Limited English Language Proficiency and Access to Health Care among U.S. Latinos

Permanent Link: http://ufdc.ufl.edu/UFE0022373/00001

Material Information

Title: Limited English Language Proficiency and Access to Health Care among U.S. Latinos
Physical Description: 1 online resource (64 p.)
Language: english
Creator: Slazinski, Karl
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2008


Subjects / Keywords: brfss, health, insurance, language, latino, proficiency
Latin American Studies -- Dissertations, Academic -- UF
Genre: Latin American Studies thesis, M.A.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation


Abstract: Numerous case studies deal with the effects of immigration status and race on access to health insurance among the Hispanic populations in the United States. However, few of them deal with English language proficiency as the main causal variable. Thus, the main objective of this study is to determine if Limited English Proficiency (LEP) is the primary or a contributing variable in determining health insurance status. The method I use to determine the effect of being an LEP Latino versus an English-speaking Latino on health insurance rates is logistic regression. The data are provided by the CDC in the form of the Behavioral Risk Factor Surveillance System. The BRFSS is the largest, ongoing, random-digit-dial telephone survey of health status and access in the United States, Puerto Rico, and Guam. It randomly selects from the population of noninstitutionalized adults 18 years of age and older. The weighted sample controls for multiple phone lines and non-response biases. Included in the sample were 18,147 mainland Latino individuals, 21,368 mainland and Puerto Rican Latino individuals and 275,491 white individuals. Respondents were permitted to answer the questions in English or Spanish. For the purposes of this study, I used the question of language choice as a proxy for dominant language. The individuals who chose to respond in Spanish were classified as people who were limited English proficient. Logistic regression analyses were performed to determine if, after controlling for years of education, sex, age, and income, English language proficiency still accounted for a substantial effect on health plan enrollment. Results showed that mainland and all Latinos who responded in English were 2.682 and 1.446 times more likely to have health insurance, respectively. Therefore, the logistic regression shows that English language proficiency is one factor in determining if Latinos have access to health insurance.
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 Karl Slazinski.
Thesis: Thesis (M.A.)--University of Florida, 2008.
Local: Adviser: Coady, Maria R.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2008
System ID: UFE0022373:00001

Permanent Link: http://ufdc.ufl.edu/UFE0022373/00001

Material Information

Title: Limited English Language Proficiency and Access to Health Care among U.S. Latinos
Physical Description: 1 online resource (64 p.)
Language: english
Creator: Slazinski, Karl
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2008


Subjects / Keywords: brfss, health, insurance, language, latino, proficiency
Latin American Studies -- Dissertations, Academic -- UF
Genre: Latin American Studies thesis, M.A.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation


Abstract: Numerous case studies deal with the effects of immigration status and race on access to health insurance among the Hispanic populations in the United States. However, few of them deal with English language proficiency as the main causal variable. Thus, the main objective of this study is to determine if Limited English Proficiency (LEP) is the primary or a contributing variable in determining health insurance status. The method I use to determine the effect of being an LEP Latino versus an English-speaking Latino on health insurance rates is logistic regression. The data are provided by the CDC in the form of the Behavioral Risk Factor Surveillance System. The BRFSS is the largest, ongoing, random-digit-dial telephone survey of health status and access in the United States, Puerto Rico, and Guam. It randomly selects from the population of noninstitutionalized adults 18 years of age and older. The weighted sample controls for multiple phone lines and non-response biases. Included in the sample were 18,147 mainland Latino individuals, 21,368 mainland and Puerto Rican Latino individuals and 275,491 white individuals. Respondents were permitted to answer the questions in English or Spanish. For the purposes of this study, I used the question of language choice as a proxy for dominant language. The individuals who chose to respond in Spanish were classified as people who were limited English proficient. Logistic regression analyses were performed to determine if, after controlling for years of education, sex, age, and income, English language proficiency still accounted for a substantial effect on health plan enrollment. Results showed that mainland and all Latinos who responded in English were 2.682 and 1.446 times more likely to have health insurance, respectively. Therefore, the logistic regression shows that English language proficiency is one factor in determining if Latinos have access to health insurance.
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 Karl Slazinski.
Thesis: Thesis (M.A.)--University of Florida, 2008.
Local: Adviser: Coady, Maria R.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2008
System ID: UFE0022373:00001

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 E20101203_AAAABS INGEST_TIME 2010-12-03T12:35:58Z PACKAGE UFE0022373_00001
24685 F20101203_AAASDT slazinski_k_Page_16.QC.jpg
6970 F20101203_AAASNN slazinski_k_Page_16thm.jpg
89408 F20101203_AAASIR slazinski_k_Page_29.jpg
18906 F20101203_AAASDU slazinski_k_Page_18.QC.jpg
27502 F20101203_AAASNO slazinski_k_Page_26.QC.jpg
85859 F20101203_AAASIS slazinski_k_Page_30.jpg
26193 F20101203_AAASDV slazinski_k_Page_46.QC.jpg
6922 F20101203_AAASNP slazinski_k_Page_44thm.jpg
86811 F20101203_AAASIT slazinski_k_Page_31.jpg
23325 F20101203_AAASDW slazinski_k_Page_52.jpg
7177 F20101203_AAASNQ slazinski_k_Page_58thm.jpg
84049 F20101203_AAASIU slazinski_k_Page_34.jpg
86348 F20101203_AAASDX slazinski_k_Page_39.jpg
5279 F20101203_AAASNR slazinski_k_Page_18thm.jpg
83169 F20101203_AAASIV slazinski_k_Page_35.jpg
28510 F20101203_AAASDY slazinski_k_Page_13.QC.jpg
27824 F20101203_AAASNS slazinski_k_Page_31.QC.jpg
83060 F20101203_AAASIW slazinski_k_Page_37.jpg
7444 F20101203_AAASNT slazinski_k_Page_20thm.jpg
86708 F20101203_AAASIX slazinski_k_Page_38.jpg
25271604 F20101203_AAASDZ slazinski_k_Page_40.tif
28087 F20101203_AAASNU slazinski_k_Page_14.QC.jpg
87847 F20101203_AAASIY slazinski_k_Page_40.jpg
6882 F20101203_AAASNV slazinski_k_Page_17thm.jpg
53130 F20101203_AAASIZ slazinski_k_Page_41.jpg
9262 F20101203_AAASNW slazinski_k_Page_06.QC.jpg
2304 F20101203_AAASGA slazinski_k_Page_01thm.jpg
7541 F20101203_AAASNX slazinski_k_Page_40thm.jpg
27781 F20101203_AAASGB slazinski_k_Page_38.QC.jpg
1051972 F20101203_AAASLA slazinski_k_Page_62.jp2
7366 F20101203_AAASNY slazinski_k_Page_59thm.jpg
27743 F20101203_AAASGC slazinski_k_Page_40.QC.jpg
2958 F20101203_AAASNZ slazinski_k_Page_10thm.jpg
7644 F20101203_AAASGD slazinski_k_Page_25thm.jpg
253783 F20101203_AAASLB slazinski_k_Page_63.jp2
1051974 F20101203_AAASGE slazinski_k_Page_40.jp2
F20101203_AAASLC slazinski_k_Page_01.tif
6378 F20101203_AAASGF slazinski_k_Page_08thm.jpg
F20101203_AAASLD slazinski_k_Page_02.tif
533391 F20101203_AAASGG slazinski_k_Page_55.jp2
F20101203_AAASLE slazinski_k_Page_04.tif
32120 F20101203_AAASGH slazinski_k_Page_32.jpg
F20101203_AAASLF slazinski_k_Page_05.tif
4029 F20101203_AAASGI slazinski_k_Page_55thm.jpg
F20101203_AAASLG slazinski_k_Page_07.tif
F20101203_AAASGJ slazinski_k_Page_63.tif
F20101203_AAASLH slazinski_k_Page_08.tif
7084 F20101203_AAASGK slazinski_k_Page_35thm.jpg
F20101203_AAASLI slazinski_k_Page_10.tif
1051961 F20101203_AAASGL slazinski_k_Page_50.jp2
F20101203_AAASLJ slazinski_k_Page_11.tif
11205 F20101203_AAASGM slazinski_k_Page_53.QC.jpg
F20101203_AAASLK slazinski_k_Page_12.tif
1051958 F20101203_AAASGN slazinski_k_Page_21.jp2
F20101203_AAASLL slazinski_k_Page_15.tif
5845 F20101203_AAASGO slazinski_k_Page_51thm.jpg
F20101203_AAASLM slazinski_k_Page_18.tif
24538 F20101203_AAASGP slazinski_k_Page_33.QC.jpg
F20101203_AAASLN slazinski_k_Page_20.tif
77230 F20101203_AAASGQ slazinski_k_Page_33.jpg
F20101203_AAASLO slazinski_k_Page_21.tif
21345 F20101203_AAASGR slazinski_k_Page_05.QC.jpg
F20101203_AAASLP slazinski_k_Page_22.tif
884625 F20101203_AAASGS slazinski_k_Page_51.jp2
F20101203_AAASLQ slazinski_k_Page_23.tif
7496 F20101203_AAASGT slazinski_k_Page_48thm.jpg
F20101203_AAASLR slazinski_k_Page_24.tif
85308 F20101203_AAASGU slazinski_k_Page_47.jpg
F20101203_AAASLS slazinski_k_Page_25.tif
5686 F20101203_AAASGV slazinski_k_Page_05thm.jpg
F20101203_AAASLT slazinski_k_Page_26.tif
26215 F20101203_AAASGW slazinski_k_Page_19.QC.jpg
F20101203_AAASLU slazinski_k_Page_27.tif
27044 F20101203_AAASGX slazinski_k_Page_30.QC.jpg
F20101203_AAASLV slazinski_k_Page_29.tif
7348 F20101203_AAASGY slazinski_k_Page_26thm.jpg
F20101203_AAASLW slazinski_k_Page_30.tif
7139 F20101203_AAASEA slazinski_k_Page_15thm.jpg
25752 F20101203_AAASGZ slazinski_k_Page_49.QC.jpg
F20101203_AAASLX slazinski_k_Page_32.tif
505710 F20101203_AAASEB slazinski_k_Page_64.jp2
F20101203_AAASLY slazinski_k_Page_33.tif
87009 F20101203_AAASEC slazinski_k_Page_22.jpg
7254 F20101203_AAASED slazinski_k_Page_28thm.jpg
33396 F20101203_AAASJA slazinski_k_Page_42.jpg
F20101203_AAASLZ slazinski_k_Page_35.tif
F20101203_AAASEE slazinski_k_Page_28.tif
80582 F20101203_AAASJB slazinski_k_Page_44.jpg
F20101203_AAASEF slazinski_k_Page_38.tif
27118 F20101203_AAASOA slazinski_k_Page_36.QC.jpg
83083 F20101203_AAASJC slazinski_k_Page_45.jpg
1051909 F20101203_AAASEG slazinski_k_Page_46.jp2
27951 F20101203_AAASOB slazinski_k_Page_24.QC.jpg
86679 F20101203_AAASJD slazinski_k_Page_48.jpg
8029 F20101203_AAASEH slazinski_k_Page_52.QC.jpg
28312 F20101203_AAASOC slazinski_k_Page_29.QC.jpg
83195 F20101203_AAASJE slazinski_k_Page_49.jpg
10349 F20101203_AAASEI slazinski_k_Page_04.QC.jpg
16665 F20101203_AAASOD slazinski_k_Page_41.QC.jpg
64892 F20101203_AAASJF slazinski_k_Page_51.jpg
26501 F20101203_AAASEJ slazinski_k_Page_35.QC.jpg
36990 F20101203_AAASJG slazinski_k_Page_53.jpg
42366 F20101203_AAASEK slazinski_k_Page_43.jpg
3711 F20101203_AAASOE slazinski_k_Page_43thm.jpg
37355 F20101203_AAASJH slazinski_k_Page_54.jpg
F20101203_AAASEL slazinski_k_Page_13.tif
6918 F20101203_AAASOF slazinski_k_Page_46thm.jpg
43610 F20101203_AAASJI slazinski_k_Page_55.jpg
10517 F20101203_AAASOG slazinski_k_Page_32.QC.jpg
86392 F20101203_AAASJJ slazinski_k_Page_57.jpg
22546 F20101203_AAASEM slazinski_k_Page_07.jpg
9287 F20101203_AAASOH slazinski_k_Page_10.QC.jpg
86269 F20101203_AAASJK slazinski_k_Page_58.jpg
781298 F20101203_AAASEN slazinski_k_Page_18.jp2
12049 F20101203_AAASOI slazinski_k_Page_42.QC.jpg
45135 F20101203_AAASJL slazinski_k_Page_60.jpg
F20101203_AAASEO slazinski_k_Page_34.tif
7332 F20101203_AAASOJ slazinski_k_Page_47thm.jpg
40636 F20101203_AAASJM slazinski_k_Page_64.jpg
77031 F20101203_AAASEP slazinski_k_Page_61.jpg
4822 F20101203_AAASOK slazinski_k_Page_41thm.jpg
235684 F20101203_AAASJN slazinski_k_Page_01.jp2
27980 F20101203_AAASEQ slazinski_k_Page_22.QC.jpg
23172 F20101203_AAASOL slazinski_k_Page_08.QC.jpg
24390 F20101203_AAASJO slazinski_k_Page_02.jp2
F20101203_AAASER slazinski_k_Page_55.tif
12845 F20101203_AAASOM slazinski_k_Page_43.QC.jpg
103214 F20101203_AAASJP slazinski_k_Page_03.jp2
1051908 F20101203_AAASES slazinski_k_Page_49.jp2
6800 F20101203_AAASON slazinski_k_Page_07.QC.jpg
404255 F20101203_AAASJQ slazinski_k_Page_04.jp2
1051981 F20101203_AAASET slazinski_k_Page_24.jp2
6609 F20101203_AAASOO slazinski_k_Page_63.QC.jpg
791238 F20101203_AAASJR slazinski_k_Page_06.jp2
4279 F20101203_AAASEU slazinski_k_Page_42thm.jpg
28696 F20101203_AAASOP slazinski_k_Page_23.QC.jpg
423543 F20101203_AAASJS slazinski_k_Page_07.jp2
78202 F20101203_AAASEV slazinski_k_Page_09.jpg
1774 F20101203_AAASOQ slazinski_k_Page_03thm.jpg
1051971 F20101203_AAASJT slazinski_k_Page_08.jp2
F20101203_AAASEW slazinski_k_Page_06.tif
26612 F20101203_AAASOR slazinski_k_Page_20.QC.jpg
314116 F20101203_AAASJU slazinski_k_Page_10.jp2
27033 F20101203_AAASEX slazinski_k_Page_50.QC.jpg
7449 F20101203_AAASOS slazinski_k_Page_14thm.jpg
F20101203_AAASJV slazinski_k_Page_11.jp2
F20101203_AAASEY slazinski_k_Page_64.tif
7716 F20101203_AAASOT slazinski_k_Page_23thm.jpg
1051956 F20101203_AAASJW slazinski_k_Page_14.jp2
344364 F20101203_AAASEZ slazinski_k_Page_42.jp2
72682 F20101203_AAASOU UFE0022373_00001.xml FULL
1051966 F20101203_AAASJX slazinski_k_Page_16.jp2
1051954 F20101203_AAASJY slazinski_k_Page_17.jp2
3209 F20101203_AAASOV slazinski_k_Page_04thm.jpg
1051952 F20101203_AAASHA slazinski_k_Page_23.jp2
F20101203_AAASJZ slazinski_k_Page_19.jp2
27352 F20101203_AAASOW slazinski_k_Page_11.QC.jpg
26891 F20101203_AAASHB slazinski_k_Page_56.QC.jpg
27221 F20101203_AAASOX slazinski_k_Page_12.QC.jpg
83177 F20101203_AAASHC slazinski_k_Page_46.jpg
F20101203_AAASMA slazinski_k_Page_36.tif
7540 F20101203_AAASOY slazinski_k_Page_12thm.jpg
1051983 F20101203_AAASHD slazinski_k_Page_57.jp2
F20101203_AAASMB slazinski_k_Page_42.tif
28733 F20101203_AAASOZ slazinski_k_Page_25.QC.jpg
F20101203_AAASHE slazinski_k_Page_41.tif
23633 F20101203_AAASHF slazinski_k_Page_62.QC.jpg
F20101203_AAASMC slazinski_k_Page_43.tif
22426 F20101203_AAASHG slazinski_k_Page_61.QC.jpg
F20101203_AAASMD slazinski_k_Page_44.tif
88880 F20101203_AAASHH slazinski_k_Page_13.jpg
F20101203_AAASME slazinski_k_Page_45.tif
F20101203_AAASMF slazinski_k_Page_46.tif
6943 F20101203_AAASHI slazinski_k_Page_33thm.jpg
F20101203_AAASMG slazinski_k_Page_47.tif
77685 F20101203_AAASHJ slazinski_k_Page_17.jpg
F20101203_AAASMH slazinski_k_Page_48.tif
F20101203_AAASHK slazinski_k_Page_09.tif
F20101203_AAASMI slazinski_k_Page_50.tif
3796 F20101203_AAASHL slazinski_k_Page_64thm.jpg
F20101203_AAASMJ slazinski_k_Page_52.tif
2286 F20101203_AAASHM slazinski_k_Page_07thm.jpg
F20101203_AAASMK slazinski_k_Page_53.tif
7472 F20101203_AAASHN slazinski_k_Page_27thm.jpg
F20101203_AAASML slazinski_k_Page_54.tif
F20101203_AAASHO slazinski_k_Page_37.jp2
F20101203_AAASMM slazinski_k_Page_57.tif
14703 F20101203_AAASHP slazinski_k_Page_03.jpg
F20101203_AAASMN slazinski_k_Page_58.tif
F20101203_AAASHQ slazinski_k_Page_49.tif
F20101203_AAASMO slazinski_k_Page_59.tif
F20101203_AAASHR slazinski_k_Page_51.tif
F20101203_AAASMP slazinski_k_Page_60.tif
F20101203_AAASHS slazinski_k_Page_37.tif
F20101203_AAASMQ slazinski_k_Page_61.tif
85117 F20101203_AAASHT slazinski_k_Page_56.jpg
F20101203_AAASMR slazinski_k_Page_62.tif
3286 F20101203_AAASHU slazinski_k_Page_53thm.jpg
169218 F20101203_AAASMS slazinski_k.pdf
84719 F20101203_AAASHV slazinski_k_Page_36.jpg
7241 F20101203_AAASMT slazinski_k_Page_19thm.jpg
86412 F20101203_AAASHW slazinski_k_Page_59.jpg
27359 F20101203_AAASMU slazinski_k_Page_39.QC.jpg
50336 F20101203_AAASHX UFE0022373_00001.mets
7395 F20101203_AAASMV slazinski_k_Page_38thm.jpg
14304 F20101203_AAASMW slazinski_k_Page_60.QC.jpg
83411 F20101203_AAASFA slazinski_k_Page_62.jpg
7523 F20101203_AAASMX slazinski_k_Page_13thm.jpg
24099 F20101203_AAASFB slazinski_k_Page_09.QC.jpg
4197 F20101203_AAASMY slazinski_k_Page_60thm.jpg
6526 F20101203_AAASFC slazinski_k_Page_09thm.jpg
1051951 F20101203_AAASKA slazinski_k_Page_20.jp2
7423 F20101203_AAASMZ slazinski_k_Page_57thm.jpg
1051970 F20101203_AAASFD slazinski_k_Page_47.jp2
1051965 F20101203_AAASKB slazinski_k_Page_22.jp2
28348 F20101203_AAASFE slazinski_k_Page_21.QC.jpg
1051949 F20101203_AAASKC slazinski_k_Page_25.jp2
F20101203_AAASFF slazinski_k_Page_56.tif
27029 F20101203_AAASPA slazinski_k_Page_27.QC.jpg
F20101203_AAASKD slazinski_k_Page_27.jp2
F20101203_AAASFG slazinski_k_Page_19.tif
7591 F20101203_AAASPB slazinski_k_Page_29thm.jpg
F20101203_AAASFH slazinski_k_Page_14.tif
7462 F20101203_AAASPC slazinski_k_Page_31thm.jpg
F20101203_AAASKE slazinski_k_Page_28.jp2
83724 F20101203_AAASFI slazinski_k_Page_19.jpg
3166 F20101203_AAASPD slazinski_k_Page_32thm.jpg
1051973 F20101203_AAASKF slazinski_k_Page_29.jp2
22951 F20101203_AAASFJ slazinski_k_Page_63.jpg
7318 F20101203_AAASPE slazinski_k_Page_34thm.jpg
1051927 F20101203_AAASKG slazinski_k_Page_30.jp2
F20101203_AAASFK slazinski_k_Page_16.tif
1051963 F20101203_AAASKH slazinski_k_Page_31.jp2
F20101203_AAASFL slazinski_k_Page_39.tif
7115 F20101203_AAASPF slazinski_k_Page_37thm.jpg
396045 F20101203_AAASKI slazinski_k_Page_32.jp2
4575 F20101203_AAASFM slazinski_k_Page_03.QC.jpg
7352 F20101203_AAASPG slazinski_k_Page_39thm.jpg
1051978 F20101203_AAASKJ slazinski_k_Page_34.jp2
F20101203_AAASFN slazinski_k_Page_03.tif
24918 F20101203_AAASPH slazinski_k_Page_44.QC.jpg
F20101203_AAASKK slazinski_k_Page_35.jp2
76863 F20101203_AAASFO slazinski_k_Page_05.jpg
7120 F20101203_AAASPI slazinski_k_Page_45thm.jpg
1051923 F20101203_AAASKL slazinski_k_Page_36.jp2
7338 F20101203_AAASFP slazinski_k_Page_50thm.jpg
27142 F20101203_AAASPJ slazinski_k_Page_47.QC.jpg
1051922 F20101203_AAASKM slazinski_k_Page_38.jp2
2837 F20101203_AAASFQ slazinski_k_Page_06thm.jpg
27526 F20101203_AAASPK slazinski_k_Page_48.QC.jpg
1051926 F20101203_AAASKN slazinski_k_Page_39.jp2
1051953 F20101203_AAASFR slazinski_k_Page_09.jp2
11035 F20101203_AAASPL slazinski_k_Page_54.QC.jpg
493588 F20101203_AAASKO slazinski_k_Page_43.jp2
3304 F20101203_AAASFS slazinski_k_Page_54thm.jpg
27572 F20101203_AAASPM slazinski_k_Page_57.QC.jpg
F20101203_AAASKP slazinski_k_Page_44.jp2
7322 F20101203_AAASFT slazinski_k_Page_36thm.jpg
27180 F20101203_AAASPN slazinski_k_Page_59.QC.jpg
1051985 F20101203_AAASKQ slazinski_k_Page_45.jp2
1051912 F20101203_AAASFU slazinski_k_Page_26.jp2
6397 F20101203_AAASPO slazinski_k_Page_61thm.jpg
1051947 F20101203_AAASKR slazinski_k_Page_48.jp2
79973 F20101203_AAASFV slazinski_k_Page_16.jpg
6792 F20101203_AAASPP slazinski_k_Page_62thm.jpg
199558 F20101203_AAASKS slazinski_k_Page_52.jp2
3069 F20101203_AAASFW slazinski_k_Page_02.QC.jpg
433514 F20101203_AAASKT slazinski_k_Page_53.jp2
7213 F20101203_AAASFX slazinski_k_Page_30thm.jpg
432216 F20101203_AAASKU slazinski_k_Page_54.jp2
86182 F20101203_AAASFY slazinski_k_Page_50.jpg
F20101203_AAASKV slazinski_k_Page_56.jp2
1051933 F20101203_AAASFZ slazinski_k_Page_12.jp2
F20101203_AAASKW slazinski_k_Page_58.jp2
1051870 F20101203_AAASKX slazinski_k_Page_59.jp2
579374 F20101203_AAASKY slazinski_k_Page_60.jp2
24078 F20101203_AAASIA slazinski_k_Page_01.jpg
1051986 F20101203_AAASKZ slazinski_k_Page_61.jp2
25850 F20101203_AAASDD slazinski_k_Page_45.QC.jpg
9975 F20101203_AAASIB slazinski_k_Page_02.jpg
7402 F20101203_AAASDE slazinski_k_Page_56thm.jpg
7426 F20101203_AAASNA slazinski_k_Page_01.QC.jpg
34122 F20101203_AAASIC slazinski_k_Page_04.jpg
27441 F20101203_AAASDF slazinski_k_Page_58.QC.jpg
31996 F20101203_AAASID slazinski_k_Page_06.jpg
F20101203_AAASDG slazinski_k_Page_31.tif
26468 F20101203_AAASNB slazinski_k_Page_28.QC.jpg
28127 F20101203_AAASIE slazinski_k_Page_10.jpg
90071 F20101203_AAASDH slazinski_k_Page_21.jpg
1362 F20101203_AAASNC slazinski_k_Page_02thm.jpg
87388 F20101203_AAASIF slazinski_k_Page_11.jpg
12693 F20101203_AAASDI slazinski_k_Page_64.QC.jpg
87050 F20101203_AAASIG slazinski_k_Page_12.jpg
713135 F20101203_AAASDJ slazinski_k_Page_41.jp2
24539 F20101203_AAASND slazinski_k_Page_17.QC.jpg
88782 F20101203_AAASIH slazinski_k_Page_14.jpg
81912 F20101203_AAASDK slazinski_k_Page_08.jpg
20865 F20101203_AAASNE slazinski_k_Page_51.QC.jpg
84368 F20101203_AAASII slazinski_k_Page_15.jpg
13140 F20101203_AAASDL slazinski_k_Page_55.QC.jpg
7385 F20101203_AAASNF slazinski_k_Page_11thm.jpg
58215 F20101203_AAASIJ slazinski_k_Page_18.jpg
1051962 F20101203_AAASDM slazinski_k_Page_15.jp2
7483 F20101203_AAASNG slazinski_k_Page_24thm.jpg
84239 F20101203_AAASIK slazinski_k_Page_20.jpg
26053 F20101203_AAASDN slazinski_k_Page_37.QC.jpg
7288 F20101203_AAASNH slazinski_k_Page_49thm.jpg
90108 F20101203_AAASIL slazinski_k_Page_23.jpg
27048 F20101203_AAASDO slazinski_k_Page_34.QC.jpg
7485 F20101203_AAASNI slazinski_k_Page_22thm.jpg
87684 F20101203_AAASIM slazinski_k_Page_24.jpg
F20101203_AAASDP slazinski_k_Page_13.jp2
2292 F20101203_AAASNJ slazinski_k_Page_63thm.jpg
90347 F20101203_AAASIN slazinski_k_Page_25.jpg
F20101203_AAASDQ slazinski_k_Page_05.jp2
26380 F20101203_AAASNK slazinski_k_Page_15.QC.jpg
87714 F20101203_AAASIO slazinski_k_Page_26.jpg
1051924 F20101203_AAASDR slazinski_k_Page_33.jp2
2538 F20101203_AAASNL slazinski_k_Page_52thm.jpg
85638 F20101203_AAASIP slazinski_k_Page_27.jpg
F20101203_AAASDS slazinski_k_Page_17.tif
7587 F20101203_AAASNM slazinski_k_Page_21thm.jpg




2 2008 Karl Slazinski


3 To my family, for their c ontinued support and love Also, to my colleagues and professors, for their challenges and endless enthusiasm for debate and discussion


4 ACKNOWLEDGMENTS I thank m y chair, Dr. Coady, and committee members, Drs. Barradas and Wood, for cultivating my pursuit of the crea tion of knowledge. Dr. Coady ha s proven to be an especially helpful and caring faculty member, and is always willing to take the time to explain the logistics of academic life. I also wish to acknowledge my parents for their constant support and their devotion to alternative-thinking scholarship. Finally, Dr. Gravlee dese rves recognition for providing valuable course conten t relevant to my research interests, and for refining my intellectual toolbox to critically evaluate the li terature surrounding heal th insurance acquisition.


5 TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES................................................................................................................. ..........6 LIST OF FIGURES.........................................................................................................................7 ABSTRACT.....................................................................................................................................9 CHAP TER 1 INTRODUCTION..................................................................................................................11 Who are U.S. Latinos?............................................................................................................12 Definitions of English Language Proficiency......................................................................... 15 2 LEP AND INSURANCE ENROLLMENT............................................................................ 19 Theoretical Framework.......................................................................................................... .19 Interpreter Use................................................................................................................ ........20 Interpreters Mitigating Language Proficiency Effects.................................................... 20 Satisfaction and Cultural Concordance........................................................................... 21 Institutional Discrimination Resultin g from Lack of Communication............................ 23 Patient Assessment of Care : Differing Expectations .............................................................25 Discontinuity of Care Leads to Health Inequalities................................................................27 Undocumented Status Affects Health Care Access......................................................... 28 Language at Home and Community Outreach ................................................................ 29 Conclusion..............................................................................................................................31 3 DATA AND METHODS....................................................................................................... 33 Sociodemographic Data.......................................................................................................... 34 Variables.................................................................................................................................37 Statistical Analysis Program................................................................................................... 40 4 LOGISTIC REGRESSI ONS AND ANALYSIS.................................................................... 44 5 CONCLUSIONS AND I MPACT........................................................................................... 56 Suggestions for Policy Making............................................................................................... 56 Questions for Further Research.............................................................................................. 58 LIST OF REFERENCES...............................................................................................................61 BIOGRAPHICAL SKETCH.........................................................................................................64


6 LIST OF TABLES Table page 3-1 Description of the Dependent and Inde pendent V ariables Used in the Analysis.............. 43 4-1 Health Insurance Regressed on English La nguage Proficiency, Sex, Age, and Incom e Logistic Regression on Main land Latinos(Odds Ratio)....................................................53 4-2 Health Insurance Regressed on English La nguage Proficiency, Sex, Age, and Incom e Logistic Regression on All Latinos(Odds Ratio)............................................................... 54 4-3 Health Insurance Regressed on English La nguage Proficiency, Sex, Age, and Incom e Logistic Regression on all Populations (Odds Ratio)........................................................ 55


7 LIST OF FIGURES Figure page 3-1 Combined Household Income Dollars (% of Respondents) ............................................. 42 3-2 Highest Educational Attainme nt, Years (% of Respondents) ............................................ 42 4-1 Mainland Latinos Odds Ratios for Hea lth Insurance Based on Incom e, Dollars (Compared to <$10,000).................................................................................................... 52


8 LIST OF ABBREVIATIONS BRFSS The Behavioral Risk Factor Surv eillance System, run by the Centers for Disease Control and Prevention. It consists of data aggregated from a phone survey conducted by all states a nd territories of the United States. CDC Center for Disease Control and Prevention, operated by the United States government in an effort to promot e public health and decrease the prevalence and/or impact of infectious disease. HIV Human Immunodeficiency Virus, a re trovirus that can lead to Acquired Immunodeficiency Syndrome (AIDS), a condition in humans in which the immune system begins to fail, lead ing to life-threatening opportunistic infections. HMO Health maintenance organization, a t ype of managed care organization that provides a form of health care coverage in the United States that is fulfilled through hospitals, doctors, a nd other providers with which the HMO has a contract. LEP Limited English Proficiency, as defined by either the respondent or a third-party observing the individual. In this pa per, LEP means that the individual is not comfortable communicating more than a few sentences in English. OLS Ordinary Least Squares, also known as regression analysis, is used to model numerical data obtained from observations by adjusting the parameters of a model so as to get an optimal fit of the data. PPO Preferred Provider Organization, a t ype of managed care organization that provides a form of health care coverage that is less restrictive than the HMO. A PPO does not require a primary care physician to prescribe specialized medical services. SES Socioeconomic status, a measure of not only the economic but the social conditions in which indivi duals and families live. Class comes into the analysis, which takes into account more measures than solely traditional economic variables. SLVR Spanish language response variable a pre-survey variable compiled from Spanish and English speakers responses to normal, everyday experiences. This accounts for any language-relat ed variation among rating schema.


9 Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Arts LIMITED ENGLISH LANGUAGE PROFICIE NCY AND ACCESS TO HEALTH CARE AMONG U.S. LATINOS By Karl Slazinski August 2008 Chair: Maria Coady Major: Latin American Studies Numerous case studies deal with the effects of immigration status and race on access to health insurance among the Hispan ic populations in the United States. However, few of them deal with English language proficiency as the main causal variable. Thus, the main objective of this study is to determine if Limited English Pr oficiency (LEP) is the primary or a contributing variable in determining health insurance status. The method I use to determine the effect of being an LEP Latino versus an Englishspeaking Latino on health insurance rates is logistic regression. The data are provided by the CDC in the form of the Behavioral Risk Factor Surveillance Syst em. The BRFSS is the largest, ongoing, random-digit-dial telephone survey of h ealth status and access in the United States, Puerto Rico, and Guam. It randomly selects from the population of noni nstitutionalized adults 18 years of age and older. The weighted sample controls for multiple phone lines and nonresponse biases. Included in the sample were 18,147 mainland Latino individuals, 21,368 mainland and Puerto Rican Latino individuals an d 275,491 white individuals. Respondents were permitted to answer the questions in English or Spanish. For the purposes of this study, I used the question of language choice as a proxy for dominant language. The individuals who chose to respond in Spanish were classifi ed as people who were limited E nglish proficient. Logistic


10 regression analyses were performed to determine if, after controlling for years of education, sex, age, and income, English language proficiency st ill accounted for a substan tial effect on health plan enrollment. The results showed that mainland and all Latinos who responded in English were 2.682 and 1.446 times more likely to have health insurance, respectively. Therefore, the logistic regression shows that English langu age proficiency is one factor in determining if Latinos have access to health insurance.


11 CHAPTER 1 INTRODUCTION The existence of health inequalities between whites and Latinos is well documented. In 2002, Hispanic or Latino individuals constitute d 13.3% of the United States population, with nearly 28.1% of all Latinos living at or under th e poverty line (Garcs et al. 2006:377). Of those individuals under sixty-five year s of age, 11% of non-Hispanic whites are uninsured, while 32% of Latinos are uninsured (Garcs et al. 2006:377-378). Moreover, about half of those AfricanAmericans or Latinos with insurance are covere d only by Medicaid, compared with only one in five whites (Nickens 1991:27). Nearly 35% of Latinos are foreign born, 67% (25,074,000) are from Mexico (Garcs et al. 2006:377-378), and a quarter of Latinos speak little or no English (Prez-Stable 1987:213). Speaking English appears to correlate to larg er rates of health insurance access. Immigration stat us seems to also play a role in whether or not Latinos have health insurance. Furthermore, having health in surance appears to be th e single most important predictor of having access to quality health care and preventive services. However, while poverty rates for Latinos and African-Americans are somewhat similar at approximately 28.1% and 32.6% (Kerner et al. 1993:357), respectively, Latinos are better off health-wiseearning the Latino mi nority the dubious title of the epidemiologic paradox (Nickens 1991:27-29). The epidem iologic paradox states that wh ile Latinos are generally as poor and have similar educational attainment profiles as African-Americans, health among Latinos is generally better. The majority of the Latino population suffers from less chronic and degenerative diseases than African-Americans, which seems counterintuitive. As observed by Dressler et al. (2005:233) in thei r review of the lite rature that establis hed the acknowledgement of health disparities between whites and minorities, there are a set of diseases that seem to account for most of the racial or ethnic disparities in mortality: hypertension, HIV, diabetes, and


12 homicide are prevalent at much higher rates in mi norities than in whites. However, the role of English language proficiency among Latinos in determining their health insurance access has been overlooked in much of the lit erature. To this end, in Chapte r 2 I will explore the relevant literature on health inequities be tween whites and Latinos, particularly those relating to language barriers to health insuran ce access, which compromise health care utilization. Who are U.S. Latinos? Latino as a term was adopted by the United States government in 1997 to be an ethnonym for Hispanic individuals. Accord ing to definition, Latino refers to any individual who is either: a Spanish speaker or person belonging to a house hold where Spanish was spoken; a person with Spanish heritage by birth location; and/or a pers on who self-identifies with Spanish ancestry or descent. While many use Hispanic and Lat ino interchangeably, scholars maintain a distinction between the terms. Hispanic come s from the Latin word for Spain, and thereby potentially encompasses all Spanish-speaking peoples in both hemispheres. The term Hispanic also emphasizes the common denominator of language and in effect remove s cultural differences from the identity. Latino, which is mo st likely a shortening of the word latinoamericano refers more exclusively to persons or communities of Latin American origin. However, to most English speakers and indeed to a majority of Latino groups, Latino means any person who is of Spanish ancestry who now resi des in the United States. In my thesis, I will be using Hispanic and Latino interchangeably. This is due to the nature of the data I am using, which was collected using the census definiti ons for ethnicity/race. However, my preferred and most frequently used term is Latino, referring to the nature of an individuals Spanish-ancestry who mainta ins a domicile in the United States. Importantly, each group of Latinos has a specif ic history and relationship to the United States. They have different legal statuses, leve ls of integration, and mo tives for being here.


13 Mexican-Americans primarily reside in Chicag o and the Southwestin California, Arizona, Texas, and surrounding areas. Puer to Ricans reside on the island of Puerto Rico, in New York Cityespecially in El Barrio and more recently have expand ed the diaspora to Orlando, Florida. Cubans who left the isla nd relocated principally to South Fl orida, particularly in Miami. More recently, this group has expanded through out Florida into Tampa and Orlando (Gonzlez 2000:86-87). Dominicans and other Latin American groups have been increasingly moving into the traditional enclave communities that each subgroup have founded and nurtured. The term Chicanoor Mexican-Americanorig inally developed as a self-designated identity for a subgroup in Chicago. The term spread into the Southwes t where many MexicanAmericans live, and now it is widely recognized by Mexican-Americans as an umbrella term. They also are, by far, the largest Spanish sp eaking minority group in th e United States. When Texas was annexed in 1845 through the signi ng of the Treaty of Gu adalupe-Hidalgo, the Mexicans who were already living there became the first Mexican-American citizens. This act established the long-standing rela tionship between the United Stat es and Mexico, with the border crossing their land and separating them from Me xico (Gonzlez 2000:105-106). Later, a migrant worker program called the Bracero Program allowed Mexicans to come into the United States to work temporarily in agriculture. The El Paso He rald Post (1956:1A-5A) printed that, More than 80,000 Braceros pass through the El Paso Center a nnually. Theyre part of an army of 350,000 or more that marches across the border each year to help plant, cultivate, and harvest cotton and other crops throughout the United States. This program established the trend of the U.S. importing a seasonal, low-wage agricu ltural labor supply from Mexico. Along with Chicanos, Puerto Ricans also have a lengthy relations hip with the United States. The island of Puerto Rico, along with the Philippines and Guam, was acquired at the end


14 of the Spanish-American War in 1898 through the si gning of the Treaty of Paris. Puerto Ricans began to move into the United States seeking work, motivated by the Puerto Rican government and Operation Bootstrap. Operatio n Bootstrap was created as a m eans to make Puerto Rico less dependent on sugar production and to increase the tax base by enticing corporations and manufacturing to the island. It also encouraged migration o ff the island, thereby shunting the amount of needed social services to the ma inland (Gonzlez 2000:246-249). This allowed the island to become increasingly i ndustrialized, shifting labor from the traditional agrarian society to an urban one. As Puerto Ricans left the isla nd, they frequently travelled and settled in New York in what became known as Spanish Harlem, simultaneously and inadvertently founding an ethnic sub-group called the Nuyoricans. As a co mmonwealth, the relationship the island has to the United States government is exceptional am ong Latino groups. Puerto Ricans do not have the same traumatic experience of immigrating ei ther legally or illegally into the country. Nonetheless, this is not to say that their expe rience and cultural transi tion is not painful and jarring, especially consideri ng the racial typing and discrimination th ey encountered upon arriving (Gonzlez 2000:250-257). While it is debated among scholars whether Island Puerto Ricans are Latinos or not, I include them in the um brella term of Latino. This is because the United States Census Bureau defines a Latino as, A person of Cuban, Mexican, Puerto Rican, South or Central American, or other Spanish culture or origin regardless of race (2001:1-2). Nonetheless, to address these scholarly concerns I perform an analysis on the Latino/Hispanic populations as both separate and aggregate groups. Another large group in the Un ited States is the Cuban-Amer ican population. The United States government has had a turbulent history wi th Cuba. While investment in Cuba was over $1.24 billion in 1924, after Castro took power in 1959, the United States steadily was pushed out


15 of the country through land expropriation, culmin ating in a communist government, the United States ceased trade with Cuba. However, the Cuban Airlift that ended in 1973 provided guaranteed asylum to 260,561 Cuban political ref ugees. Moreover, the Catholic Church, and later the U.S. government, helped them accultu rate through providing English language courses and loans to start their own small businesse s. Later, the Mari el Boatlift and the Balsero Rafter Movement allowed many immigrantslegal and otherwiseto enter the United States (Gonzlez 2000:108-116). Each of these subgroups has its own cultu ral background and unique history relating to the United States. These experiences may contribute to differences in English language abilities and cultural norms for seeking out medical care and insurance. Definitions of English Language Proficiency Language proficiency is difficult to measure or classify. The four areas usually used to determine an individuals leve l of language proficiency are: listening, reading, writing, and speaking. Quantifying constructs and capturing th e level of proficienc y is difficult. The literature contains numerous definitions for la beling certain levels of language fluency. Accordingly, each author of the studies I use in Chapter 1 and 2 has chosen to use their own definition(s) of English language proficiency. Proficiency and fluency is also used interchangeably throughout. In each case, I have maintained the original authors terminology. Monolingual, English or Spanish dominant, academic language proficient versus conversational proficiency, and Limited English Proficient (LEP) are some of the terms used by these authors. In this study, I understand that pe ople are in different stages of acquiring English proficiency, but have are not fully proficient. After reviewing these studies, LEP individuals are not proficient enough to hold a conversation in a specialized, hi ghly-demanding context. However, LEP is a


16 government term that is not widely used in the fi eld, and is usually consid ered by scholars to be pejorative. In an effort to clarify the different levels of foreign or second la nguage proficiency, I use the Interagency Language Roundtable (ILR) scale. It is a set of descriptions of abilities to communicate in a language originally developed by the United States Foreign Service Institute. It is commonly known as the FSI scale, and contai ns five levels of flue ncy. It was last updated in 1998, and has existed since 1984. There are many other systems used to determine levels of proficiency by individual author s, schools of thought, and even pa rticular state school boards. However, it is arguably the most consistent a nd well-known of these scales, and therefore can give a broad understanding of the commonly accep ted different levels of fluency (Higgs 1984:217). The first level is S-1, or Elementary Proficie ncy. It is characteriz ed by the ability to satisfy routine travel needs and minimum courte sy requirements; can as k and answer questions on very familiar topics, within the scope of very limited langu age experience; can understand simple questions and statements, allowing for sl owed speech, repetition, or paraphrasing; and has a speaking vocabulary which is inad equate to express anything but the most elementary of needs, making frequent errors in pronunciation and grammar, but can be understood by a native speaker. Level two, or S-2, is Limited Working Proficie ncy. It is characteri zed by the ability to satisfy routine social demands and limited work requirements; can handle with confidence, but not facility, most social situa tions including introductions and cas ual conversations about current events, as well as work, family, and autobiogra phical information; can handle limited work requirements, needing help in handling any complica tions or difficulties; can get the gist of most


17 conversations on non-technical subject, and ha s a speaking vocabulary sufficient to respond simply with some circumlocutions; has an accent which, though often quite faulty, is intelligible; and can usually handle elementary constructions quite accurately but does not have thorough or confident control of grammar. Level three is Professional Working Proficie ncy, and is demonstrated through the ability to speak the language with suffici ent structural accuracy and voca bulary to participate effectively in most formal and informal conversations on practical, social and pr ofessional topics; can discuss particular interests a nd special fields of competence with reasonable ease; has comprehension which is quite complete for a normal rate of speech; has a general vocabulary which is broad enough that he or she rarely has to grope for a word; and has an accent which may be obviously foreign, but has a good control of grammar and whose errors virtually never interfere with understand or di sturb the native speaker. S-4 is Full Professional Proficiency, and is able to use the language fluently and accurately on all levels normally pertinent to professional need s; can understand and participate in any conversations within th e range of own personal and prof essional experience with a high degree of fluency and precision of vocabulary; w ould rarely be taken for a native speaker, but can respond appropriately even in an unfamiliar situation; makes quite rare and unpatterned errors of pronunciation and grammar; and can handl e interpreting informally from and into the language. The fifth level is native or bilingual profic iency. It is characterized by a speaking proficiency equivalent to that of an educated native speaker; and has complete fluency in the language, such that speech on all levels is fully accepted by native educated speakers in all of its


18 features, including breadth of vocabulary and idiom, colloquia lisms, and pertinent cultural references. Choosing to take a questionnaire in Spanis h rather than English does not necessarily indicate actual language ability le vel. The individual could have chosen to take the survey in Spanish simply because it was offered or because he/she was not familiar with the specialized English terminology associated with medical care. Thus, due to the level of technical language used when describing health care and access to insurance and medical care, I believe that LEP refers to a certain level of dayto-day fluency without mastery of technical, specialized language. Therefore, while the respondent may be able to navigate his/her daily life in English, Spanish may remain the dominant language, and he/she may not have a high level of specialized language acquisition. Relating this to the ILR sc ales, LEP could encompass an individual from S-1 to S-3. However, I consider most indivi duals who are capable of S-4 or above English proficiency may feel comfortable to respond to the oral questionnair e in English. Therefore, in this study, LEP refers to individuals characterized by S-1 to S-3 proficiency in English and who would not normally be able to complete the survey without interpreter or aid of a dictionary.


19 CHAPTER 2 LEP AND INSURANCE ENROLLMENT Theoretical Framework In this review of the literature, I attempt to answer the question: Are Latinos enrolled in health insurance programs at a lower rate than whites due to limited English language proficiency? Understanding the causes of La tino underutilization and la ck of access to health insurance and medical services is a difficult task. While there have been many proposed explanatory modelsthe racial-gen etic model, health-behavior m odel, and socioeconomic status modelthe most promising one seems to be the social structural mode l. Dressler et al. (2005:231-237) states that the soci al structure model needs to be revised to reflect two distinct frameworks that better explore mi nority health inequalities. In updating this model, Dressler et al. have identified the psychosocial stress model and the structural-constructivist model. The psychosoc ial stress model explicitly attempts to include the biological pathways along whic h perceived discrimination and st ress from not being a part of the dominant social group may resu lt in unequal and negative hea lth outcomes. The structuralconstructivist model proposes that social, psychol ogical, and biological pr ocesses take place at the intersection between external superstructu res and the social relationships in which individuals are embedded. The constructivist pe rspective refers to a socially-amalgamated, shared understanding of the reality of life. Without these models, it is impossible to accurately evaluate the mechanisms through which et hnic health inequalities may occur. The first salient argument is one that fa lls under the socioeconomic status model for understanding ethnic disparities in health outcomes. Without bei ng able to speak English, Latino immigrants and Latinos who maintain Spanish as their primary language through community and home monolinguism are at a severe disadvantag e in their possible earning power. Without


20 speaking English fluently and bei ng able to, at the very least, graduate from high school, Latinos are less likely to earn enough money to be able to afford health insurance plans while supporting their households (Dressler et al. 2005:237-239) Moreover, undocumented Latino immigrants are de facto not able to enroll in any such plan regardless of th eir desire to, simply because of their illegal status. The second argument consists of a struct ural-construc tivist mode of questioning, considering that English is the dominant language in the United States. Spanish speakers, even when they have access to medical care and insuran ce, may be more likely to be dissatisfied with their health care experiences. The structural-constructivist m odel attributes this to the combination of cultural and la nguage barriers between monolingual Spanish speakers and their English-speaking doctors and medical staff. La tinos responding in English experience cultural barriers but not severe language barriers, leading to their rela tive satisfaction level being intermediary between whites res ponding in English and Latinos responding in Spanish (Dressler et al. 2005:239-243). Latinos responding in Spanish frequently face cultural and language barriers to communication, which may make them most at risk to be the most dissatisfied with their health care. Interpreter Use Interpreters Mitigating Language Proficiency Effects Jacobs et al. (2004:866-869) examined the co sts and benefits of pr ofessional interpreter use among Spanish speakers in four hospitals in Massachusetts. Because most persons who do not speak English well are provided no interpreter services, they do not rece ive needed or quality health care. Reliance on ad hoc untrained interpreters, such as friends, family members, children, or non-clinical employees, who have very little technical knowledge, is shown to have negative health consequen ces rather than positive effects. To create a cost-benefit analysis, the


21 authors of the study created a group that utilized an interp reter at all hospital visits for one year, while comparing them to a control group. The authors found that while the cost for providing interpreter services significantly increased the cost per person for treatment in the interpreter group, it also was followed by increases in preven tive care visits, number of physician visits, and utilization of prescriptio n drugs. This indicates that the limit ed English speakers access to care was enhanced by the interpreter services at a mo derate cost, which was then associated with better health outcomes. The aut hors of the study concluded that th e cost was justified in relation to normal reimbursement procedures during the years of their study, a nd note that the increase in preventive care utilization may very well save money in the long run. Thus, according to this argument, Latinos with limited English language proficiency are being structurally excluded from receiving effective, quality health care, even when they are enrolled in a health insurance plan. This, in turn, may lead to higher rates of preventable disease a nd health consequences, leading to higher insurance co mpany costs in the long term. Satisfaction and Cultural Concordance In a related article by Morales et al. (1999:409-417), the author s investigated how satisfied Latinos are with the communication they receive from their health care providers. The objective of their study was to de termine if Latinos who respond in Spanish or English will be more or less likely to be satisfied in compar ison with English-speaking whites. The authors controlled for not only the normal socioeconomic a nd demographic variables, but also a Spanish language response variable (SLVR). The SLVR statistically controls for differences in ratings between Spanish and English language respondent s attributable to linguistic and cultural differences in using the response scale alone with out any other mitigating reference source. The study found results that would generally satisf y the socioeconomic models viewpoint, with Spanish-speaking Latinos reporting lower educati onal attainment, lower annual income, larger


22 family size, younger mean age, fewer mean numb er of co-morbid conditions, lower private health insurance coverage enrollment, the highe st rates of reporting no insurance, and more likely to be married. Moreover, there were no sta tistically significant differences in physical and mental health indices reported across the gr oups, although Spanish-speaking Latinos reported the fewest number of health condi tions. This aspect of Spanis h-speaking Latinos demographics could be attributable to a high number of undia gnosed conditions, especially when taking into account the fact that they are mo re dissatisfied than the other st udy groups with their health care providers communication. Ultimately, the Morales study found that Spanish-speaking Latino respondents were significantly more dissati sfied with provider communications than English-speaking Latinos and white respondents. Greater barriers between so cial class and subsequent difficulties in health care providers ability to relate to the patient may at least partially account for this disparity in satisfaction with communicati on. The language-related communi cation barrier is important on many levels. First, it establishes that Spanish-spea king Latinos seem to perc eive their health care providers communication as less than satisfactory, either because th ey are in fact receiving subpar care, there is an interpreterrelated problem, or the patients ha ve different expectations than an United States-born, English-speaking individual. Secondly, Spanish-speaking Latinos health outcomes may be affected because the doctor ma y order excessive tests attempting to diagnose the patients condition(s) without a proper history, or the patients may inadvertently misuse their prescribed treatments, thereby becoming accide ntally noncompliant. Many of the proposed solutions to this communication-based problem ar e that providing profes sional interpreters or bilingual doctors with adequate fluency in both languages helps faci litate better patient understanding of their disease and the necessary treatment regimen. However, the study


23 conducted by Morales et al. (1999) dealt only with Mexican-Americans, thereby making the applicability of inferred gene ralizations for all Latino groups from its results suspect. Institutional Discrimination Resulting from Lack of Communication Going beyond the simple use and cost-benefit analysis of utilizing interpreters, Sarver and Baker (2000:256-264) seek to determine whether patients who encountered language barriers during an emergency department visit were less likely to be referred for a follow-up appointment, less likely to be aware that such an appointment had been scheduled and/or less likely to actually attend that recommended appoint ment. One major difference in this study from the others is that the responde nts were asked whether an inte rpreter was used and if the respondent thought that he/she s hould have used one or not. If used, the interpreters relationship to the patient was determined. The study found that language-concordant patientdoctor combinations were more likely to discharge with a follow-up appointment scheduled, as well as the patient being more compliant. This is contrasted with both alternative circumstances, where an interpreter was used, and the patient stating that they should have used an interpreter when one was not utilized. However, nearly all the groups were awar e of their scheduled appointments, with only minor diffe rences being attributable to the location of the appointment being different than the emergency department. This was not due to a miscommunication, but to the lack of comprehensive comm unication available to both patie nts and health care providers. Because the medical personnel could not effectiv ely communicate, they could not accurately relay the location of the follow up visit to their patie nt. Therefore, this seems to at least partially debunk the idea that patient compliance will be enha nced by proper interpreter use or culturallyand language-concordant medical staff being used. However, it is obvious that to increase overall patient-doctor satisfacti on and the highest rate s of compliance possible, professionallytrained interpreters should be used whenever possible.


24 Sarver and Baker (2000) suggested that po ssible explanations for the differences in follow-up referrals for non-English speaking Latinos could be due to physician perception of the patients lack of compliance due to poverty or language barriers. Physicians may also have an incomplete understanding of the problems being suffered by their patients and therefore may believe that a follow-up appointment is not necessary. Doctors may also be struggling with other aspects of their care plan, with the strain to communicate effectively making care providers liable to not refer the patient for a follow-up appointment. The overarching benefits of concordant staff and/or properl y trained interpreters remain compelling and well supported by the literature and this article in particular. The socioeconomic barrier seems to be refu ted by Sarver and Bakers study (2000), with Ordinary Least Squares regression models showi ng that there were no statistically significant differences between the three study groups (Eng lish-speaking whites, En glish-speaking Latinos, and Spanish-speaking Latinos) in compliance with fo llow-up visits, as long as the individuals are aware of the appointment at the time of discharg e. However, another possible explanation could be that the passage of Proposition 187 in the area that this study was conducted had created an anti-Spanish-speaker bias. The Proposition mandates that all California hospitals refuse treatment to undocumented immigrants, and that they be reported to government officials, possibly engendering a racist attitude in the popu lation being surveyed. Sarver and Baker pose these questions for further resear ch, stating, Further studies are required to determine if Latino patients are less likely to receiv e other types of medical care; whether care patterns differ because of communication barriers, lack of cultura l awareness, or patient behaviors; and whether these differences can be reduced through program s that increase the availability of properlytrained interpreters and teaching physicians how to handle cultural and linguistic barriers to care


25 (2000:264). Their study raises im portant questions about why La tinos are treated differently than whites and if this inequality is du e to limited English language proficiency. Patient Assessment of Care: Differing Expectations Intrinsic to providing quality health caremore than simply providing interpreter servicesis having health care providers who ar e culturally aware (Prez-Stable 1987:216-217). For example, personalismo or polite friendliness, is consid ered a staple mode of interaction for Latino patients with their doctor. Moreover, a level of dignidad or the dignity and decorum of the patient-doctor relationship, is expected, and the rapport can quickly dissolve if this is overlooked. Part of cultural awareness is either being bilingual to the extent of being able to communicate well and accurately, correctly choosi ng an interpreter if a professional is not available, and/or also physically situating yourself in a way that will allow the interpreter to translate without obstructing patie nt-doctor eye contact and body language/.contact exchange. According to Hubbell et al., Latinos are less likely than any other ethnic group to have health insurance (1991:414-417). In the Dr. Robert Wood Johnson Foundation national telephone survey in 1986, blacks were twice as likely as Latinos to have health insurance, and whites were three times as likely. These statisti cs are surprising consid ering that blacks and Latinos are both minorities that have long-standing histories w ithin the United States. One would think that they should be at similar leve ls of incorporation and reflect similar health inequalities (at least for some Latino groups). However, according to Hubbell et al., 38% of Latinos and 60% of undocumented Latinos report ed not having health insurance (1991:417). Hubbell et al. also reported defi ciencies greater than blacks in other important measures of access to health care, such as having a regular so urce of care or at least one office visit a year (1991:417-418). Something distinct still occurs within Latino populat ions that restricts access to health insurance and medical care, which may be related to limited English language proficiency.


26 While the article by Hubbell et al. does not specifically address the issue of English language proficiency, the survey was administered by spec ially-trained, bilingual interviewers from the University of California, Irvine. The respondent s were given the choice of taking the survey in English or Spanish. This effectively limits th e applicability of the surveys findings on the structural impediments of m ono-linguistic health insurance systems in the United States influencing Latinos access to medical care. Patient assessments of health care are linked to service utilization, the decision to switch health plans, and treatment compliance. Therefore, it is important to accurately assess Latino trust in the health care systems. To this end, Hunt et al. (2005:555-558) constructed trust indices and conducted a telephone su rvey in both English and Span ish, attempting to use the data to show relative trust levels. The subjects trust in their health care providers varied widely on an individual basis, and Spanis h-speaking Latinos had the lowest overall trust compared to whites, English-speaking Latinos, and African-Ame ricans. One possible reason behind the lower levels of trust exhibited by minorities is that they are disproportionately enrolled in more restrictive health care plans th an are whites, thereby accounting for some of the heretofore racially-attributed variance in trust. More re strictive health plans include HMOs and PPOs, where access to primary care physicians is dictat ed by the companys participation with medical care providers, and all speci alist visits must first be prescribed by a primary care provider. These plans are also associated with high deductibles and co-pay fees, ra ising the overall cost of care when added to the plans enrollment costs. Ordinary Least Squares regression revealed a correlation between restrictiveness of the insuranc e policy and trust in the health care provider on a case-by-case basis, but these trends break dow n when looked at across race lines. There are variables in the survey that confounds the evidence that perhaps ther e is a structural or perceived


27 social cause that makes non-English speakers distru st the health care system, or to have a less satisfying health care situation. However, it may be that more restrictiv e plans further exacerbate the prevalent problems of language and cultural barriers a nd tie the hands of the providers to give the best possible carethereby resulting in Latin os lower trust in the health care system. Discontinuity of Care Lead s to He alth Inequalities Discontinuity of care is associated with lowe r levels of access to health care and health insurance, as well as negative health outcomes. Doescher et al. (2001:7889) show that the site where care is administered is critical to the co ntinuity of care with the same doctor and with regular follow-up and preventive care visits. Ha ving a sustained relationship with a doctor is considered to be a core component of quality primary care and preventive medicine; improves compliance; reduces number and duration of hos pital admissions; reduces the use of primary care resources; and increases pati ent and provider satisfa ction. Therefore, it makes sense that insurance companies, policymakers, health car e providers, and the patient would all perhaps wish to create a rapport which results in a lasting relationship. In this particular survey, Spanishspeaking Hispanics were most likely to reside in low-income areas, be non-smokers, be less educated, and lack insurance coverage. They were also unlikely to identify a regular site of care, thereby signaling that they were suffering from an exacerbated health ine quality. Moreover, the study found that non-English-speaking Latinos were disproportionately more likely to be seen at a clinic or hospital, thereby ma king it impossible to see one docto r consistently. High turnover rates among trained physicians working in health centers and physicians-i n-training in hospital clinics account for some of the provider discontinuity based on site of treatment. There remain inequities based on structural ba rriers likely created by language differences between patients and health care providers.


28 Morales et al. discovered that when worki ng with HIV patients, having health insurance was not statistically correlated to English language proficie ncy (2004:1119-1121). This result surprised the study authors, with weak associa tions between access to ca re and acculturation indices, survey language, and citizenship status The authors cite th e possibility of underrepresentation of Hispanics in th e portion of the survey which c onsisted of patients who have no insurance or are incarcerated a nd therefore did not receive care for their HIV infection. There are several alternative explanations. For one, mo st HIV patient will quickly establish a primary care physician/specialist who ma y adequately address their n eeds. Secondly, community organizations, if not family members or friends, are more likely to provide interpreter services for non-English speakers. Thirdly, there are many community-based groups that seek to help HIV sufferers who do not have access to health insurance or their medicines because of the prohibitive costs associated with them. This support structure could very well pr ovide a network of continuity of care and mitigat e the effects that lack of English language proficiency normally would have on individuals suffering this particular disease. Undocumented Status Affects Health Care Access The effects of language barriers and undocum ented status are again seen in the study performed by Granados et al (2001:1806-1807). The rationale fo r the study was that Latinos constitute approximately 8% of the US population and nearly 25% of Latino children in the US are not covered. The study surveyed a pr edominantly Latino community and broke the respondents down into three categor ies: both parents were Unite d States-born, one parent was United States-born, and both parent s were foreign-born. Language was strongly associated with the child having or not having insurance, but was also a confounding factor associated with nation of origin.


29 Findings showed that United States-born chil dren with two immigrant parents were more than twice as likely as mixed-citizenship homes to not have health insuran ce, with their lack of access being six times more likely than United States-born children with United States-born parents. This could be explained by the fact th at immigrants were forced into low-paying jobs without fringe benefits such as insurance access due to their lo w educational attainment, or it could be viewed as immigrant st atus being associated, in some cases, with undocumented status. Therefore, it would be unlikely for undocumented parents to provide health insurance to their children even if they can afford to do so, lead ing to intergenerational transfer and compounding of health inequalities through Geronimus weathering mechanisms (1992:207-218). The weathering mechanism is a manner in which the ramifications of sociallyconstructed concepts of race and inherent genetic difference based on superficial appearance are transmuted into biological and chemical changes. This is most clearly demonstrated through the womb environment. If a woman is malnourished or not properly educated on eating a healthful diet all her life, and especially during pre gnancy, her womb will be a subpar environment compared to higher SES women (Geronimus 199 2:210-218). Now her child will most likely grow up in a similar low-SES environment, and will have a comparatively more sub-optimal womb environment for her children. This cy cle continues, constant ly exacerbating the gap between the poor and rich individuals. The Gran ados et al. 2001 studys implications will, in effect, begin to remedy this intergener ational transfer of inequality. Language at Home and Community Outreach Flores et al. (2005:418-425) extend prior research by includ ing home language use. The authors state, Speaking a nonEnglish language at home or having parents who choose to be interviewed in a non-English langua ge are associated with impaired health status, a lower likelihood of having an usual [or no] source of me dical care, and impaired access to care for


30 children with special needs (2005:419). The study ove rtly demonstrates that there is an inverse relationship between English language proficiency and insurance enroll ment rates: the better the parents report speaking English, th e higher the enrollment in a heal th insurance plan. Moreover, there was a direct dose response relationship to parental English language proficiency and private health insurance enrollment, with individuals who speak English very well at the top with over a quarter of the respondents enrolled. Most of the discussion of parental characteristics serves to reinforce the preconceptions that the majority of anthropologists accept of the Latino population. Some of these are the higher the English language proficiency, the more likely they were to graduate from high school, and undocumented or expired visa status was an indicator of lower educ ational attainment and lower English proficiency, as well as lower income levels. In all analysis models employed by Flores et al., LEP parents were associated with a three hundred percent in crease in their offspring having a fair or poor overall health status, and a two hundred percent increas e in the odds of the child having to spend at least one day in bed due to illness in the last year (2005:422). Overall, their children were at an elevated risk of not receiving necessary treatment due to cost issues, transportation problems, difficulty making appointments, being uninsured, the medical staff not understanding the familys culture, and/or the health care facility be ing inaccessible due to distance. There seems to be hope in our efforts to a ddress these language and cultural barriers to access to health care and quality preventive me dicine. According to Baughman (2007:1-22), Individuals in the Hispanic communities who are exposed to Spanish-language outreach programs are able to recognize when they are elig ible for certain public h ealth initiatives even when they did not think they had access before. This is due to a series of laws that were passed


31 in an effort to decouple welfare eligibility from government-sponsored Medicare equivalents in some states. Thus, according to Baughman (2007), once Latinos were aware that even though they were not eligible for welfare or they might have been undocumented, the state would provide their children with enrollm ent in a health insurance plan fo r little or no cost to them, they quickly took advantage of it. While there may be language barriers, access to resources in patients first language may incr ease participation in availabl e programsthereby lessening the impact of Latino-white health inequalities. Conclusion Language barriers, cultural impediments, and discrimination are perceived to exist (American Anthropology Associ ation 1996:569-570). We can empirically recognize whiteminority health inequalities which need to be ad dressed. English language proficiency seems to be one of the strongest predictors for health insurance enrollment among Latinos and their children. It also corresponds to a more critical interpretation of their received health care services, if they are accessible in the first place. However, la nguageand culturally-concordant physicians and medical support staff, or interpreter-facilitated intera ctions, appear to increase the level of satisfaction as well as reduce misco mmunication-caused lapses in compliance with follow-up care and visits. While Latinos have similar socioeconomic profiles to blacks, the challenges that face them are quite different, and some scholars would argue that these barriers are more pronounced due to the language differen ces creating concrete walls between Latinos, potentially higher-earning jobs a nd better educational attainment, and health care providers. Further compounding these difficulties is the radically diverse backgrounds and circumstances surrounding each Latino groups immigration status. Most of the possible explanations seem to fall in the structural-c onstructivist mode of understanding the situation, with most of the inequalities occurring because of the lack of interpreter se rvices, lack of English


32 language proficiency and socioeconomic factors, as well as the dominant governmental agencies controlling which programs Latinos may enroll in, i rrespective of immigrat ion status or language ability. However, psychosocial stress models pervade the literature as well, with trust in the health care system and satisfaction with office vi sits going down as English language proficiency decreases, showing perceived di scrimination or perhaps a form of marginalization due to the patients emergent ability to speak English. One thing is evident: at the confluence of these two phenomena, we see unequal health outcomes that need to be addressed.


33 CHAPTER 3 DATA AND METHODS The source of the data used in the subsequent an alyses was the Behavioral Risk Factor Surveillance System (BRFSS), a yearly survey conducted throughout the c ountry by each state, and carried out in coordination with the Cent er for Disease Control and Preventions (CDC) National Center for Health Statistics. The BRFSS is the worlds la rgest ongoing telephone health survey system, tracking health conditions and risk behaviors of noninstitutionalized adults in the United States since 1984. The history of the BRFSS refl ects the evolution of health-rel ated research. In the early 1980s, studies clearly showed th at personal health behaviors played a role in premature morbidity and mortality, yet there were few national-level data sets that addressed this issue. This deficiency represented a critical gap in kno wledge for state health agencies whose primary responsibility was reducing behavior al risks and their consequent i llnesses. As the significance of personal health behaviors received wide r recognition, telephone surveys emerged as an acceptable method for determining the prevalence of many health risk behaviors. Telephone surveys were especially desirable at the state a nd local levels, where the necessary expertise and resources for conducting area probability sampli ng for in-person household interviews were not available. The objective was to collect data on actual behaviors, rather than on attitudes or knowledge, that would be especially useful for health policy planning ( http://www.cdc.gov/brfss ). When sa mpling weights are used, the BRFSS da ta can be generalized to the entire population of the United States (including each ethnic/racial gr oup, given that the BRFSS is a random-digit dial sample of all households in the 50 states, Puerto Ri co, the District of


34 Columbia, and Guam). The sampling weights al so adjust for households with multiple phone lines and for non-response bias. To conduct this study, I downloaded the data for 2005 from the BRFSS website as an ASCII file and then imported the data into the Statistical Package for the Social Science (SPSS 15.0). For each variable used in the multivariate analyses, all cases that contained refusal to respond or non-response answers were excluded from the data set. The three groups I analyzed are Hispanic (referred to as Latino), non-Latino whites, and to tal respondents. The Latino group was analyzed excluding Puerto Ricans who live on th e island of Puerto Rico. I excluded them from the data set because all their communication with the insurance providers was in Spanish. Specifically, speaking Spanish on the island of Puerto Rico circumvented the confounding experience of not being able to speak the nece ssary language to have enrolled in health insurance. Therefore their experience, while va lid, was not needed in the regression except to provide a whole country l ook at the experience of all individuals of Hispanic origin associated with the United States. However, it was inform ative to compare island Puerto Ricans to their mainland counterparts. The demographic data and the regression resu lts provided base line comparisons between the two populations, and co rrespond to the findings reported in Chapter 1 and 2. Sociodemographic Data Table 3-1 shows selected sociodemographic characteristics of e ach population. Central to the analysis is the fact that national rates of medical insurance enro llment are corroborated by the sample in the 2005 BRFSS. Mainland Latinos, at 33% uninsured, were far less likely to have a health plan of some typeeither Medicare, Medicaid, an HM O-PPO, or private insurance than their white counterparts. Moreover, only 59% of mainland Latinos chose to answer the questionnaire in English, while 100% of Non-La tino whites chose to do so. Interestingly, the


35 mainland Latino category did not obviously differ fr om the Puerto Ricans on the island-inclusive category, All Latinos. In the mainland Lati no category, there were 18,147 respondents. The mean age was 42.32 years, with an average of 12 .47 years of schooling. The mean income range was $25-35,000. Figure 3-1 shows the distribu tion of income among mainland Latinos compared to whites. In the aggregate Latino category, 28% reported that they were uninsured, 56% took the questionnaire in English, there were 21,368 re spondents, mean age was 43.75 years, average years of schooling was 12.5 years, and the average family income was also $2535,000 a year. In contrast to the Latino population, the Non-Hispanic white category contained 275,491 individuals, with an average age of 51.3 years, and 14.04 years of schooling. The mean income range was $35-50,000, which is substantially hi gher than the comparable value among both mainland and island Latinos. The total category combined the two groups in an effort to determine if, controlling for all the socioeconomic and educationa l inequalities between Latinos and Non-Hispanic whites, there was still a measur able difference in insurance rates associated with English language proficiency. There were 297,777 total respondents, with a mean age of 50.78 years. They had an average education of 13.93 years, 97% took the questionnaire in English, the mean family income range was $3550,000, and 88% had a health care plan of some type. The proportion of the sample that was male is roughly 40% for all groups. This is interesting because it shows that most of the men may have been out of the house when the telephone survey was conducted. If this were the case, the st udy may have been positively selective for female respondents depending on the tim e of day the data are collected. Also, mean age was very high in the Non-Hispanic whites cate gory, at 51.3 years. It was similarly possible


36 that the study positively selected for older respon dents, which have traditionally been shown to be more likely to have health insurance. Howeve r, the logistic regression statistically controlled for the effects of the potential age disparity. The educational attainment of Latinos comp ared to Non-Hispanic whites is found in Figure 3-2. This creates a baseline for th e socioeconomic inequalities between the two populations. While mean education for Latinos is 12.47 years, implying that most Latinos graduate high school, the Figure shows otherwis e. While 70% of main land Latinos surveyed have completed high school or higher, roughly 30% of Latinos dr opped out before completing 12 years of schooling. Moreover, 9.7% of whites do not graduate from high school. In contrast, Latinos report a rate of 15.4% only completing the 8th grade. This shows a gap in scholastic achievement, compared to only 3.2% of whites. 90.2% of whites complete high school or higher. Moreover, 60% of whites attained at leas t an associates degree or equivalent, with 33% completing a bachelors degree or more. The marked differences between Non-Hisp anic whites and Latinos educational attainment is commonly associated with gaps in income, health indicators, and access to health insurance/care. As shown in Fi gure 3-2, educational disparities ar e reflected in the measure of combined household income. Here, the magnitude of the differences is even more evident. While the BRFSS does not ask for an exact income amount, it codes income into categories. This is not as useful as an exact mean, but the fact that the average mainland Latino family makes between $25-35,000 and the average white family makes between $35-50,000 illustrates a large wage gap that has repercussions in other asp ects of life. As the Figure shows, compared to Latinos, almost three times as many wh ite respondents were earned over $75,000. Approximately 20.6% of white families reporte d living under the fede ral poverty line of $20,200,


37 while 36.3% of mainland Latinos were also in this category. While lowering poverty rates is important, it appears that these efforts may need to be doubled to benefit the Latino community. Similarly, more than of the Non-Hispanic whites surveyed had incomes $50,000 or above, while only 22.8% of mainland Latino respondents experienced similar inco mes. In fact, the opposite is true: 80% of all mainland Latino re spondents fall in the $35,000 or under categories, with half of them at or below the poverty line. Detailed descriptions of the methods, de sign, and operation of the BRFSS are provided on the CDCs website (http://www.cdc.gov/brfss ). Variables The m ain dependent variable is the respondent s enrollment in any health insurance plan (coded for yes and for no). The exact wording of the question is: Do you have any kind of health care coverage, including health insurance, prepaid plans such as HMOs, or government plans such as Medicare? This is a broad, all-encompassi ng question designed to measure the insured status of each population. If the survey is administered in Spanish, an equivalent version is administer ed but is not available on the we bsite. In the case of Latinos, approximately 67% responded Yes, or S, to that question. While many would claim that this is a high proportion, the average age of th e sample was also 42.32 years. The literature provides many studies that show a positive correl ation between increasing age and having health insurance coverage. Therefore, the results ma y be positively skewed towards having health insurance. The results vary according to state of reside nce. The highest proportion of individuals with health insurance in any of the populations surveyed was recorded in Puerto Rico, where 94.5% of the respondents indicated that they had so me form of health coverage. This is in contrast to many areas throughout the United Stat es. While Puerto Rico has a high enrollment


38 rate, states like Nevada (56.1%) and Texas (55.4%) have exceptionally low rates of enrollment. A possible explanation for the unusually high enrollment in Puerto Rico, even compared to NonHispanic whites in the United States, is that the Puerto Rican government sponsors a program called La Tarjetita, or the Little Card. La Tarj etita is a state-run form of Medicare for which more than half of the Puerto Rican population qu alifies for coverage. In addition, it is important to note that all forms that people have to complete in order to obtain health insurance in Puerto Rico are written in Spanish, thereby facilita ting enrollment among that group. Compared with this, only 40.1% of the Latinos from Nevada who chose to take the ques tionnaire in Spanish had health insurance, compared to 72% of their E nglish speaking counterparts. To illustrate this pattern, Maine has a 91.4% Latino enrollment in health care coverage, but none of these selfidentified Latinos took the survey in Spanish. Thes e statistics show us that even before logistic regression, there is a disparity between Englis h and non-English-speaki ng Latinos access to health insurance. The objective of the multivariate analysis was to test whether the probability of having health insurance is affected by the responde nts proficiency in the English language. Unfortunately, the BRFSS questionnaire does not include an item that directly measures the ability to speak English, as does the demographic census. Despite this limitation, it is possible to derive a proxy measure, based on the language that the respondent chose to use to respond to the telephone survey. Hence, for the purposes of this study, the primary assumption was that individuals who preferred to an swer in Spanish are those who are less proficient in English (compared to individuals who completed the su rvey in English). The variable is named Questionnaire Language, and the description is Language identifier, which is coded for English and for Spanish. Normally, this variab le would be suspect due to the preferences of


39 any given individual to use their native language if provided the option. Put another way, the indicator of English language proficiency does not distinguish among individuals with different levels of ability. In addition, it is possible that some people ma y have preferred to answer in Spanish, even though they do speak English. Thes e possibilities speak to the validity of the measure of language ability used in this study and indicate potential caveat s in the interpretation of the results. However, being described by the CDC as Language identifier, it stands to reason that the interviewer iden tified the primary language of th e respondent and provided them with the option to take the survey in Spanish. Since the BRFSS questionnaire uses specialized English and high level medical te rminology, it is reasonable to pr esume that if the individual took the survey in Spanish, it was likely that he or she did not have the level of vocabulary necessary to respond in English. I hypothesized a positive correla tion between taking the survey in English and health insurance enrollment. Other demographic variables used in the regr essions were highest e ducational attainment (quantitative variable), age of the respondent (quantitative va riable), gender (dichotomous variable), and combined household income (ordinal variable). Education and age were reported in years, and males were coded (females ). Household income was reported in eight different categorical groups (coded 1 to 8), rang ing from <10,000 to >75,000. In order to include these categories in the regressions, eight dummy vari ables were constructed (y1 to y8). By treating each value as a separate category, the regression analysis will show the degree to which the probability of having health in surance is associated with each income level compared to the reference category of <10,000 a year. As shown in the literature, I expected a pos itive relationship between higher income and higher probability of enrollment in a health insura nce plan. The literature also revealed that as


40 age and educational attainment increase, so do th e rates of insurance coverage in any specific group compared to their younger, less educated c ounterparts. Gender was included to determine if there were significant differences in insura nce enrollment between males and females. I anticipated that males are less likely to be insure d compared to females. This expectation was based on the plausible assumption that women, by vi rtue of their role as mothers, may be more concerned about their health and about covering costs of childbeari ng and rearing children. Men, on the other hand, especially young men, may be mo re inclined to forgo a health insurance enrollment in order to save their money. Statistical Analysis Program Because the dependent v ariable is binary, meaning that the variable has only two possible outcomes (the respondent has health in surance, coded or the respondent does not have health insurance, coded ), the appropriate an alysis technique is logistic regression. Logistic regression applies the maximum likelih ood estimation after transforming the dependent into a logit variable (the natura l log of the odds of the dependent variable occurring or not). In this way, logistic regression estimates the odds of a certain event occu rring. Note that the logistic regression calculates ch anges in the log odds of the depe ndent variable, not changes in the dependent itself as Ordinary Least Squares (OLS) regression does. Logistic regression is analogous to OLS regression: log it coefficients correspond to be coefficients in the logistic regression equation, the standardized logit coefficients correspond to beta weights, and a pseudo R2 statistic is available to summarize the streng th of the relationship. Unlike OLS regression, logistic regression does not assu me linearity of relationship betw een the independent variables and the dependent, does not require normally distributed variables, does not assume homoscedasticity, and in general has less stringent requirements. Logistic regression is also preferred because, in the case of a dichotomous dependent variable (whose variation is limited to


41 1 and 0), OLS estimates can produce predictions that are higher or lower than the possible values. Logistic regression tests the impact of each unit change in the independent variables on the probability associated with a unit change of the dependent variable. In the following regressions, the change is from not having a he alth insurance plan to having one. The main variable being analyzed is the language in which the res pondent chose to answer the questionnaire. Those who opted to respond in Span ish are assumed to be less proficient in the use of English. When we focus on the language proficiency variable, the odds ratio (Exp(B)) value shows the increase in the probability of ha ving health insurance that is associated with English competence, net of the effects of th e sociodemographic variables included in the equation. For example, if questionnaire langua ge is the primary independent variable, the dependent variable is enrollment in a health pl an. The odds ratio is then 1.5. All else being equal, taking the questionnaire in English makes the respondent 1.5 tim es as likely to have health insurance as compared to a pers on with all the same characteristi cs who responded in Spanish.


42 Figure 3-1. Combined Household In come, Dollars (% of Respondents) Figure 3-2. Highest Educational Atta inment, Years (% of Respondents)


43 Table 3-1. Description of the Dependent and Independent Variables Used in the Analysis Population Subgroups Variable Mainland Latinos All Latinos Non-Hispanic whites Total Mean age (years) 42.32 43.75 51.3 50.78 Mean education (years) 12.47 12.5 14.04 13.93 Male (%) 40 39 40 40 English questionnaire (%) 59 56 100 97 Mean family income (range, $) 25-35,000 2535,000 35-50,000 3550,000 Income (%), <10,000 $ 10.2 13.8 5.3 6.0 10-15,000 10.6 11.7 6.0 6.4 15-20,000 15.5 15.0 7.6 8.2 20-25,000 15.5 15.0 9.8 10.2 25-35,000 15.3 14.5 13.6 13.7 35-50,000 12.9 12.1 17.3 16.9 50-75,000 9.9 8.9 17.6 16.9 >75,000 10.2 9.1 22.8 21.8 Have health care plan (%) 67 72 89 88 Married (%) 66 60 67 67 Sample size (N) 18,147 21,368 275,491 297,777 Data source: BRFSS 2005


44 CHAPTER 4 LOGISTIC REGRESSIONS AND ANALYSIS Chapter 4 pr esents the results of a multivariate analysis. The objective was to test the hypothesis that the probability of having health insurance is partly contingent upon peoples ability to speak English. Speci fically, the hypothesis was that, net of the effects of other factors that are associated with the outcome variab le, individuals who opted to respond to the questionnaire in English are more likely to have health insurance. Hence, I used the questionnaire language as an in dicator of the respondents Eng lish proficiency. I interpreted the results in light of the caveats associated with this indicator of language ability, as noted in Chapter 3. The data analysis proceeded in several st eps, beginning with an analysis that deals exclusively with the Latino populat ion, not including Puerto Rican s who reside on the island of Puerto Rico (as shown in Table 4-1). Models 1, 2, and 3 incr ementally add variables to the equation. This procedure allowed me to first examine the effect of language ability on the dependent variable, and then to determine if and the degree to which that relationship is affected by the introduction of additional variables (Models 2 and 3). The values in Model 3 are particularly important because they provide the main test of the research hypothesis. A second step in the analysis was to e xpand the scope of the data to model the determinants of health insurance within the popula tion of Non-Hispanic whites. In this case, I did not include the variable on language proficiency because, within the sample, all of the respondents chose to answer the que stionnaire in English. The analysis is none theless of interest because it enabled me to compare the results fo r the two populations. Sp ecifically, it provided insight into the similarities and differences in th e magnitude of the effect of each variable on the


45 probability of having health insurance for both the Latino and the Non-Hispanic white populations. As illustrated in Table 4-1 and 4-3, logistic regression analyses yielded results that are consistent with the research hypothesis. M odel 1 includes only one variable, whether the respondent answered the questionn aire in English or Spanish. The results were significant: without taking into account a ny other variables, mainland La tinos who responded in English were 5.18 times more likely to own health insurance compared to those who did not. The relationship is statistically significant, and the estimate of variance explained is 17.4%. The second equation, Model 2, added important variables that are commonly thought to influence the probability of havi ng health insurance: education, sex, and age. The findings show that each of these variables is associated with the outcome variab le. For example, the odds ratio for the education variable is 1.128. This can be interpreted to mean that for every increase in years of schooling, the respondent is 12.8% more likely to have health insurance. The odds ratio associated with being male is .827, which means that males are 17.3% less likely to have health insurance as women. This may be explained by the fact that wo men are usually the members of the family who are responsible for taking care of children. Adult males may have less reason to pay for health insurance without families, hoping to escape harm while living from paycheck to paycheck. Women may therefore be more preo ccupied with providing preventive and regular health care to her children. As discussed in the previous analysis of the literature, studies have shown that as an individual ages he/she is more likely to have health insurance coverage. The logistic regression corroborates this conclusion, with an odds ratio of 1.036. English language questionnaire has an odds ratio of 3.601, which is sm aller than Model 1. However, this effect is


46 still statistically significant at the .001 level, and the directi on of the effect corroborates the hypothesis. Moreover, the R Squa re value increases to .245. Model 3 adds household income into the regres sion. I expected the odds ratios associated with English language questionnaire, sex, and ed ucation to decrease but remain in the same direction. This is because income is a confounding factor that is interrela ted to ability to speak English and years of education. The odds rati o for males dropped to .726, or a difference of 10%. This may indicate a cultur al or ideological di fference between health seeking behavior between men and women, or perhaps shows the resu lts of a female-specific, targeted health insurance outreach program. Educations odds ratio value dropped from 1.128 to 1.06, which is rather small. However, if one considers an in dividual that, all else being equal, completes a bachelors degree instead of only high school, he or she will be 24% more likely to have health insurance. In Model 3, I did not expect the odds ratio for age to necessarily decrease, considering that the literature has shown that as one ages, net of all other va riables, an individual is more likely to have health care coverage. Accordingl y, ages odds ratio increased from 1.036 to 1.037. The effect of English proficiency, illustrate d by the questionnaire language, dropped to 2.682. The decline in the size of the coefficient is du e to the effects of the other variables in the equation. However the ability to speak English is still by far the larges t single variable with respect to health insurance coverage. The income categories show the odds ratios compared to an individual reporting making less than $10,000 a year. It is worth noting that, compared to the lowest income category (which serves as a reference group in the analysis), the odds ratio for the first two groups ($10-15,000 and $15-20,000) are not statis tically significant. These findings i ndicate that, with respect to the


47 probability of ownership of health insurance, there are very little differences among lower income families (i.e. earning $20,000 or less). Income becomes a statistically significant pred ictor of the probability of having health insurance when income exceeds $20,000. The odds ratio for families that earn $20-25,000 is 1.179, indicating that, compared to the lowest income stratum, they are 17.9% more likely to own health insurance, net of the effect s of the other variables in the equation. A second observation concerns the pattern of the odds ratio at other levels of income. As the figures in Table 4-1 show, there is a steady increase in income effect. It increases from 1.734 in the $25-35,000 category, to 6.915 in the highest in come strata. The pattern of increase is shown in Figure 4-1, which graphically disp lays the odds ratios by income category. The Figure clearly shows that the probability of having a health insurance policy remains low and fairly constant among families with a total annual income of $25,000 or less. Increases in income are associated with higher odds ratios. As evident in the gra ph, the largest jump is evident when we compare the odds ratio fo r the $35-50,000 (2.98) with the odds ratio for $5075,000 (5.288). All else being equal, a person making more than $75,000 a year is 6.915 times as likely as a person in the lowest income category to have health insurance. Table 4-2 presents results similar to those in Table 4-1, but allows a comparison between islandand mainland-dwelling Lati nos in the United States. The central difference among these groups is that this regression in cludes island-dwelling Puerto Ricans, who do not need to speak English, in the analysis. I incl ude them to give a country-wide view of the aggregate Latino populations health insurance rate and the factors leading to that rate. Because Puerto Ricans who live on the island of Puerto Rico are offered health insurance enrollment in their own language, the effect of questi onnaire language in Model 1 is immediately reduced (2.547)


48 compared to only analyzing their mainland Latino counterparts. Also, the R Square value drops from .174 to .061, thereby demonstrating the lack of need of English proficiency on the island. However, English language proficiency on the is land has been shown to be associated with higher levels of education and income, there by confounding their ability to procure health insurance, even if the forms are in Spanish. In Model 2 of Table 4-2, the odds ratio for questionnaire language (1.957) is again lessened due to the interaction of other variables. Ages odds ratio is 1.046 as compared to mainland Latinos 1.036, thereby de monstrating that age on the island is significantly more important to having health insurance. Being ma les odds ratio is associated with a decrease compared to Table 4-1, from .827 to .797, again demonstrating that on the island being male makes one significantly less likely to be enrolled in health insurance. For each year of education, the respondent is 1.186 times more like ly to have health insurance. This is a significant increase over mainland-only Latinos value at 1.128. Finally, the R Square climbs to .204, illustrating the importance of these other variables on the odds ratios for Puerto Rican islanders. Model 3 includes the categories for househol d income, like Table 4-1. Again, all the coefficients for each of the variables except age d eclined, as they did in Table 4-1 from Model 2 to Model 3. Ages odds ratio moved from 1.046 to 1.045, but being males changed from .797 to .73. This signifies that even with the same amo unt of income, education, and age, a male would be 27% less likely than a female to enroll in heal th insurance. Interestingly, this coefficient is the same for the regression concerning only mainland Latinos. The impact of years of education on the odds of having health in surance declines to 1.133 from 1.186. However, compared to 1.06 for mainland Latinos, this remains a strong pr edictor of health insurance status on the island. This effect may be explained by th e targeted insurance programs on the mainland


49 positively selecting for lower education, or perhaps it may be a cultural priority for those individuals living on the mainla nd. The coefficients for income categories show a marked difference between Tables 4-1 and 4-2. Mainland Latinos in the $10-15,000 $15-20,000, and $20-25,000 income categories are about as or more likely than someone making <$10,000 a year to have health insurance. Adding in the influe nce of island-dwelling Puer to Ricans lowers the odds ratios to .728, .656, and .787, respectively. This shows that the indigent Puerto Ricans on the island are receiving some form of free health care insurance, while many with only slightly higher incomes are not. However, as income in creases in the aggregat e Latino regression, the odds ratios increase as well, albeit more slowly than in the mainland La tino regression. Again, this suggests that there may be a cultural bias selecting for health in surance enrollment among the stateside Latinos. Table 4-3 includes regression re sults for Non-Hispanic whit es as well as all Latinos, mainland Latinos, and Total respondents. It is no t appropriate to include the language variable for Non-Hispanic whites because all the respondents answered in English. However, it is still important to determine if the ot her variables behave in the same way as they did for Latinos. Hence, Table 4-3 reports regr essions on all the populations. This provides us with the opportunity to look at patterns that arise between the groups, and to identify inequities. For NonHispanic whites, the two lowest income groupings were not statis tically significant due to the low number of respondents that fell into thes e two categories. Moreover, while for islanddwelling Puerto Ricans English la nguage proficiency is not importa nt to having health insurance, and they seem to be poorer on average than th eir mainland counterparts, island-dwelling Puerto Ricans seem to have nearly identical values for all other variables in the regression.


50 An interesting quirk revealed by the regr essions is the fact that among mainland, all Latinos, and Total respondents, those ma king between $10,000 and $25,000 are less likely to have health insurance than those making less than $10,000. This may well be due to the fact that there are several federaland state-funded initiatives to prov ide health care to indigent individuals and their families. There are several targeted, subsidized health insurance programs in larger cities, but the problem is the enro llment of qualifying indi viduals. Whether the confounding factor is their immigr ation status, or if they simply do not know enough English to find information about the programs, these issues n eed to be addressed. Also, it is possible that individuals just at or above the poverty line may not qualify for any governme ntal aid, leading to lower rates of enrollment overall. While the Non-Hispanic whites group is not statistically significant in this area of income due to lo w numbers of people falling in these income groupings, those individuals right below the pover ty line of $20,200 dollars seem to be just as likely as lower income respondents to have heal th insurance, lending cr edence to the targeted health insurance program hypothesis. Each year of added education among all Latino respondents seems to have a slightly larger impact, with an odds ratio of 1.133 comp ared to whites ratio of 1.094. However, for mainland Latinos alone, years of education, at 1. 06, does not play a large role in determining health insurance enrollment. This may be due to the fact that most mainland Latinos are perhaps first generation formal students, even if their families have lived in the United States for decades. whites, on the other hand, are sometimes multi-generation descendents that are already well established economically and socially. Non-La tino whites generally strive to have health insurance, education, and income but the odds of a white indivi dual having health insurance are better, all other circumstances bei ng equal, compared to Latinos.


51 Analyzing the high end of the income spectrum for Latinos and Non-Latino whites reveals an interesting pattern. A mainland Latino individual making over $50,000 is about five to seven times as likely as a Latino in a lower income grouping to have health insurance. In contrast, a white in the same income category is six to twelve times as likely to have health insurance as one of his compat riots making less than $10,000. This shows that comparatively, whites are much more likely to have health insura nce, at any income level, than a Latino. Again, this may be due to the fact that insurance comp any regulations are difficult to navigate, and that completing the requirements to gain access to physicians and services, as well as benefits, is difficult. Ultimately, it is easy to tease out ine qualities between mainland and all Latinos, NonHispanic whites, and total respondents. The impo rtant observation to note is that, all else being equal, increasing years of education, age, inco me, and being female all have the effect of increasing an individuals odds of having health insurance. Ho wever, a strong association for Latinos compared to each other is questionnaire language. Base d on the odds ratios reported in Table 4-3, mainland Latinos need to speak English more than island-dwelling Puerto Ricans. On the other hand, any Latino speaking English is associated with a substa ntial increase in the probability for health insurance enrollment versus a monolingual person with the same characteristics.


52 Figure 4-1. Mainland Latinos Odds Ratios for Health Insurance Based on Income, Dollars (Compared to <$10,000)


53 Table 4-1. Health Insurance Regressed on Englis h Language Proficiency, Sex, Age, and Income Logistic Regression on Mainland Latinos(Odds Ratio) Variable Model 1 Model 2 Model 3 English language questionnaire (1=yes) 5.18 3.601 2.682 Education (in years completed) 1.128 1.060 Sex (1=male) 0.827 0.726 Age (in years) 1.036 1.037 Income (reference = $<10,000) 10-15,000 0.910* 15-20,000 0.962* 20-25,000 1.179 25-35,000 1.734 35-50,000 2.984 50-75,000 5.288 >75,000 6.915 Constant 1.7420.095 1.792 Nagelkerke R Square Value 0.1740.245 0.303 Total respondents (N) 18,14718,147 18,147 Data source: BRFSS2005 *=Not significant; all others are significant at <.001


54 Table 4-2. Health Insurance Regressed on Englis h Language Proficiency, Sex, Age, and Income Logistic Regression on All Latinos(Odds Ratio) Variable Model 1 Model 2 Model 3 English language questionnaire (1=yes) 2.547 1.957 1.446 Education (in years completed) 1.186 1.133 Sex (1=male) 0.797 0.730 Age (in years) 1.046 1.045 Income (reference = $<10,000) 10-15,000 0.728 15-20,000 0.656 20-25,000 0.787 25-35,000 1.179* 35-50,000 2.066 50-75,000 3.563 >75,000 4.326 Constant 2.4850.046 0.279 Nagelkerke R Square Value 0.0610.204 0.253 Total respondents (N) 21,36821,36821,368 Data source: BRFSS2005 *=Significant <.01; all othe rs are significant at <.001


55 Table 4-3. Health Insurance Regressed on Englis h Language Proficiency, Sex, Age, and Income Logistic Regression on all Populations (Odds Ratio) Population Subgroups Variable Mainland Latinos All Latinos Non-Hispanic whites Total English language questionnaire (1=yes) 2.682 1.446 1.564 Education (year) 1.060 1.133 1.095 1.103 Sex (1=male) 0.726 0.730 0.721 0.721 Age (year) 1.037 1.045 1.045 1.045 Income <$10,000=reference 10-15,000 0.910* 0.728 1.009* 0.951 15-20,000 0.962* 0.656 0.992* 0.921 20-25,000 1.179 0.787 1.228 1.137 25-35,000 1.734 1.179 2.036 1.868 35-50,000 2.984 2.066 3.650 3.356 50-75,000 5.288 3.563 6.799 6.240 >75,000 6.915 4.326 12.318 11.050 Constant 1.792 0.279 2.856 1.625 Nagelkerke R Square Value 0.303 0.253 0.206 0.230 Total respondents (N) 18,147 21,368 275,491 297,777 Data source: BRFSS 2005 *=Not significant; all others are significant at <.001


56 CHAPTER 5 CONCLUSIONS AND IMPACT Roughly 44% of Latino respondents to th e 2005 BRFSS surv ey chose to respond in Spanish rather than English. This suggests that language is a impor tant factor to consider in a multilingual country such as the United States. Teaching Latinos and immi grants in general to speak English, while repressing their native culture and language, is detrimental to their integration into American society. While many scholars argue the opposite, studies have shown that additive-bilingual education shows positive l earning effects for bilingual education versus standard ESOL education (Bamford and Mizokawa 1991:413-429). The Latino population group surveyed in the BRFSS illustrates this patt ern. While Puerto Ricans on the island have over a 90% enrollment rate in a health insuranc e plan of some type, which is higher than NonHispanic whites rates, Latinos on the mainland in some areas do not reach even 40% enrollment. Moreover, logistic regression analysis of BRFSS data from 2005 shows that, all else being equal, mainland Latinos who speak Englis h are 2.682 times more likely to have health insurance than their non-English-speaking counterparts. That odds ratio for the complete model is statistically significant to a level <.001, and the R Square value is .303. These values reflect the fact that the odds ratio is statistically significant, the eff ect is large, and the regression accounts for a large portion of the variance. In e ffect, English language ab ility is a determining factor, even after including many cont rol variables in the regression. Suggestions for Policy Making Results of the regression anal ysis indicate two possible explan ations. The first is that Spanish speakers experience institutionalized di scrimination because the steps needed to obtain health insurance are only available in English. Hence, their emergence of English proficiency may discourage people from enrolling in insurance, even if they can afford to do so. If this is so,


57 then the first set of policy implications would be to create bilingual enrollment forms and to provide interpreters at appropr iate points during the enrollment process. Bilingual health insurance representatives, bilingual forms and paperwork once enrolled, and bilingual advertising, are key to the success of increasi ng health insurance enrollment. Targeted, subsidized insurance programs should also be advertised in the targ eted neighborhoods and areas. While this initiative may be labeled by some to be a racist policy, the United States Census website is capable of producing maps showing populati on density for each minority, including Latinos. Distributing fliers or putting up a billboard in a predominantly-Spanishspeaking area may yield a positive response from the community. As shown in Chapter 1, Spanish-language advertising leads to a significant increase in response by individuals who did not realize that the program existed or that they qualified. The second possible explanation is that ther e is a potential cultural difference between Latinos and the rest of the United States populati on. It could be the case that individuals of Hispanic origin are less inclined to invest their money into health insurance. This could be for cultural reasons, or for reasons a ssociated with different and possi bly negative life experiences in Latin America. If this is the case, then the policy implications are different. The appropriate method of mitigating this phenomenon would be to focus on educating Latinos as to the value of having health insurance. Putti ng hospital stays and even preven tive care into a dollars-and-cents framework while showing the pot ential benefitsof adding nearly twenty years to their life expectancywould perhaps make health insu rance enrollment a priority for most. Steps should also be taken to alleviate th e strain of health insurance premiums and associated costs for those families just above th e poverty line. While prov iding health insurance at no or little cost is obviously an important step to eliminatin g the disparities between high and


58 low socioeconomic status families, those families th at are close to but not considered living in poverty also need to be cared for. If they were to pay several thousand dollars a year for insurance premiums, co-pays, and prescription me dicine costs out of their total income, nearpoverty individuals will be well unde r the poverty line, yet still be paying out of pocket for all medical care. Therefore, creating a program for at-risk families to be placed on, at the least, subsidized health insura nce programs is one way to remedy the situation. Further prospective plans of action could be to provide at-risk, poor families children with free preventive health care visits. Most re search, including this thesis review of the literature, demonstrates the neces sity of preventive care to make a difference in this generations educational and earning potentials. In a true effort to mitigate the effects of the ever-widening gap between high and low socioeconomic status familie s, this is the first possible step that could be taken. Questions for Further Research This study probes a central question regard ing Latino populations throughout the United States: that of English language proficiency and access to health care insurance. However, there are additional issues to explore before leading to a full understanding of th e possible pitfalls and logistical problems in enacting legislation to increase insurance enrollment among at-risk families. For one, does having access to a bilingual insurance enrollment process and bilingual staff allow for health insurance utilization at higher rates? This may show insurance companies the efficacy and relatively low costs of enacting such a policy, as well as the effectiveness thereof. If new enrollment were shown to be an effect of instituting bilingual insurance programs, then insu rance companies may be more likely to implement them. Another question for further res earch is, if parents were in formed of a program providing preventive and emergency health care to their ch ildren, regardless of th eir legal status, would


59 they enroll? It is possible that an undocumented parent would stil l not enroll his or her children in such programs because of their fears about deportation. These fears are valid, considering the fact that California and other states have im plemented legislation designed to limit access to emergency health care for undocumented immigrants, or to create an aggressive and anti-Spanish environment in professional specialties. Further research may include a qualitative i nvestigation of indivi duals who responded to the BRFSS. This may include conducting inte rviews (individual and focus groups) with respondents who do not have access to health insu rance and investigating their English language ability levels in various contexts (including medicine). It is also possible to then better direct the response in policy so as to help enroll the most individuals. Moreover, it is important to detail the different experiences among various Latino communities in the Unit ed States and on the island of Puerto Rico. Ultimately, because this study is statistical and quantitative, it is difficult to know for sure the reasons behind these social inequalities. Qualita tive follow-up work should be a staple part of any further research on this topi c, but lies outside of the scope of this thesis. An important part of many studies is con tinuity. Because the BRFSS is a continuous study, it is possible to use my statistical analysis and techniques to continue my research with new data sets as they are made available. This may be done in an effort to further corroborate my findings, or to track any changes or similari ties between each years language proficiency and access to health insurance. It also may serve to demonstrate the ever-widening divide between Latinos and Non-Latino whites access to care. The BRFSS could be expanded to include more detailed questions and to have more states use the optional m odules. For example, an exact number for income could be helpful in teasing out more subtle patterns in income relati ve to other variables. Or perhaps additional


60 questions could be utilized to ask for their immigr ation status, or to creat e an access to care index which could then point out at-risk groups. As it is, not every state asks extra modules about health status, health care access, continuity of health care, and other risk behaviors, like smoking or drinking. If this module could be used more uniformly, then further research could be performed to look at health care access a nd utilization among people throughout the United States. My original research que stions was limited due to this problem. In conclusion, this study was a beginning, quantitative investigation into th e relationship between English language ability (as determined through respondents language choi ce in a telephone survey) and access to health insurance. This study found a strong, positive correlation between English language choice and access to health care. Further studies may investigate this topic using alternative, qualitative research methods.


61 LIST OF REFERENCES Am erican Anthropology Association 1996 Statement on Race. AJPA 101:569-570. Baughman, R. 2007 Expanding public health insurance for ch ildren: do local economic conditions matter? International Journal of Health Care Finance and Economics 7(1):1-22. Bamford, K. W., and D. T. Mizokawa 1991 Additive-Bilingual (Immersion) Educati on: Cognitive and Language Development. Language Learning 41(3):413-429. Centers for Disease Control and Prevention (CDC) 2005 Behavioral Risk Factor Surveillance System Survey Data Electronic document, http://www.cdc.gov/brfss accessed April 13, 20 06. Doescher, M. P., with B. G. Saver, K. Fiscella, and P. Franks 2001 Racial/Ethnic inequities in continuity and site of care: location, location, location. HSR: Health Services Research 36(6):78-89. Dressler, W. W., with K. S. Oths, and C. C. Gravlee 2005 Race and ethnicity in public health rese arch: Models to explain health disparities. Annual Review of Anthropology 34(1):231-252. Editorial 1956 Bracero Program. El Paso Post, April 1956: 1A-5A. Flores, G., with M. Abreu, and S. C. Tomany-Korman 2005 Limited English proficiency, primary lan guage at home, and disparities in childrens health care: how language barriers are measur ed matters. Public Health Reports 120:418430. Flores, G., and S. C. Tomany-Korman 2008 Racial and Ethnic Disparities in Medical and Dental Health, Acce ss to Care, and Use of Services in US Children. Pediatrics 121:286-298. Garcs, I., with I. Scarinci, and L. Harrison 2006 An Examination of Sociocultural Factor s Associated with Health and Health Care Seeking among Latina Immigrants. J Immigrant Health 8:377-385. Geronimus, A. 1992 The weathering hypothesis and the health of African-A merican women and infants. Ethnicity & Disease 2:207-221.


62 Gonzlez, Juan 2000 Harvest of Empire: A History of Latinos in America. New York, NY: Penguin Group. Granados, G., with J. Puvvula, N. Berman, and P. T. Dowling 2001 Health care for Latino children: impact of child and parental birthplace on insurance status and access to health services. Amer ican Journal of Public Health 91(11):18061807. Higgs, T. 1984 Teaching for Proficiency, the Organizing Principle. Lincolnwood, IL: National Textbook Company. Hubbell F. A., with H. Waitzkin, S. Mi shra, J. Dombrink, and L. R. Chavez 1991 Access to medical care for documented and undocumented Latinos in a southern California county. West J Med 154:414-417. Hunt, K. A., with A. Gaba, and R. Lavizzo-Mourey 2005 Racial and ethnic disparities and percep tions of health care: doe s health plan type matter? HSR: Health Services Research 40(2):551-576. Jacobs, E. A., with D. S. Shepard, J. A. Suaya., and E. Stone 2004 Overcoming language barriers in health car e: costs and benefits of interpreter services. American Journal of Pub lic Health 94(5):866-869. Kerner, J. F., with L. Dusenbury, and J. S. Mandelblan 1993 Poverty and Cultural Diversity: Chal lenges for Health Promotion Among the Medically Underserved. Annu. Rev. Pub. Health 14:355-377. Morales, L. S., with W. E. Cunningham, J. A. Brown, H. Liu, and R. D. Hays 1999 Are Latinos less satisfied with communi cation by health care providers? J Gen Intern Med 14:409-417. Morales, L. S., with W. E. Cunningham, F. H. Galvan, R. M. Andersen, T. T. Nakazono, and M. F. Shapiro 2004 Sociodemographic differences in acce ss to care among Hispanic patients who are HIV infected in the United States. American Journal of Public Health 94(7):1119-1121. Nickens, H. W. 1991 The health status of minority populations in the United States. West J Med 155:27-32. Prez-Stable, E. J. 1987 Issues in Latino Health Care-Medical Staff Conference. University of California, San Francisco. West J Med 146:213-218.


63 Sarver, J., and D. W. Baker 2000 Effect of language barriers on followup appointments after an emergency department visit. J Gen Intern Med 15:256-264. United States Census Bureau 2001 Overview of Race and Hispan ic Origin. Electronic document, http://www.census.gov/prod/2001pubs/c2kbr01-1.pdf accessed March 1 5, 2008.


BIOGRAPHICAL SKETCH Karl Slazinski was born in Sarasota, Florida, where he attended Pine View School for the Gi fted from 2nd to 12th grades. After completing high school, he went on to New York University as a National Merit Finalist for his Bach elor of Arts in Spanish. After completing his degree in three years, he conti nued for a year of post-baccalau reate work in science at the University of Florida to complete the pre-medicine requirements. He fini shed his last semester in the Master of Arts in Latin American Studies with a specialization in Latino Studies at the Center for Latin American Studies at the Univer sity of Florida in the spring semester of 2008. Karl is interviewing for admission to medical sch ool throughout Florida and in Puerto Rico, and plans to matriculate into the Doctor of Medicine program in the fall of 2008.