944e2b2
10e23b63d2f151c0506e335060e25093eb5a666a
6886 F20101108_AABTOT watcharathanakij_s_Page_67.QC.jpg
b50759e88d45b421f582cd0f929646bf
0f65ad8f7f427c4f56820670d8bf5163f9b21470
11842 F20101108_AABTOU watcharathanakij_s_Page_68.QC.jpg
51e5a84bd3aa93bea00d3a4fde99a64b
3d29a8475967c830f21bce5390c9ebe63ce13e2e
8887 F20101108_AABTPJ watcharathanakij_s_Page_90thm.jpg
8643ce4e74545b9a4fb5b146c2ed8948
e36833c47425281a2924e6a17e66f745e1b9a6ef
3600 F20101108_AABTOV watcharathanakij_s_Page_70thm.jpg
e58d58edcf3e22b4d60c48ad1f1b11a6
29345c57e824c78a831c4aada4eac3d25f192013
9030 F20101108_AABTPK watcharathanakij_s_Page_92thm.jpg
2d47bc850abc23774246b72dc2167f24
ca58082bfb75a4dd747a141290ff119c5f4db60b
12472 F20101108_AABTOW watcharathanakij_s_Page_71.QC.jpg
f6edcd49053200541b76723d4a95d7e3
3e91f5437fa16c7dd49bd7c253081020e2f8cfa8
26855 F20101108_AABTPL watcharathanakij_s_Page_93.QC.jpg
7b5a08806cf15d194ab87f229d8be459
e340fe91a290f762c869770c1f95f76d8e7e9dd7
3141 F20101108_AABTOX watcharathanakij_s_Page_73thm.jpg
4217a588d0c6813001a2a6cb51319b7f
8c0186bdff91ab4b82e09469b1362e039a9eb224
12131 F20101108_AABTPM watcharathanakij_s_Page_94.QC.jpg
a38618e893c41cf398462ee2d778f1c3
53fb7aa8bdf31e542b08681ee29a9b09ca9830a8
8717 F20101108_AABTOY watcharathanakij_s_Page_75.QC.jpg
d5f38aae041fccaa14d43a14aec845c1
3e117dbac8492a25fa0b753a7af449bb07e634fc
2896 F20101108_AABTOZ watcharathanakij_s_Page_75thm.jpg
169590a854326fa11ae6f7bff6f753a0
7d81462c61e2d959b633c0cc8a10c4ca7f1bda25
72650 F20101108_AABSQA watcharathanakij_s_Page_06.pro
9387ea375a70f1490a821480410d8860
0270958e9f7dc4a8c6568f8ceb753d9a7de3200b
7780 F20101108_AABSQB watcharathanakij_s_Page_41thm.jpg
6be873ce490c608b6de2a85accf4e5e5
06f849a91c4b0b14911c569fd81c6058e2bcea27
28900 F20101108_AABSQC watcharathanakij_s_Page_42.QC.jpg
5b65a95b6031836bebaf74e6701c2733
6de858ee44d0be276bac4fe4858673795c6b3220
7198 F20101108_AABSQD watcharathanakij_s_Page_85thm.jpg
c9da183e99e8c3d0ead962ce8894cd95
6e4700e41e2bad96445ace6d091e804dec7029e6
87026 F20101108_AABSQE watcharathanakij_s_Page_18.jp2
bcac86502fd693d2c566a8ef3bb2fe0e
ac49cf93f71ffd50995fdd55eb07840d3cbd274e
3804 F20101108_AABSQF watcharathanakij_s_Page_88thm.jpg
279067c97bb679e51d42e0ea0f7727ca
97a27779efac078541dd4a62386e14b385998b34
9024 F20101108_AABSQG watcharathanakij_s_Page_72.QC.jpg
e1477e8da8d317d92dd2787a0c5f4921
7cf58f5dacb6fe5d00470a954a38bf477ba39372
103329 F20101108_AABSQH watcharathanakij_s_Page_38.jpg
255279462014584eab5d1cc8633888a2
cf8cf50c34cbf8398dc7f4ec385103d096106e83
35972 F20101108_AABSQI watcharathanakij_s_Page_82.QC.jpg
c3e2deb9d2f11fea5a767ab556f7c1e1
fc7e00b6a3c1550355fd9b21321058ee369ecb83
F20101108_AABSQJ watcharathanakij_s_Page_17.tif
46dd2de15d248e9eec5c705a7416b617
27e8e83adee4e10a55a97e5a8fd9747d64203459
1977 F20101108_AABSQK watcharathanakij_s_Page_50.txt
f808bfce2fe699df24d8c7839de0be9c
7eab7182a85b4b8f72ae8a7f94c97fa9e0c17ddd
92647 F20101108_AABSRA watcharathanakij_s_Page_41.jpg
ed65b161b4f8b8a1bd976f5dd8030304
d7820a880a33a2cb0e5c97467168f54351f89be6
44563 F20101108_AABSQL watcharathanakij_s_Page_41.pro
16a46c500c8a51d601e5a7bd006adc82
a0a4144de0ffd841bb8e3759ac3f1fec48baf413
110120 F20101108_AABSRB watcharathanakij_s_Page_15.jpg
49edebd389c597d025ec7bd8d34ea083
9d8cf97fa5ad431f37a13753517d267bd8e42f80
61300 F20101108_AABSQM watcharathanakij_s_Page_90.pro
d9e2e1c823c160f6385b928b83538e88
f8bf705c33acdfe624034b995ed8b3e279aefb77
21331 F20101108_AABSRC watcharathanakij_s_Page_43.QC.jpg
e435e003d242ef30680d8bea7f6363f2
2dad174e9f74e54f3951d039758596266b5c59de
8694 F20101108_AABSQN watcharathanakij_s_Page_30thm.jpg
f91bc7973f5fb6fa1da9c260096d3647
a2ea5f18bea8587b342aeac310b13a67fdc8b441
50865 F20101108_AABSRD watcharathanakij_s_Page_63.jpg
8068d3528d84a34727a95262718531a3
09f267dd7aff43a915aa7ef768d1811db8729887
730 F20101108_AABSQO watcharathanakij_s_Page_31.txt
fc5cab41053375517efff303296041d2
aafd8018b3277ad056f8d666d1174a619123eab7
595 F20101108_AABSRE watcharathanakij_s_Page_60.txt
9435dfb06025f2ee75da1453347e3416
f50e96245aa4bc9861370ca3975128faed75e9b3
83229 F20101108_AABSQP watcharathanakij_s_Page_49.jp2
fb0e4a07cc3e7656deffc1008a1f4f01
63f8d322da412b935d35ddeabd23db8652e81676
101045 F20101108_AABSRF watcharathanakij_s_Page_54.jpg
ec30d9f4c6b0aba84a4d62f306f1c675
bd86af8d14bf13e1e9e053bec68a5db2ecea1a18
107108 F20101108_AABSQQ watcharathanakij_s_Page_51.jp2
c11468e70bcf63b468f24fb530638a6d
f7d66a8c5cfa542e2feb9d788b54676885cb1b13
205899 F20101108_AABSRG watcharathanakij_s_Page_77.jp2
6ab4a8ee2925e70f42f08d60ce746842
a4d8fc38cca4cc5fab5406d9633adba3c2f426f8
91376 F20101108_AABSQR watcharathanakij_s_Page_11.jpg
b13ae7da35b550ffb220147777e43004
8c817bc8f0b920c452367829f689d3c8fbd105eb
42271 F20101108_AABSRH watcharathanakij_s_Page_85.pro
e1ec0b7987ba52e43d1bcf1f9d79982f
89bbb005cc7c88c2e3de322420d2fb79337b913d
97378 F20101108_AABSQS watcharathanakij_s_Page_33.jpg
5139727d90222b1b34514631bd0c75c5
8033391de470ca6e44326acf6f1af80255b45fd6
1563 F20101108_AABSRI watcharathanakij_s_Page_71.pro
9ff54d73044084f3162dc23fade3ca27
c879aa743744e67942ec82c91a7e4d6eeade1363
2483 F20101108_AABSQT watcharathanakij_s_Page_35thm.jpg
214b8d040898720a2eb2fa21e1aea598
ed29b047cd09eb6fa08d925abef84d356161045c
F20101108_AABSRJ watcharathanakij_s_Page_93.tif
1e5d90f5c2c5047482187102b64a0639
d966fadb7e1b9fbf78143611cd9d6357fcfc991c
F20101108_AABSQU watcharathanakij_s_Page_35.tif
abaef9c2ee35532515e180b2acbaaba8
309b6705193425f0730fc19d4299a4ecf5acd4ce
8595 F20101108_AABSRK watcharathanakij_s_Page_51thm.jpg
a81b2fd83abc6f6e739ebd07a4099b3b
e75f89b28f85befeebc3cff64310a44cd4f420d2
2013 F20101108_AABSQV watcharathanakij_s_Page_53.txt
85183286a3c4ed8f4a60220bad930373
3e9a2f22af4df4c22a133842b3a61731ebc17fde
34836 F20101108_AABSRL watcharathanakij_s_Page_30.QC.jpg
3bba60e908595f1c972362e5aee2f1bd
bbbbcd003cead2e0452010d000866f5ccbaa87e1
35182 F20101108_AABSQW watcharathanakij_s_Page_70.jpg
20c574533bffc66de2d13445e3ab27f5
17718a1a4532a8a0cdbf5b275996df58162115d7
103186 F20101108_AABSSA watcharathanakij_s_Page_51.jpg
96ffa5f11797a0a8e95be998cfce6271
5442490b3a60333c8e65db43af9f65447a9e5edd
6069 F20101108_AABSQX watcharathanakij_s_Page_06thm.jpg
623c26794ec9d3f34399459bbcd5505d
75ef02013db27c4949e13695c46acc2d0e0e1d16
18329 F20101108_AABSSB watcharathanakij_s_Page_59.pro
09160ba3a69d735db077d24c266a19c0
6b5c3fe605a77bf63e5d574084946e734058e560
2119 F20101108_AABSRM watcharathanakij_s_Page_81.txt
c2434311d0c41109a4c8cd7066d978ec
614bd0459fa63e72d5e89288a4885786b60bc4ec
41591 F20101108_AABSQY watcharathanakij_s_Page_31.jpg
b3c1229321c27d7b3e0e8e96dc194f21
1308804883d4c5006625f130e7847f07573631c6
9534 F20101108_AABSSC watcharathanakij_s_Page_73.QC.jpg
d4b3088650284c441fa8d82dd0f86e8e
b3231da535dfc5fceb1087e9536e8d791b4f70f6
105527 F20101108_AABSRN watcharathanakij_s_Page_50.jp2
1ce756ba968a8e9ece9c72d79490f6a5
81ad9c7aec85955af3a7be950d155383ab65f76c
F20101108_AABSQZ watcharathanakij_s_Page_68.tif
426dd08e04fa6a42ef248702990c4f27
ef32fdfab8cd2f69a791397700a6902cc2148a19
F20101108_AABSSD watcharathanakij_s_Page_21.tif
612d8d2b7431b84d99f0bcae4c67ab16
76383d52ee8ac20222d296c2c93a82c489a7cdfd
12621 F20101108_AABSRO watcharathanakij_s_Page_74.QC.jpg
b6c33132a30e7c66c4a6b617ae63e350
0efa529ea9724c667439b9989df1352b5ddfc40e
2530 F20101108_AABSSE watcharathanakij_s_Page_78thm.jpg
593a5faf3e44b50225584ba75925711a
cea3df4c6e20148ac08a91bac97a32d136cd523c
125799 F20101108_AABSRP watcharathanakij_s_Page_92.jpg
9c8349f715c676e0b47d7287da09007b
5693b6d3b1452acf825a369b9c3a2513d6158dfc
1099 F20101108_AABSSF watcharathanakij_s_Page_02.pro
2bd2b398578010d65404025a76b3c87d
d03e01cabb80f119cc4e5da1421339d77f2be036
35038 F20101108_AABSRQ watcharathanakij_s_Page_92.QC.jpg
c004c97b04e947f89f4d8343b51ec096
6b9e1dd80bb9e267da9a35bd265308889024e909
8733 F20101108_AABSSG watcharathanakij_s_Page_82thm.jpg
d0dfb0a13d8b092b563c9b641dbfdd1c
f7d7d8fea635e6fff1ce94ac1ebcd826462332c1
F20101108_AABSRR watcharathanakij_s_Page_32.tif
9ded40f3bfb511f515a2114f218fa117
523c0cf389adefada091b8041a150e293edb7232
F20101108_AABSSH watcharathanakij_s_Page_73.tif
64d5f64a7d52648926bc79ef6b0f4ff5
5e0e4e09dddf94a97528099f7614bba38225a3d4
481771 F20101108_AABSRS watcharathanakij_s.pdf
546858e08d4c8f77605d5c473b324542
0721e9be554c5251686805747cbe158423b9ffec
53415 F20101108_AABSSI watcharathanakij_s_Page_27.pro
b9acf65b7d07228ef089d7f99daf9490
26326171d432c6b01b31926e0fb3dfd2a0b7d96b
32952 F20101108_AABSRT watcharathanakij_s_Page_22.QC.jpg
abbb60a40495a219d570523360bddd51
105fa7a31ed6506f9a428b4c9bc6c103946c1f25
7649 F20101108_AABSSJ watcharathanakij_s_Page_78.QC.jpg
390bc4cedc2ac52efb5892ef302577ad
cab86a42bf3005381c0ec5819a2e0983c69acae8
64518 F20101108_AABSRU watcharathanakij_s_Page_66.jp2
be94a883aeb14f351783d90c1e35d0f4
163170aec5ce436cd1ed50ed651a3b91d7f132c4
F20101108_AABSSK watcharathanakij_s_Page_30.tif
7cb485a37b79ac5dd5659c0fb6f4f803
32b11414dbdb3084b20905b8ac911aaf738ee425
F20101108_AABSRV watcharathanakij_s_Page_12.tif
7a93c6c8b5c55d4da236007ef94f3f07
0d5c8a4ff4535f9fc514e7a132565f9aef18ce1e
35117 F20101108_AABSSL watcharathanakij_s_Page_55.QC.jpg
a1227274d3e727971637083976380650
c18e498790196704e26d17a3ad98a8f1541fe42f
F20101108_AABSRW watcharathanakij_s_Page_79.tif
46f20fc8249d0a7c0668be9bc714d1c0
515c2980eea545de3f5de5a5c518aa662fae61ad
508 F20101108_AABSSM watcharathanakij_s_Page_01.txt
2552bdf769f0c23d3ffa87f0453e6d29
8af7c83480a35102c76566d786e09fc060dea7b2
202 F20101108_AABSRX watcharathanakij_s_Page_72.txt
8104c94e84cc3d979bb414755895204d
c7fbe89be01e029168cce2b3c5a77b905d6b730d
8179 F20101108_AABSTA watcharathanakij_s_Page_23thm.jpg
55962f0474844070cefdce20bd32af01
b74ca95ce39b1a3fbca531a64fdc8232cba8fef8
2080 F20101108_AABSRY watcharathanakij_s_Page_13.txt
8001847bf25401a21639bed855c0d03b
a9a9fcb6b814b37aaa4b91842a24801da5671649
12592 F20101108_AABSTB watcharathanakij_s_Page_69.QC.jpg
76d873904f3af525222ba9d52f2434ae
d3d782c392125853b1be153b87477b879ecd7c23
72692 F20101108_AABSSN watcharathanakij_s_Page_44.jpg
daad992e77ae06ba79f05b5f3b6dafc3
3d478982eeaccf15806dcaf8c71644a63db46e97
52199 F20101108_AABSRZ watcharathanakij_s_Page_25.pro
b6300eaa62424a4ce80993c2546416d0
11f9cbd5702675b9af6dd06259811f7b8b299761
2203 F20101108_AABSTC watcharathanakij_s_Page_16.txt
76e2515aefef8924666b99f884084849
6b2c289108c31ff47feda290a4a73145443267d1
F20101108_AABSSO watcharathanakij_s_Page_64.tif
be1ab2d0b7ff2fb522eee3847e1c2fa3
52f6187237540bc21d512c60a0f6de31e969f712
F20101108_AABSTD watcharathanakij_s_Page_28.tif
327a760750ad900ed2e56a613f7561c8
c9653bca0db7d9914c811232f5193f4ee4dd1550
F20101108_AABSSP watcharathanakij_s_Page_78.tif
dfe541882cefc54308b6423fa94dbda3
b96b702b76bab515d1037625e9398b0e83d79232
42440 F20101108_AABSTE watcharathanakij_s_Page_42.pro
950bc8166f5c9021030d0e6665358699
f4cad3692cb636c2a4bd60316bffb4e7694f1779
23450 F20101108_AABSSQ watcharathanakij_s_Page_36.jp2
ab11481ad28aab02f712453b66a3a8f1
44b77ad60c41ace3bf617024314b6ba13936ba99
1862 F20101108_AABSTF watcharathanakij_s_Page_80.txt
f3bcc5e4d2f081b271d83a58d6824366
b7ca246909906b99b36ab97c154adcf386132104
102104 F20101108_AABSSR watcharathanakij_s_Page_40.jp2
68b6939aa641910960e23782411b601a
21fee7576f486e3697841b3e562d3b6e121123ec
34045 F20101108_AABSTG watcharathanakij_s_Page_57.QC.jpg
09ea68593439158c399526d8e4bc19ab
201d7245ba87abca2b53577822617b0101ca9947
31573 F20101108_AABSSS watcharathanakij_s_Page_35.jp2
5f42880ea90b39d2cb57067530dad8aa
97e5cbb7236a232dc555ad56dc2566357fbc0bd2
43503 F20101108_AABSTH watcharathanakij_s_Page_12.jpg
4366ef8327d9fa2db824014ee2ba5720
4cf47dda6a2e987faab71f7cb3584d5bdb647680
39463 F20101108_AABSST watcharathanakij_s_Page_65.jpg
afd2a2b2072447f124646ecbb5171175
0b235c53d20680e08b8ff813096aea2d135dc891
274 F20101108_AABSTI watcharathanakij_s_Page_09.txt
ad7352815f4cbb42e9f538b704b60691
3457af7860d42a2c747a6c96008b24c094deba8d







TIME LAG MODEL FOR QUALITY OF LIFE AS SES SMENT INT HIV-INFECTED
PATIENTS WITH HIGHLY ACTIVE ANTIRETROVIRAL THERAPY






















By

SAWAENG WATCHARATHANAKIJ


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

2007

































O 2007 Sawaeng Watcharathanakij

































To my beloved parents and family









ACKNOWLEDGMENTS

First, I sincerely thank my dissertation chair, Professor Richard Segal, for his invaluable

guidance and support through out dissertation period. I also thank Professor Carole Kimberlin,

co-chair, and Professor L. Douglas Ried for their recommendation and encouragement. In

addition, I extend special thanks to my external committee member, Professor Michael Daniels,

for his invaluable advice on the data analysis. Finally, I express thanks to all my friends for their

friendship.












TABLE OF CONTENTS


page

ACKNOWLEDGMENTS .............. ...............4.....


LIST OF TABLES ............ ...... ._ ...............8....


LIST OF FIGURES .............. ...............10....


AB S TRAC T ............._. .......... ..............._ 1 1..


CHAPTERS


1 INTRODUCTION ................. ...............13.......... ......


Problem Statement ................. ...............13.................

Background .................. .. ......... ...............14.......
HIV Infection and Treatments ................. ........... ...............14. ....
Conceptual Model for Quality of Life Assessment ................. ............................15
The Time Reference in QOL Measurement ................ ................. .. ........ ......... ......1
Methodological Limitations of Previous Research in HAART and QOL .....................16
Research Questions and Hypotheses .........._.... ...............17..__._. .....
Research Question 1 .............. ...............17....
Research Hypothesis 1 .............. ...............17....
Research Question 2 .........._.... ...............17...__........
Research Hypothesis 2 .............. ...............18....
Research Question 3 .............. ...............18....
Research Hypothesis 3 .............. ...............18....
Research Question 4 .........._.... ...............18...__........
Research Hypothesis 4 .............. ...............18....
Research Question 5 .............. ...............18....
Research Hypothesis 5 .............. ...............19....
Research Question 6 .........._.... ...............19...__........
Research Hypothesis 6 .............. ...............19....
Research Question 7 .........._.... ...............19...__........
Research Hypothesis 7 .............. ...............19....
Research Question 8 .........._.... ...............19...__........
Research Hypothesis 8 .............. ...............20....
Si gnificance of Research .........___ ....... ....... ......._.. ...._._................20

2 LITERATURE REVIEW ................. ...............22................


Epidemiology of HIV/AID S ................. ...............22..__._ .....
Classification of HIV Infection .........._.... ...............22._.._. .....
Antiretroviral Agents ............... .. .......... ... .. ...............23.......
Conceptual Model for Quality of Life Assessment ................. ...............23........... ..
Biological Function .............. ...............24....












Symptoms ............ _...... ._ ...............24....
Functional Status .............. ...............24....
General Health Perception ............ ..... ..__ ...............25...
Overall Quality of Life .............. .......__ .. .. ....__ ...........2
Time Reference in MO S SF-3 6 Instrument for Current QOL ......____ ....... ...__...........25
Validity of Quality of Life Assessment Model .................. ........... ..... .......... ..... 2
CD4+ Cell counts, Viral load, and QOL in HIV-infected Persons .............. ....................27


3 METHODS .............. ...............37....


Conceptual Framework............... ...............3
D ata..................... ... .............3
HAART Definition................ ..............3
Health-Related Quality of Life ............ _...... ...............39...
CD4+ Cell Count ........._...... ...............39...__........
V iral Load ....................... .............4
Inclusion and Exclusion Criteria .............. ...............40....
Data Analysis Process............... ...............40
Data Preparation ............... .. .. ......... ... ... .......4
Missing data and dropping out in the longitudinal study ................. ................ ...41
How to handle dropouts in longitudinal study .............. ...............41....
Data Analysis............... ...............42
IRB Approval............... ...............44


4 RE SULT S .............. ...............49....


Merging MACS Database............... ...............49
Drug Table............... ...............49.
QOL Table............... ...............49.
Lab Test Table ................. ...............49........... ....
Missing Data ................. ...............50.................
Exploratory Analysis .............. ...............50....
Demographic Data............... ...............50..
Quality of Life Trajectory .............. .. ...............50..
CD4+ Cell Count and Viral Load Traj ectory ................ ...............51..............
Correlation among PHC and MHC .............. ...............51....
Time-lag and Non Time-lag Models ..................... ...............52.
Time-lag and Non Time-lag Models for PHC ...._ ......_____ ...... .....__........5
Time-lag and Non Time-lag Models for MHC .............. ...............55....












5 DI SCUS SSION ............ ..... ._ ............... 0....


QOL Trajectory .............. ...............80....
PHC Trajectory............... ...............8
1VHC Trajectory .............. ...............80....
Clinical Lab Test Traj ectory ................. ...............80........... ...
CD4+ Cell Count Traj ectory ................. ...............80.......... ....
Viral Load Traj ectory .............. ... .. ......... ........ ...... ...........8
Relationship between CD4+ Cell Traj ectory and QOL Trajectory ................ ................ ..81
Relationship between Viral Load Traj ectory and QOL Traj ectory ................ ................ ..82
Limitations ................. ...............82.................
Future Research .............. ...............8 5....
Conclusion ................. ...............85.......... .....


APPENDIX


A SQL SYNTAX FOR HAART ............ ......__ ...............86..

LIST OF REFERENCES .............. ...............89__. ......


BIOGRAPHICAL SKETCH .............. ...............94....










LIST OF TABLES


Table page

2-1 The estimated numbers and percentage of HIV/AIDS and AIDS cases by diagnosis, age,
gender and race in 1981-2004 ........... .......__ ...............32.

2-2 CDC HIV Infection Categories by Clinical Conditions ........._.._.. ...._... ......._.._......33

2-3 CDC Classification System for HIV Infection ...._.._.._ ........._.._......_ ..........3

3-1 Missing Data Mechanisms of CD4+ Cell count ........._.._.. ....._.. ......__. ........4

3-2 Advantages and disadvantages of selection and pattern-mixture models .............. ................46

3-2 Independent Structure of QOL Measurements ...._.._.._ ........._.._......_ ...........4

3-3 Exchangeable Structure of QOL Measurements .............. ...............48....

3-4 m-dependent Structure of QOL Measurement .............. ...............48....

3-5 Autoregressive Correlation Structure of QOL Measurements .............. ....................4

4-1 Time patients started HAART ........... ...... .__ ...............59.

4-2 Average age at HAART initiation ..........._ .......__ ...............59.

4-3 Age of patients at HAART initiation.. ....._ ....__.......___ ......__ ..............59

4-4 Average follow up time after HAART initiation............... ...............5

4-5 The number of patients followed up until last clinic visit after HAART initiation. ...............60

4-6 Education level of patients who received HAART ........._ ....... __ ......_ ............60

4 -7 Average physical health component score by time since HAART .............. ...................61

4-8 Average mental health component score by time since HAART.............__ .........___......61

4-9 Average CD4+ cell count at each visit .........._.... ...............62...._... ..

4-10 Average log viral load at each visit .........._.... ...............62...__._.

4-11 Correlation coefficients among PHC measured at different time............... ..................6

4-12 Correlation coefficients among MHC measured at different time .............. ...................63

4-13 Models for PHC ........... __..... ._ ...............64..











4-14 Models for MHC .............. ...............67....


































































9











LIST OF FIGURES


Figure page

1-1 Wilson and Cleary's Conceptual Model for QOL Assessment ......____ ..... .....__ ..........21

2-1 Ferrans's Conceptual Model for QOL Assessment ................. ...............35........... .

2-2 Modified Version of Wilson' s Model ................. ...............36.......... ...

3-1 Time-lag Conceptual Model for Quality of Life Assessment ..........._......._ ..............46

3-2 Conceptual Diagram for Traditional Model .............. ...............47....

3-3 Conceptual Diagram for Time-lag Model .............. ...............47....

4-1 Physical Health Component Score after HAART Initiation of 10 patients.............._.._. ..........70

4-2 Mental Health Component Score after HAART Initiation of 10 patients .............. ..............71

4-3 Average PHC score over time after HAART initiation............... ...............7

4-4 Average MHC score over time after HAART initiation .............. ...............73....

4-5 CD4+ cell count over time after HAART initiation for 10 patients .............. ............. ..74

4-6 Average CD4+ cell count over time since HAART initiation. ......____ ...... ...__ ...........75

4-7 Log viral load over time after HAART initiation for 10 patients.............__ .........___......76

4-8 Average log viral loads over time after HAART initiation ......____ ........_ ..............77

4-9 PHC residuals for model 2............... ...............78...

4-10 MHC residuals for model 2 .............. ...............79....









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

TIME LAG MODEL FOR QUALITY OF LIFE AS SES SMENT
IN HIV-INFECTED PATIENTS
WITH HIGHLY ACTIVE ANTIRETROVIRAL THERAPY



By

Sawaeng Watcharathanakij

December 2007

Chair: Professor Richard Segal
Major Department: Pharmacy Health Care Administration

Quality of life (QOL) assessment plays a pivotal role in determining the longitudinal effect

of highly active antiretroviral therapy (HAART) in HIV-disease. This research addresses how

the timing of clinical lab tests, such as CD4+ count and viral load, affect patients' self-reported

quality of life.

This obj ectives of this study are to: (1) verify whether a lag time between clinical lab test

results (CD4+ cell count and viral load) and QOL exists by comparing two statistical models,

time-lag and non-time lag model, and (2) determine how well change in CD4+ cell count and

change in viral load over time can predict change in QOL over time. Subj ects treated a drug

regimen called HAART were selected from a secondary database called the Multicenter AIDS

Cohort study (MACS). The MACS is a prospective observational cohort study of the natural and

treated histories of HIV-1 infection in homosexual and bisexual men.

Data were analyzed with both time lag and non time-lag random coefficient models

because each patient had a unique CD4+ cell count, viral load and QOL trajectory. The effect of

CD4+ cell count and viral load on two dimensions of QOL physical health component (PHC)









and mental health component (MHC) was examined by comparing the time lag and non time-

lag random coefficient models with the model fit statistics, Akaike information criterion (AIC).

PHC and MHC in HIV-infected patients who were on HAART slightly decreased over

time. The change in viral load over time significantly predicts change in PHC and MHC over

time, whereas the change in CD4+ cell count significantly predicts PHC over time only. CD4+

cell count has a positive longitudinal relationship with PHC, whereas viral load has a negative

longitudinal relationship with both PHC and MHC. Overall, time-lag models were not different

from non time-lag models in terms of the model fit statistics and regression coefficients.









CHAPTER 1
INTTRODUCTION

Problem Statement

Quality of life (QOL) is an outcome that is generally viewed as important for measuring

the impact of a health care intervention because it takes patients' perspectives into account.

While examining the effect of interventions on QOL for patients who are treated for acute

conditions is important, measuring QOL is especially of interest for patients who have chronic

medical conditions since other health outcomes may be influenced by their perceptions of their

functioning and well-being. This research examines one chronic medical condition, Human

Immunodeficiency Virus (HIV) infection, for which measuring QOL may be of particular

importance in understanding patients' decisions about their HIV care.

Understanding the relationship between a patient' s perception of their QOL and the

decisions they make when managing their condition is not straight-forward. Although many

studies support a negative relationship between HIV symptoms and QOL, taking anti-retroviral

medications often leads to side effects, such as nausea, pain, and anemia, which can also lessen

QOL. Consequently, patients may discontinue taking antiretroviral drugs in order to avoid those

side effects, which may have the effect of lowering the effectiveness of the treatment strategy

and, thereby, increasing their HIV symptoms. Further complicating the prediction of QOL is the

availability of information from clinical lab tests for HIV disease including CD4+ cell count and

viral load. At the time of a clinic visit, the findings from clinical lab tests are usually

unavailable since it takes at least one week to process the blood samples. Therefore, measures of

QOL taken at the time of a clinic visit is likely based on results from earlier lab tests along with

current symptoms, and current anti-retroviral side effects. This study intends to examine









whether the lag time relationship between clinical lab tests and current QOL exists by comparing

two different models, time-lag model and non time-lag model.

Background

HIV Infection and Treatments

Human Immunodeficiency Virus (HIV) infection is one of the maj or health problems in

United States and worldwide. The Centers for Disease Control and Prevention (CDC) reported

on a study conducted by Glynn and his colleagues that over one million persons were infected

with HIV infection in United States in 2006 and approximately 40,000 new cases were

diagnosed in 2006 (1).

Many antiretroviral agents are available in the market and treatment guidelines are also

available for the clinicians to treat HIV-infected persons. The treatment goals of HIV infection

are to reach "maximal and durable suppression of viral load, restoration and preservation of

immunologic function, improvement of quality of life, and reduction of HIV-related morbidity

and mortality" (2). The most updated treatment guidelines recommend some combination of

anti-retroviral drugs, called Highly Active Anti-retroviral Treatment (HAART), to treat HIV-

infected patients in order to achieve the treatment goals mentioned above.

Managing symptoms in HIV-infected patients is very complicated because patients in

different stages of the disease may respond to treatment in different ways. For example, for

symptomatic patients, the use of antiretroviral agents may adequately control symptoms for a

patient who has HIV-infection, which leads to an increase in their QOL. However, because

antiretroviral agents may lead to side effects, QOL may be affected negatively (3). In contrast,

asymptomatic patients may behave differently from symptomatic patients in terms of how they

manage their disease because side effects of anti-retroviral drugs may be of greater concern than

symptoms from HIV-infection. Consequently, these patients may stop taking anti-retroviral









medications, which can lead to faster disease progression and shorter life expectancy. Therefore,

the focus on maximizing QOL in HIV-infected persons is not only important immediately after

treatment initiation, but also for the long term. In other words, it is essential for healthcare

provider to maximize patients' QOL and maintain maximum QOL over time.

Conceptual Model for Quality of Life Assessment

The conceptual model used in this research to assess QOL was first proposed in 1995 by

Wilson and Cleary (4). It describes the relationship among five levels of health outcomes: (1)

biological and physiological factors, (2) symptom status, (3) functioning status, (4) general

health perception and (5) overall quality of life, as well as characteristics of the individual and

environment as in Figure 1-1. Based upon this conceptual model, QOL may be affected by a

patient' s interpretation of their response to treatment based on CD4+ cell count and viral load,

symptoms from HIV disease such as diarrhea lasting greater than one month, and medication-

related side effects.

The Time Reference in QOL Measurement

A validated and widely used quality of life instrument, the Medical Outcome Study Short

Form 36 (MOS SF-36), asks respondents to assess their QOL in the last four weeks e.g. during

the past four weeks, have you had any of the following problems with your work or other regular

daily activities as a result of your physical health. Based upon this time reference and the

conceptual model for QOL assessment, the MOS SF-36 places a time boundary around the

information that should be used by the patient as he or she responds to each question that is,

information available to patients during the four week period prior to completing the survey,

including their HIV symptoms, experiences with drug side effects, and information about viral

load and CD4+ cell count. However, one possible limitation with using the four-week time

frame may be that biological factors, such as viral load results and CD4+ cell count, may have









been last measured more than four weeks ago. Consequently, it is possible that patients assess

their current QOL based on using CD4+ cell count and viral load from three or more months

earlier. For example, patients with severe symptoms may assess their current QOL higher than it

should be because they know that CD4+ cell count and viral load from their last clinic visit

indicated they were well controlled. However, if they knew that their current CD4+ cell count

and viral load were not in control, they may assess their current QOL lower than the situation

above. In other words, patients will assess their current QOL based upon the findings from

clinical lab test result that are available to them regardless of when the tests were administered.

Methodological Limitations of Previous Research in HAART and QOL

Although numerous studies focused on either the effect of symptoms such as diarrhea or

the effect of anti-retroviral drugs on QOL in HIV-infected patients, only a few explored the

relationship between HAART, HIV-related symptoms, side-effect of anti-retroviral medication,

clinical lab tests, and QOL. Among those studies, most were crossectional, which suffered from

limitations that may be overcome with the use of a longitudinal design. Some of the studies

which used a longitudinal design examined the relationship between HAART, HIV-related

symptoms, side-effect of anti-retroviral medication, clinical lab tests and QOL, but the authors

defined the meaning of longitudinal relationship differently than what we proposed in this study.

For example, in a study conducted at a VA hospital, the authors described their study as a

longitudinal study because the patients were followed up to 12 months after enrollment (5).

QOL, the outcome in this study, was measured at 2 different points in time, baseline and at one

year. In addition, the authors used baseline QOL and CD4 cell count, depression and other

factors to predict QOL at one year. Although another study conducted by the same researchers

had fewer limitations because they measured predictors such as CD4+ cell counts, coping, and









comorbidity at baseline and one year (6), it was different from the methodological design

proposed in the present study.

Based upon the Wilson and Cleary conceptual model to measure QOL, time reference in

QOL assessment, and to maximize QOL in HIV-infected patients with HAART, the present

study proposes to investigate whether the lag time relationship between previous clinical lab

tests, and current QOL in patients with HAART exists by comparing two models

Research Questions and Hypotheses

The purpose of this study is to determine whether patient' s assessment of their current

QOL over time (trend of QOL) may be better predicted from clinical lab test results over time,

measured from previous clinic visit, compared to only using clinical lab test results performed at

the same clinical visit where QOL is measured. Since it is possible that the lag-time relationship

may differ for predicting different domains in the MOS SF-36 QOL scale, research questions are

offered for two maj or domains associated with the SF-36, physical and mental subscales.

Research Question 1

How well does the change in current CD4+ cell count over time predict change in current

physical health in HIV-infected patients over time?

Research Hypothesis 1

The null hypothesis is that change in current CD4+ cell count over time cannot predict

change in current physical health in HIV-infected patients over time whereas the alternative

hypothesis is that change in current CD4+ cell count over time can predict change in current

physical health in HIV-infected patients over time.

Research Question 2

How well does the change in previous CD4+ cell count over time predict change in current

physical health in HIV-infected patients over time?









Research Hypothesis 2

The null hypothesis is that change in previous CD4+ cell count over time cannot predict

change in current physical health in HIV-infected patients over time whereas the alternative

hypothesis is that change in previous CD4+ cell count over time can predict change in current

physical health in HIV-infected patients over time.

Research Question 3

How well does change in current viral loads over time predict change in current physical

health in HIV-infected patients over time?

Research Hypothesis 3

The null hypothesis is that change in current viral load over time cannot predict change in

current physical health in HIV-infected patients over time whereas the alternative hypothesis is

that change in current viral load over time can predict change in current physical health in HIV-

infected patients over time.

Research Question 4

How well does change in previous viral loads over time predict change in current physical

health in HIV-infected patients over time?

Research Hypothesis 4

The null hypothesis is that change in previous viral load over time cannot predict change in

current physical health in HIV-infected patients over time whereas the alternative hypothesis is

that change in previous viral load over time can predict change in current physical health in HIV-

infected patients over time.

Research Question 5

How well does the change in current CD4+ cell count over time predict change in current

mental health in HIV-infected patients over time?









Research Hypothesis 5

The null hypothesis is that change in current CD4+ cell count over time cannot predict

change in current mental health in HIV-infected patients over time whereas the alternative

hypothesis is that change in current CD4+ cell count over time can predict change in current

mental health in HIV-infected patients over time.

Research Question 6

How well does the change in previous CD4+ cell count over time predict change in current

mental health in HIV-infected patients over time?

Research Hypothesis 6

The null hypothesis is that change in previous CD4+ cell count over time cannot predict

change in current mental health in HIV-infected patients over time whereas the alternative

hypothesis is that change in previous CD4+ cell count over time can predict change in current

mental health in HIV-infected patients over time.

Research Question 7

How well does change in current viral loads over time predict change in current mental

health in HIV-infected patients over time?

Research Hypothesis 7

The null hypothesis is that change in current viral load over time cannot predict change in

current mental health in HIV-infected patients over time whereas the alternative hypothesis is

that change in current viral load over time can predict change in current mental health in HIV-

infected patients over time.

Research Question 8

How well does change in previous viral loads over time predict change in current mental

health in HIV-infected patients over time?









Research Hypothesis 8

The null hypothesis is that change in previous viral load over time cannot predict change in

current mental health in HIV-infected patients over time whereas the alternative hypothesis is

that change in previous viral load over time can predict change in current mental health in HIV-

infected patients over time.

Significance of Research

Understanding how patients assess their QOL in chronic diseases such as HIV disease is

very important because patients have to take medication continuously. In addition, clinical lab

tests are performed to help clinicians diagnose patients' condition and decide whether to treat or

prescribe medication to the patients. The research will provide two significant contributions,

clinical care and methodological contribution. For clinical care, this research will help clinicians

profoundly understand how patients assess their current QOL and what factors that patients take

into account when they assess their QOL. Specifically, it aims to understand whether patients

will consider all the information currently available at hand to assess their current QOL or they

will consider a combination of past and current information to assess their current QOL (e.g., use

current symptoms and past clinical lab test).

For methodological contribution, this study will help researcher design study about the

timing to measure patients' QOL that really capture their current QOL. For example, if the

patients assess their current QOL regard to symptoms and previous clinical lab test result

because it takes a few days or a week to know the current lab test result from current clinic visit,

researcher should postpone collecting QOL data until patients receive their current lab test result.

Moreover, if lag-time between clinical lab test and QOL really exists (e.g., 4 months), patients

should be monitored for their QOL at least every 4 months, but no longer than 4 months.











Characteristics of
the Individual





Symptom Personality Values
Amplification Motivation Preferences


Psychological Social and
Supports Economic


Social and
Psychological
Supports


Figure 1-1 Wilson and Cleary's Conceptual Model for QOL Assessment









CHAPTER 2
LITERATURE REVIEW

In this chapter, literature related to the epidemiology of HIV/AIDS and its classification

and treatments are reviewed. In addition, this review offers (1) a conceptual model for QOL

assessment, (2) an analysis of instruments used for measuring QOL, with an emphasis on the

time frame used when measuring QOL, and (3) findings about the relationship between CD4+

cell count, viral load and QOL.

Epidemiology of HIV/AIDS

The first five cases ofPneumocystis carinii pneumonia, soon after called Acquired

Immunodeficiency Syndromes (AIDS), were reported in 1981 (7, 8). By 1989, approximately

one million people in United States were infected with HIV (9). The peak of AIDS in the United

States was reached in 1992, and, fortunately, the number of AIDS cases decreased 47% from

1992 to 1998 (1). The numbers of new AIDS cases continued to decrease during 1996 to 2004.

During 1981-1995, most AIDS cases were white. Since 1996, however, most AIDS cases were

African/nonhispanic. Table 2-1 shows the estimated numbers and percentage of HIV/AIDS and

AIDS cases by diagnosis, age, gender and other characteristics from 1981-2004.

Classification of HIV Infection

Two maj or HIV classification systems are currently available for clinicians to diagnose and

classify HIV-infected patients, the U.S. Centers for Disease Control and Prevention (CDC)

classification system and the World Health Organization (WHO) Clinical Staging and Disease

Classification System. The WHO system classifies HIV disease based upon clinical

manifestations that can be identified and treated by clinicians in practice settings where

laboratory tests are not available whereas CDC classification system uses both CD4+ cell count

and HIV-related conditions.










According to the 1993 Revised Classification System HIV infection and Expanded

Surveillance Case Definition for AIDS among Adolescents and Adults, patients can be classified

into nine categories by clinical conditions (Table 2-2) and CD4+ cell count as shown in Table

2-3.

Antiretroviral Agents

Many antiretroviral drugs are available in the market. Generally, there are four classes of

anti-retroviral agents; (1) nucleoside/nucleotide reverse transcriptase inhibitors (NRTIs), (2) non-

nucleoside reverse transcriptase inhibitors (NNRTIs), (3) protease inhibitors (PIs) and (4) fusion

inhibitors (10, 11). Nucleoside/nucleotide analogue includes Zidovudine, Didanosine,

Abacarvia, and Tenofovir. NNRTIs include Nevirapine, Efavirenz, and Delavirdine. Protease

Inhibitors, which are more potent than the first two groups, include Saquinavir, Ritonavir and

Indinavir. Fusion inhibitor includes Enfuvirtide.

Conceptual Model for Quality of Life Assessment

A model to assess QOL was first proposed in 1995 by Wilson and Cleary (4). It includes

five maj or components which are discussed more fully below: (1) biological and physiological

variables, (2) symptom status, (3) functional status, (4) general health perception and (5) overall

quality of life. Also included in the model are characteristics of the environment and the

individual. In 2005, this model was revised by Ferrans and colleagues by integrating the

ecological model into Wilson and Cleary's model as shown in Figure 2-1 (12). The revised

model indicates that the characteristics of the individual and environment can also influence all

the components in main path of the model and the possible direction in the main path goes from

biological function, symptoms, functional status, general health perception, and overall quality of

life.









Biological Function

Biological function, previously known as biological and physiological variables in

Wilson's model, includes the processes at the molecular and cellular level. It also focuses on the

function of an organ or an organ system. For example, in the circulatory system, systolic and

diastolic blood pressure are measurable biological functions in hypertension, fasting blood sugar

in diabetes, and viral load and CD4+ cell count in HIV infection disease.

Symptoms

Unlike the collection of measures at the molecular and cellular level, symptoms are

observed directly by patients and/or providers because they are related to a "patient' s perception

of an abnormal physical, emotional or cognitive state" as stated by Wilson (4). However,

symptoms may be categorized into either physical, psychological or psychophysical from

Ferrans's perspective. In HIV infection disease, symptoms such as diarrhea and myalgia,

whether they are HIV disease-related or HAART-related, are meaningful because these

symptoms may affect a patient' s evaluation of his or her QOL.

Functional Status

Wilson and Cleary defined functional status as "the ability to perform a particular defined

task" in multiple domains such as physical function, social function, role function, and

psychological function whereas Ferrans defined functional status differently from Wilson and

Cleary by including Leidy's framework (4, 12). In Leidy's framework, functional status includes

four dimensions: (1) functional capacity, (2) functional performance, (3) functional capacity

utilization, and (4) functional reserve. Details about the difference of functional status between

these two concepts are discussed elsewhere (4, 12).

Functional status in HIV infection measured by MOS SF-36 includes physical functioning,

role functioning, and social functioning. For role functioning, patients are asked whether they









are able to perform regular activities normally such as "during the past 4 weeks, have you had

any of the following problems with your work or any regular activity as a result of physical

health?." For physical functioning, patients are asked whether how much their health is limited

in vigorous activities such as running, lifting heavy objects or participate in strenuous sports.

General Health Perception

Ferrans agreed with Wilson and Cleary that general health perception is different from

other components on the left side of the model in Figure 2-1. From Wilson and Cleary's

perspective, general health perception includes all the components that come earlier in the model

and they all represent subjective measures. For example, general health perception in MOS SF-

36 includes 5 items. One item asks the patients "in general, would you say your health is" and

the answers range from "excellent/very good/good/fair/poor."

Overall Quality of Life

Wilson and Cleary defined overall quality of life as "subj ective well-being related to how

happy and satisfied someone is with life as a whole." It is a multidimensional construct. Ferrans

concurred with Wilson and Cleary in this concept and he explained that Wilson and Cleary's

concept of overall quality of life and how it is influenced by patient' s value and preferences was

concordant with Campbell's concept (12). For example, being blind may be viewed by one

person as a disability that is not worth living with, but might be considered only moderately

bothersome for another person. In the case of HIV, another example might be where

lipodystrophy might be considered less important in an HIV-infected male patient compared with

an HIV-infected female patient.

Time Reference in MOS SF-36 Instrument for Current QOL

Some questions in the MOS SF-36 ask patients to assess their current QOL regarding their

ability to perform some activities such as running, and walking one block. Other questions









impose a time frame to help patients assess their current quality of life, but this time frame varies

from question to question. For example, one item in the 'general health domain' asks patients to

"compare to one year ago, how would you rate your health in general now?," but another

question asks patients to respond based on their experiences during the past 4 weeks, "During the

past 4 weeks, have you had any of the following problems with your work or other regular daily

activities as a result of your physical health?." Thus, the time frame imposed on the questions in

the MOS SF-36 varies from one week to four weeks. Four weeks to one year?

In addition, the responses to the rating scale used for scoring each question in the MOS SF-

36 is are summed to create a summated score that reflects each patient' s current QOL. This

implies that the current QOL is composed of the attributes of these questions within the past 4

weeks to the day that patients assess their QOL. Patients' QOL beyond one month ago is not

assessed in current QOL.

Validity of Quality of Life Assessment Model

Wilson and Cleary's QOL model was tested and validated in several studies in different

diseases (13-16). For example, Sousa and Kwok investigated the five major components of

Wilson and Cleary's QOL model simultaneously by using structural equation modeling (SEM) in

patients living with AIDS from the AIDS Time-Oriented Health Outcomes Study (13), whereas

Wettergren and colleagues examined the relationship of those components in Hodgkin' s disease

(17).

As stated by Wilson and Clearly, the non-existence of an arrow in the model and its

direction between those components doesn't mean it doesn't exist. However, this model was

developed from the biomedical model developed to show the causal relationship between the

construct on the left and the construct on the right and the finding from Sousa' s study showed

that the model was valid (13).









In regard to the components in the model, for the purpose of the present study Wilson and

Cleary's model was modified as shown in Figure 2-2 because functional status is a part of

general health perception in Wettergren' study, and functional status, general health perception

and overall quality of life are included in MOS SF-36 as QOL. This modified model is similar to

the conceptual model to assess QOL in HIV/AIDS population proposed by Vidrine and

colleagues (18).

CD4+ Cell counts, Viral load, and QOL in HIV-infected Persons

Numerous studies investigated the association of viral loads, CD4+ cell counts, symptoms,

and QOL in patients with HAART (19-25). Most of these studies compared QOL among

different treatment regimens or among groups of patients categorized by either CDC

classification or CD4+ cell counts or viral load level. Some addressed the difference in change

in QOL among treatment regimens. For example, Nieuwkerk and colleagues examined the

difference in change in HRQOL between two regimens, ritronavir (RQV)/saquinavir (SQV)

versus RQV/SQV/stavudine (d4T), in asymptomatic and symptomatic patients (23). QOL was

measured by MOS-HIV at baseline and after 12, 24, 36 and 48 weeks of follow up. The MOS

HIV was developed specifically for Multicenter AIDS Cohort study (MACS). Mean change in

QOL between the two regimens was compared by repeated measured analysis of variance

(repeated ANOVA). This statistical model used regimens as a between subject factor and time

as a within subject factor, adjusting for baseline CD4+ cell counts and viral loads. The results

showed no difference in change in QOL from baseline between the two regimens, but QOL

statistically increased from baseline for all dimensions.

Gill conducted a cross-sectional study to investigate the relationship between viral load,

CD4+ cell counts, HAART use, and HRQOL by using baseline data from 513 participants in

Nutrition for Healthy Living (NFHL) conducted in Boston and Providence (26). Four domains









in HRQOL, physical functioning (PF), role functioning (RF), energy levels (EL), and general

health perception (HP) were selected and obtained from the HIV Patients Assessed Report of

Status and Experience (HIV-PARSE). CD4+ cell counts and viral load were categorized by

clinically meaningful cut off points: CD4 > 500, 200-500, and < 200 cells/mL; VL < log 2.6

(undetectable, < 400 copies/mL), log 2.6 to 4.0 (400-10,000 copies/mL) and > log 4.0 (10,000

copies/mL). The results showed that HAART and viral load level had a significant effect on PH

only, whereas CD4+ cell counts had the significant effect on PF, RF and HP.

In the COMBINE-QoL substudy, Casado et al. assessed the effect of HAART regimens -

zidovudine (ZDV), lamivudine (3TC) and either nelfinavir (NFV) or nevirapine (NVP) on QOL

in HIV-infected naive patients. They found a statistically significant correlation between the

MHS score at 12 months and a decrease in viral load in only the ZDV/3TC/NFV arm, whereas

PHS at 12 month and a decrease in viral load were statistically correlated in the ZDV/3TC/NVP

arm (22).

Globe et al. conducted a cross-sectional survey and reviewed medical records to

investigate the association between clinical parameters and HRQOL in hospitalized persons with

HIV disease (27). Data retrieved from medical records included length of stay during index

admission, CD4+ cell count during index admission, AIDS-related diagnoses at admission, the

number of comorbid medical conditions at admission, and the number of presenting symptoms.

Outcomes were measured by the specific HRQOL questionnaire which was modified from the

Medical Outcome Study HIV (MOS-HIV), HIV Outcomes Study (HOS) and HIV-PARSE.

The results showed that CD4+ cell count did not have a significant relationship with most of

HRQOL dimensions i.e. physical, role and social function, but it had a significant negative

association with emotional well being and cognitive function.









Eriksson and colleagues investigated the association between CD4+ cell counts and

HRQOL in 72 HIV-infected Swedish men (28). HRQOL was measured by The Swedish

HRQOL questionnaire (SWED-QUAL) which comprised of 13 dimensions. Because the sample

size is small for both the symptomatic and AIDS groups, non-parametric analyses, such as the

Kruskal-Wallis or the Mann-Whitney U-test, were used. They found a statistically significant

difference between asymptomatic, symptomatic, and AIDS patients in physical functioning,

mobility, satisfaction with physical ability and role limitations because of physical health. In

addition, post-hoc comparisons revealed a significant difference between asymptomatic patients

and AIDS in those dimensions.

Preau et al. examined the longitudinal association between CD4+ cell count, viral loads,

clinical stage, the numbers of self-reported symptoms, other factors such as depression, HIV

transmission, and HRQOL measured by MOS SF-36 from baseline to 3 years in 360 patients

(29). Data were analyzed by regression analysis with Generalize Estimation Equations (GEE)

which accounted for the correlation between HRQOL measured at different time points. Only

the number of self-reported symptoms was significantly associated with MCS and PCS.

In the ACTG 175 substudy, Justice et al. examined whether physician-reported symptoms

were a clinically important subset of patient-reported symptoms in HIV infection. Both

physicians and patients used a similar format of questions. They reported that physicians

underreported the prevalence of symptoms compared with patient reports. The researchers also

noted that the physician reports of symptoms, using patient reports of symptoms as a gold

standard, had poor sensitivity, good specificity, moderate positive predictive value and a poor

negative predictive value (30). In addition, symptom severity, not the symptoms itself, was









associated with HRQOL measured General Health-assessment Questionnaire for Clinical Trials.

Both CD4+ cell count and viral load were not associated with HRQOL.

All of the studies mentioned previously differ from those studies that focus on how a

change in CD4+ cell counts and viral load over time predict change in QOL over time,

particularly when each individual has his/her own growth curve trajectory. Further details about

how individual change is different from group change can be found in Clarke's article (31).

Examples of those research findings are as follows.

Chu and co-researchers examined the rate of change of CD4+ cell counts over time, also

called CD4+ cell count traj ectory, at both the population and individual level by using two

different databases, Multicenter AIDS Cohort Study (MACS) and Women's Interagency HIV

Study (WIHS) (32). CD4+ cell counts were modeled by using a Bayesian change point model.

The results showed that in the population model, both men and women had a significant change

in CD4+ cell counts within 2 years after HAART initiation. However, in the individual model,

both men and women gained significant change in CD4+ cell counts after 7 years of HAART

initiation.

Weinfurt et al. investigated the relationship between a change in CD4+ cell counts, viral

load, HRQOL and time, as well as how the change of these factors correlated to one another in a

double-blinded randomized clinical trial (33). This trial compared the effect of 2 regimens, ddl

or ddl and delavirdine mesylate. The results revealed that MCS and PCS statistically decreased

over time. However, they did not investigate how CD4+ cell counts and viral load can predict

HRQOL over time. Liu investigated the predictors for lower QOL in HAART among HIV-

infected men (34). Predictors were educational level, individual risk behaviors, social support,

biological markers, HIV-related medications, and clinical outcomes. The time-lag model,









predictors at time (t-1) and outcome at time (t), was used. CD4+ cell count was independently

associated with PHS. Clinical outcomes were significant predictors for PHS, but not MHS.

The most recent five-year longitudinal study, called The French APROCO-COPILOTE

(ANR CO-8) multicenter cohort study, investigated the relationship between numerous variables

including the number of self-reported side effects, depression, clinical disease stage, CD4+ cell

count and HRQOL measured by MOS SF-36 in HIV-1 infected patients on HAART (35). The

results showed that both the number of self-reported side effects and CD4+ cell count had a

negative association with PCS and MCS in the first year.










Table 2-1 The estimated numbers and percentage of HIV/AIDS and AIDS cases by diagnosis, age, gender and race in 1981-2004
AIDS HIVIAIDS
1981-1995 1996-2000 2001-2004 2001-2004
N % N %N %N %
Sex
Male 467,286 84.7 173,608 75.9 120,242 73.4 112,237 71.3
Female 84,229 15.3 55,253 24.1 43,576 26.6 45,231 28.7
Age Groups
<13 7,668 1.4 1,426 0.6 341 0.2 1,025 0.7
13-19 2,748 0.5 1,659 0.7 1,480 0.9 4,336 2.8
20-29 98,990 18.0 30,161 13.2 19,632 12.0 31,503 20.0
30-44 336,967 61.1 137,963 60.3 90,581 55.3 80,063 50.8
45-59 89,530 16.2 49,658 21.7 44,862 27.4 34,882 22.2
260 15,612 2.8 7,996 3.5 6,921 4.2 5,660 3.6
Race
White/non-Hispanic 256,460 46.5 72,314 31.6 46,325 28.3 45,497 28.9
Black/non-Hispanic 190,561 34.6 107,618 47.0 81,057 49.5 80,310 51.0
Hispanic 98,438 17.9 45,529 19.9 33,185 20.3 28,725 18.2
Asian/Pacific Islander 3,660 0.7 1,868 0.8 1,788 1.1 1,360 0.9
American Indian/Alaska Native 1,490 0.3 858 0.4 736 0.5 768 0.5










Table 2-2 CDC HIV Infection Categories by Clinical Conditions
Category A consists of one or more of the conditions listed below in an adolescent or adult
(greater than or equal to 13 years) with documented HIV infection. Conditions listed in
Categories B and C must not have occurred.
-Asymptomatic HIV infection
-Persistent generalized lymphadenopathy
-Acute (primary) HIV infection with accompanying illness or history of acute HIV
infection (29,30) Category B

Category B consists of symptomatic conditions in an HIV-infected adolescent or adult that are
not included among conditions listed in clinical Category C and that meet at least one of the
following criteria: a) the conditions are attributed to HIV infection or are indicative of a defect
in cell-mediated immunity; or b) the conditions are considered by physicians to have a clinical
course or to require management that is complicated by HIV infection. Examples of conditions
in clinical Category B include, but are not limited to:
-Bacillary angiomatosis
-Candidiasis, oropharyngeal (thrush)
-Candidiasis, vulvovaginal; persistent, frequent, or poorly responsive to therapy
-Cervical dysplasia (moderate or severe)/cervical carcinoma in situ
-Constitutional symptoms, such as fever (38.5 C) or diarrhea lasting greater than 1 month
-Hairy leukoplakia, oral
-Herpes zoster (shingles), involving at least two distinct episodes or more than one
dermatome
diopathic thrombocytopenic purpura
-Listeriosis
-Pelvic inflammatory disease, particularly if complicated by tubo-ovarian abscess
-Peripheral neuropathy
-For classification purposes, Category B conditions take precedence over those in
Category A. For example, someone previously treated for oral or persistent vaginal
candidiasis (and who has not developed a Category C disease) but who is now
asymptomatic should be classified in clinical Category B.

Category C includes the clinical conditions listed in the AIDS surveillance case definition
(Appendix B). For classification purposes, once a Category C condition has occurred, the person
will remain in Category C.










Table 2-3 CDC Classification System for HIV Infection
Clinical Categories
CD4+ Cell Categories A B C
Asymptomatic Symptom atic AIDS
S500 cells/CLL Al B1 Cl*
200-499 cells/CLL A2 B2 C2*
<: 200 cells/CLL A3* B3* C3*
Patients in A3, B3 and C1-C3 are considered as AIDS


































Characteristics of
the environment


Figure 2-1 Ferrans's Conceptual Model for QOL Assessment

































Characteristics of
the environment


Figure 2-2 Modified Version of Wilson' s Model









CHAPTER 3
METHOD S

This chapter describes the conceptual framework for the research, information about the

secondary database that will be used (Multicenter AIDS Cohort Study database), the independent

and dependent variables with their operationalization and how the variables were transformed to

fit the assumptions of statistical methods for data analysis. In addition, the plan for handling

missing data, as well as the techniques that will be used to handle the correlation of multiple

QOL measurements in this longitudinal study will be summarized. Finally, two statistical

models, traditional and time-lag, will also be described and tested.

Conceptual Framework

The modified QOL assessment model that is used in this dissertation is comprised of two

components: biological function and QOL. The modified model is used in this study for the

following reasons. First, measuring QOL is very important since in order to achieve the

treatment goals specified in the clinical practice guidelines for treating HIV-infection, "to reach

maximal and durable suspension of viral load, restoration and preservation of immunologic

function, improvement of quality of life, and reduction of HIV-related morbidity and mortality,"

it is important to study how QOL in HIV-infected patients treated with HAART changes over

time.

Second, the temporal relationship from Wilson and Clearly's model implies that change in

biological function occurs before change in symptoms and change in symptoms occurs before

change in QOL. From a pharmacological perspective, HAART suppresses viral load and

increases CD4+ cell count. Viral load and CD4+ cell count are related to both symptoms and

QOL. Therefore, the modified version of Wilson' s model was used to investigate the

longitudinal relationship between the trend over time of the previous lab tests or current lab test









results with current QOL. Specifically, this modified model was used to empirically validate

whether the lag time relationship between previous lab test results and current QOL exists by

comparing time-lag model with non time-lag model. Figure 3-1 shows the modified conceptual

model for the lag time longitudinal relationship between previous biological function and current

QOL.

Data

Public Data Set from the Multicenter AIDS Cohort Study (MACS) associated with Johns

Hopkins University will be used in this study. MACS, initiated in 1984, is a study of the natural

and treated histories of HIV-1 infection in homosexual and bisexual men conducted by sites

located in Los Angeles, Chicago, Pittsburgh and Baltimore. It is a prospective cohort study with

semi-annual visits. At each visit, descriptions of the medication, medication adherence, physical

examination, HIV-related symptoms, side effects of antiretroviral medication, and QOL are

collected. By 2007, 6,972 patients were enrolled in MACS, accounting for 74,536 person-years.

Unfortunately, not all of the MACS data is publicly available. This study uses the MACS Public

Data Set version Pl5 which includes patients enrolled until October 2002.

The MACS database comprises several data tables, but only three data tables were used in

this study; (1) drug, (2) quality of life and (3) lab tests. In the drug table, antiretroviral drugs

prescribed at each clinic visit were recorded. This table was used to identify patients prescribed

HAART therapy.

HAART Definition

The definition of HAART used by the MACS study was based on the DHHS/Kaiser Panel

[DHHS/Kaiser 2005] guidelines. The guidelines define HAART as: (a) two or more NRTIs in

combination with at least one PI or one NNRTI (89% of observations classified as HAART); (b)

one NRTI in combination with at least one PI and at least one NNRTI (6%); (c) a regimen









containing ritonavir and saquinavir in combination with one NRTI and no NNRTIs (1%); and (d)

an abacavir or tenofovir containing regimen of three or more NRTIs in the absence of both PIs

and NNRTIs (4%), except for the three-NRTI regimens consisting of: abacavir + tenofovir +

lamivudine OR didanosine + tenofovir + lamivudine. Combinations of zidovudine (AZT) and

stavudine (d4T) with either a PI or NNRTI were not considered as HAART.

Health-Related Quality of Life

Patients were asked at each visit to complete the self-administered Medical Outcome Study

Short Form 36 (MOS SF-36) health survey. This instrument was first incorporated in the MACS

protocol in 1994 (19). MOS SF-36 consists of eight HRQOL domains: general health

perception, physical functioning, role limitation caused by physical health problems, role

limitation caused by mental health problems, emotional well-being, social functioning,

energy/fatigue, and pain. MOS SF-36 domains were classified into two major components,

physical and mental health component. Physical component summary scores (PCS) include

general health perception, physical functioning, role limitation caused by physical health

problems and pain, whereas mental component summary scores (MCS) include role limitation

caused by mental health problems, emotional well-being, social functioning and energy/fatigue.

This study will use the PCS and MCS as the dependent variable.

CD4+ Cell Count

In the MACS study, CD4 cell lymphocytes were quantified by flow cytometry in the

laboratories certified by Flow Cytometry Quality Assessment Program of the National Institute

of Allergy and Infectious Diseases (34, 36). CD4+ cell count was measured as percent CD4+

cell.









Viral Load

HIV-1 RNA level was measured by Nuclisens in the laboratories certified by the Virology

Quality Assurance Laboratory proficiency testing program of the National Institute of Health

(34, 36). Viral load was measured in term of the number of copies per millimeter. Because viral

load is not normally distributed, viral load levels were transformed into loglo scale (33).

Inclusion and Exclusion Criteria

Patients included in this study are all those who were prescribed HAART. The index date

was defined as the first clinic visit patients received HAART regimen. Patients were followed

until the cut-off date in the database. Patients with only one laboratory test were excluded from

data analysis because insufficient data is available for the time-lag model.

Data Analysis Process

This study used a secondary database called MACS. Data preparation was the first step in

working with MACS. The data were modeled in the second step. Time-lag and non time-lag

models were compared in the final step.

Data Preparation

A descriptive analysis was conducted for each variable, e.g., average and standard

deviation for continuous variables and percent for categorical variables. In addition, missing

data associated with QOL were tested to determine whether it was missing completely at random

(MCAR) or missing at random (MAR). A strategy for handling dropouts and missing values was

employed, in order to increase the sample size because some patients dropped out of the MACS

study and some had missing values for study variables. If missing data were MAR, multiple

imputation was used. If they were MNAR, selection or pattern mixture model was used. The

strategy for handling missing data is presented in more detail in the next section.









Missing data and dropping out in the longitudinal study

Missing data and subject drop outs are common in longitudinal studies. Dropping out in a

longitudinal study is theoretically equivalent to either unit non-response or item non-response in

cross-sectional surveys (37, 38). Unit non-response occurs when the survey from a sampled

person is incomplete whereas item non-response takes place when a participant does not answer

one or more items in the questionnaire.

Several review articles related to missing data mechanisms and how to handle missing data

are available (37-39). Generally, three different "missingness" mechanisms are noted: (1)

missing completely at random (MCAR), (2) missing at random (MAR) and (3) missing not at

random (MNAR). In MAR, if R is an indicator for the missingness of data and Yeom is the

complete data. Yeom can be partitioned into Yobs and Ymis, where Yobs and Ymis are the observed

and missing parts respectively. The relationship between Yeom, Yobs and Ymis can be presented by

the following equation because the distribution of missingness does not depend on Ymis, but on

Yobs (38).

P(R/ Y, ,)= P(R/ Yobs

For MCAR, the distribution of missingness does not depend on Yobs.

P(R / Ys, ) = P(R)

When P(R / Ys, )= P(R /Yobs) is Violated, the distribution of missingness depends on Yobs,

and it is called MNAR. Table 3-1 shows missing patterns of CD4+ cell count measured in two

consecutive months (X and Y).

How to handle dropouts in longitudinal study

Several techniques or models to handle missing data were developed and tested, including

mean of series method, hot-deck method, last value carry forward method, regression imputation,









multiple imputation and weighting (37, 38, 40-42). When data were missing not at random, two

types of models were developed and tested: selection models and pattemn-mixture models (39,

43, 44). For selection models, Y and the probability that Y is missing are modeled

simultaneously as shown in equation 1.

f (y,r) = f (y) P(r/y) (1)

Pattern mixture models can be expressed by equation 2.

f (y,r) = f (y/r) P(r) (2)

These models do not have any implicit untestable assumptions and they have

computational advantages. An example of random a pattern mixture model is given below (43).

Advantages and disadvantages of both models are summarized in Table 3-2.

QOLlt = aa + a,, (Visit) + al2 13etmn) (Gender) + PI, + ul + e

Log(DropoutTime) = a20 21a, (Treatment) + bul + ei,

Where P, and u, are the subject and pattern level random intercepts.

Data Analysis

In this step, health-related quality of life was modeled using a random effect model, also

called multilevel model. Multilevel models have been used in studying functional impairment in

HIV-infected persons (45) and relationships between CD4+ cell count, viral load, and quality of

life over time in HIV-1-infected patients (33). These models have an advantage over repeated-

measured ANOVA and other statistical analyses because a balanced data structure is not required

(46-48) and it takes into account unit of measurement, correlation among measurements and

time-varying covariates (49). The statistical model is presented below and the conceptual

diagram for this model is found in Figure 3-2.

Y, = or, + C a,X,* 2tt wf











Y, I N(X, ~2,c





Where YI, is PHC or IVHC for individual i measured at time t, orI is the random intercept,


XI, is the independent variable, CD4+ cell count or viral load, for subject i at time t, P,I is the

random regression coefficient for independent variable j, t is time, P22 is the random regression

coefficient to time and E,, is the error for subject i at time t. Time was coded as 1 for an index

date and coded as 2 to 11 corresponding to each visit to the clinic scheduled every 6 months

according to the study protocol.

In addition, PHC and MHC were modeled by a combination of time-lag model and random

coefficient models when knowledge of HIV test results from a previous visit was taken into

account. Consequently, the model mentioned above was partially modified as follow.




Where all variables and parameters are the same as the previous model, except XI,(t-i is the

independent variable j for subject i at time t-1. The time-lag conceptual model is shown in

Figure 3-3.

In addition, the correlation between QOL measurements will be taken into account based

upon one of these assumptions.

Independent Structure

In this structure, the correlations between subsequent QOL measurements are assumed to

be zero as show in Table 3-2.









Exchangeable Structure

In this structure, the correlations between subsequent QOL measurements are the same,

regardless of the time as shown in Table 3-3

m-dependent Structure

In this structure, it is assumed that the correlations < m measurements apart are equal and

correlations > m measurements apart are assumed to be zero as shown in Table 3-4 when m = 2.

Autoregressive Correlation Structure

In this structure, the correlation between two consecutive QOL is p, the correlation

between t measurement apart is p" as in Table 3-5.

Maximum likelihood estimation (MLE) method was used to fit the model. Using this

method, both the fixed part (coeffieients) and the random part (variance) between the models can

be compared. For both non time-lag and time-lag models, if the model didn't converge, grand-

mean centering was used to help achieve the model conversion. For nested models, model fit

was compared by the log likelihood ratio test (LR test). Non-nested models were compared by

Akaike information criterion (AIC). Model residuals were also examined to ascertain whether

the data were normally distributed. Type II error was set at 5%. All statistical analyses were

conducted by statistical program, STATA.

IRB Approval

This study was approved by Health Center Institutional Review Board, University of

Florida.










Table 3-1 Missing Data Mechanisms of CD4+ Cell count
Y
X Complete MCAR MAR MNAR
168 148 148 148 148


103 106 106
78 74 74
151 113 113
Mean (SD)
125.7 121.9 108.6 138.3 153.4
(23.0) (24.7) (25.1) (21.1) (7.5)













I ~Current


SSymptoms I IQuality of
I ~Life
I I,, ,, ,, ,


Table 3-2 Advantages and disadvantages of selection and pattern-mixture models
Selection Models Pattern Mixture Models
Models include parameters of interest -Models exclude parameter of interest
Easy to formulate hypothesis about drop -Make explicit assumptions about
out process unobserved responses
Difficult to infer how assumptions on -Implied drop out process is not
drop out process translate into immediately transparent
assumption about distribution of
unobserved responses
Difficult to determine model -Straightforward to determine model
identifiability identifiability
Computational intractable -Computational simple


Biological
Function


Lag Time


Figure 3-1 Time-lag Conceptual Model for Quality of Life Assessment












Lati -- Labt2 -- Labt3 -- Labt4 -- Labts












QOhi -- QOLt2 -- QOL3t3-- QOLt4-- QOL5ts



Figure 3-2 Conceptual Diagram for Traditional Model






Labto Lat -- Labt2 -- Labt3 -- Labt4












QOhi -- QOLt2 -- QOL3t3-- QOLt4-- QOL5ts



Figure 3-3 Conceptual Diagram for Time-lag Model









Table 3-2 Independent Structure of QOL Measurements
QOLtl QOLt2 QOLw QOLt4 QOLts
QOLtl 0 0 0
QOLt2 0 0 0
QOLw 0 0 0
QOLt4 0 0 0 -
QOLts 0 0 0 0


Table 3-3 Exchangeable Structure of QOL Measurements
QOLtl QOLt2 QOLw QOLt4 QOLts
QOLtl p p pp
QOLt2 p p pp
QOLw p p pp
QOLt4 p p p -p
QOLts p p p p


Table 3-4 m-dependent Structure of QOL Measurement
QOLtl QOLt2 QOLw QOLt4 QOLts
QOLtl -1 p, O 0
Q OL t2 p1 -1 p, O
Q OLw Pt P1 p,
QOLt4 0 Pt P1 -P1
QOLts 0 0 Pt P1


Table 3-5 Autoregressive Correlation Structure of QOL Measurements
QOLtl QOLt2 QOLw QOLt4 QOLts
QOLti pp2 P 4
QOLt2 pl p p2
QOLt3 P2 pl 1 P2
QOLt4 P3 p2 1 1
QOLts p4 p 2 1'









CHAPTER 4
RESULTS

This chapter is divided into three sections. First, the result from merging the three data

tables from MACS database is described. Second, the result from exploratory analysis of PHC

and MHC is individually presented. Finally, the results from traditional and time-lag models are

presented.

Merging MACS Database

Data from the following tables were merged by using case number and visit as a merging

key variable.

Drug Table

The drug table has 13,668 observations from 1,599 patients. Any observation before visit

210 was dropped because QOL was first collected in MACS at visit 210, resulting in 8,677

observations. Following this step, the visit with the first HAART was determined and any visit

before the first HAART was dropped. Consequently, 5,701 observations were used to merge

with QOL and lab test tables.

QOL Table

The QOL table contains 19,913 observations from 2,858 patients. This table was merged

with the drug table, resulting in 20,509 observations. 14,808 observations were dropped because

these observations were only available in the QOL table. Finally, 5,701 observations were left to

merge with the lab test table.

Lab Test Table

The lab test table contains 22,886 observations from 2,701 patients. Merging data from the

previous section with the lab test table resulted in 3,452 observations from 490 patients for data

analy si s.









Missing Data

After merging the three data tables together, 18 observations (0.52%) had a missing value

in CD4+ cell count and 34 observations (0.98%) had a missing value in viral load. Because

missing data is less than 1%, all missing values were substituted with the average of that variable

for a particular patient.

Exploratory Analysis

Demographic Data

Table 4-1 depicts clinic visits when patients started HAART. The average age of the 490

patients when they started HAART (index date) was approximately 45 years (Table 4-2), with

most patients between 31 to 60 years old (Table 4-3). The average follow up time for all patients

after HAART initiation was approximately 46 months (Table 4-4). Most patients were followed

up at least 5 years (Table 4-5), while some patients were followed up only 2 years because they

entered the study later than the others. In other words, some patients entered the study as little as

2 years before the cut-off date (2002). Most patients had a college degree (Table 4-6).

Quality of Life Traj ectory

Figure 4-1 and Figure 4-2 show PHC and MHC score traj ectory of 10 randomly selected

patients respectively. In both figures, each patient has a different baseline PHC and MHC level.

In addition, each patient has a different rate of change in QOL over time. The average PHC

score after patients started HAART was 76.03 whereas the average MHC score was 73.39 (Table

4-7 and Table 4-8). Average PHC scores ranged from 74.03 to 76.54, whereas average MHC

scores ranged from 71.52 to 73.50.

The standard error of the mean for PHC at 6 months after HAART initiation was 1.00 and

increased thereafter. Average PHC at 6 months after initiation ranged from 74.06 to 78.01. The

standard error of the mean for PHC increased over time. The standard error of the mean for









MHC at 6 months after HAART initiation ranged from 1.04 to 1.57 and increased over time.

The average PHC and average MHC scores fluctuated during the follow-up period as shown in

Figure 4-3 and Figure 4-4.

CD4+ Cell Count and Viral Load Trajectory

Figure 4-5 shows the change in CD4+ cell count over time for 10 patients. These changes

in CD4+ cell count over time indicate a unique traj ectory of CD4+ cell count for each individual.

However, the average change in CD4+ cell count over time increases as shown in Figure 4-6 and

Table 4-9. CD4+ cell count 6 months after HAART initiation was 220.68 and increased over

time except for the last observation where the CD4+ cell count dropped. The standard error of

the mean for CD4+ cell count also decreased over time. The average change in lag CD4+ cell

count also improved overtime as shown in Figure 4-7.

Figure 4-8 shows the change in viral load over time for 10 patients. Again, each patient

has a different initial viral load level and has a different rate of change over time. Table 4-10 and

Figure 4-9 depict the change in average viral load over time for all patients. The average viral

load at 6 months after HAART initiation was 30,398.02 copies/mL. Although viral load

fluctuated, overall the trend of average viral load decreased over time. The average change in

lag viral load over time also was lower as shown in Figure 4-10.

Correlation among PHC and MHC

Table 4-11 shows the correlation coefficients between PHC measured at different follow-

up times. The correlation coefficient ranges from 0.57 to 0.83. The highest correlation

coefficient of PHC is between index 9 (4 years after HAART initiation) and index 11 (5 years

after HAART initiation), whereas the lowest correlation coefficient of PHC is between index 2

(6 months after HAART initiation) and index 10 (54 months after HAART initiation).









Table 4-12 also shows the correlation coefficients between IVHC measured at different

follow-up times. The highest correlation coefficient of 1VHC (0.74) is between index 2 (6

months after HAART initiation) and index 9 (4 years after HAART initiation), whereas the

lowest correlation coefficient of 1VHC (0.59) is between index 4 (24 months after HAART

initiation) and index 9 (4 years after HAART initiation).

Time-Lag and Non Time-Lag Models

In this part, the time-lag and non-time lag models for PHC and IVHC are presented

individually.

Time-lag and Non Time-lag Models for PHC

Table 4-13 shows all models for PHC, However only the residuals for model 2 is shown in

Figure 4-9 since the residuals for other models were also normally distributed. The first model is

a null model which has no predictor. The log likelihood statistic (LL) was -13,931.55. For all

models, time has an inverse relationship with PHC (P = -0.21 to -0.44, p<0.05). In other words,

PHC scores significantly decreased over time. CD4+ cell count and previous CD4+ cell count

have a positive relationship with PHC in all models. Regression coefficients of previous CD+

cell count are the same as those of CD+ cell count.

The second and third models have time as the predictor. In the second model, time was

treated as a fixed effect. The log likelihood (LL) significantly increased from -13,931.55 to -

13,927.27 from model 1 to model 2 (p<0.05). This implied that, on average, PHC significantly

decrease over time. However, time was treated as a random effect in the third model. The log

likelihood statistic (LL) also increased from -13,927.27 in model 2 to -13,863.02 in model 3. In

other words, the difference in the log likelihood statistic (-2LL) of these two models was 128.50,

which indicated that the two models were statistically different (p<0.05). That means the effect









of time on PHC was different from patient to patient. In other words, each patient had his/her

own PHC trajectory.

Model 4 to model 7 answered research question 1. When CD4+ cell count was added into

the second model and treated for fixed effect, the log likelihood statistics (LL) increased from -

13,927.27 to -13,909.52. The difference in the log likelihood statistics (-2LL) was 35.50 which

was statistically different (p<0.05). CD4+ cell count significantly predicted PHC (P = 0.03,

p<0.05). When time was treated for random effect (model 5), the model fit statistic statistically

increased from -13,909.52 to -13,848.84, which implied that the random model was better than

the fixed model. In this model, CD4+ cell count significantly predicted PHC (P = 0.03, p<0.05).

When CD4+ cell count was treated for random effect (model 6), compared with model 4,

the log likelihood statistic increased from -13,909.52 to -13,889.82, which was statistically

significant (p<0.05). This indicated that the effect of CD4+ cell count on PHC varied among

patients. When both time and CD4+ cell counts were treated for random effect (model 7), the

log likelihood statistic increased from -13,909.52 to -13,844.13, which was statistically

significant (p<0.05). This indicated that PHC traj ectory and the effect of CD4+ cell count on

PHC were different among patients.

Model 8 to model 11 answered research question 2. When previous CD4+ cell count was

added into the second model and treated for fixed effect, the log likelihood statistic increased

from -13927.27 to -13905.95 (model 8). This showed statistical difference between two models

(p<0.05). In this model, previous CD4+ cell count significantly predicted PHC (P = 0.03,

p<0.05).

In model 9 where time was treated for random effect, the log likelihood statistic

significantly increased from -13,905.95 to -13845.09 (p<0.05). This showed that PHC trajectory









was different among patients. When previous CD4+ cell count was treated for random effect

(model 10), the log likelihood statistic significantly increased from -13,905.95 to -13,894.39

(p<0.05). This result indicated that the effect of previous CD4+ cell count on PHC varied among

patients. When both time and previous CD4+cell count were treated for random effect (model

11), the log likelihood statistic significantly increased from -13,905.95 to -13,844. 11 (p<0.05).

Again, this showed that PHC traj ectory and the effect of previous CD4+ cell count on PHC

varied among patients.

Model 12 to model 15 answered research question 3. When viral load was added into the

second model and treated for fixed effect, the log likelihood statistics (LL) increased from -

13,927.27 to -13,921.08 (model 12). The difference in the log likelihood statistics (-2LL) was

12.38 which was statistically different (p<0.05). In this model, viral load significantly predicted

PHC (P = 0.96, p<0.05). When time was treated for random effect (model 13), the log likelihood

statistic statistically increased from -13,921.08 to -13,857.86, which implied that the PHC

trajectory was different among patients. In this model, viral load significantly predicted PHC

(p = 0.89, p<0.05).

When viral load was treated for random effect (model 14), compared with model 12, the

log likelihood statistic increased from -13,921.08 to -13,903.09, which was statistically

significant (p<0.05). This indicated that the effect of viral load on PHC varied among patients.

When both time and CD4+ cell counts were treated for random effect (model 15), the log

likelihood statistic increased from -13,921.08 to -13,841.43, which was statistically significant

(p<0.05). This implied that the effect of CD4+ cell count on PHC and the trend of PHC varied

among patients.









Model 16 to model 19 answered research question 4. When previous viral load was added

into the second model and treated for fixed effect, the log likelihood statistic increased from -

13927.27 to -13920.39 (model 16). This showed statistical difference between two models

(p<0.05). In this model, previous viral load significantly predicted PHC (P = 0.98, p<0.05).

In model 17, time was treated for random effect, the log likelihood statistic significantly

increased from -13,920.39 to -13858.41 (p<0.05). This showed that PHC trajectory was different

among patients. When previous viral load was treated for random effect (model 18), the log

likelihood statistic significantly increased from -13,920.39 to -13,909.31 (p<0.05). This result

indicated that the effect of previous viral load on PHC varied among patients. When both time

and previous viral load were treated for random effect (model 19), the log likelihood statistic

significantly increased from -13,920.39 to -13,849.05 (p<0.05). Again, this showed that PHC

traj ectory and the effect of previous viral load on PHC varied among patients. In this model,

previous viral load also significantly predicted PHC (P = 0.75, p<0.05).

Comparing non-nested models (model 4 vs. model 8, model 5 vs. model 9, model 6 vs.

model 10, model 7 vs. model 11, model 12 vs. model 16, model 13 vs. model 17, model 14 vs.

model 18, model 15 vs. model 19), AIC among these pairs were not different. In other words,

there were no different between time-lag and non time-lag models.

Time-lag and Non Time-lag Models for MHC

Models for MHC are shown in Table 4-14. However, only the residuals for model 2 was

shown in Figure 4-10 because the residuals for other models were also normally distributed. The

first model was also the null model, no predictor, as in the previous section. The log likelihood

statistic (LL) for this model was -14,320.45. For all converged models, MHC scores










significantly decreased over time. In other words, time significantly predicted MHC scores (P= -

0.14 to -0.27, p<0.05).

The second and third models have time as the predictor. In the second model, time was

treated as a fixed effect but as a random effect in model 3. The log likelihood (LL) significantly

increased from -14,320.45 to 14,318.51 from model 1 to model 2 (p<0.05). This indicated that

on average MHC decreased over time. The log likelihood statistic (LL) also increased from -

14,318.51 in model 2 to -14,291.46 in model 3. In other words, the difference in the log

likelihood statistic (-2LL) of these two models was 54.08, which indicated that the two models

were statistically different (p<0.05). This implied that patients had different rate of change of

MHC scores over time.

Model 4 to model 7 answered research question 5. When CD4+ cell count was added into

the second model and treated for fixed effect, the log likelihood statistic (LL) increased from -

14,318.51 to -14,317.07. The difference in the log likelihood statistics (-2LL) was 2.87, which

was not different. When time was treated for random effect (model 5), the model fit statistic

statistically increased from -14,317.07 to -14,290.65 (p<0.05). This implied that MHC trajectory

varied among patients. In this model, time significantly predicted MHC (P = -0.25, p<0.05).

When CD4+ cell count was treated for random effect (model 6), compared with model 4,

the log likelihood statistic increased from -14,317.07 to -14,314.00, which was statistically

significant (p<0.05). This indicated that the effect of CD4+ cell count on MHC varied among

patients. When both time and CD4+ cell counts were treated for random effect (model 7), the

log likelihood statistic increased from -14,317.07 to -14,289.78 which was statistically

significant (p<0.05). This indicated that MHC traj ectory and the effect of CD4+ cell count on

MHC were different among patients.









Model 8 to model 11 answered research question 6. When previous CD4+ cell count was

added into the second model and treated for Eixed effect, the log likelihood statistic increased

from -14,318.51 to -14,316.31 (model 8). This showed statistical difference between two models

(p<0.05). In this model, previous CD4+ cell count significantly predicted MHC (P = 0.01,

p<0.05).

In model 9, time was treated for random effect, the log likelihood statistic significantly

increased from -14,316.31 to -14,289.98 (p<0.05). This showed that MHC trajectory was

different among patients. When previous CD4+ cell count was treated for random effect (model

10), the log likelihood statistic increased, but not significant, from -14,316.31 to -14,315.46.

This result indicated that the effect of previous CD4+ cell count on MHC did not vary among

patients. When both time and previous CD4+cell count were treated for random effect (model

11), the model did not converge.

Model 12 to model 15 answered research question 7. When viral load was added into the

second model and treated for Eixed effect, the log likelihood statistics (LL) increased from -

14,318.51 to -14,314.36 (model 12). The difference in the log likelihood statistics (-2LL) was

16.70, which was statistically different (p<0.05). In this model, viral load significantly predicted

MHC (P = 0.89, p<0.05). When time was treated for random effect (model 13), the log

likelihood statistic statistically increased from -14,314.36 to -14,288.02 (p<0.05). This implied

that the MHC trajectory was different among patients. In this model, viral load significantly

predicted MHC (P = 0.82, p<0.05).

When viral load was treated for random effect (model 14), compared with model 12, the

log likelihood statistic increased from -14,314.36 to -14,306.01 which was statistically

significant (p<0.05). This indicated that the effect of viral load on MHC varied among patients.









When both time and CD4+ cell counts were treated for random effect (model 15), the log

likelihood statistic increased from -14,314.36 to -14,277.60, which was statistically significant

(p<0.05). This implied that the effect of CD4+ cell count on MHC and the trend of MHC varied

among patients.

Model 16 to model 19 answered research question 8. When previous viral load was added

into the second model and treated for Eixed effect, the log likelihood statistic increased from

-14,318.51 to -14,315.46 (model 16). This showed statistical difference between two models

(p<0.05). In this model, previous viral load significantly predicted MHC (P = 0.73, p<0.05).

In model 17, time was treated for random effect, the log likelihood statistic significantly

increased from -14,315.46 to -14,289.30 (p<0.05). This showed that MHC trajectory was

different among patients. When previous viral load was treated for random effect (model 18),

the log likelihood statistic significantly increased from -14,315.46 to -14,310.87 (p<0.05). This

result indicated that the effect of previous viral load on MHC varied among patients. When both

time and previous viral load were treated for random effect (model 19), the log likelihood

statistic significantly increased from -14,315.46 to -14,284.22 (p<0.05). Again, this showed that

MHC traj ectory and the effect of previous viral load on MHC varied among patients. In this

model, previous viral load also significantly predicted MHC (P = 0.66, p<0.05).

Comparing non-nested models for both CD4+ cell count and viral load (model 4 vs. model

8, model 5 vs. model 9, model 6 vs. model 10, model 7 vs. model 11, model 12 vs. model 16,

model 13 vs. model 17, model 14 vs. model 18, model 15 vs. model 19), AIC among these pairs

were not different. In other words, there were no different between time-lag and non time-lag

models.





Table 4-2 Average age at HAART initiation
Variable N Mean SEM 95% CI
Age 490 44.91 0.37 44.18 45.63

Table 4-3 Age of patients at HAART initiation
Age N Percent

<30 4 0.82
31-40 147 30.00
41-50 245 50.00
51-60 73 14.90
61-70 16 3.26
70-80 1 0.20
>81 4 0.82
Total 490 100.00


95% CI
44.69 47.53


Table 4-1 Time patients started HAART
Index
220
230
240
250
260
270
280
290
300
310
320
330
340
350
Total


N
1
4
26
141
115
89
65
18
8
3
8
4
2
6
490


Percent
0.20
0.82
5.31
28.78
23.47
18.16
13.27
3.67
1.63
0.61
1.63
0.82
0.41
1.22
100.00


Table 4-4 Average follow up time after HAART
Variable N Mean
Follow- up time 490 46.11


initiation
SEM
0.72









Table 4-5 The number of patients followed up until last clinic visit after HAART initiation
Follow up Time (Months) N Percent
12 33 6.73
24 55 11.22
36 56 11.43
48 74 15.10
60 272 55.51




Table 4-6 Education level of patients who received HAART
Education N Percent
Less than 12th grade 4 0.89
12th grade 42 9.35
At least one year college, but no degree 139 30.96
For year college/got a degree 106 23.61
Some graduate work 61 13.59
Post graduate degree 97 21.60










Table 4 -7 Average physical health I


Table 4-8 Average mental health component score by time since HAART


component score by time since HAART
95% CI
Mean SEM
L U
76.03 1.00 74.06 78.01
74.18 1.09 72.05 76.32
75.53 1.02 73.53 77.55
75.52 1.06 73.44 77.59
75.38 1.11 73.19 77.56
76.06 1.14 73.82 78.30
76.54 1.17 74.25 78.94
75.69 1.24 73.25 78.14
74.73 1.30 72.16 77.31
74.03 1.45 71.17 76.88
Index = Time since started HAART with 6 months interval


Variable Index


PHC 2 425
PHC 3 416
PHC 4 401
PHC 5 386
PHC 6 373
PHC 7 343
PHC 8 318
PHC 9 297
PHC 10 280
PHC 11 213
PHC = Physical health component


95% CI


Variable Index


Mean

73.39
71.99
71.52
71.94
72.76
72.62
73.78
72.91
73.50
73.49


SEM

1.04
1.05
1.14
1.14
1.13
1.22
1.26
1.32
1.35
1.57


L
71.35
69.93
69.27
69.69
70.53
70.23
71.29
70.31
70.85
70.40


U
75.42
74.05
73.77
74.19
74.99
75.01
76.27
75.51
76.15
76.58


MHC
MHC
MHC
MHC
MHC
MHC
MHC
MHC
MHC
MHC


MHC = Mental health component Index = Time since started HAART with 6 months interval










Table 4-9 Average CD4+ cell count at each visit


95% CI


Index

2
3
4
5
6
7
8
9
10
11


N

425
416
401
386
373
343
318
297
280
213


Mean

220.68
231.13
239.65
249.62
250.73
251.42
259.93
259.08
260.97
255.10


SEM

5.18
5.24
5.14
5.30
5.41
5.56
5.87
6.06
6.39
6.76


L
210.52
220.86
229.58
239.23
240.12
240.51
248.43
247.19
248.43
241.84


U
230.85
241.40
249.72
260.02
261.35
262.33
271.43
270.96
273.50
268.36


Table 4-10


Average log viral load at each visit


95% CI


Index

2
3
4
5
6
7
8
9
10
11


N

425
416
401
386
373
343
318
297
280
213


Mean

2.74764
2.75666
2.65515
2.58011
2.51395
2.48340
2.36459
2.37969
2.30658
2.31888


SEM

0.06111
0.05980
0.06358
0.06432
0.06242
0.06392
0.06328
0.06559
0.06518
0.07861


L
2.62782
2.63941
2.53048
2.45399
2.39156
2.35806
2.24052
2.25109
2.17880
2.16475


U
2.86746
2.87390
2.77982
2.70622
2.63633
2.60873
2.48866
2.50828
2.43437
2.47301






























Table 4-12 Correlation coeffieients among MHC measured at different time
Index Index3 Index4 Index5 Index6 Index7 Index8 Index9 Index10 Index11


Table 4-11 Correlation coeffieients among PHC measured at different time
Index2 Index3 Index4 Index5 Index6 Index7 Index8 Index9 Index10 Index11


Index2 1.0000
Index 0.7315
Index 0.8130
Index 0.6600
Index6 0.7398
Index 0.6653
Index 0.6515
Index9 0.6576
Index10 0.5718
Index11 0.6077


1.0000
0.6778
0.6662
0.6638
0.6351
0.6272
0.6061
0.6109
0.6090


1.0000
0.6823 1.0000
0.7225 0.7355 1.0000
0.6977 0.6841 0.7804
0.7046 0.7137 0.7180
0.6821 0.7571 0.7347
0.6490 0.6476 0.7012
0.6516 0.6882 0.6984


1.0000
0.7656
0.8032
0.7614
0.7130


1.0000
0.8059
0.7630
0.7378


1.0000
0.8235 1.0000
0.8276 0.8004


1.000


Index 1.0000
Index3 0.7355
Index4 0.6541
Index Os.6653
Index 0.6708
Index7 017324
Index8 0.6336
Index 0.7449
Index10 0.6073
Index11 0.6286


1.0000
0.6745
Os.6544
0.6500
016420
0.6109
0.6277
0.6034
0.5846


1.0000
Os.7071 1.0000
0.6744 0.6850 1.0000
0.5881 0.6882 016907
0.6328 0.6626 0.6481
0.5827 0.6719 0.7103
0.6061 0.6165 0.6667
0.6210 0.6094 0.6513


1.0000
0.7071
0.7348
0.6942
0.7109


1.0000
0.7079
0.7004
0.7277


1.0000
0.6888 1.0000
0.7136 0.7100


1.000





Table 4-13 Models for PHC

Null


2
Index

75.66*
-0.21*






356.27






122.65

-13927.27
27854.54
27862.54


3
Index

75.92*
-0.27*






334.13
1.84





106.97

-13863.02
27726.04
27736.05


Fixed Effect
Intercept
Index
CD4+ cell^
Log VL
Lag CD4+ cell^
Log Lag VL
Random Effect#


74.46*







354.92






123.08

-13931.55
27863.10
27869.10


Intercept
Index
CD4+ cell^
Log VL
Lag CD4+ cell~
Log Lag VL
Residual"
Fit Statistics
LL
-2LL
AIC
* p<0.05 v


variance


^ grand mean-centered


LL = Log likelihood


-2LL = -2 Log likelihood


AIC = Akaike Information Criterion












Table 4-13 Models for PHC (continued)
4 5 6 7 8 9 10 11
Index & Index & Index & Index & Index & Index & Index & Index &
CD4+ CD4+ CD4+ CD4+ Lag CD4+ Lag CD4+ Lag CD4+ Lag CD4+
Fixed Effect


ariance ^ grand mean-centered


Intercept
Index
CD4+ cell
Log VL
Lag CD4+ cell
Log Lag VL
Random Effect#
Intercept
Index
CD4+ cell
Log VL
Lag CD4+ cell
ul Log Lag VL
Residual"
Fit Statistics
LL
-2LL
AIC
p<0.05 v


76.51*
-0.34*
0.03*





331.39






122.56

-13909.52
27819.04
27829.03


76.67*
-0.38*
0.03*





312.51
1.75





107.38

-13848.84
27697.68
27709.69


76.82*
-0.34*
0.03*


76.85*
-0.38*
0.03*


76.81*
-0.39*


0.03*


331.96


76.98*
-0.44*


0.03*


312.22
1.74


77.16*
-0.41*


0.03*


321.27


77.11*
-0.44*


0.03*


305.73
1.68


0.0007

106.32

-13844.11
27688.22
27702.23


318.18

0.004


332.86
1.59
0.002


0.003


115.53

-13889.82
27779.64
27791.64


105.59

-13844.13
27688.26
27702.26


122.24

-13905.95
27811.90
27821.89


107.13

-13845.29
27690.58
27702.58


117.24

-13894.39
27788.78
27800.77


LL = Log likelihood -2LL = -2 Log likelihood


AIC = Akaike Information Criterion








































AIC = Akaike Information Criterion


ed)


Table 4-13 Models for PHC continueu
12
Index &
Log VL


13
Index &
Log VL

78.41*
-0.31*

-0.89*




327.57
1.82





106.97


14
Index &
Log VL

78.49*
-0.26*

-0.93*


15
Index &
Log VL

78.51*
-0.31*


16
Index & Log
Lag VL

78.57*
-0.28*


17
Index & Log
Lag VL

78.30*
-0.32*




-0.80*

327.18
1.80


18
Index & Log
Lag VL

78.58*
-0.29*




-0.90*

314.86


19
Index & Log
Lag VL

75.35*
-0.33*




-0.75*

280.96
1.75


Fixed Effect
Intercept
Index
CD4+ cell^
Log VL
Lag CD4+ cell^
Log Lag VL
Random Effect#
Intercept
Index
CD4+ cell^
Log VL
Lag CD4+ cell^
Log Lag VL
Residual"
Fit Statistics
LL
-2LL
AIC
* p<0.05 v


78.40*
-0.26*

-0.96*




349.24






122.52


-0.87*


-0.98*

348.73


305.35


7.63


117.71

-13903.09
27806.18
27818.19


276.50
1.76


7.12


5.92
118.47

-13909.31
27818.62
27830.62


5.58
104.48

-13849.05
-27698.10
27712.10


103.44

-13841.43
27682.86
27696.87


122.57

-13920.39
27840.78
27850.77


107.13

-13858.41
27716.82
27828.81


-13921.08 -13857.86
27842.16 27715.72
27852.15 27727.72
ariance ^ grand mean-centered


LL = Log likelihood -2LL = -2 Log likelihood





Table 4-14 Models for MHC
1
Null


2
Index

72.32*
-0.16*






365.72






158.65

-14318.51
28637.02
28645.01


3
Index

72.56*
-0.22*






329.97
1.41





147.41

-14291.46
28582.92
28592.93


Fixed Effect
Intercept
Index
CD4+ cell^
Log VL
Lag CD4+ cell^
Log Lag VL
Random Effect#
Intercept
Index
CD4+ cell^
Log VL
Lag CD4+ cell^
Log Lag VL
Residual"
Fit Statistics
LL
-2LL
AIC
* p<0.05 v


71.40*







364.35






158.95

-14320.45
28640.90
28646.90


ariance


^ grand mean-centered


LL = Log likelihood


-2LL = -2 Log likelihood


AIC = Akaike Information Criterion










































AIC = Akaike Information Criterion


Table 4-14 Models for MHC (continued)
4 5 6 7 8 9 10 11
Index & Index & Index & Index & Index & Index & Index & Index &
CD4+ CD4+ CD4+ CD4+ Lag CD4+ Lag CD4+ Lag CD4+ Lag CD4+


Fixed Effect
Intercept
Index
CD4+ cell
Log VL
Lag CD4+ cell
Log Lag VL
Random Effect#
Intercept
Index
CD4+ cell
Log VL
Lag CD4+ cell
00 Log Lag VL
Residual"
Fit Statistics
LL
-2LL
AIC
p<0.05 v


72.58*
-0.21*
0.009





361.57






158.77


72.75*
-0.25*
0.007





327.56
1.40





147.61


72.66*
-0.19*
0.008


72.80*
-0.24*
0.006


72.73*
-0.23*


0.01*


361.42


72.89*
-0.27*


0.009


327.03
1.39


72.74*
-0.22*


0.01


356.89


349.69

0.002


319.56
1.37
0.001


0.001


155.78

-14314.00
28628.00
28640.00


146.49

-14289.78
28579.56
28593.56


158.70

-14316.31
28633.62
28642.62


147.59

-14289.98
28579.96
28591.96


157.21

-14315.46
28630.82
28642.92


-14317.07 -14290.65
28634.14 28581.30
28864.14 28593.30
ariance ^ grand-mean centered


LL = Log likelihood -2LL = -2 Log likelihood








































LL = Log likelihood -2LL = -2 Log likelihood


AIC = Akaike Information Criterion


ued)


Table 4-14 Models for MHC contain ~
12
Index &
Log VL


13
Index &
Log VL

74.86*
-0.25*

-0.82*




325.35
1.39





147.48


14
Index &
Log VL

74.96*
-0.20*

-0.90*


15
Index &
Log VL

75.06*
-0.26*


16
Index & Log
Lag VL

74.50*
-0.21*


17
Index & Log
Lag VL

74.40*
-0.26*




-0.62*

326.53
1.39


18
Index & Log
Lag VL

74.63*
-0.21*




-0.75*

335.14


19
Index & Log
Lag VL

74.61*
-0.26*




-0.66*

291.79
1.41


Fixed Effect
Intercept
Index
CD4+ cell^
Log VL
Lag CD4+ cell^
Log Lag VL
Random Effect#
Intercept
Index
CD4+ cell^
Log VL
Lag CD4+ cell^
Log Lag VL
$ Residual
Fit Statistics
LL
-2LL
AIC
p<0.05 v


74.83*
-0.20*

-0.89*




360.54






158.56


-0.86*


-0.73*

361.55


323.26


6.30


154.61

-14306.01
28612.02
28624.01


275.49
1.46


7.10


4.08
155.86

-14310.87
28621.74
28633.73


4.39
144.90

-14284.22
28568.44
28582.44


143.22

-14277.60
28555.20
28569.19


158.61

-14315.46
28630.92
28640.92


147.57

-14289.30
28578.60
28590.60


-14314.36 -14288.02
28628.72 28576.04
28638.71 28588.04
ariance ^ grand mean-centered











i~YO~~


2 68101
Inde
Fiue41Pyia elhCmoetSor fe A R ntaino 0ptet































Ine

Fgre etlHat opnn cr fe AR ntaino 0ptet

































2 4 6 8 10 12
Index


Figure 4-3 Average PHC score over time after HAART initiation





























Ine









Figure 4-4 Average MHC score over time after HAART initiation































2 01
Ine

Fiur 4- D+cl on vrtm ferHATiiito o0ptet












O






+3








2 68101


Ine


Fiue46AeaeC4+cl on vrtm sneH ATiiito



















O













2 4 6 8 10 12
Index


Figure 4-7 Log viral load over time after HAART initiation for 10 patients




























2 68101
Inde



Fiue48Aeaelg ia od vrtm atrH A Tiiito


























IO









-60 -40 -20 0 20 40 60
PHC Residuals



Figure 4-9 PHC residuals for model 2






























-6 4 2 02 06
MH eiul










Figure ~ ~ ~ MH 4-0 HCreidal fr odl









CHAPTER 5
DISCUSSION

In this chapter, discussion about PHC and MHC traj ectory is presented first, followed by a

discussion of the CD4+ cell count and viral load traj ectory, and then the relationship between

clinical lab test trajectory and QOL trajectory. This chapter also includes limitations,

recommendation for future research, and a summary.

QOL Trajectory

PHC Trajectory

Average PHC scores marginally decreased every 6 months as shown in Figure 4-3.

Although PHC statistically decreased over time (Table 4-13, model 2 and model 3), it decreased

approximately 0.2 for every 6 months. This result was similar to the study conducted by

Weinfurt and colleagues that found PHC declined over time (33). However in their study, PHC

weekly decreased by 0.09. In other words, it declined by 2.16 every 6 months.

MHC Trajectory

Average MHC scores also slightly decreased every 6 months (Figure 4-4). Although MHC

statistically decreased over time, it decreased only 0.16 or 0.22 for every 6 months as shown in

Table 4-14, model I and II respectively. This also implied that HAART can clinically maintain

mental health in HIV-infected patients. This result was similar to the finding from Weinfurt and

colleagues' study that showed MHC weekly declined by 0.09 or 2. 16 point every 6 months (33).

Clinical Lab Test Trajectory

CD4+ Cell Count Trajectory

On average, CD4+ cell count increased over time as expected, which is consistent with

many studies that showed that HAART helps to improve the immune system. This result was

consistent with other previously conducted studies (32, 36, 50-55). However, the finding from









Binquet and friends showed that CD4+ cell count increased from baseline to 12 months only

(50). After 12 months, CD4+ cell count was consistent. The result from Chu was slightly

different in that CD4+ cell count increased from baseline to 24 months (32). The result from this

study indicated the plateau period of CD4+ cell count at 48 months, which was similar to the

Ending from Moore and Garcia's study (51, 52).

Viral Load Trajectory

It is not surprising that the average viral load decreased over time after taking HAART

(Figure 4-9). Many studies show HAART suppresses viral load to an undetectable level (55, 56).

The Einding from Burgoyne indicated that average viral load in patients who received HAART

regimen decreased from 7,943 to 316 copies/mL in 4 years, whereas the result from Low-Beer

showed a reduction from 11,000 to 499 copies/mL in one year. Although 5 NRTIs and 3 PIs

were available during Low-Beer's study period, patients included in Low-Beer study at least

received PI. Therefore, whether these patients received HAART or not cannot be determined.

Relationship between CD4+ Cell Trajectory and QOL Trajectory

Change in CD4+ cell count can slightly predict change in PHC when time was controlled

as shown in Table 4-13 (model 4 to model 11). The underlying reason for this is because PHC

decreased little by little over time. When there is less variation of PHC, CD4+ cell count cannot

predict PHC well. However, the Einding from this study was consistent with the result from a

longitudinal clinical trial where there was a positive relationship between change in PHC and

change in CD4+ cell count (33). Comparing time-lag model with non time-lag model, the model

fit statistics showed that time-lag model with random coefficients were better than non time-lag

model with random coefficients.

The result from this study confirmed that CD4+ cell count had a positive relationship with

MHC as with several other cross-sectional and longitudinal studies (33, 57) that demonstrated a









similar relationship. Comparing time-lag and non time-lag models, only lag CD4+ cell count

significantly predicted MHC (Table 4-14, model 4 to model 10). However, the model fit

statistics of non time-lag model were better than those of time-lag model.

Relationship between Viral Load Trajectory and QOL Trajectory

An inverse relationship between viral load and PHC in this study was similar to the result

from Weinfurt' s study. Previous viral load also had an inverse relationship with PHC. In other

words, when viral load decreased, PHC increased. However, the magnitude of the association

between current viral load and previous viral load and PHC were different as shown in Table 4-

13. Based on model 12 to model 19 in Table 4-13, viral load should be fitted into the model with

random effect. This implied that the effect of viral load on PHC was different from patient to

patient. Viral load and previous viral load had an inverse relationship with MHC (Table 4-14,

model 12 to model 19). That means when viral load increased, MHC decreased. However, viral

load and previous viral load predicted MHC in different magnitude.

Limitations

This study has several limitations. First, it is possible that patients were not only informed

by physicians about their current CD4+ cell count and/or viral load, but also notified about the

evaluation of that lab test value (e.g., whether it is good or bad). Patients may consider their lab

test result as good/bad. Therefore, dichotomizing or categorizing CD4+ cell count and viral load

may possibly reflect the way patients used the lab test results to assess their QOL. Using a

different scale of measurement of the independent variables (e.g., continuous vs. binary variable)

may change the magnitude of the relationship between CD4+ cell count and viral load and QOL

because when there is less variation of the independent when it was dichotomized, compared

with when it was measured as a continuous variable. However, we do not know how physicians









evaluate CD4+ cell count and viral load and what they told about CD4+ cell count and viral load

to the patients. Qualitative research such as an interview may help clarify this.

Second, it is possible that patients may not only assess their own current QOL about the

questions in the MOS SF-36, but also use other information such as CD4+ cell count. If they use

CD4+ cell count to assess their current QOL and this information is not available, it is possible

that patients will use previous lab test result to assess their current QOL. Once more, qualitative

research by interviewing patients will help determine whether patients use lab test result from

last clinic visit to asses their current QOL or not.

Third, MACS is an observational study, collecting data from HIV-infected patients from 4

maj or cities, patients included in MACS visit clinic biannually only. Therefore, data from

MACS are available to prove whether lag time at 6 months between either CD4+ cell count or

viral load and QOL exists. It cannot be used to prove or validate whether lag time between

CD4+ cell count/viral load and QOL is 3 or 2 months. In other words, if the lag time between

CD4+ cell count and QOL existed and was about 3 months, the time-lag model of 3 months

would better predict QOL than a time-lag of 6 months or a non time-lag model. In other words,

QOL measured at an improper timing is considered as an invalid measure of the effect of CD4+

cell count.

Fourth, HIV-related symptoms (e.g., fatigue, pain and diarrhea) were not included in this

study. Those symptoms were reported as the significant predictors in QOL in HIV-infected

patients (29, 30, 58-61). For those studies, only the study by Johnson et al. included patients

treated with HAART and only two are longitudinal studies with an aim to investigate the

longitudinal relationship between HIV-related symptoms and QOL (29, 59). Including HIV-

infected symptoms in the model will possibly help explain change in QOL over time.









Fifth the effect of HIV-infected symptoms on QOL could change over time (33). It is

possible that patients get used to HIV-related symptoms over time, so the way patients assessed

their QOL regarding to symptoms may changed over time. For example, when asymptomatic

patients became symptomatic, symptoms such as diarrhea may considerably bother them.

However, when they were tolerated to diarrhea, the way they assess how diarrhea effect QOL

changed.

Sixth, this study included all patients who were on HAART. Naive and previous HAART

users were not separately analyzed. In the most recent study, patients included had no prior

HAART experiences (62). Both PHC and MHC increased over time which was contradict to the

finding from this study. However, the researchers found a positive relationship between CD4+

cell count and QOL as in this study.

This study compared time-lag and non time-lag models only in HIV-infected patients.

HIV is one of many chronic diseases where lab test results are important for determining

patients' condition and clinical status. Lab test results also help physician to determine when to

start treatment as stated in disease guidelines. Lab test results are also important to patients

because it help patients monitor their own condition. This study investigated the lag time

between clinical lab tests and QOL in HIV-infected patients who were on HAART only. The

results from this study can not be applied to those with other chronic diseases. Chronic diseases

that the lag time between clinical lab results and QOL might exist include diabetes,

hyperlipidemia. For these diseases, symptoms do not play an important role in assessing QOL.

Patients with high LDL can do regular activities (e.g., walk 1 to 2 miles). However, when they

do know that the LDL level was too high for two previous consecutive weeks, they might use










those lab tests results to assess their current QOL because they didn't know the current lab test

result.

Future Research

This study has two different implications for future research; (1) QOL measurement and

(2) clinical application. Researchers interested in QOL measurement should investigate the

timing of QOL measurement because patients may differ in their QOL assessments depending

upon whether they know their previous or current clinical lab test results. In other words,

researchers can compare QOL measured when patients had a clinic visit and QOL measured a

week later when patients knew their current lab test results. It is also recommended that other

databases be used to establish the lag time between CD4+ cell counts, viral load and QOL in

HIV-infected patients in order to confirm the finding from this study. For clinical application, if

the lag time relationship between either viral load or CD4+ cell count and QOL truly exists (e.g.,

3 months), it is recommended that patients be monitored their QOL at least every 3 months.

Conclusion

PHC and MHC in HIV-infected patients who were on HAART slightly decreased over

time. The change in viral load over time significantly predicts change in PHC and MHC over

time, whereas the change in CD4+ cell count significantly predicts PHC over time only. CD4+

cell count has a positive longitudinal relationship with PHC, whereas viral load has a negative

longitudinal relationship with both PHC and MHC. Overall, time-lag models were not different

from non time-lag models in terms of the model fit statistics and regression coefficients.









APPENDIX A
SQL SYNTAX FOR HAART

SELECT DISTINCT DRUGFl.CASEID, DRUGFl.VISIT, DRUGFl.AVQY,
Switch(C.CASEID,"Y"',DRUGFl.CASEID, "N") AS HAART
FROM DRUGF 1 LEFT JOIN [SELECT CASEID, VISIT
FROM


SELECT CASEID, VISIT, 1 AS FLAG
FROM

SELECT CASEID, VISIT
FROM (SELECT CASEID,VISIT
FROM DRUGF 1 INNER JOIN Drug_Grp_Code ON DRUGF l.DRGAV =
Drug_Grp_C ode.Drug_C ode
WHERE DRUG GRP = "NRTI"
GROUP BY CASEID, VISIT
HAVING COUNT(*) >=2
) AS A
WHERE EXISTS

SELECT 1
FROM DRUGF 1 INNER JOIN Drug_Grp_Code ON DRUGF l.DRGAV =
Drug_Grp_C ode.Drug_C ode
WHERE A.CASEID = DRUGFl1.CASEID AND A.VISIT = DRUGFl1.VISIT AND
(DRUGGRP = "NNRTI" OR DRUGGRP = "PI")


UNION

SELECT CASEID, VISIT
FROM

SELECT DISTINCT CASEID,VISIT
FROM DRUGF 1 INNER JOIN Drug_Grp_Code ON DRUGF l.DRGAV =
Drug_Grp_C ode.Drug_C ode
WHERE DRUG GRP = "NRTI"
UNION ALL
SELECT DISTINCT CASEID,VISIT
FROM DRUGF 1 INNER JOIN Drug_Grp_Code ON DRUGF l.DRGAV =
Drug_Grp_C ode.Drug_C ode
WHERE DRUG GRP = "NNRTI"
UNION ALL
SELECT DISTINCT CASEID,VISIT
FROM DRUGF 1 INNER JOIN Drug_Grp_Code ON DRUGF l.DRGAV =
Drug_Grp_C ode.Drug_C ode









WHERE DRUG GRP = "PI"
) AS A
GROUP BY CASEID, VISIT
HAVING COUNT(*) = 3

UNION

SELECT CASEID, VISIT
FROM (
SELECT CASEID,VISIT
FROM

SELECT DISTINCT CASEID,VISIT
FROM DRUGF 1 INNER JOIN Drug_Grp_Code ON DRUGF l.DRGAV =
Drug_GrpCode.Drug_C ode
WHERE DRUG CODE = 210
UNION ALL
SELECT DISTINCT CASEID,VISIT
FROM DRUGF 1 INNER JOIN Drug_Grp_Code ON DRUGF l.DRGAV =
Drug_Grp_Code.Drug_Code
WHERE DRUG CODE = 211
UNION ALL
SELECT DISTINCT CASEID,VISIT
FROM DRUGF 1 INNER JOIN Drug_Grp_Code ON DRUGF l.DRGAV =
Drug_GrpCode.Drug_C ode
WHERE DRUG GRP = "NRTI"
) AS B
GROUP BY CASEID,VISIT
HAVING COUNT(*) = 3
) AS A
WHERE NOT EXISTS

SELECT 1
FROM DRUGF 1 INNER JOIN Drug_Grp_Code ON DRUGF l.DRGAV =
Drug_GrpCode.Drug_C ode
WHERE A.CASEID = DRUGF l.CASEID AND A.VISIT = DRUGF l.VISIT AND
DRUG GRP = "NNRTI"


UNION SELECT CASEID, VISIT
FROM

SELECT CASEID,VISIT
FROM

SELECT DISTINCT CASEID,VISIT









FROM DRUGF 1
WHERE DRGAV
) AS A
WHERE EXISTS


218


SELECT 1
FROM DRUGF 1 INNER JOINT Drug_Grp_Code ON DRUGF l.DRGAV =
Drug_GrpCode.Drug_C ode
WHERE DRUG GRP = "NRTI" AND DRUGF l.CASEID = A.CASEID AND
DRUGFl1.VISIT = A.VISIT
GROUP BY DRUGFl1.CASEID,DRUGFl1.VISIT
HAVING COUNT(*) >= 4

) AS B
WHERE NOT EXISTS

SELECT 1
FROM DRUGF 1 INNER JOINT Drug_Grp_Code ON DRUGF l.DRGAV =
Drug_GrpCode.Drug_C ode
WHERE (DRUGGRP = "NNRTI" OR DRUGGRP = "PI") AND DRUGF l.CASEID
B.CASEID AND DRUGFl.VISIT = B.VISIT





GROUP BY CASEID,VISIT
HAVING COUNT(*) = 1 AND MAX(FLAG) = 1

]. AS C ON (DRUGFl.VISIT = C.VISIT) AND (DRUGFl.CASEID = C.CASEID);









LIST OF REFERENCES


1. Twenty-five years ofHIV/AIDS--United States, 1981-2006. Morb Mortal Wkly Rep
2006;55:585-589

2. Dybul M, Fauci AS, Bartlett JG, et al. Guidelines for using antiretroviral agents among
HIV-infected adults and adolescents. Ann Intern Med 2002;13 7:3 81-43 3

3. Sax PE, Gathe JC, Jr. Beyond efficacy: the impact of combination antiretroviral therapy
on quality of life. AIDS Patient Care STDS 2005;19:563-576

4. Wilson IB, Cleary PD. Linking clinical variables with health-related quality of life. A
conceptual model of patient outcomes. JAMA 1995;273:59-65

5. Jia H, Uphold CR, Wu S, et al. Predictors of changes in health-related quality of life
among men with HIV infection in the HAART era. AIDS Patient Care STDS
2005;19:395-405

6. Jia H, Uphold CR, Zheng Y, et al. A further investigation of health-related quality of life
over time among men with HIV infection in the HAART era. Qual Life Res 2007

7. Pneumocystis pneumonia--Los Angeles. 1981. MMWR Morb Mortal Wkly Rep
1996;45:729-733

8. Pneumocystis pneumonia--Los Angeles. MMWR Morb Mortal Wkly Rep 1981;30:250-
252

9. Estimates of HIV prevalence and proj ected AID S cases: summary of a workshop,
October 31i-November 1, 1989. MMWR Morb Mortal Wkly Rep 1990;39: 110-112, 117-
119

10. McNicholl I. Adverse Events of Antiretroviral Drugs/July 2006. 2006. Available at:
http://hivinsite.ucsf. edu/pdf/ar-05-01l.pdf. Accessed 01/25, 2007

11. Shibuyama S, Gevorkyan A, Yoo U, et al. Understanding and avoiding antiretroviral
adverse events. Curr Pharm Des 2006; 12: 1075-1090

12. Ferrans CE, Zerwic JJ, Wilbur JE, et al. Conceptual model of health-related quality of
life. J Nurs Scholarsh 2005;37:336-342

13. Sousa KH, Kwok OM. Putting Wilson and Cleary to the test: analysis of a HRQOL
conceptual model using structural equation modeling. Qual Life Res 2006; 15:725-737

14. Chrischilles EA, Rubenstein LM, Voelker MD, et al. Linking clinical variables to health-
related quality of life in Parkinson's disease. Parkinsonism Relat Disord 2002;8:199-209









15. Sousa KH, Holzemer WL, Henry SB, et al. Dimensions of health-related quality of life in
persons living with HIV disease. J Adv Nurs 1999;29: 178-187

16. Wilson IB, Cleary PD. Clinical predictors of functioning in persons with acquired
immunodeficiency syndrome. Med Care 1996;34:610-623

17. Wettergren L, Bj orkholm M, Axdorph U, et al. Determinants of health-related quality of
life in long-term survivors of Hodgkin's lymphoma. Qual Life Res 2004; 13: 1369-1379

18. Vidrine DJ, Amick BC, 3rd, Gritz ER, et al. Assessing a conceptual framework of health-
related quality of life in a HIV/AIDS population. Qual Life Res 2005;14:923-933

19. Bing EG, Hays RD, Jacobson LP, et al. Health-related quality of life among people with
HIV disease: results from the Multicenter AIDS Cohort Study. Qual Life Res 2000;9:55-


20. Call SA, Klapow JC, Stewart KE, et al. Health-related quality of life and virologic
outcomes in an HIV clinic. Qual Life Res 2000;9:977-985

21. Carrieri P, Spire B, Duran S, et al. Health-related quality of life after 1 year of highly
active antiretroviral therapy. J Acquir Immune Defic Syndr 2003;32:3 8-47

22. Casado A, Badia X, Consiglio E, et al. Health-related quality of life in HIV-infected
naive patients treated with nelfinavir or nevirapine associated with ZDV/3TC (the
COMBINE-QoL substudy). HIV Clin Trials 2004;5:132-139

23. Nieuwkerk PT, Gisolf EH, Colebunders R, et al. Quality of life in asymptomatic- and
symptomatic HIV infected patients in a trial of ritonavir/saquinavir therapy. The
Prometheus Study Group. AIDS 2000; 14: 181-187

24. Liu C, Ostrow D, Detels R, et al. Impacts of HIV infection and HAART use on quality of
life. Qual Life Res 2006; 15:941-949

25. Brechtl JR, Breitbart W, Galietta M, et al. The use of highly active antiretroviral therapy
(HAART) in patients with advanced HIV infection: impact on medical, palliative care,
and quality of life outcomes. J Pain Symptom Manage 2001;21:41-51

26. Gill CJ, Griffith JL, Jacobson D, et al. Relationship of HIV viral loads, CD4 counts, and
HAART use to health-related quality of life. J Acquir Immune Defic Syndr 2002;30:485-
492

27. Globe DR, Hays RD, Cunningham WE. Associations of clinical parameters with health-
related quality of life in hospitalized persons with HIV disease. AIDS Care 1999; 11:71-
86









28. Eriksson LE, Nordstrom G, Berglund T, et al. The health-related quality of life in a
Swedish sample of HIV-infected persons. J Adv Nurs 2000;32: 1213-1223

29. Preau M, Leport C, Salmon-Ceron D, et al. Health-related quality of life and patient-
provider relationships in HIV-infected patients during the first three years after starting
PI-containing antiretroviral treatment. AIDS Care 2004; 16:649-661

30. Justice AC, Chang CH, Rabeneck L, et al. Clinical importance of provider-reported HIV
symptoms compared with patient-report. Med Care 2001;39:397-408

31. Clarke PS. Causal analysis of individual change using the difference score. Epidemiology
2004;15:414-421

32. Chu H, Gange SJ, Yamashita TE, et al. Individual variation in CD4 cell count trajectory
among human immunodeficiency virus-infected men and women on long-term highly
active antiretroviral therapy: an application using a Bayesian random change-point
model. Am J Epidemiol 2005;162:787-797

33. Weinfurt KP, Willke RJ, Glick HA, et al. Relationship between CD4 count, viral burden,
and quality of life over time in HIV-1-infected patients. Med Care 2000;38:404-410

34. Liu C, Johnson L, Ostrow D, et al. Predictors for Lower Quality of Life in the HAART
Era Among HIV-Infected Men. J Acquir Immune Defic Syndr 2006

35. Protopopescu C, Marcellin F, Spire B, et al. Health-related quality of life in HIV-1-
infected patients on HAART: a five-years longitudinal analysis accounting for dropout in
the APROCO-COPILOTE cohort (ANRS CO-8). Qual Life Res 2007; 16:577-591

36. Anastos K, Barron Y, Cohen MH, et al. The prognostic importance of changes in CD4+
cell count and HIV-1 RNA level in women after initiating highly active antiretroviral
therapy. Ann Intern Med 2004; 140:256-264

37. Raghunathan TE. What do we do with missing data? Some options for analysis of
incomplete data. Annual review of public health 2004;25:99-1 17

38. Schafer JL, Graham JW. Missing data: our view of the state of the art. Psychological
methods 2002;7: 147-177

39. Hogan JW, Roy J, Korkontzelou C. Handling drop-out in longitudinal studies. Stat Med
2004;23:1455-1497

40. Twisk J, de Vente W. Attrition in longitudinal studies. How to deal with missing data. J
Clin Epidemiol 2002;55:329-337

41. Kristman VL, Manno M, Cote P. Methods to account for attrition in longitudinal data: do
they work? A simulation study. Eur J Epidemiol 2005;20:657-662











42. Ironson G, O'Cleirigh C, Fletcher MA, et al. Psychosocial factors predict CD4 and viral
load change in men and women with human immunodeficiency virus in the era of highly
active antiretroviral treatment. Psychosomatic medicine 2005;67:1013-1021

43. Wensheng G, Sarah JR, Thomas TTH. A Random Pattern-Mixture Model for
Longitudinal Data With Dropouts. Journal of the American Statistical Association
2004;99:929

44. Kenward MG, Molenberghs G. Parametric models for incomplete continuous and
categorical longitudinal data. Statistical methods in medical research 1999;8:51-83

45. Crystal S, Sambamoorthi U. Functional impairment traj ectories among persons with HIV
disease: a hierarchical linear models approach. Health Serv Res 1996;31:469-488

46. Wu YW, Clopper RR, Wooldridge PJ. A comparison of traditional approaches to
hierarchical linear modeling when analyzing longitudinal data. Res Nurs Health
1999;22:421-432

47. Nich C, Carroll K. Now you see it, now you don't: a comparison of traditional versus
random-effects regression models in the analysis of longitudinal follow-up data from a
clinical trial. J Consult Clin Psychol 1997;65:252-261

48. Wu YW. An application of hierarchical linear models to longitudinal studies. Res Nurs
Health 1996; 19:75-82

49. O'Connell AA, McCoach DB. Applications of hierarchical linear models for evaluations
of health interventions: demystifying the methods and interpretations of multilevel
models. Eval Health Prof 2004;27: 119-151

50. Binquet C, Chene G, Jacqmin-Gadda H, et al. Modeling changes in CD4-positive T-
lymphocyte counts after the start of highly active antiretroviral therapy and the relation
with risk of opportunistic infections: the Aquitaine Cohort, 1996-1997. Am J Epidemiol
2001;153:386-393

51. Garcia F, de Lazzari E, Plana M, et al. Long-term CD4+ T-cell response to highly active
antiretroviral therapy according to baseline CD4+ T-cell count. J Acquir Immune Defic
Syndr 2004;36:702-713

52. Moore RD, Keruly JC, Gebo KA, et al. An improvement in virologic response to highly
active antiretroviral therapy in clinical practice from 1996 through 2002. J Acquir
Immune Defic Syndr 2005;39: 195-198

53. Moore RD, Keruly JC. CD4+ cell count 6 years after commencement of highly active
antiretroviral therapy in persons with sustained virologic suppression. Clin Infect Dis
2007;44:441-446











54. Viard JP, Burgard M, Hubert JB, et al. Impact of 5 years of maximally successful highly
active antiretroviral therapy on CD4 cell count and HIV-1 DNA level. AIDS 2004; 18:45-
49

55. Low-Beer S, Chan K, Wood E, et al. Health related quality of life among persons with
HIV after the use of protease inhibitors. Qual Life Res 2000;9:941-949

56. Burgoyne RW, Rourke SB, Behrens DM, et al. Long-term quality-of-life outcomes
among adults living with HIV in the HAART era: the interplay of changes in clinical
factors and symptom profile. AIDS Behav 2004;8:151-163

57. Ruiz Perez I, Rodriguez Bano J, Lopez Ruz MA, et al. Health-related quality of life of
patients with HIV: impact of sociodemographic, clinical and psychosocial factors. Qual
Life Res 2005;14:1301-1310

58. Johnson MO, Stallworth T, Neilands TB. The drugs or the disease? Causal attributions of
symptoms held by HIV-positive adults on HAART. AIDS Behav 2003;7: 109-117

59. Lorenz KA, Cunningham WE, Spritzer KL, et al. Changes in symptoms and health-
related quality of life in a nationally representative sample of adults in treatment for HIV.
Qual Life Res 2006; 15:951-958

60. Lorenz KA, Shapiro MF, Asch SM, et al. Associations of symptoms and health-related
quality of life: findings from a national study of persons with HIV infection. Ann Intern
Med 2001; 13 4:8 54-8 60

61. Tramarin A, Parise N, Campostrini S, et al. Association between diarrhea and quality of
life in HIV-infected patients receiving highly active antiretroviral therapy. Qual Life Res
2004;13:243-250

62. Stangl AL, Wamai N, Mermin J, et al. Trends and predictors of quality of life among
HIV-infected adults taking highly active antiretroviral therapy in rural Uganda. AIDS
Care 2007; 19:626-636









BIOGRAPHICAL SKETCH

Sawaeng Watcharathanakij is an assistant professor at Faculty of Pharmacy, Ubon

Ratchathani Universityn (UBU), Thailand. He was born and raised in Chiangmai province. He

earned Bachelor of Pharmacy from Mahidol Unversity (MU). He worked as a hospital

pharmacist at Ramathibodi Hospital. Then he pursued graduate study and earned Master of

Pharmacy from Mahidol University and worked as faculty at Faculty of Pharmacy, Ubon

Ratchathani University before j oining the doctoral program at the Department of Pharmacy

Health Care Administration, University of Florida in 2002.





PAGE 1

1 TIME LAG MODEL FOR QUALITY OF LI FE ASSESSMENT IN HIV-INFECTED PATIENTS WITH HIGHLY ACTI VE ANTIRETROVIRAL THERAPY By SAWAENG WATCHARATHANAKIJ 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 2007

PAGE 2

2 2007 Sawaeng Watcharathanakij

PAGE 3

3 To my beloved parents and family

PAGE 4

4 ACKNOWLEDGMENTS First, I sincerely thank my di ssertation chair, Professor Rich ard Segal, for his invaluable guidance and support through out dissertation period. I also thank Professor Carole Kimberlin, co-chair, and Professor L. Douglas Ried for their recommendation and encouragement. In addition, I extend special thanks to my external committee member, Professor Michael Daniels, for his invaluable advice on the data analysis. Fi nally, I express thanks to all my friends for their friendship.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES................................................................................................................. ..........8 LIST OF FIGURES................................................................................................................ .......10 ABSTRACT....................................................................................................................... ............11 CHAPTERS 1 INTRODUCTION..................................................................................................................13 Problem Statement.............................................................................................................. ....13 Background..................................................................................................................... ........14 HIV Infection and Treatments.........................................................................................14 Conceptual Model for Quality of Life Assessment.........................................................15 The Time Reference in QOL Measurement....................................................................15 Methodological Limitations of Prev ious Research in HAART and QOL......................16 Research Questions and Hypotheses......................................................................................17 Research Question 1........................................................................................................17 Research Hypothesis 1....................................................................................................17 Research Question 2........................................................................................................17 Research Hypothesis 2....................................................................................................18 Research Question 3........................................................................................................18 Research Hypothesis 3....................................................................................................18 Research Question 4........................................................................................................18 Research Hypothesis 4....................................................................................................18 Research Question 5........................................................................................................18 Research Hypothesis 5....................................................................................................19 Research Question 6........................................................................................................19 Research Hypothesis 6....................................................................................................19 Research Question 7........................................................................................................19 Research Hypothesis 7....................................................................................................19 Research Question 8........................................................................................................19 Research Hypothesis 8....................................................................................................20 Significance of Research....................................................................................................... .20 2 LITERATURE REVIEW.......................................................................................................22 Epidemiology of HIV/AIDS...................................................................................................22 Classification of HIV Infection..............................................................................................22 Antiretroviral Agents.......................................................................................................... ....23 Conceptual Model for Quality of Life Assessment................................................................23 Biological Function.........................................................................................................24

PAGE 6

6 Symptoms....................................................................................................................... .24 Functional Status.............................................................................................................24 General Health Perception...............................................................................................25 Overall Quality of Life....................................................................................................25 Time Reference in MOS SF-36 Instrument for Current QOL................................................25 Validity of Quality of Life Assessment Model.......................................................................26 CD4+ Cell counts, Viral load, a nd QOL in HIV-infected Persons........................................27 3 METHODS........................................................................................................................ .....37 Conceptual Framework...........................................................................................................37 Data........................................................................................................................... ..............38 HAART Definition..........................................................................................................38 Health-Related Quality of Life........................................................................................39 CD4+ Cell Count.............................................................................................................39 Viral Load..................................................................................................................... ...40 Inclusion and Exclusion Criteria............................................................................................40 Data Analysis Process.......................................................................................................... ...40 Data Preparation..............................................................................................................40 Missing data and dropping out in the longitudinal study.........................................41 How to handle dropouts in longitudinal study.........................................................41 Data Analysis.................................................................................................................. .42 IRB Approval................................................................................................................... .......44 4 RESULTS........................................................................................................................ .......49 Merging MACS Database.......................................................................................................49 Drug Table..................................................................................................................... ..49 QOL Table...................................................................................................................... .49 Lab Test Table.................................................................................................................49 Missing Data................................................................................................................... ........50 Exploratory Analysis........................................................................................................... ...50 Demographic Data...........................................................................................................50 Quality of Life Trajectory...............................................................................................50 CD4+ Cell Count and Viral Load Trajectory..................................................................51 Correlation among PHC and MHC.................................................................................51 Time-lag and Non Time-lag Models......................................................................................52 Time-lag and Non Time-lag Models for PHC.................................................................52 Time-lag and Non Time-lag Models for MHC...............................................................55

PAGE 7

7 5 DISCUSSION..................................................................................................................... ....80 QOL Trajectory................................................................................................................. .....80 PHC Trajectory................................................................................................................80 MHC Trajectory..............................................................................................................80 Clinical Lab Test Trajectory...................................................................................................80 CD4+ Cell Count Trajectory...........................................................................................80 Viral Load Trajectory......................................................................................................81 Relationship between CD4+ Cell Trajectory and QOL Trajectory........................................81 Relationship between Viral Load Trajectory and QOL Trajectory........................................82 Limitations.................................................................................................................... ..........82 Future Research................................................................................................................ ......85 Conclusion..................................................................................................................... .........85 APPENDIX A SQL SYNTAX FOR HAART................................................................................................86 LIST OF REFERENCES............................................................................................................. ..89 BIOGRAPHICAL SKETCH.........................................................................................................94

PAGE 8

8 LIST OF TABLES Table page 2-1 The estimated numbers and percentage of HIV/AIDS and AIDS cases by diagnosis, age, gender and race in 1981-2004............................................................................................32 2-2 CDC HIV Infection Categor ies by Clinical Conditions..........................................................33 2-3 CDC Classification Syst em for HIV Infection........................................................................34 3-1 Missing Data Mechanisms of CD4+ Cell count......................................................................45 3-2 Advantages and disadvantages of selection and pattern-mixture models...............................46 3-2 Independent Structure of QOL Measurements........................................................................48 3-3 Exchangeable Structure of QOL Measurements.....................................................................48 3-4 m-dependent Structur e of QOL Measurement........................................................................48 3-5 Autoregressive Correlation Structure of QOL Measurements................................................48 4-1 Time patients started HAART................................................................................................ .59 4-2 Average age at HAART initiation...........................................................................................59 4-3 Age of patients at HAART initiation.......................................................................................59 4-4 Average follow up time after HAART initiation.....................................................................59 4-5 The number of patients followed up until last clinic visit after HAART initiation.................60 4-6 Education level of patients who received HAART.................................................................60 4 -7 Average physical health comp onent score by time since HAART........................................61 4-8 Average mental health com ponent score by time since HAART............................................61 4-9 Average CD4+ cell count at each visit....................................................................................62 4-10 Average log viral lo ad at each visit...................................................................................... .62 4-11 Correlation coefficients among PH C measured at different time..........................................63 4-12 Correlation coefficients among MH C measured at different time........................................63 4-13 Models for PHC............................................................................................................ .........64

PAGE 9

9 4-14 Models for MHC............................................................................................................ .......67

PAGE 10

10 LIST OF FIGURES Figure page 1-1 Wilson and Clearys Conceptual Model for QOL Assessment...............................................21 2-1 Ferranss Conceptual Model for QOL Assessment.................................................................35 2-2 Modified Version of Wilsons Model.....................................................................................36 3-1 Time-lag Conceptual Model for Quality of Life Assessment.................................................46 3-2 Conceptual Diagram for Traditional Model............................................................................47 3-3 Conceptual Diagram for Time-lag Model...............................................................................47 4-1 Physical Health Component Scor e after HAART Initiatio n of 10 patients.............................70 4-2 Mental Health Component Score after HAART Initiation of 10 patients...............................71 4-3 Average PHC score over time after HAART initiation...........................................................72 4-4 Average MHC score over ti me after HAART initiation.........................................................73 4-5 CD4+ cell count over time after HAART initiation for 10 patients........................................74 4-6 Average CD4+ cell count ove r time since HAART initiation.................................................75 4-7 Log viral load over time after HAART initiation for 10 patients............................................76 4-8 Average log viral loads over time after HAART initiation.....................................................77 4-9 PHC residuals for model 2.................................................................................................. .....78 4-10 MHC residuals for model 2................................................................................................. ..79

PAGE 11

11 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 TIME LAG MODEL FOR QUALITY OF LIFE ASSESSMENT IN HIV-INFECTED PATIENTS WITH HIGHLY ACTIVE ANTIRETROVIRAL THERAPY By Sawaeng Watcharathanakij December 2007 Chair: Professor Richard Segal Major Department: Pharmacy Health Care Administration Quality of life (QOL) assessment plays a pivotal role in determining the longitudinal effect of highly active antiretroviral th erapy (HAART) in HIV-disease. This research addresses how the timing of clinical lab tests, such as CD4+ c ount and viral load, affect patients self-reported quality of life. This objectives of this study are to: (1) verify whether a lag ti me between clinical lab test results (CD4+ cell count and viral load) and QOL exists by comparing tw o statistical models, time-lag and non-time lag model, and (2) determ ine how well change in CD4+ cell count and change in viral load over time can predict cha nge in QOL over time. Subjects treated a drug regimen called HAART were select ed from a secondary database called the Multicenter AIDS Cohort study (MACS). The MACS is a prospective observational c ohort study of the natural and treated histories of HIV-1 infecti on in homosexual and bisexual men. Data were analyzed with both time lag and non time-lag random coefficient models because each patient had a unique CD4+ cell count, viral load and QOL trajectory. The effect of CD4+ cell count and viral load on two dimensi ons of QOL physical health component (PHC)

PAGE 12

12 and mental health component (MHC) was ex amined by comparing the time lag and non timelag random coefficient models with the model fit statis tics, Akaike information criterion (AIC). PHC and MHC in HIV-infected patients w ho were on HAART slightly decreased over time. The change in viral load over time signi ficantly predicts change in PHC and MHC over time, whereas the change in CD4+ cell count sign ificantly predicts PHC over time only. CD4+ cell count has a positive longitudi nal relationship with PHC, wher eas viral load has a negative longitudinal relationship with bot h PHC and MHC. Overall, timelag models were not different from non time-lag models in terms of the mode l fit statistics and re gression coefficients.

PAGE 13

13 CHAPTER 1 INTRODUCTION Problem Statement Quality of life (QOL) is an outcome that is generally viewed as important for measuring the impact of a health care intervention because it takes patients perspectives into account. While examining the effect of interventions on QOL for patients who are treated for acute conditions is important, measuring QOL is especial ly of interest for pa tients who have chronic medical conditions since other he alth outcomes may be influenced by their perceptions of their functioning and well-being. This research ex amines one chronic medical condition, Human Immunodeficiency Virus (HIV) infection, for wh ich measuring QOL may be of particular importance in understanding patients decisions about their HIV care. Understanding the relationship between a pati ents perception of their QOL and the decisions they make when managing their condi tion is not straight-f orward. Although many studies support a negative rela tionship between HIV symptoms and QOL, taking anti-retroviral medications often leads to side effects, such as nausea, pain, and anemia, which can also lessen QOL. Consequently, patients may discontinue taki ng antiretroviral drugs in order to avoid those side effects, which may have the effect of lo wering the effectiveness of the treatment strategy and, thereby, increasing their HIV symptoms. Furt her complicating the prediction of QOL is the availability of information from clinical lab tests for HIV disease including CD4+ cell count and viral load. At the time of a clinic visit, the findings from clinical lab tests are usually unavailable since it takes at least one week to process the blood samp les. Therefore, measures of QOL taken at the time of a clinic visit is likely based on results from earlier lab tests along with current symptoms, and current anti-retroviral si de effects. This study intends to examine

PAGE 14

14 whether the lag time relationship between clinical lab tests and cu rrent QOL exists by comparing two different models, time-lag model and non time-lag model. Background HIV Infection and Treatments Human Immunodeficiency Virus (HIV) infection is one of the major health problems in United States and worldwide. The Centers fo r Disease Control and Prevention (CDC) reported on a study conducted by Glynn and his colleagues th at over one million persons were infected with HIV infection in United States in 2006 and approximately 40,000 new cases were diagnosed in 2006 (1). Many antiretroviral agents are available in the market and treatment guidelines are also available for the clinicians to treat HIV-infected persons. The treatment goals of HIV infection are to reach maximal and durable suppression of viral load, restoration and preservation of immunologic function, improvement of quality of life, and reduction of HIV-related morbidity and mortality (2). The most updated treatm ent guidelines recommend some combination of anti-retroviral drugs, called Highly Active Anti -retroviral Treatment (HAART), to treat HIVinfected patients in order to achieve the treatment goals mentioned above. Managing symptoms in HIV-infected patients is very complicated because patients in different stages of the disease may respond to tr eatment in different ways. For example, for symptomatic patients, the use of antiretroviral agents may adequately control symptoms for a patient who has HIV-infection, which leads to an increase in their QOL. However, because antiretroviral agents may lead to side effects, QOL may be affected negatively (3). In contrast, asymptomatic patients may behave differently fr om symptomatic patients in terms of how they manage their disease because side effects of anti-retroviral drugs may be of greater concern than symptoms from HIV-infection. Consequently, these patients may stop taking anti-retroviral

PAGE 15

15 medications, which can lead to faster disease prog ression and shorter life e xpectancy. Therefore, the focus on maximizing QOL in HIV-infected pe rsons is not only important immediately after treatment initiation, but also for the long term. In other words, it is essential for healthcare provider to maximize patients QOL a nd maintain maximum QOL over time. Conceptual Model for Quality of Life Assessment The conceptual model used in this research to assess QOL was first proposed in 1995 by Wilson and Cleary (4). It describes the relations hip among five levels of health outcomes: (1) biological and physiological factors, (2) sympto m status, (3) functioni ng status, (4) general health perception and (5) overall quality of life, as well as characteristics of the individual and environment as in Figure 1-1. Based upon this conceptual model, QOL may be affected by a patients interpretation of their response to tr eatment based on CD4+ ce ll count and viral load, symptoms from HIV disease such as diarrhea lasting greater than one month, and medicationrelated side effects. The Time Reference in QOL Measurement A validated and widely used quality of lif e instrument, the Medical Outcome Study Short Form 36 (MOS SF-36), asks respondents to assess their QOL in the last four weeks e.g. during the past four weeks, have you had any of the fo llowing problems with your work or other regular daily activities as a result of your physical h ealth. Based upon this time reference and the conceptual model for QOL assessment, the MOS SF-36 places a time boundary around the information that should be used by the patient as he or she responds to each question that is, information available to patients during the four week period prior to completing the survey, including their HIV symptoms, experiences with drug side effects, and information about viral load and CD4+ cell count. However, one po ssible limitation with us ing the four-week time frame may be that biological fact ors, such as viral load result s and CD4+ cell count, may have

PAGE 16

16 been last measured more than four weeks ago. Consequently, it is possibl e that patients assess their current QOL based on using CD4+ cell count and viral load from th ree or more months earlier. For example, patients with severe symp toms may assess their current QOL higher than it should be because they know that CD4+ cell count and viral load from thei r last clinic visit indicated they were well contro lled. However, if they knew that their current CD4+ cell count and viral load were not in control, they may assess their current QOL lower than the situation above. In other words, patients will assess their current QOL based upon the findings from clinical lab test result that are available to them regardless of when the tests were administered. Methodological Limitations of Prev ious Research in HAART and QOL Although numerous studies focused on either the e ffect of symptoms such as diarrhea or the effect of anti-retroviral dr ugs on QOL in HIV-infected pati ents, only a few explored the relationship between HAART, HIV-related symptoms side-effect of anti-retroviral medication, clinical lab tests, and QOL. Am ong those studies, most were cr ossectional, which suffered from limitations that may be overcome with the use of a longitudinal design. Some of the studies which used a longitudinal design examined the relationship between HAART, HIV-related symptoms, side-effect of anti-retroviral medicatio n, clinical lab tests a nd QOL, but the authors defined the meaning of longitudina l relationship differently than wh at we proposed in this study. For example, in a study conducted at a VA hosp ital, the authors described their study as a longitudinal study because the patients were follo wed up to 12 months after enrollment (5). QOL, the outcome in this study, was measured at 2 different points in time, baseline and at one year. In addition, the author s used baseline QOL and CD4 ce ll count, depression and other factors to predict QOL at one year. Although another study conducted by the same researchers had fewer limitations because they measured pr edictors such as CD4+ cell counts, coping, and

PAGE 17

17 comorbidity at baseline and one year (6), it was different from the methodological design proposed in the present study. Based upon the Wilson and Cleary conceptual model to measure QOL, time reference in QOL assessment, and to maximize QOL in HIVinfected patients with HAART, the present study proposes to investigate whether the lag time relationship between previous clinical lab tests, and current QOL in patients with HAART exists by comparing two models Research Questions and Hypotheses The purpose of this study is to determine wh ether patients assessment of their current QOL over time (trend of QOL) may be better predicted from clinical lab test results over time, measured from previous clinic visit, compared to only using clinical lab te st results performed at the same clinical visit where QOL is measured. Since it is possible that the lag-time relationship may differ for predicting different domains in the MOS SF-36 QOL scale, research questions are offered for two major domains associated w ith the SF-36, physical a nd mental subscales. Research Question 1 How well does the change in current CD4+ cell count over time predict change in current physical health in HIV-infected patients over time? Research Hypothesis 1 The null hypothesis is that ch ange in current CD4+ cell c ount over time cannot predict change in current physical health in HIV-infected patients over time whereas the alternative hypothesis is that change in current CD4+ cell count over time can pred ict change in current physical health in HIV-infected patients over time. Research Question 2 How well does the change in pr evious CD4+ cell c ount over time predict change in current physical health in HIV-infected patients over time?

PAGE 18

18 Research Hypothesis 2 The null hypothesis is that change in prev ious CD4+ cell count over time cannot predict change in current physical health in HIV-infected patients over time whereas the alternative hypothesis is that change in previous CD4+ cell count over time can pred ict change in current physical health in HIV-infected patients over time. Research Question 3 How well does change in current viral loads over time predict change in current physical health in HIV-infected patients over time? Research Hypothesis 3 The null hypothesis is that cha nge in current viral load over time cannot predict change in current physical health in HIV-infected patients over time whereas the al ternative hypothesis is that change in current viral load over time can pr edict change in current physical health in HIVinfected patients over time. Research Question 4 How well does change in previous viral loads ov er time predict change in current physical health in HIV-infected patients over time? Research Hypothesis 4 The null hypothesis is that change in previous viral load over time cannot predict change in current physical health in HIV-infected patients over time whereas the al ternative hypothesis is that change in previous viral load over time can pr edict change in current physical health in HIVinfected patients over time. Research Question 5 How well does the change in current CD4+ cell count over time predict change in current mental health in HIV-infected patients over time?

PAGE 19

19 Research Hypothesis 5 The null hypothesis is that ch ange in current CD4+ cell c ount over time cannot predict change in current mental health in HIV-inf ected patients over time wh ereas the alternative hypothesis is that change in current CD4+ cell count over time can pred ict change in current mental health in HIV-infected patients over time. Research Question 6 How well does the change in pr evious CD4+ cell c ount over time predict change in current mental health in HIV-infected patients over time? Research Hypothesis 6 The null hypothesis is that change in prev ious CD4+ cell count over time cannot predict change in current mental health in HIV-inf ected patients over time wh ereas the alternative hypothesis is that change in previous CD4+ cell count over time can pred ict change in current mental health in HIV-infected patients over time. Research Question 7 How well does change in current viral loads over time predict change in current mental health in HIV-infected patients over time? Research Hypothesis 7 The null hypothesis is that cha nge in current viral load over time cannot predict change in current mental health in HIV-in fected patients over time whereas the alternative hypothesis is that change in current viral load over time can predict change in current mental health in HIVinfected patients over time. Research Question 8 How well does change in previous viral loads ov er time predict change in current mental health in HIV-infected patients over time?

PAGE 20

20 Research Hypothesis 8 The null hypothesis is that change in previous viral load over time cannot predict change in current mental health in HIV-in fected patients over time whereas the alternative hypothesis is that change in previous viral load over time can predict change in curren t mental health in HIVinfected patients over time. Significance of Research Understanding how patients assess their QOL in chronic diseases such as HIV disease is very important because patients have to take me dication continuously. In addition, clinical lab tests are performed to help clin icians diagnose patients condition and decide whether to treat or prescribe medication to the patients. The rese arch will provide two significant contributions, clinical care and methodologi cal contribution. For clinical care, th is research will help clinicians profoundly understand how patients assess their current QOL and what factors that patients take into account when they assess their QOL. Sp ecifically, it aims to unde rstand whether patients will consider all the information currently availa ble at hand to assess their current QOL or they will consider a combination of past and current information to assess their current QOL (e.g., use current symptoms and past clinical lab test). For methodological contributi on, this study will help res earcher design study about the timing to measure patients QOL that really cap ture their current QOL. For example, if the patients assess their current QOL regard to symp toms and previous clin ical lab test result because it takes a few days or a week to know the cu rrent lab test result from current clinic visit, researcher should postpone collecting QOL data until patients receive their cu rrent lab test result. Moreover, if lag-time between clin ical lab test and QOL really exists (e.g., 4 months), patients should be monitored for their QOL at least every 4 months, but no longer than 4 months.

PAGE 21

21 Figure 1-1 Wilson and Clearys Con ceptual Model for QOL Assessment Biological and Physiological Variables Symptom Status Functional Status General Health Perceptions Overall Quality of Life Characteristics of the Individual Characteristics of the Environment Symptom Amplification Personality Motivation Values Preferences Psychological Supports Social and Economic Supports Social and Psychological Supports Nonmedical Factors

PAGE 22

22 CHAPTER 2 LITERATURE REVIEW In this chapter, literature related to the epidemiology of HIV/AIDS and its classification and treatments are reviewed. In addition, this review offers (1) a conceptual model for QOL assessment, (2) an analysis of instruments us ed for measuring QOL, with an emphasis on the time frame used when measuring QOL, and (3) findings about the rela tionship between CD4+ cell count, viral load and QOL. Epidemiology of HIV/AIDS The first five cases of Pneumocystis carinii pneumonia, soon after called Acquired Immunodeficiency Syndromes (AIDS), were repo rted in 1981 (7, 8). By 1989, approximately one million people in United States were infected with HIV (9). The peak of AIDS in the United States was reached in 1992, and, fortunately, th e number of AIDS cases decreased 47% from 1992 to 1998 (1). The numbers of new AIDS cas es continued to decrease during 1996 to 2004. During 1981-1995, most AIDS cases were white. Since 1996, however, most AIDS cases were African/nonhispanic. Table 2-1 shows the estima ted numbers and percentage of HIV/AIDS and AIDS cases by diagnosis, age, gender a nd other characteristics from 1981-2004. Classification of HIV Infection Two major HIV classification systems are currently available for clinicians to diagnose and classify HIV-infected patients, the U.S. Ce nters for Disease Control and Prevention (CDC) classification system and the World Health Or ganization (WHO) Clinical Staging and Disease Classification System. The WHO system cl assifies HIV disease based upon clinical manifestations that can be identified and tr eated by clinicians in practice settings where laboratory tests are not availabl e whereas CDC classification syst em uses both CD4+ cell count and HIV-related conditions.

PAGE 23

23 According to the 1993 Revised Classifica tion System HIV infection and Expanded Surveillance Case Definition for AIDS among Adol escents and Adults, patie nts can be classified into nine categories by clinical conditions (Table 2-2) and CD4+ cell count as shown in Table 2-3. Antiretroviral Agents Many antiretroviral drugs are available in the ma rket. Generally, there are four classes of anti-retroviral agents; (1 ) nucleoside/nucleotide reverse transc riptase inhibitors (NRTIs), (2) nonnucleoside reverse transcriptase i nhibitors (NNRTIs), (3) protease inhibitors (PIs) and (4) fusion inhibitors (10, 11). Nucleo side/nucleotide analogue in cludes Zidovudine, Didanosine, Abacarvia, and Tenofovir. NNRTIs include Nevira pine, Efavirenz, and Delavirdine. Protease Inhibitors, which are more potent than the firs t two groups, include Saquinavir, Ritonavir and Indinavir. Fusion inhibitor includes Enfuvirtide. Conceptual Model for Quality of Life Assessment A model to assess QOL was first proposed in 1995 by Wilson and Cleary (4). It includes five major components which are discussed more fully below: (1) biological and physiological variables, (2) symptom status, (3 ) functional status, (4) general he alth perception and (5) overall quality of life. Also include d in the model are characterist ics of the environment and the individual. In 2005, this model was revised by Ferrans and colleagues by integrating the ecological model into Wilson and Clearys model as shown in Figure 2-1 (12). The revised model indicates that the characteristics of the individual and environment can also influence all the components in main path of the model and th e possible direction in the main path goes from biological function, symptoms, functional status, general health perceptio n, and overall quality of life.

PAGE 24

24 Biological Function Biological function, previously known as bi ological and physiol ogical variables in Wilsons model, includes the processes at the mol ecular and cellular level. It also focuses on the function of an organ or an organ system. For ex ample, in the circulatory system, systolic and diastolic blood pressure are m easurable biological functions in hypertension, fasting blood sugar in diabetes, and viral load and CD4+ cell count in HIV infection disease. Symptoms Unlike the collection of measures at the mo lecular and cellular level, symptoms are observed directly by patients and/ or providers because they are re lated to a patients perception of an abnormal physical, emotional or cognitive state as stated by Wilson (4). However, symptoms may be categorized into either physical, psychological or psychophysical from Ferranss perspective. In HIV infection dise ase, symptoms such as diarrhea and myalgia, whether they are HIV disease-related or HAART-related, are meaningful because these symptoms may affect a patients evaluation of his or her QOL. Functional Status Wilson and Cleary defined functiona l status as the ability to perform a particular defined task in multiple domains such as physical function, social func tion, role function, and psychological function whereas Fe rrans defined functional status differently from Wilson and Cleary by including Leidys framewor k (4, 12). In Leidys framewor k, functional status includes four dimensions: (1) functional capacity, (2) functional performa nce, (3) functional capacity utilization, and (4) functional rese rve. Details about the differen ce of functional status between these two concepts are disc ussed elsewhere (4, 12). Functional status in HIV infection measur ed by MOS SF-36 include s physical functioning, role functioning, and social functioning. For ro le functioning, patients are asked whether they

PAGE 25

25 are able to perform regular activities normally su ch as during the past 4 weeks, have you had any of the following problems with your work or any regular activity as a result of physical health?. For physical functioning, patients are as ked whether how much their health is limited in vigorous activities such as running, lifting heavy objects or pa rticipate in strenuous sports. General Health Perception Ferrans agreed with Wilson and Cleary that ge neral health perception is different from other components on the left side of the mode l in Figure 2-1. From Wilson and Clearys perspective, general health perc eption includes all the components th at come earlier in the model and they all represent subjective measures. Fo r example, general heal th perception in MOS SF36 includes 5 items. One item asks the patients in general, would you sa y your health is and the answers range from excel lent/very good/good/fair/poor. Overall Quality of Life Wilson and Cleary defined overall quality of lif e as subjective well-being related to how happy and satisfied someone is with life as a whole. It is a multidimensional construct. Ferrans concurred with Wilson and Cleary in this con cept and he explained that Wilson and Clearys concept of overall quality of life and how it is influenced by patients value and preferences was concordant with Campbells concept (12). Fo r example, being blind may be viewed by one person as a disability that is not worth living with, but might be considered only moderately bothersome for another person. In the case of HIV, another example might be where lipodystrophy might be considered le ss important in an HIV-infected male patient compared with an HIV-infected female patient. Time Reference in MOS SF36 Instrument for Current QOL Some questions in the MOS SF-36 ask patients to assess their current QOL regarding their ability to perform some activities such as r unning, and walking one block. Other questions

PAGE 26

26 impose a time frame to help patients assess their cu rrent quality of life, but this time frame varies from question to question. For exam ple, one item in the general he alth domain asks patients to compare to one year ago, how would you rate y our health in genera l now?, but another question asks patients to respond based on their experiences during the past 4 weeks, During the past 4 weeks, have you had any of the following problems with your work or other regular daily activities as a result of your phys ical health?. Thus, the time frame imposed on the questions in the MOS SF-36 varies from one week to f our weeks. Four weeks to one year? In addition, the responses to th e rating scale used for scori ng each question in the MOS SF36 is are summed to create a summated score that reflects each patients current QOL. This implies that the current QOL is composed of the attributes of these questions within the past 4 weeks to the day that patients assess their QOL. Patients QOL beyond one month ago is not assessed in current QOL. Validity of Quality of Life Assessment Model Wilson and Clearys QOL model was tested and validated in several studies in different diseases (13-16). For example, Sousa and Kw ok investigated the five major components of Wilson and Clearys QOL model simultaneously by using structural equation modeling (SEM) in patients living with AIDS from the AIDS Time -Oriented Health Outcomes Study (13), whereas Wettergren and colleagues examined the relation ship of those components in Hodgkins disease (17). As stated by Wilson and Clearly, the non-exis tence of an arrow in the model and its direction between those components doesnt mean it doesnt exist. Howe ver, this model was developed from the biomedical model developed to show the causal relationship between the construct on the left and the c onstruct on the right and the find ing from Sousas study showed that the model was valid (13).

PAGE 27

27 In regard to the components in the model, for the purpose of the present study Wilson and Clearys model was modified as shown in Figure 2-2 because functional status is a part of general health perception in Wettergren study, a nd functional status, gene ral health perception and overall quality of life are included in MOS SF-36 as QOL. Th is modified model is similar to the conceptual model to assess QOL in HI V/AIDS population proposed by Vidrine and colleagues (18). CD4+ Cell counts, Viral load, and QOL in HIV-infected Persons Numerous studies investigated the association of viral loads, CD4+ cell counts, symptoms, and QOL in patients with HAART (19-25). Most of these studies compared QOL among different treatment regimens or among groups of patients categorized by either CDC classification or CD4+ cell counts or viral load level. Some a ddressed the difference in change in QOL among treatment regimens. For example, Nieuwkerk and colleagues examined the difference in change in HRQOL between two regimens, ritronavir (RQV)/saquinavir (SQV) versus RQV/SQV/stavudine (d4T), in asymptoma tic and symptomatic pati ents (23). QOL was measured by MOS-HIV at baseline and after 12, 24, 36 and 48 weeks of follow up. The MOS HIV was developed specifically for Multicenter AIDS Cohort study (MACS). Mean change in QOL between the two regimens was compared by repeated measured analysis of variance (repeated ANOVA). This statisti cal model used regimens as a between subject factor and time as a within subject factor, adjusting for baseline CD4+ cell counts and viral loads. The results showed no difference in change in QOL from baseline between the two regimens, but QOL statistically increased from baseline for all dimensions. Gill conducted a cross-sectional study to inve stigate the relationship between viral load, CD4+ cell counts, HAART use, and HRQOL by usi ng baseline data from 513 participants in Nutrition for Healthy Living (N FHL) conducted in Boston and Pr ovidence (26). Four domains

PAGE 28

28 in HRQOL, physical functioning (PF), role func tioning (RF), energy levels (EL), and general health perception (HP) were selected and obt ained from the HIV Patients Assessed Report of Status and Experience (HIV-PARSE). CD4+ ce ll counts and viral load were categorized by clinically meaningful cut off points: CD4 > 500, 200-500, and < 200 cells/mL; VL < log 2.6 (undetectable, < 400 copies/mL), log 2.6 to 4.0 (400-10,000 copies/mL) and > log 4.0 (10,000 copies/mL). The results showed that HAART and viral load level had a significant effect on PH only, whereas CD4+ cell counts had the si gnificant effect on PF, RF and HP. In the COMBINE-QoL substudy, Casado et al. assessed the effect of HAART regimens zidovudine (ZDV), lamivudine (3TC) and either nelf inavir (NFV) or nevirapine (NVP) on QOL in HIV-infected nave patients. They found a statistically significant correlation between the MHS score at 12 months and a decrease in viral load in only the ZDV/3TC/NFV arm, whereas PHS at 12 month and a decrease in viral load were statistically correlated in the ZDV/3TC/NVP arm (22). Globe et al. conducted a cross-sectional su rvey and reviewed medical records to investigate the association between clinical parameters and HRQOL in hospitalized persons with HIV disease (27). Data retrie ved from medical records includ ed length of stay during index admission, CD4+ cell count during index admissi on, AIDS-related diagnoses at admission, the number of comorbid medical c onditions at admission, and the numb er of presenting symptoms. Outcomes were measured by the specific HRQO L questionnaire which was modified from the Medical Outcome Study HIV (MOS-HIV), HI V Outcomes Study (HOS) and HIV-PARSE. The results showed that CD4+ cell count did not have a significant relati onship with most of HRQOL dimensions i.e. physical, role and so cial function, but it had a significant negative association with emotional well being and cognitive function.

PAGE 29

29 Eriksson and colleagues investigated the association between CD4+ cell counts and HRQOL in 72 HIV-infected Swedish men ( 28). HRQOL was measured by The Swedish HRQOL questionnaire (SWED-QUAL) which compri sed of 13 dimensions. Because the sample size is small for both the symptomatic and AIDS groups, non-parametric analyses, such as the Kruskal-Wallis or the Mann-Whitney U-test, were used. They found a statistically significant difference between asymptomatic, symptomatic and AIDS patients in physical functioning, mobility, satisfaction with physical ability and role limitations because of physical health. In addition, post-hoc comparisons re vealed a significant difference between asymptomatic patients and AIDS in those dimensions. Prau et al. examined the longitudinal associ ation between CD4+ ce ll count, viral loads, clinical stage, the numbers of self-reported symptoms, other fact ors such as depression, HIV transmission, and HRQOL measured by MOS SF36 from baseline to 3 years in 360 patients (29). Data were analyzed by regression analysis with Generalize Estimation Equations (GEE) which accounted for the correlation between HRQOL measured at different time points. Only the number of self-reported symptoms was si gnificantly associated with MCS and PCS. In the ACTG 175 substudy, Justice et al. ex amined whether physician-reported symptoms were a clinically important subset of patient -reported symptoms in HIV infection. Both physicians and patients used a similar format of questions. They reported that physicians underreported the prevalence of symptoms compared with patient reports. The researchers also noted that the physician reports of symptoms, using patient reports of symptoms as a gold standard, had poor sensitivity, good specificity, moderate positiv e predictive value and a poor negative predictive value (30). In addition, sy mptom severity, not the symptoms itself, was

PAGE 30

30 associated with HRQOL measured General Health-a ssessment Questionnaire for Clinical Trials. Both CD4+ cell count and viral load were not associated with HRQOL. All of the studies mentioned previously di ffer from those studies that focus on how a change in CD4+ cell counts and viral load over time predict change in QOL over time, particularly when each individual has his/her own growth curve traj ectory. Further details about how individual change is diffe rent from group change can be f ound in Clarkes article (31). Examples of those research findings are as follows. Chu and co-researchers examined the rate of change of CD4+ cell counts over time, also called CD4+ cell count trajecto ry, at both the population and individual level by using two different databases, Multicenter AIDS Cohor t Study (MACS) and Womens Interagency HIV Study (WIHS) (32). CD4+ cell counts were mode led by using a Bayesian change point model. The results showed that in the population model, both men and women had a significant change in CD4+ cell counts within 2 years after HAART in itiation. However, in the individual model, both men and women gained significant change in CD4+ cell counts after 7 years of HAART initiation. Weinfurt et al. investigated the relationship between a change in CD4+ cell counts, viral load, HRQOL and time, as well as how the change of these factors correlated to one another in a double-blinded randomized clinical tr ial (33). This trial compared the effect of 2 regimens, ddI or ddI and delavirdine mesylate. The results re vealed that MCS and PCS statistically decreased over time. However, they did not investigate how CD4+ cell counts and viral load can predict HRQOL over time. Liu investigated the pr edictors for lower QOL in HAART among HIVinfected men (34). Predictors were educational le vel, individual risk be haviors, social support, biological markers, HIV-related medications, a nd clinical outcomes. The time-lag model,

PAGE 31

31 predictors at time (t-1) and outco me at time (t), was used. CD4+ cell count was independently associated with PHS. Clinical outcomes were significant predictors for PHS, but not MHS. The most recent five-year longitudinal study, called The French APROCO-COPILOTE (ANR CO-8) multicenter cohort st udy, investigated the relations hip between numerous variables including the number of se lf-reported side effects, depression, clinical di sease stage, CD4+ cell count and HRQOL measured by MOS SF-36 in HIV1 infected patients on HAART (35). The results showed that both the number of self-re ported side effects and CD4+ cell count had a negative association with PCS and MCS in the first year.

PAGE 32

32Table 2-1 The estimated numbers and percentage of HIV/AIDS and AIDS cases by diagnosis, age, gender and race in 1981-2004 AIDS HIV/AIDS 1981-1995 1996-2000 2001-2004 2001-2004 N % N % N % N % Sex Male 467,286 84.7 173,608 75.9 120,242 73.4 112,237 71.3 Female 84,229 15.3 55,253 24.1 43,576 26.6 45,231 28.7 Age Groups <13 7,668 1.4 1,426 0.6 341 0.2 1,025 0.7 13-19 2,748 0.5 1,659 0.7 1,480 0.9 4,336 2.8 20-29 98,990 18.0 30,161 13.2 19,632 12.0 31,503 20.0 30-44 336,967 61.1 137,963 60.3 90,581 55.3 80,063 50.8 45-59 89,530 16.2 49,658 21.7 44,862 27.4 34,882 22.2 60 15,612 2.8 7,996 3.5 6,921 4.2 5,660 3.6 Race White/non-Hispanic 256,460 46.5 72,314 31.6 46,325 28.3 45,497 28.9 Black/non-Hispanic 190,561 34.6 107,618 47.0 81,057 49.5 80,310 51.0 Hispanic 98,438 17.9 45,529 19.9 33,185 20.3 28,725 18.2 Asian/Pacific Islander 3,660 0.7 1,868 0.8 1,788 1.1 1,360 0.9 American Indian/Alaska Native 1,490 0.3 858 0.4 736 0.5 768 0.5

PAGE 33

33 Table 2-2 CDC HIV Infection Cate gories by Clinical Conditions Category A consists of one or more of the conditio ns listed below in an adolescent or adult (greater than or equal to 13 years) with documented HIV infection. Conditions listed in Categories B and C must not have occurred. Asymptomatic HIV infection Persistent generalized lymphadenopathy Acute (primary) HIV infection with accompanying illness or history of acute HIV infection (29,30) Category B Category B consists of symptomatic conditions in an HIV-infected adolescent or adult that are not included among conditions listed in clinical Ca tegory C and that meet at least one of the following criteria: a) the conditions are attributed to HIV infection or are indicative of a defect in cell-mediated immunity; or b) the conditions ar e considered by physicians to have a clinical course or to require management that is complicated by HIV infection. Examples of conditions in clinical Category B include, but are not limited to: Bacillary angiomatosis Candidiasis, oropharyngeal (thrush) Candidiasis, vulvovaginal; persistent, fre quent, or poorly responsive to therapy Cervical dysplasia (moderate or severe)/cervical carcinoma in situ Constitutional symptoms, such as fever (38.5 C) or diarrhea lasting greater than 1 month Hairy leukoplakia, oral Herpes zoster (shingles), involving at least two distinct episodes or more than one dermatome Idiopathic thrombocytopenic purpura Listeriosis Pelvic inflammatory disease, particularly if complicated by tubo-ovarian abscess Peripheral neuropathy For classification purposes, Category B c onditions take precedence over those in Category A. For example, someone previously treated for oral or persistent vaginal candidiasis (and who has not developed a Category C disease) but who is now asymptomatic should be classified in clinical Category B. Category C includes the clinical conditions listed in the AIDS surveillance case definition (Appendix B). For classification purposes, once a Ca tegory C condition has occurred, the person will remain in Category C.

PAGE 34

34 Table 2-3 CDC Classification System for HIV Infection Clinical Categories CD4+ Cell Categories A Asymptomatic B Symptomatic C AIDS 500 cells/ L A1 B1 C1* 200-499 cells/ L A2 B2 C2* 200 cells/ L A3* B3* C3* Patients in A3, B3 and C1-C3 are considered as AIDS

PAGE 35

35 Figure 2-1 Ferranss Conceptual Model for QOL Assessment Biological function Symptoms Functional status General health perception Overall quality of life Characteristics of the individual Characteristics of the environment

PAGE 36

36 Figure 2-2 Modified Version of Wilsons Model Biological function Symptoms Quality of life Characteristics of the individual Characteristics of the environment

PAGE 37

37 CHAPTER 3 METHODS This chapter describes the conceptual framework for the research, information about the secondary database that will be used (Multicente r AIDS Cohort Study database), the independent and dependent variables with thei r operationalization and how the variables were transformed to fit the assumptions of statistica l methods for data analysis. In addition, the plan for handling missing data, as well as the techniques that will be used to handle the correlation of multiple QOL measurements in this longitudinal study wi ll be summarized. Finally, two statistical models, traditional and time-lag, w ill also be described and tested. Conceptual Framework The modified QOL assessment model that is used in this dissertation is comprised of two components: biological function and QOL. The modi fied model is used in this study for the following reasons. First, measuring QOL is ve ry important since in order to achieve the treatment goals specified in the clinical practice guidelines for treating HIV-infection, to reach maximal and durable suspension of viral loa d, restoration and pres ervation of immunologic function, improvement of quality of life, and redu ction of HIV-related morbidity and mortality, it is important to study how QOL in HIV-inf ected patients treated with HAART changes over time. Second, the temporal relationship from Wilson a nd Clearlys model implies that change in biological function occurs before change in symp toms and change in symptoms occurs before change in QOL. From a pharmacological pe rspective, HAART suppresses viral load and increases CD4+ cell count. Viral load and CD4+ cell count are related to both symptoms and QOL. Therefore, the modified version of Wilsons model was used to investigate the longitudinal relationship between th e trend over time of the previous lab tests or current lab test

PAGE 38

38 results with current QOL. Specifically, this mo dified model was used to empirically validate whether the lag time relationship between previo us lab test results and current QOL exists by comparing time-lag model with non time-lag mode l. Figure 3-1 shows the modified conceptual model for the lag time longitudinal relationship be tween previous biologi cal function and current QOL. Data Public Data Set from the Multicenter AIDS Cohort Study (MACS) a ssociated with Johns Hopkins University will be used in this study. MACS, initiated in 1984, is a study of the natural and treated histories of HIV1 infection in homosexual and bisexual men conducted by sites located in Los Angeles, Chicago, Pittsburgh and Ba ltimore. It is a prospective cohort study with semi-annual visits. At each visit, descriptions of the medication, medication adherence, physical examination, HIV-related symptoms, side eff ects of antiretroviral medication, and QOL are collected. By 2007, 6,972 patients were enrolled in MACS, accounting for 74,536 person-years. Unfortunately, not all of the MACS data is publicly available. This study uses the MACS Public Data Set version P15 which includes patients enrolled until October 2002. The MACS database comprises several data tables but only three data tables were used in this study; (1) drug, (2) quality of life and (3) lab tests. In th e drug table, antiretroviral drugs prescribed at each clinic visit we re recorded. This table was used to identify patients prescribed HAART therapy. HAART Definition The definition of HAART used by the MACS study was based on the DHHS/Kaiser Panel [DHHS/Kaiser 2005] guidelines. The guidelines define HAART as : (a) two or more NRTIs in combination with at least one PI or one NNRTI (89% of observations classified as HAART); (b) one NRTI in combination with at least one PI and at least one NNRTI (6%); (c) a regimen

PAGE 39

39 containing ritonavir and saquinavir in combina tion with one NRTI and no NNRTIs (1%); and (d) an abacavir or tenofovir containing regimen of th ree or more NRTIs in the absence of both PIs and NNRTIs (4%), except for the three-NRTI re gimens consisting of: abacavir + tenofovir + lamivudine OR didanosine + tenofovir + lamivudi ne. Combinations of zidovudine (AZT) and stavudine (d4T) with either a PI or NNRTI were not considered as HAART. Health-Related Quality of Life Patients were asked at each visit to comple te the self-administered Medical Outcome Study Short Form 36 (MOS SF-36) health survey. This instrument was first incorporated in the MACS protocol in 1994 (19). MOS SF-36 consists of eight HRQOL domains: general health perception, physical functioning, role limitation caused by physical health problems, role limitation caused by mental health problems, emotional well-bei ng, social functioning, energy/fatigue, and pain. MOS SF-36 domains were classified into two major components, physical and mental health component. Physic al component summary scores (PCS) include general health perception, physical functioni ng, role limitation caused by physical health problems and pain, whereas mental component su mmary scores (MCS) include role limitation caused by mental health problems, emotional we ll-being, social functioning and energy/fatigue. This study will use the PCS and MC S as the dependent variable. CD4+ Cell Count In the MACS study, CD4 cell lymphocytes were quantified by flow cytometry in the laboratories certified by Flow Cy tometry Quality Assessment Progr am of the National Institute of Allergy and Infectious Diseases (34, 36). CD4+ cell count was measured as percent CD4+ cell.

PAGE 40

40 Viral Load HIV-1 RNA level was measured by Nuclisens in the laboratories ce rtified by the Virology Quality Assurance Laboratory proficiency testing program of the National Institute of Health (34, 36). Viral load was measured in term of th e number of copies per millimeter. Because viral load is not normally distributed, viral load levels were transformed into log10 scale (33). Inclusion and Exclusion Criteria Patients included in this study are all those who were prescr ibed HAART. The index date was defined as the first clinic visit patient s received HAART regimen. Patients were followed until the cut-off date in the database. Patients with only one laboratory test were excluded from data analysis because insufficient data is available for the time-lag model. Data Analysis Process This study used a secondary database called MACS Data preparation was the first step in working with MACS. The data were modeled in the second step. Time-lag and non time-lag models were compared in the final step. Data Preparation A descriptive analysis was conducted for each variable, e.g., average and standard deviation for continuous variable s and percent for categorical va riables. In addition, missing data associated with QOL were tested to dete rmine whether it was missing completely at random (MCAR) or missing at random (MAR). A stra tegy for handling dropouts and missing values was employed, in order to increase th e sample size because some patients dropped out of the MACS study and some had missing values for study vari ables. If missing data were MAR, multiple imputation was used. If they were MNAR, sel ection or pattern mixture model was used. The strategy for handling missing data is presente d in more detail in the next section.

PAGE 41

41 Missing data and dropping out in the longitudinal study Missing data and subject drop out s are common in longitudinal studies. Dropping out in a longitudinal study is theoretically equivalent to either unit non -response or item non-response in cross-sectional surveys (37, 38). Unit non-response occurs when the survey from a sampled person is incomplete whereas item non-response ta kes place when a participant does not answer one or more items in the questionnaire. Several review articles related to missing da ta mechanisms and how to handle missing data are available (37-39). Generally, three diffe rent missingness mechanisms are noted: (1) missing completely at random (MCAR), (2) miss ing at random (MAR) and (3) missing not at random (MNAR). In MAR, if R is an indicator for the missingness of data and Ycom is the complete data. Ycom can be partitioned into Yobs and Ymis, where Yobs and Ymis are the observed and missing parts respectively. The relationship between Ycom, Yobs and Ymis can be presented by the following equation because the distribution of missingness does not depend on Ymis, but on Yobs (38). (/)(/)comobsPRYPRY For MCAR, the distribution of missingness does not depend on Yobs. (/)()comPRYPR When (/)(/)comobsPRYPRY is violated, the distribu tion of missingness depends on Yobs, and it is called MNAR. Table 3-1 shows missing pattern s of CD4+ cell count measured in two consecutive months (X and Y). How to handle dropouts in longitudinal study Several techniques or models to handle missi ng data were developed and tested, including mean of series method, hot-deck method, last valu e carry forward method, regression imputation,

PAGE 42

42 multiple imputation and weighting (37, 38, 40-42). When data were missing not at random, two types of models were developed and tested: se lection models and pattern-mixture models (39, 43, 44). For selection models, Y and the pr obability that Y is missing are modeled simultaneously as shown in equation 1. (y,r)(y)P(r/y) ff (1) Pattern mixture models can be expressed by equation 2. (y,r)(y/r)P(r) ff (2) These models do not have any implicit untestable assumptions and they have computational advantages. An example of random a pattern mixture model is given below (43). Advantages and disadvantages of both models are summarized in Table 3-2. itQOL = ij i ije u Gender Treatment Visit ) ( ) ( ) (13 12 11 10 ) ( e DropoutTim Log= j ii bu Treatment ) (21 20 Where ij and iu are the subject and pattern level random intercepts. Data Analysis In this step, health-re lated quality of life was modeled us ing a random effect model, also called multilevel model. Multilevel models have been used in studying functional impairment in HIV-infected persons (45) and relationships betw een CD4+ cell count, vira l load, and quality of life over time in HIV-1-infected patients (33). These models have an advantage over repeatedmeasured ANOVA and other statisti cal analyses because a balanced data structure is not required (46-48) and it takes into account unit of measurement, corre lation among measurements and time-varying covariates (49). The statistical model is presented below and the conceptual diagram for this model is found in Figure 3-2. 012 itiiittitYXt

PAGE 43

432(,)itiiYNXI (,)iN (0,)itiN Where itY is PHC or MHC for individual i measured at time t i0 is the random intercept, it X is the independent variable, CD4+ cell count or viral load, for subject i at time t 1i is the random regression coefficient for independent variable j t is time, i2 is the random regression coefficient to time and it is the error for subject i at time t Time was coded as 1 for an index date and coded as 2 to 11 corresponding to each visit to the clinic scheduled every 6 months according to the study protocol. In addition, PHC and MHC were modeled by a combination of time-lag model and random coefficient models when knowledge of HIV test results from a previous visit was taken into account. Consequently, the model mentioned above was partially modified as follow. 01(1)2itiiitititYX Where all variables and parameters are the same as the previous model, except (1)itX is the independent variable j for subjec t i at time t-1. The time-lag conceptual model is shown in Figure 3-3. In addition, the correlation between QOL measurements will be taken into account based upon one of these assumptions. Independent Structure In this structure, the correl ations between subsequent QOL measurements are assumed to be zero as show in Table 3-2.

PAGE 44

44Exchangeable Structure In this structure, the correl ations between subsequent QOL measurements are the same, regardless of the time as shown in Table 3-3 m-dependent Structure In this structure, it is assumed that the co rrelations < m measuremen ts apart are equal and correlations > m measurements apart are assumed to be zero as shown in Table 3-4 when m = 2. Autoregressive Correlation Structure In this structure, th e correlation between two consecutive QOL is the correlation between t measurement apart is t as in Table 3-5. Maximum likelihood estimation (MLE) method was used to fit the model. Using this method, both the fixed part (coeffi cients) and the random part (var iance) between the models can be compared. For both non time-lag and time-lag models, if the model didnt converge, grandmean centering was used to help achieve the m odel conversion. For nested models, model fit was compared by the log likelihood ratio test (LR test). Non-nested models were compared by Akaike information criterion (AIC). Model residuals were also examined to ascertain whether the data were normally distributed Type II error was set at 5%. All statistical analyses were conducted by statistical program, STATA. IRB Approval This study was approved by Hea lth Center Institutional Revi ew Board, University of Florida.

PAGE 45

45 Table 3-1 Missing Data Mechanisms of CD4+ Cell count Y X Complete MCAR MAR MNAR 168 148 148 148 148 126 123 132 149 149 160 169 169 105 138 116 102 125 88 112 100 133 150 150 94 113 109 96 109 78 176 137 137 128 155 155 131 131 130 101 101 145 155 155 155 136 140 146 134 134 111 129 97 85 85 134 124 124 153 112 112 118 118 137 122 122 101 119 103 106 106 78 74 74 151 113 113 Mean (SD) 125.7 121.9 108.6 138.3 153.4 (23.0) (24.7) (25.1) (21.1) (7.5)

PAGE 46

46 Figure 3-1 Time-lag Conceptual Mode l for Quality of Life Assessment Table 3-2 Advantages and di sadvantages of selection a nd pattern-mixture models Selection Models Pattern Mixture Models Models include parameters of interest Models exclude parameter of interest Easy to formulate hypothesis about drop out process Make explicit assumptions about unobserved responses Difficult to infer how assumptions on drop out process translate into assumption about distribution of unobserved responses Implied drop out process is not immediately transparent Difficult to determine model identifiability Straightforward to determine model identifiability Computational intractable Computational simple Current Biological Function Symptoms Quality of Life Lag Time

PAGE 47

47 Figure 3-2 Conceptual Diagram for Traditional Model Figure 3-3 Conceptual Diagram for Time-lag Model QOLt1 QOLt2 QOL3t3 QOLt4 QOL5t5 Labt0 Labt1 Labt2 Labt3 Labt4 QOLt1 QOLt2 QOL3t3 QOLt4 QOL5t5 Labt1 Labt2 Labt3 Labt4 Labt5

PAGE 48

48 Table 3-2 Independent Struct ure of QOL Measurements QOLt1 QOLt2 QOLt3 QOLt4 QOLt5 QOLt1 0 0 0 0 QOLt2 0 0 0 0 QOLt3 0 0 0 0 QOLt4 0 0 0 0 QOLt5 0 0 0 0 Table 3-3 Exchangeable Structure of QOL Measurements QOLt1 QOLt2 QOLt3 QOLt4 QOLt5 QOLt1 QOLt2 QOLt3 QOLt4 QOLt5 Table 3-4 m-dependent Stru cture of QOL Measurement QOLt1 QOLt2 QOLt3 QOLt4 QOLt5 QOLt1 1 2 0 0 QOLt2 1 1 2 0 QOLt3 2 1 1 2 QOLt4 0 2 1 1 QOLt5 0 0 2 1 Table 3-5 Autoregressive Correlatio n Structure of QOL Measurements QOLt1 QOLt2 QOLt3 QOLt4 QOLt5 QOLt1 1 2 3 4 QOLt2 1 1 2 3 QOLt3 2 1 1 2 QOLt4 3 2 1 1 QOLt5 4 3 2 1

PAGE 49

49 CHAPTER 4 RESULTS This chapter is divided into three sections. First, the result from merging the three data tables from MACS database is described. Second, the result fr om exploratory analysis of PHC and MHC is individually presented. Finally, th e results from traditional and time-lag models are presented. Merging MACS Database Data from the following tables were merged by using case number and visit as a merging key variable. Drug Table The drug table has 13,668 observations from 1, 599 patients. Any obs ervation before visit 210 was dropped because QOL was first collected in MACS at visit 210, resulting in 8,677 observations. Following this step, the visit with the first HAART was determined and any visit before the first HAART was dropped. Consequen tly, 5,701 observations were used to merge with QOL and lab test tables. QOL Table The QOL table contains 19,913 obs ervations from 2,858 patients. This table was merged with the drug table, resulting in 20,509 observa tions. 14,808 observations were dropped because these observations were only availa ble in the QOL table. Finally, 5,701 observations were left to merge with the lab test table. Lab Test Table The lab test table contains 22, 886 observations from 2,701 patient s. Merging data from the previous section with the lab test table result ed in 3,452 observations from 490 patients for data analysis.

PAGE 50

50Missing Data After merging the three data tables together, 18 observations (0.52%) had a missing value in CD4+ cell count and 34 observations (0.98% ) had a missing value in viral load. Because missing data is less than 1%, all missing values were substituted with the average of that variable for a particular patient. Exploratory Analysis Demographic Data Table 4-1 depicts clinic visits when patients started HAART. The average age of the 490 patients when they started HAART (index date) was approximately 45 years (Table 4-2), with most patients between 31 to 60 years old (Table 4-3). The average follow up time for all patients after HAART initiation was approximately 46 months (Table 4-4). Most patients were followed up at least 5 years (Table 4-5) while some patients were followed up only 2 years because they entered the study later than the othe rs. In other words, some patient s entered the study as little as 2 years before the cut-off date (2002). Most patients had a college degree (Table 4-6). Quality of Life Trajectory Figure 4-1 and Figure 4-2 show PHC and MHC score trajectory of 10 randomly selected patients respectively. In both figures, each patien t has a different baseline PHC and MHC level. In addition, each patient has a different rate of change in QOL over time. The average PHC score after patients started HAA RT was 76.03 whereas the averag e MHC score was 73.39 (Table 4-7 and Table 4-8). Average PHC scores rang ed from 74.03 to 76.54, whereas average MHC scores ranged from 71.52 to 73.50. The standard error of the mean for PHC at 6 months after HAART initiation was 1.00 and increased thereafter. Average PHC at 6 months after initiation ranged fr om 74.06 to 78.01. The standard error of the mean for PHC increased ov er time. The standard error of the mean for

PAGE 51

51 MHC at 6 months after HAART initiation ranged fr om 1.04 to 1.57 and increased over time. The average PHC and average MHC scores fluctu ated during the follow-up period as shown in Figure 4-3 and Figure 4-4. CD4+ Cell Count and Viral Load Trajectory Figure 4-5 shows the change in CD4+ cell count over time for 10 patients. These changes in CD4+ cell count over time indicat e a unique trajectory of CD4+ cell count for each individual. However, the average change in CD4+ cell count over time increases as shown in Figure 4-6 and Table 4-9. CD4+ cell count 6 months af ter HAART initiation was 220.68 and increased over time except for the last observation where the CD 4+ cell count dropped. The standard error of the mean for CD4+ cell count also decreased ove r time. The average change in lag CD4+ cell count also improved overtime as shown in Figure 4-7. Figure 4-8 shows the change in viral load ove r time for 10 patients. Again, each patient has a different initial viral load level and has a different rate of change over time. Table 4-10 and Figure 4-9 depict the change in average viral load over time for all patients. The average viral load at 6 months after HAART initiation was 30,398.02 copies/mL. Although viral load fluctuated, overall the trend of average viral load decreased over time. The average change in lag viral load over time also wa s lower as shown in Figure 4-10. Correlation among PHC and MHC Table 4-11 shows the correlation coefficients between PHC measured at different followup times. The correlation coefficient ranges fr om 0.57 to 0.83. The highest correlation coefficient of PHC is between index 9 (4 year s after HAART initiation) and index 11 (5 years after HAART initiation), whereas the lowest correlation coefficient of PHC is between index 2 (6 months after HAART initiation) and index 10 (54 months after HAART initiation).

PAGE 52

52 Table 4-12 also shows the correlation coeffi cients between MHC measured at different follow-up times. The highest correlation coeffi cient of MHC (0.74) is between index 2 (6 months after HAART initiation) and index 9 (4 years after HAART in itiation), whereas the lowest correlation coefficient of MHC (0.59) is between index 4 (24 months after HAART initiation) and index 9 (4 years after HAART initiation). Time-Lag and Non Time-Lag Models In this part, the time-lag and non-time la g models for PHC and MHC are presented individually. Time-lag and Non Time-lag Models for PHC Table 4-13 shows all models for PHC, Howeve r only the residuals for model 2 is shown in Figure 4-9 since the residuals for other models were also normally distribut ed. The first model is a null model which has no predictor. The log likelihood statistic (LL) was -13,931.55. For all models, time has an inverse relationship with PHC ( = -0.21 to -0.44, p<0.05). In other words, PHC scores significantly decreased over time. CD4+ cell count and prev ious CD4+ cell count have a positive relationship with PHC in all mode ls. Regression coefficients of previous CD+ cell count are the same as those of CD+ cell count. The second and third models have time as the predictor. In the second model, time was treated as a fixed effect. Th e log likelihood (LL) significantly increased from -13,931.55 to 13,927.27 from model 1 to model 2 (p<0.05). Th is implied that, on average, PHC significantly decrease over time. However, time was treated as a random effect in the third model. The log likelihood statistic (LL) al so increased from -13,927.27 in mode l 2 to -13,863.02 in model 3. In other words, the difference in th e log likelihood statistic (-2LL) of these two models was 128.50, which indicated that the two models were statistically different (p<0.05). That means the effect

PAGE 53

53 of time on PHC was different from patient to patien t. In other words, each patient had his/her own PHC trajectory. Model 4 to model 7 answered re search question 1. When CD 4+ cell count was added into the second model and treated for fi xed effect, the log likelihood stat istics (LL) increased from 13,927.27 to -13,909.52. The difference in the log likelihood statistics (-2LL) was 35.50 which was statistically different (p <0.05). CD4+ cell count si gnificantly predicted PHC ( = 0.03, p<0.05). When time was treated for random effect (model 5), the model fit statistic statistically increased from -13,909.52 to -13,848.84, which im plied that the random model was better than the fixed model. In this model, CD4+ cell count significan tly predicted PHC ( = 0.03, p<0.05). When CD4+ cell count was treated for random effect (model 6), compared with model 4, the log likelihood statistic increased from -13, 909.52 to -13,889.82, which was statistically significant (p<0.05). This indicated that the effect of CD4+ cell c ount on PHC varied among patients. When both time and CD4+ cell counts were treated for random effect (model 7), the log likelihood statisti c increased from -13,909.52 to 13,844.13, which was statistically significant (p<0.05). This indicated that PHC trajectory and the effect of CD4+ cell count on PHC were different among patients. Model 8 to model 11 answered research question 2. When previous CD4+ cell count was added into the second model and treated for fixe d effect, the log likeli hood statistic increased from -13927.27 to -13905.95 (model 8). This show ed statistical difference between two models (p<0.05). In this model, previous CD 4+ cell count significan tly predicted PHC ( = 0.03, p<0.05). In model 9 where time was treated for random effect, the log likelihood statistic significantly increased from -13,905. 95 to -13845.09 (p<0.05). This showed that PHC trajectory

PAGE 54

54 was different among patients. When previous CD4+ cell count was treated for random effect (model 10), the log likelihood stat istic significantly increased from -13,905.95 to -13,894.39 (p<0.05). This result in dicated that the effect of previous CD4+ cell count on PHC varied among patients. When both time and previous CD4+cel l count were treated for random effect (model 11), the log likelihood statistic significantly increased from -1 3,905.95 to -13,844.11 (p<0.05). Again, this showed that PHC trajectory and the effect of pr evious CD4+ cell count on PHC varied among patients. Model 12 to model 15 answered re search question 3. When vi ral load was added into the second model and treated for fixed effect, the lo g likelihood statistics (LL) increased from 13,927.27 to -13,921.08 (model 12). The difference in the log lik elihood statistics (-2LL) was 12.38 which was statistically differe nt (p<0.05). In this model, viral load significantly predicted PHC ( = 0.96, p<0.05). When time was treated fo r random effect (model 13), the log likelihood statistic statistically increased from -13,921. 08 to -13,857.86, which implied that the PHC trajectory was different am ong patients. In this model, viral load significantly predicted PHC ( = 0.89, p<0.05). When viral load was treated for random eff ect (model 14), compared with model 12, the log likelihood statisti c increased from -13,921.08 to 13,903.09, which was statistically significant (p<0.05). This indicated that the eff ect of viral load on PHC varied among patients. When both time and CD4+ cell counts were tr eated for random effect (model 15), the log likelihood statistic increa sed from -13,921.08 to -13,841.43, wh ich was statistically significant (p<0.05). This implied that the effect of CD 4+ cell count on PHC and the trend of PHC varied among patients.

PAGE 55

55 Model 16 to model 19 answered re search question 4. When pr evious viral load was added into the second model a nd treated for fixed effect, the log likelihood statistic increased from 13927.27 to -13920.39 (model 16). This showed st atistical difference between two models (p<0.05). In this model, previous viral load significantly predicted PHC ( = 0.98, p<0.05). In model 17, time was treated for random eff ect, the log likelihood st atistic significantly increased from -13,920.39 to -13858.41 (p<0.05). Th is showed that PHC tr ajectory was different among patients. When previous viral load was treated for random effect (model 18), the log likelihood statistic significantly in creased from -13,920.39 to -13, 909.31 (p<0.05). This result indicated that the effect of previous viral load on PHC va ried among patients. When both time and previous viral load were treated for random effect (model 19), the log likelihood statistic significantly increased from -13,920. 39 to -13,849.05 (p<0.05). Again, this showed that PHC trajectory and the effect of prev ious viral load on PHC varied among patients. In this model, previous viral load also significantly predicted PHC ( = 0.75, p<0.05). Comparing non-nested models (model 4 vs. mo del 8, model 5 vs. model 9, model 6 vs. model 10, model 7 vs. model 11, model 12 vs. model 16, model 13 vs. model 17, model 14 vs. model 18, model 15 vs. model 19), AIC among these pairs were not different. In other words, there were no different between time-lag and non time-lag models. Time-lag and Non Time-lag Models for MHC Models for MHC are shown in Table 4-14. However, only the residuals for model 2 was shown in Figure 4-10 because the residuals for othe r models were also normally distributed. The first model was also the null model, no predictor, as in the previous section. The log likelihood statistic (LL) for this model was -14,320.45. For all converged models, MHC scores

PAGE 56

56 significantly decreased over time. In other wo rds, time significantly predicted MHC scores ( = 0.14 to -0.27, p<0.05). The second and third models have time as the predictor. In the second model, time was treated as a fixed effect but as a random effect in model 3. The log likelihood (LL) significantly increased from -14,320.45 to 14,318.51 from model 1 to model 2 (p<0.05). This indicated that on average MHC decreased over time The log likelihood statistic (LL) also increased from 14,318.51 in model 2 to -14,291.46 in model 3. In other words, the difference in the log likelihood statistic (-2LL) of these two models was 54.08, whic h indicated that the two models were statistically different (p<0.05). This implie d that patients had different rate of change of MHC scores over time. Model 4 to model 7 answered re search question 5. When CD 4+ cell count was added into the second model and treated for fi xed effect, the log likelihood st atistic (LL) increased from 14,318.51 to -14,317.07. The difference in the lo g likelihood statistics (-2LL) was 2.87, which was not different. When time was treated for ra ndom effect (model 5), the model fit statistic statistically increased from -14,317.07 to -14,290.65 (p<0.05). This implied that MHC trajectory varied among patients. In this model, time significantly predicted MHC ( = -0.25, p<0.05). When CD4+ cell count was treated for random effect (model 6), compared with model 4, the log likelihood statistic increased from -14, 317.07 to -14,314.00, which was statistically significant (p<0.05). This indicated that the effect of CD4+ cell c ount on MHC varied among patients. When both time and CD4+ cell counts were treated for random effect (model 7), the log likelihood statistic incr eased from -14,317.07 to -14, 289.78 which was statistically significant (p<0.05). This indicated that MHC trajectory and the effect of CD4+ cell count on MHC were different among patients.

PAGE 57

57 Model 8 to model 11 answered research question 6. When previous CD4+ cell count was added into the second model and treated for fixe d effect, the log likeli hood statistic increased from -14,318.51 to -14,316.31 (model 8). This sh owed statistical differe nce between two models (p<0.05). In this model, previous CD 4+ cell count significan tly predicted MHC ( = 0.01, p<0.05). In model 9, time was treated for random eff ect, the log likelihood statistic significantly increased from -14,316.31 to -14,289.98 (p<0.05) This showed that MHC trajectory was different among patients. When previous CD4+ cell count was tr eated for random effect (model 10), the log likelihood statistic in creased, but not significant, fr om -14,316.31 to -14,315.46. This result indicated that the effect of previous CD4+ cell count on MHC did not vary among patients. When both time and previous CD4+cel l count were treated for random effect (model 11), the model did not converge. Model 12 to model 15 answered re search question 7. When vi ral load was added into the second model and treated for fixed effect, the lo g likelihood statistics (LL) increased from 14,318.51 to -14,314.36 (model 12). The difference in the log lik elihood statistics (-2LL) was 16.70, which was statistically different (p<0.05). In this model, viral load significantly predicted MHC ( = 0.89, p<0.05). When time was treate d for random effect (model 13), the log likelihood statistic statis tically increased from -14,314.36 to -14,288.02 (p<0.05). This implied that the MHC trajectory was diffe rent among patients. In this m odel, viral load significantly predicted MHC ( = 0.82, p<0.05). When viral load was treated for random eff ect (model 14), compared with model 12, the log likelihood statistic incr eased from -14,314.36 to -14, 306.01 which was statistically significant (p<0.05). This indicated that the eff ect of viral load on MHC varied among patients.

PAGE 58

58 When both time and CD4+ cell counts were tr eated for random effect (model 15), the log likelihood statistic increa sed from -14,314.36 to -14,277.60, wh ich was statistically significant (p<0.05). This implied that the effect of CD 4+ cell count on MHC and the trend of MHC varied among patients. Model 16 to model 19 answered re search question 8. When pr evious viral load was added into the second model a nd treated for fixed effect, the log likelihood statistic increased from -14,318.51 to -14,315.46 (model 16). This showed statistical differen ce between two models (p<0.05). In this model, previous viral load significantly predicted MHC ( = 0.73, p<0.05). In model 17, time was treated for random eff ect, the log likelihood st atistic significantly increased from -14,315.46 to -14,289.30 (p<0.05) This showed that MHC trajectory was different among patients. When previous viral load was treate d for random effect (model 18), the log likelihood statistic signifi cantly increased from -14,315.46 to -14,310.87 (p<0.05). This result indicated that the effect of previous viral load on MHC varied among patients. When both time and previous viral load were treated fo r random effect (model 19), the log likelihood statistic significantly increased from -14,315.46 to -14,284.22 (p<0.05) Again, this showed that MHC trajectory and the effect of previous viral load on MHC varied among patients. In this model, previous viral load also significantly predicted MHC ( = 0.66, p<0.05). Comparing non-nested models fo r both CD4+ cell count and vi ral load (model 4 vs. model 8, model 5 vs. model 9, model 6 vs. model 10, model 7 vs. model 11, model 12 vs. model 16, model 13 vs. model 17, model 14 vs. model 18, model 15 vs. model 19), AIC among these pairs were not different. In other words, there were no different between time-lag and non time-lag models.

PAGE 59

59 Table 4-1 Time patients started HAART Index N Percent 220 1 0.20 230 4 0.82 240 26 5.31 250 141 28.78 260 115 23.47 270 89 18.16 280 65 13.27 290 18 3.67 300 8 1.63 310 3 0.61 320 8 1.63 330 4 0.82 340 2 0.41 350 6 1.22 Total 490 100.00 Table 4-2 Average age at HAART initiation Variable N Mean SEM 95% CI Age 490 44.91 0.37 44.18 45.63 Table 4-3 Age of patients at HAART initiation Age N Percent <30 4 0.82 31-40 147 30.00 41-50 245 50.00 51-60 73 14.90 61-70 16 3.26 70-80 1 0.20 >81 4 0.82 Total 490 100.00 Table 4-4 Average follow up time after HAART initiation Variable N Mean SEM 95% CI Followup time 490 46.11 0.72 44.69 47.53

PAGE 60

60 Table 4-5 The number of patients followed up until last clinic visit after HAART initiation Follow up Time (Months) N Percent 12 33 6.73 24 55 11.22 36 56 11.43 48 74 15.10 60 272 55.51 Table 4-6 Education level of patients who received HAART Education N Percent Less than 12th grade 4 0.89 12th grade 42 9.35 At least one year college, but no degree 139 30.96 For year college/got a degree 106 23.61 Some graduate work 61 13.59 Post graduate degree 97 21.60

PAGE 61

61 Table 4 -7 Average physical health component score by time since HAART 95% CI Variable Index N Mean SEM L U PHC 2 425 76.03 1.00 74.06 78.01 PHC 3 416 74.18 1.09 72.05 76.32 PHC 4 401 75.53 1.02 73.53 77.55 PHC 5 386 75.52 1.06 73.44 77.59 PHC 6 373 75.38 1.11 73.19 77.56 PHC 7 343 76.06 1.14 73.82 78.30 PHC 8 318 76.54 1.17 74.25 78.94 PHC 9 297 75.69 1.24 73.25 78.14 PHC 10 280 74.73 1.30 72.16 77.31 PHC 11 213 74.03 1.45 71.17 76.88 PHC = Physical health component Index = Time since started HAART with 6 months interval Table 4-8 Average mental health component score by time since HAART 95% CI Variable Index N Mean SEM L U MHC 2 425 73.39 1.04 71.35 75.42 MHC 3 416 71.99 1.05 69.93 74.05 MHC 4 401 71.52 1.14 69.27 73.77 MHC 5 386 71.94 1.14 69.69 74.19 MHC 6 373 72.76 1.13 70.53 74.99 MHC 7 343 72.62 1.22 70.23 75.01 MHC 8 318 73.78 1.26 71.29 76.27 MHC 9 297 72.91 1.32 70.31 75.51 MHC 10 280 73.50 1.35 70.85 76.15 MHC 11 213 73.49 1.57 70.40 76.58 MHC = Mental health component Index = Time since started HAART with 6 months interval

PAGE 62

62 Table 4-9 Average CD4+ cell count at each visit 95% CI Index N Mean SEM L U 2 425 220.68 5.18 210.52 230.85 3 416 231.13 5.24 220.86 241.40 4 401 239.65 5.14 229.58 249.72 5 386 249.62 5.30 239.23 260.02 6 373 250.73 5.41 240.12 261.35 7 343 251.42 5.56 240.51 262.33 8 318 259.93 5.87 248.43 271.43 9 297 259.08 6.06 247.19 270.96 10 280 260.97 6.39 248.43 273.50 11 213 255.10 6.76 241.84 268.36 Table 4-10 Average log viral load at each visit 95% CI Index N Mean SEM L U 2 425 2.74764 0.06111 2.62782 2.86746 3 416 2.75666 0.05980 2.63941 2.87390 4 401 2.65515 0.06358 2.53048 2.77982 5 386 2.58011 0.06432 2.45399 2.70622 6 373 2.51395 0.06242 2.39156 2.63633 7 343 2.48340 0.06392 2.35806 2.60873 8 318 2.36459 0.06328 2.24052 2.48866 9 297 2.37969 0.06559 2.25109 2.50828 10 280 2.30658 0.06518 2.17880 2.43437 11 213 2.31888 0.07861 2.16475 2.47301

PAGE 63

63 Table 4-11 Correlation coefficients am ong PHC measured at different time Index2 Index3 Index4 Index5 Index6 Index7 Index8 Index9 Index10 Index11 Index2 1.0000 Index3 0.7315 1.0000 Index4 0.8130 0.6778 1.0000 Index5 0.6600 0.6662 0.6823 1.0000 Index6 0.7398 0.6638 0.7225 0.7355 1.0000 Index7 0.6653 0.6351 0.6977 0.6841 0.7804 1.0000 Index8 0.6515 0.6272 0.7046 0.7137 0.7180 0.7656 1.0000 Index9 0.6576 0.6061 0.6821 0.7571 0.7347 0.8032 0.8059 1.0000 Index10 0.5718 0.6109 0.6490 0.6476 0.7012 0.7614 0.7630 0.8235 1.0000 Index11 0.6077 0.6090 0.6516 0.6882 0.6984 0.7130 0.7378 0.8276 0.8004 1.000 Table 4-12 Correlation coefficients am ong MHC measured at different time Index2 Index3 Index4 Index5 Index6 Index7 Index8 Index9 Index10 Index11 Index2 1.0000 Index3 0.7355 1.0000 Index4 0.6541 0.6745 1.0000 Index5 0.6653 0.6544 0.7071 1.0000 Index6 0.6708 0.6500 0.6744 0.6850 1.0000 Index7 0.7324 0.6420 0.5881 0.6882 0.6907 1.0000 Index8 0.6336 0.6109 0.6328 0.6626 0.6481 0.7071 1.0000 Index9 0.7449 0.6277 0.5827 0.6719 0.7103 0.7348 0.7079 1.0000 Index10 0.6073 0.6034 0.6061 0.6165 0.6667 0.6942 0.7004 0.6888 1.0000 Index11 0.6286 0.5846 0.6210 0.6094 0.6513 0.7109 0.7277 0.7136 0.7100 1.000

PAGE 64

64Table 4-13 Models for PHC 1 2 3 Null Index Index Fixed Effect Intercept 74.46* 75.66* 75.92* Index -0.21* -0.27* CD4+ cell^ Log VL Lag CD4+ cell^ Log Lag VL Random Effect# Intercept 354.92 356.27 334.13 Index 1.84 CD4+ cell^ Log VL Lag CD4+ cell^ Log Lag VL Residual# 123.08 122.65 106.97 Fit Statistics LL -13931.55 -13927.27 -13863.02 -2LL 27863.10 27854.54 27726.04 AIC 27869.10 27862.54 27736.05 p<0.05 # variance ^ grand mean-centered LL = Log likelihood -2LL = -2 Log likelihood AIC = Akaike Information Criterion

PAGE 65

65 Table 4-13 Models for PHC (continued) 4 5 6 7 8 9 10 11 Index & CD4+ Index & CD4+ Index & CD4+ Index & CD4+ Index & Lag CD4+ Index & Lag CD4+ Index & Lag CD4+ Index & Lag CD4+ Fixed Effect Intercept 76.51* 76.67* 76.82* 76.85* 76.81* 76.98* 77.16* 77.11* Index -0.34* -0.38* -0.34* -0.38* -0.39* -0.44* -0.41* -0.44* CD4+ cell^ 0.03* 0.03* 0.03* 0.03* Log VL Lag CD4+ cell^ 0.03* 0.03* 0.03* 0.03* Log Lag VL Random Effect# Intercept 331.39 312.51 318.18 332.86 331.96 312.22 321.27 305.73 Index 1.75 1.59 1.74 1.68 CD4+ cell^ 0.004 0.002 Log VL Lag CD4+ cell^ 0.003 0.0007 Log Lag VL Residual# 122.56 107.38 115.53 105.59 122.24 107.13 117.24 106.32 Fit Statistics LL -13909.52 -13848.84 -13889.82 -13844.13 -13905.95 -13845.29 -13894.39 -13844.11 -2LL 27819.04 27697.68 27779.64 27688.26 27811.90 27690.58 27788.78 27688.22 AIC 27829.03 27709.69 27791.64 27702.26 27821.89 27702.58 27800.77 27702.23 p<0.05 # variance ^ grand mean-centered LL = Log likelihood -2LL = -2 Log likelihood AIC = Akaike Information Criterion

PAGE 66

66Table 4-13 Models for PHC (continued) 12 13 14 15 16 17 18 19 Index & Log VL Index & Log VL Index & Log VL Index & Log VL Index & Log Lag VL Index & Log Lag VL Index & Log Lag VL Index & Log Lag VL Fixed Effect Intercept 78.40* 78.41*78.49*78.51*78.57*78.30*78.58*75.35* Index -0.26* -0.31*-0.26*-0.31*-0.28*-0.32*-0.29*-0.33* CD4+ cell^ Log VL -0.96* -0.89*-0.93*-0.87* Lag CD4+ cell^ Log Lag VL -0.98*-0.80*-0.90*-0.75* Random Effect# Intercept 349.24 327.57305.35276.50348.73327.18314.86280.96 Index 1.821.761.801.75 CD4+ cell^ Log VL 7.637.12 Lag CD4+ cell^ Log Lag VL 5.925.58 Residual# 122.52 106.97117.71103.44122.57107.13118.47104.48 Fit Statistics LL -13921.08 -13857.86-13903.09-13841.43 -13920.39-13858.41-13909.31-13849.05 -2LL 27842.16 27715.7227806.1827682.8627840.7827716.8227818.62-27698.10 AIC 27852.15 27727.7227818.1927696.8727850.7727828.8127830.6227712.10 p<0.05 # variance ^ grand mean-centered LL = Log likelihood -2LL = -2 Log likelihood AIC = Akaike Information Criterion

PAGE 67

67Table 4-14 Models for MHC 1 2 3 Null Index Index Fixed Effect Intercept 71.40* 72.32* 72.56* Index -0.16* -0.22* CD4+ cell^ Log VL Lag CD4+ cell^ Log Lag VL Random Effect# Intercept 364.35 365.72 329.97 Index 1.41 CD4+ cell^ Log VL Lag CD4+ cell^ Log Lag VL Residual# 158.95 158.65 147.41 Fit Statistics LL -14320.45 -14318.51 -14291.46 -2LL 28640.90 28637.02 28582.92 AIC 28646.90 28645.01 28592.93 p<0.05 # variance ^ grand mean-centered LL = Log likelihood -2LL = -2 Log likelihood AIC = Akaike Information Criterion

PAGE 68

68 Table 4-14 Models for MHC (continued) 4 5 6 7 8 9 10 11 Index & CD4+ Index & CD4+ Index & CD4+ Index & CD4+ Index & Lag CD4+ Index & Lag CD4+ Index & Lag CD4+ Index & Lag CD4+ Fixed Effect Intercept 72.58* 72.75* 72.66* 72.80* 72.73* 72.89* 72.74* Index -0.21* -0.25* -0.19* -0.24* -0.23* -0.27* -0.22* CD4+ cell^ 0.009 0.007 0.008 0.006 Log VL Lag CD4+ cell^ 0.01* 0.009 0.01 Log Lag VL Random Effect# Intercept 361.57 327.56 349.69 319.56 361.42 327.03 356.89 Index 1.40 1.37 1.39 CD4+ cell^ 0.002 0.001 Log VL Lag CD4+ cell^ 0.001 Log Lag VL Residual# 158.77 147.61 155.78 146.49 158.70 147.59 157.21 Fit Statistics LL -14317.07 -14290.65 -14314.00 -14289.78 -14316.31 -14289.98 -14315.46 -2LL 28634.14 28581.30 28628.00 28579.56 28633.62 28579.96 28630.82 AIC 28864.14 28593.30 28640.00 28593.56 28642.62 28591.96 28642.92 p<0.05 # variance ^ grand-mean centered LL = Log likelihood -2LL = -2 Log likelihood AIC = Akaike Information Criterion

PAGE 69

69Table 4-14 Models for MHC (continued) 12 13 14 15 16 17 18 19 Index & Log VL Index & Log VL Index & Log VL Index & Log VL Index & Log Lag VL Index & Log Lag VL Index & Log Lag VL Index & Log Lag VL Fixed Effect Intercept 74.83* 74.86* 74.96* 75.06* 74.50* 74.40* 74.63* 74.61* Index -0.20* -0.25* -0.20* -0.26* -0.21* -0.26* -0.21* -0.26* CD4+ cell^ Log VL -0.89* -0.82* -0.90* -0.86* Lag CD4+ cell^ Log Lag VL -0.73* -0.62* -0.75* -0.66* Random Effect# Intercept 360.54 325.35 323.26 275.49 361.55 326.53 335.14 291.79 Index 1.39 1.46 1.39 1.41 CD4+ cell^ Log VL 6.30 7.10 Lag CD4+ cell^ Log Lag VL 4.08 4.39 Residual# 158.56 147.48 154.61 143.22 158.61 147.57 155.86 144.90 Fit Statistics LL -14314.36 -14288.02 -14306.01 -14277.60 -14315.46 -14289.30 -14310.87 -14284.22 -2LL 28628.72 28576.04 28612.02 28555.20 28630.92 28578.60 28621.74 28568.44 AIC 28638.71 28588.04 28624.01 28569.19 28640.92 28590.60 28633.73 28582.44 p<0.05 # variance ^ grand mean-centered LL = Log likelihood -2LL = -2 Log likelihood AIC = Akaike Information Criterion

PAGE 70

70 Figure 4-1 Physical Health Component Sc ore after HAART Initiation of 10 patients

PAGE 71

71 Figure 4-2 Mental Health Component Score after HAART Initiation of 10 patients

PAGE 72

72 Figure 4-3 Average PHC score over time after HAART initiation

PAGE 73

73 Figure 4-4 Average MHC score over time after HAART initiation

PAGE 74

74 Figure 4-5 CD4+ cell count over time after HAART initiation for 10 patients

PAGE 75

75 Figure 4-6 Average CD4+ cell count over time since HAART initiation

PAGE 76

76 Figure 4-7 Log viral load over time after HAART initiation for 10 patients

PAGE 77

77 Figure 4-8 Average log viral loads over time after HAART initiation

PAGE 78

78 Figure 4-9 PHC residuals for model 2

PAGE 79

79 Figure 4-10 MHC residuals for model 2

PAGE 80

80 CHAPTER 5 DISCUSSION In this chapter, discussion about PHC and MH C trajectory is presented first, followed by a discussion of the CD4+ cell count and viral load trajectory, an d then the relationship between clinical lab test trajectory and QOL trajectory. This chapter also includes limitations, recommendation for future research, and a summary. QOL Trajectory PHC Trajectory Average PHC scores marginally decreased ev ery 6 months as shown in Figure 4-3. Although PHC statistically decreased over time (Tab le 4-13, model 2 and model 3), it decreased approximately 0.2 for every 6 months. This result was similar to the study conducted by Weinfurt and colleagues that found PHC declined over time (33). However in their study, PHC weekly decreased by 0.09. In other wo rds, it declined by 2.16 every 6 months. MHC Trajectory Average MHC scores also slig htly decreased every 6 months (Figure 4-4). Although MHC statistically decreased over time it decreased only 0.16 or 0.22 for every 6 months as shown in Table 4-14, model I and II respectively. This al so implied that HAART can clinically maintain mental health in HIV-infected patients. This re sult was similar to the finding from Weinfurt and colleagues study that showed MH C weekly declined by 0.09 or 2. 16 point every 6 months (33). Clinical Lab Test Trajectory CD4+ Cell Count Trajectory On average, CD4+ cell count increased over ti me as expected, which is consistent with many studies that showed that HAART helps to improve the immune system. This result was consistent with other previously conducted studi es (32, 36, 50-55). However, the finding from

PAGE 81

81 Binquet and friends showed that CD4+ cell coun t increased from baseline to 12 months only (50). After 12 months, CD4+ ce ll count was consistent. The re sult from Chu was slightly different in that CD4+ cell count increased from baseline to 24 m onths (32). The result from this study indicated the plateau period of CD4+ cell count at 48 months, which was similar to the finding from Moore and Ga rcias study (51, 52). Viral Load Trajectory It is not surprising that the average viral load decreased over time after taking HAART (Figure 4-9). Many studies show HAART suppresses viral load to an undetectable level (55, 56). The finding from Burgoyne indica ted that average viral load in patients who received HAART regimen decreased from 7,943 to 316 copies/mL in 4 years, whereas the result from Low-Beer showed a reduction from 11,000 to 499 copies/mL in one year. Although 5 NRTIs and 3 PIs were available during Low-Beers study period, patients included in Low-Beer study at least received PI. Therefore, whet her these patients re ceived HAART or not cannot be determined. Relationship between CD4+ Cell Trajectory and QOL Trajectory Change in CD4+ cell count can slightly predict change in PHC when time was controlled as shown in Table 4-13 (model 4 to model 11). The underlying r eason for this is because PHC decreased little by little over ti me. When there is less variatio n of PHC, CD4+ cell count cannot predict PHC well. However, the finding from th is study was consistent w ith the result from a longitudinal clinical trial where there was a pos itive relationship between change in PHC and change in CD4+ cell count (33). Comparing tim e-lag model with non time-lag model, the model fit statistics showed that time-lag model with ra ndom coefficients were better than non time-lag model with random coefficients. The result from this study conf irmed that CD4+ cell count had a positive relationship with MHC as with several other crosssectional and longitudinal studies (33, 57) that demonstrated a

PAGE 82

82 similar relationship. Comparing time-lag and non time-lag models, onl y lag CD4+ cell count significantly predicted MHC (Table 4-14, mode l 4 to model 10). However, the model fit statistics of non time-lag model were better than those of time-lag model. Relationship between Viral Load Trajectory and QOL Trajectory An inverse relationship between viral load and PHC in this st udy was similar to the result from Weinfurts study. Previous viral load also had an inverse re lationship with PHC. In other words, when viral load decreased, PHC increased However, the magnitude of the association between current viral load and previous viral lo ad and PHC were different as shown in Table 413. Based on model 12 to model 19 in Table 4-13, vi ral load should be fitted into the model with random effect. This implied that the effect of viral load on PH C was different from patient to patient. Viral load and previous viral load ha d an inverse relationship with MHC (Table 4-14, model 12 to model 19). That m eans when viral load increased, MHC decreased. However, viral load and previous viral load pred icted MHC in different magnitude. Limitations This study has several limitations. First, it is possible that patients were not only informed by physicians about their current CD4+ cell count and/or viral load, but also notified about the evaluation of that lab test value (e.g., whether it is good or bad). Patients may consider their lab test result as good/bad. Therefor e, dichotomizing or categorizing CD4+ cell count and viral load may possibly reflect the way patie nts used the lab test results to assess their QOL. Using a different scale of measurement of the independent variables (e.g., continuous vs. binary variable) may change the magnitude of the relationship be tween CD4+ cell count a nd viral load and QOL because when there is less variation of the independent when it was dichotomized, compared with when it was measured as a continuous va riable. However, we do not know how physicians

PAGE 83

83 evaluate CD4+ cell count and vira l load and what they told about CD4+ cell count and viral load to the patients. Qualitative research such as an interview may help clarify this. Second, it is possible that patients may not only assess their own current QOL about the questions in the MOS SF-36, but al so use other informa tion such as CD4+ cell count. If they use CD4+ cell count to assess their current QOL and th is information is not available, it is possible that patients will use previous lab test result to assess their current QOL. Once more, qualitative research by interviewing patients will help determine whether patie nts use lab test result from last clinic visit to asses their current QOL or not. Third, MACS is an observationa l study, collecting data from HIV-infected patients from 4 major cities, patients included in MACS visit clinic biannually only. Therefore, data from MACS are available to prove whether lag time at 6 months between eith er CD4+ cell count or viral load and QOL exists. It cannot be used to prove or va lidate whether lag time between CD4+ cell count/viral load and QOL is 3 or 2 mont hs. In other words, if the lag time between CD4+ cell count and QOL existed and was about 3 months, the time-lag model of 3 months would better predict QOL than a ti me-lag of 6 months or a non tim e-lag model. In other words, QOL measured at an improper timing is considered as an invalid measure of the effect of CD4+ cell count. Fourth, HIV-related symptoms (e.g., fatigue, pa in and diarrhea) were not included in this study. Those symptoms were reported as the si gnificant predictors in QOL in HIV-infected patients (29, 30, 58-61). For those studies, only the study by J ohnson et al. included patients treated with HAART and only tw o are longitudinal studies with an aim to investigate the longitudinal relationship between HIV-related sy mptoms and QOL (29, 59). Including HIVinfected symptoms in the model will possibly help explain change in QOL over time.

PAGE 84

84 Fifth the effect of HIV-infected symptoms on QOL could change over time (33). It is possible that patients get used to HIV-related symptoms over time, so the way patients assessed their QOL regarding to symptoms may changed over time. For example, when asymptomatic patients became symptomatic, symptoms such as diarrhea may consid erably bother them. However, when they were tolerated to diarrh ea, the way they assess how diarrhea effect QOL changed. Sixth, this study included all patients who were on HAART. Nav e and previous HAART users were not separately analyzed. In the most recent study, patients included had no prior HAART experiences (62). Both PHC and MHC incr eased over time which was contradict to the finding from this study. However, the resear chers found a positive relationship between CD4+ cell count and QOL as in this study. This study compared time-lag and non time-lag models only in HIV-in fected patients. HIV is one of many chronic diseases where la b test results are important for determining patients condition and clinical stat us. Lab test results also help physician to determine when to start treatment as stated in disease guidelines. Lab test results are also important to patients because it help patients monito r their own condition. This study investigated the lag time between clinical lab tests and QOL in HIV-in fected patients who were on HAART only. The results from this study can not be applied to thos e with other chronic diseases. Chronic diseases that the lag time between clinical lab re sults and QOL might exist include diabetes, hyperlipidemia. For these diseases, symptoms do not play an important role in assessing QOL. Patients with high LDL can do regular activities (e.g., walk 1 to 2 miles). However, when they do know that the LDL level was t oo high for two previous consecu tive weeks, they might use

PAGE 85

85 those lab tests results to assess their current QOL because they didnt know the current lab test result. Future Research This study has two different implications fo r future research; (1) QOL measurement and (2) clinical application. Re searchers interested in QOL m easurement should investigate the timing of QOL measurement because patients may differ in their QOL assessments depending upon whether they know their previous or current clinical lab test resu lts. In other words, researchers can compare QOL measured when pa tients had a clinic visit and QOL measured a week later when patients knew th eir current lab test results. It is also recommended that other databases be used to establish the lag time be tween CD4+ cell counts, viral load and QOL in HIV-infected patients in order to confirm the finding from this study. For clinical application, if the lag time relationship between ei ther viral load or CD4+ cell count and QOL truly exists (e.g., 3 months), it is recommended that patients be monitored their QOL at least every 3 months. Conclusion PHC and MHC in HIV-infected patients who were on HAART slightly decreased over time. The change in viral load over time signi ficantly predicts change in PHC and MHC over time, whereas the change in CD4+ cell count sign ificantly predicts PHC over time only. CD4+ cell count has a positive longitudi nal relationship with PHC, wher eas viral load has a negative longitudinal relationship with bot h PHC and MHC. Overall, timelag models were not different from non time-lag models in terms of the mode l fit statistics and re gression coefficients.

PAGE 86

86 APPENDIX A SQL SYNTAX FOR HAART SELECT DISTINCT DRUGF1.CASEID DRUGF1.VISIT, DRUGF1.AVQY, Switch(C.CASEID,"Y",DRUGF1.CASEID, "N") AS HAART FROM DRUGF1 LEFT JOIN [SELECT CASEID, VISIT FROM ( SELECT CASEID, VISIT, 1 AS FLAG FROM ( SELECT CASEID, VISIT FROM (SELECT CASEID,VISIT FROM DRUGF1 INNER JOIN Drug_G rp_Code ON DRUGF1.DRGAV = Drug_Grp_Code.Drug_Code WHERE DRUG_GRP = "NRTI" GROUP BY CASEID, VISIT HAVING COUNT(*) >=2 ) AS A WHERE EXISTS ( SELECT 1 FROM DRUGF1 INNER JOIN Dr ug_Grp_Code ON DRUGF1.DRGAV = Drug_Grp_Code.Drug_Code WHERE A.CASEID = DRUGF1.CASEID AND A.VISIT = DRUGF1.VISIT AND (DRUG_GRP = "NNRTI" OR DRUG_GRP = "PI") ) UNION SELECT CASEID, VISIT FROM ( SELECT DISTINCT CASEID,VISIT FROM DRUGF1 INNER JOIN Dr ug_Grp_Code ON DRUGF1.DRGAV = Drug_Grp_Code.Drug_Code WHERE DRUG_GRP = "NRTI" UNION ALL SELECT DISTINCT CASEID,VISIT FROM DRUGF1 INNER JOIN Dr ug_Grp_Code ON DRUGF1.DRGAV = Drug_Grp_Code.Drug_Code WHERE DRUG_GRP = "NNRTI" UNION ALL SELECT DISTINCT CASEID,VISIT FROM DRUGF1 INNER JOIN Dr ug_Grp_Code ON DRUGF1.DRGAV = Drug_Grp_Code.Drug_Code

PAGE 87

87 WHERE DRUG_GRP = "PI" ) AS A GROUP BY CASEID, VISIT HAVING COUNT(*) = 3 UNION SELECT CASEID, VISIT FROM ( SELECT CASEID,VISIT FROM ( SELECT DISTINCT CASEID,VISIT FROM DRUGF1 INNER JOIN Drug_Grp_Code ON DRUGF1.DRGAV = Drug_Grp_Code.Drug_Code WHERE DRUG_CODE = 210 UNION ALL SELECT DISTINCT CASEID,VISIT FROM DRUGF1 INNER JOIN Drug_Grp_Code ON DRUGF1.DRGAV = Drug_Grp_Code.Drug_Code WHERE DRUG_CODE = 211 UNION ALL SELECT DISTINCT CASEID,VISIT FROM DRUGF1 INNER JOIN Drug_Grp_Code ON DRUGF1.DRGAV = Drug_Grp_Code.Drug_Code WHERE DR UG_GRP = "NRTI" ) AS B GROUP BY CASEID,VISIT HAVING COUNT(*) = 3 ) AS A WHERE NOT EXISTS ( SELECT 1 FROM DRUGF1 INNER JOIN Drug_Grp_Code ON DRUGF1.DRGAV = Drug_Grp_Code.Drug_Code WHERE A.CASEID = DRUGF1.CASEI D AND A.VISIT = DRUGF1.VISIT AND DRUG_GRP = "NNRTI" ) UNION SELECT CASEID, VISIT FROM ( SELECT CASEID,VISIT FROM ( SELECT DISTINCT CASEID,VISIT

PAGE 88

88 FROM DRUGF1 WHERE DRGAV = 218 ) AS A WHERE EXISTS ( SELECT 1 FROM DRUGF1 INNER JOIN Drug_Grp_Code ON DRUGF1.DRGAV = Drug_Grp_Code.Drug_Code WHERE DRUG_GRP = "NRTI" AND DRUGF1.CASEID = A.CASEID AND DRUGF1.VISIT = A.VISIT GROUP BY DRUGF1.CASEID,DRUGF1.VISIT HAVING COUNT(*) >= 4 ) ) AS B WHERE NOT EXISTS ( SELECT 1 FROM DRUGF1 INNER JOIN Dr ug_Grp_Code ON DRUGF1.DRGAV = Drug_Grp_Code.Drug_Code WHERE (DRUG_GRP = "NNRTI" OR DRUG_GRP = "PI") AND DRUGF1.CASEID = B.CASEID AND DRUGF1.VISIT = B.VISIT ) ) ) GROUP BY CASEID,VISIT HAVING COUNT(*) = 1 AND MAX(FLAG) = 1 ]. AS C ON (DRUGF1.VISIT = C.VISIT) AND (DRUGF1.CASEID = C.CASEID);

PAGE 89

89 LIST OF REFERENCES 1. Twenty-five years of HIV/AIDS--Unite d States, 1981-2006. Morb Mortal Wkly Rep 2006;55:585-589 2. Dybul M, Fauci AS, Bartlett JG, et al. Guidelines for usi ng antiretroviral agents among HIV-infected adults and adolesce nts. Ann Intern Med 2002;137:381-433 3. Sax PE, Gathe JC, Jr. Beyond efficacy: the im pact of combination antiretroviral therapy on quality of life. AIDS Patie nt Care STDS 2005;19:563-576 4. Wilson IB, Cleary PD. Linking clinical variables with health-related quality of life. A conceptual model of patient outcomes. JAMA 1995;273:59-65 5. Jia H, Uphold CR, Wu S, et al. Predictors of changes in health -related quality of life among men with HIV infection in the HAART era. AIDS Patient Care STDS 2005;19:395-405 6. Jia H, Uphold CR, Zheng Y, et al. A further investigation of health-related quality of life over time among men with HIV infection in the HAART era. Qual Life Res 2007 7. Pneumocystis pneumonia--Los Angeles. 1981. MMWR Morb Mortal Wkly Rep 1996;45:729-733 8. Pneumocystis pneumonia--Los Angeles. MMWR Morb Mortal Wkly Rep 1981;30:250252 9. Estimates of HIV prevalence and project ed AIDS cases: summary of a workshop, October 31-November 1, 1989. MMWR Mo rb Mortal Wkly Rep 1990;39:110-112, 117119 10. McNicholl I. Adverse Events of Antiret roviral Drugs/July 2006. 2006. Available at: http://hivinsite.ucsf.edu/pdf/ ar-05-01.pdf. Accessed 01/25, 2007 11. Shibuyama S, Gevorkyan A, Yoo U, et al. Understanding and a voiding antiretroviral adverse events. Curr Ph arm Des 2006;12:1075-1090 12. Ferrans CE, Zerwic JJ, Wilbur JE, et al. Conceptual model of h ealth-related quality of life. J Nurs Scholarsh 2005;37:336-342 13. Sousa KH, Kwok OM. Putting Wilson and Cl eary to the test: analysis of a HRQOL conceptual model using stru ctural equation modeling. Qual Life Res 2006;15:725-737 14. Chrischilles EA, Rubenstein LM, Voelker MD, et al. Linking c linical variables to healthrelated quality of lif e in Parkinson's disease. Park insonism Relat Disord 2002;8:199-209

PAGE 90

90 15. Sousa KH, Holzemer WL, Henr y SB, et al. Dimensions of h ealth-related quality of life in persons living with HIV dis ease. J Adv Nurs 1999;29:178-187 16. Wilson IB, Cleary PD. Clinical predictors of functioning in persons with acquired immunodeficiency syndrome Med Care 1996;34:610-623 17. Wettergren L, Bjorkholm M, Axdorph U, et al Determinants of health-related quality of life in long-term survivor s of Hodgkin's lymphoma. Qu al Life Res 2004;13:1369-1379 18. Vidrine DJ, Amick BC, 3rd, Gritz ER, et al Assessing a conceptual framework of healthrelated quality of life in a HIV/AIDS population. Qual Life Res 2005;14:923-933 19. Bing EG, Hays RD, Jacobson LP, et al. Heal th-related quality of life among people with HIV disease: results from the Multicenter AIDS Cohort Study. Qual Life Res 2000;9:5563 20. Call SA, Klapow JC, Stewart KE, et al. H ealth-related quality of life and virologic outcomes in an HIV clinic. Qual Life Res 2000;9:977-985 21. Carrieri P, Spire B, Duran S, et al. Health -related quality of life after 1 year of highly active antiretroviral ther apy. J Acquir Immune Defic Syndr 2003;32:38-47 22. Casado A, Badia X, Consiglio E, et al. Health-related quality of life in HIV-infected naive patients treated with nelfinavir or nevirapine associated with ZDV/3TC (the COMBINE-QoL substudy). HIV Clin Trials 2004;5:132-139 23. Nieuwkerk PT, Gisolf EH, Colebunders R, et al. Quality of life in asymptomaticand symptomatic HIV infected patients in a tr ial of ritonavir/saqu inavir therapy. The Prometheus Study Group. AIDS 2000;14:181-187 24. Liu C, Ostrow D, Detels R, et al. Impact s of HIV infection and HAART use on quality of life. Qual Life Res 2006;15:941-949 25. Brechtl JR, Breitbart W, Galietta M, et al. The use of highly active antiretroviral therapy (HAART) in patients with advanced HIV infection: impact on medical, palliative care, and quality of life outcomes. J Pain Symptom Manage 2001;21:41-51 26. Gill CJ, Griffith JL, Jacobson D, et al. Re lationship of HIV viral loads, CD4 counts, and HAART use to health-related quality of life. J Acquir Immune Defic Syndr 2002;30:485492 27. Globe DR, Hays RD, Cunningha m WE. Associations of clini cal parameters with healthrelated quality of lif e in hospitalized pers ons with HIV disease. AIDS Care 1999;11:7186

PAGE 91

91 28. Eriksson LE, Nordstrom G, Berglund T, et al. The health-related quality of life in a Swedish sample of HIV-infected persons. J Adv Nurs 2000;32:1213-1223 29. Preau M, Leport C, Salmon-Ceron D, et al Health-related quality of life and patientprovider relationships in HIV-in fected patients during the first three years after starting PI-containing antiretroviral trea tment. AIDS Care 2004;16:649-661 30. Justice AC, Chang CH, Rabeneck L, et al. Clinical importance of provider-reported HIV symptoms compared with patient -report. Med Care 2001;39:397-408 31. Clarke PS. Causal analysis of individual change using th e difference score. Epidemiology 2004;15:414-421 32. Chu H, Gange SJ, Yamashita TE, et al. Indi vidual variation in CD4 cell count trajectory among human immunodeficiency virus-infect ed men and women on long-term highly active antiretroviral th erapy: an application using a Bayesian random change-point model. Am J Epidemiol 2005;162:787-797 33. Weinfurt KP, Willke RJ, Glick HA, et al. Relationship between CD4 count, viral burden, and quality of life over time in HIV-1infected patients. Med Care 2000;38:404-410 34. Liu C, Johnson L, Ostrow D, et al. Predic tors for Lower Quality of Life in the HAART Era Among HIV-Infected Men. J Acquir Immune Defic Syndr 2006 35. Protopopescu C, Marcellin F, Spire B, et al. Health-related quality of life in HIV-1infected patients on HAART: a five-years l ongitudinal analysis acc ounting for dropout in the APROCO-COPILOTE cohort (ANRS CO-8). Qual Life Res 2007;16:577-591 36. Anastos K, Barron Y, Cohen MH, et al. Th e prognostic importance of changes in CD4+ cell count and HIV-1 RNA level in women af ter initiating highly active antiretroviral therapy. Ann Intern Med 2004;140:256-264 37. Raghunathan TE. What do we do with missi ng data? Some options for analysis of incomplete data. Annual review of public health 2004;25:99-117 38. Schafer JL, Graham JW. Missing data: our vi ew of the state of the art. Psychological methods 2002;7:147-177 39. Hogan JW, Roy J, Korkontzel ou C. Handling drop-out in lo ngitudinal studies. Stat Med 2004;23:1455-1497 40. Twisk J, de Vente W. Attrition in longitudi nal studies. How to deal with missing data. J Clin Epidemiol 2002;55:329-337 41. Kristman VL, Manno M, Cote P. Methods to account for attrition in longitudinal data: do they work? A simulation study. Eur J Epidemiol 2005;20:657-662

PAGE 92

92 42. Ironson G, O'Cleirigh C, Flet cher MA, et al. Psychosocial factors predict CD4 and viral load change in men and women with human im munodeficiency virus in the era of highly active antiretroviral treatment. Ps ychosomatic medicine 2005;67:1013-1021 43. Wensheng G, Sarah JR, Thomas TTH. A Random Pattern-Mixture Model for Longitudinal Data With Dropouts. Journal of the American Sta tistical Association 2004;99:929 44. Kenward MG, Molenberghs G. Parametric models for incomplete continuous and categorical longitudinal data Statistical methods in me dical research 1999;8:51-83 45. Crystal S, Sambamoorthi U. Functional impairment trajectories among persons with HIV disease: a hierarchical linear models approach. Health Serv Res 1996;31:469-488 46. Wu YW, Clopper RR, Wooldridge PJ. A comparison of traditional approaches to hierarchical linear modeling when analyz ing longitudinal data Res Nurs Health 1999;22:421-432 47. Nich C, Carroll K. Now you see it, now you don't: a comparison of traditional versus random-effects regression models in the analysis of longit udinal follow-up data from a clinical trial. J Consu lt Clin Psychol 1997;65:252-261 48. Wu YW. An application of hi erarchical linear models to longitudinal studies. Res Nurs Health 1996;19:75-82 49. O'Connell AA, McCoach DB. Applications of hierarchical linear models for evaluations of health interventions: demystifying the methods and interpretations of multilevel models. Eval Health Prof 2004;27:119-151 50. Binquet C, Chene G, Jacqmin-Gadda H, et al. Modeling changes in CD4-positive Tlymphocyte counts after the start of highly ac tive antiretroviral ther apy and the relation with risk of opportunistic infections: the Aquitaine Cohort, 19961997. Am J Epidemiol 2001;153:386-393 51. Garcia F, de Lazzari E, Plana M, et al. Long-term CD4+ T-cell response to highly active antiretroviral therapy accordi ng to baseline CD4+ T-cell c ount. J Acquir Immune Defic Syndr 2004;36:702-713 52. Moore RD, Keruly JC, Gebo KA, et al. An improvement in virologic response to highly active antiretroviral therapy in clinical practice from 1996 through 2002. J Acquir Immune Defic Syndr 2005;39:195-198 53. Moore RD, Keruly JC. CD4+ cell count 6 years after commencement of highly active antiretroviral therapy in pers ons with sustained virologic suppression. Clin Infect Dis 2007;44:441-446

PAGE 93

93 54. Viard JP, Burgard M, Hubert JB, et al. Im pact of 5 years of maximally successful highly active antiretroviral therapy on CD4 cell count and HI V-1 DNA level. AIDS 2004;18:4549 55. Low-Beer S, Chan K, Wood E, et al. Hea lth related quality of life among persons with HIV after the use of protease inhi bitors. Qual Life Res 2000;9:941-949 56. Burgoyne RW, Rourke SB, Behrens DM, et al. Long-term quality-of-life outcomes among adults living with HIV in the HAART er a: the interplay of changes in clinical factors and symptom prof ile. AIDS Behav 2004;8:151-163 57. Ruiz Perez I, Rodriguez Bano J, Lopez Ruz MA, et al. Health-related quality of life of patients with HIV: impact of sociodemographic, clinical a nd psychosocial factors. Qual Life Res 2005;14:1301-1310 58. Johnson MO, Stallworth T, Neilands TB. The dr ugs or the disease? Causal attributions of symptoms held by HIV-positive adu lts on HAART. AIDS Behav 2003;7:109-117 59. Lorenz KA, Cunningham WE, Spritzer KL, et al. Change s in symptoms and healthrelated quality of life in a nationally representa tive sample of adults in treatment for HIV. Qual Life Res 2006;15:951-958 60. Lorenz KA, Shapiro MF, Asch SM, et al. Associations of symptoms and health-related quality of life: findings from a national study of persons with HIV in fection. Ann Intern Med 2001;134:854-860 61. Tramarin A, Parise N, Campostrini S, et al. Association between di arrhea and quality of life in HIV-infected patients receiving highly ac tive antiretroviral therapy. Qual Life Res 2004;13:243-250 62. Stangl AL, Wamai N, Mermin J, et al. Tr ends and predictors of quality of life among HIV-infected adults taking hi ghly active antiretroviral therapy in rural Uganda. AIDS Care 2007;19:626-636

PAGE 94

94 BIOGRAPHICAL SKETCH Sawaeng Watcharathanakij is an assistant professor at Faculty of Pharmacy, Ubon Ratchathani Universityn (UBU), Thailand. He wa s born and raised in Chiangmai province. He earned Bachelor of Pharmacy from Mahidol Unversity (MU). He worked as a hospital pharmacist at Ramathibodi Hospital. Then he pursued graduate study and earned Master of Pharmacy from Mahidol University and worked as faculty at Faculty of Pharmacy, Ubon Ratchathani University before joining the docto ral program at the Department of Pharmacy Health Care Administration, Un iversity of Florida in 2002.