Antiretroviral therapy (ART) has led to marked reductions in mortality and morbidity among HIV-positive individuals, yet from the outset optimal adherence to prescribed regimens has remained essential to its success.1,2 Identifying subgroups at risk for nonadherence and tailoring treatment to address their needs or to otherwise address factors causing the disparities may help optimize treatment success.
Racial/ethnic minority group status has been linked repeatedly to lower levels of ART adherence and viral suppression among HIV-positive cohorts in the United States.3–10 For example, in a large sample of patients enrolled in an integrated health care system (thus equalizing access to care), Silverberg et al11 reported that adjusted mean refill adherence, operationalized as the percentage of days with antiretroviral (ARV) medication coverage between refills, was lower among patients who were African American (64%) or Hispanic (65%) compared with white (70%). Based on data from the Multicenter AIDS Cohort Study, Lee et al12 reported that after controlling for site, age, financial difficulties, popper use, crack use, and viral load, Hispanics were 2.2 (95% CI: 1.5 to 3.2) and African Americans were 1.4 (95% CI: 1.1 to 1.8) times more likely than whites to report having taken less than 100% of all their prescribed HIV medication doses over the previous 4 days.
Most prior studies have important methodological limitations. Some have small samples, which limits statistical power, and others use only self-reports of adherence, which can overestimate actual adherence. In addition, few prior studies adequately controlled for potential confounders known to be important in racially and ethnically diverse populations including depression13 and substance use.14–16 Psychiatric disorder and drug dependence are highly prevalent among individuals with HIV, with positive screening for each as high at 50% in the HIV Cost and Services Utilization Study.17 Although data on racial/ethnic differences in these potential barriers to adherence are mixed,11,17,18 racial/ethnic minority groups in the United States seem to have less access to high quality treatment for these conditions, which may exacerbate their effects on nonadherence.19,20
Most of the studies examining racial/ethnic difference in adherence included African Americans, with smaller numbers of and fewer comparisons with Latinos. Those including Latinos generally showed them to be at greater risk for nonadherence than Whites.21 Waldrop-Valverde et al22 concluded that there is no consensus on whether African Americans adhere worse than individuals from other racial/ethnic groups, with demonstrated differences possibly due to adherence measurement error or comorbidities such as depression and substance abuse.
In the present study, we analyzed data from the Multi-site Adherence Collaboration on HIV (MACH14) to examine the association of race/ethnicity to adherence. Advantages of MACH14 were that adherence was measured electronically, the population was large relative to most prior studies, African Americans and Latinos were well represented, and we were able to control for key potential confounders including depressive symptomatology and substance use.
Data came from a National Institute of Mental Health–funded project combining data from 16 studies of electronically monitored ART adherence conducted at 14 sites in the United States.23 Demographic, psychosocial, and biological data assessed across the studies were combined in the pooled data via the creation of new variables incorporating data across studies. The present study focused on 13 studies conducted at 11 research sites, excluding the 3 studies that collected no data on income or education level. Only individuals self-identifying as African American, Latino, or white were included. Prospective predictors were taken from the baseline assessment (ie, the first assessment point at which they were collected). As indicated below, we used a missing indicator approach to deal with missing data.24
These included site, gender, age in years, income (percent reporting an annual household income of more than $10,000), and education level (percent graduating high school). Race/ethnicity was determined by self-identification as white, African American, or Latino.
Adherence was measured with the Medication Event Monitoring System (MEMS; http://www.aardex.ch), a pill bottle and cap that electronically records the date and time of each opening as a presumptive dose. The adherence outcome was operationalized as percent adherence across the entire period the participant was observed, up to 12 months. Specifically, adherence was calculated separately for each of the first 12 months after a participant's baseline assessment. Each month was defined as a 4-week period with the percent adherence calculated across monitored ARV medications. Percent ART adherence was calculated as the number of doses taken divided by the total number of doses prescribed during each 4-week period. An overall index of ART adherence was derived by averaging adherence for each of the monthly intervals during which a participant was monitored. For studies that followed participants for less than 12 months, the adherence outcome reflected the average for the monitored months.
Different measures of depression were administered across studies as follows: 4 used the Beck Depression Inventory version II25; 1 used the original version of the Beck Depression Inventory26; 3 used the Center for Epidemiological Studies Depression Scale27; 2 used the depression subscale of the Brief Symptom Inventory28; and 1 used the Primary Care Screener for Affective Disorders.29 To aggregate data from these different scales, we first calculated standardized Z scores (ie, subtracting the normative—not the study—mean of the scale from the sum of the scale items in the particular study and then dividing by the normative standard deviation).30 This approach, which we used successfully in a prior study (Wagner et al31) circumvents the potential bias of participants in each study receiving Z scores based on their level of depression relative to the other participants in that particular study rather than the wider population.
The participant's Z score was denoted as missing if more than half of the depression scale items were missing for that individual. Based on this continuous measure, we determined a 3-category diagnosis of “minimal, mild/moderate, or severe” depression, reflecting established diagnostic cutoffs for each depression measure. A fourth category of “missing” was included for participants who were administered a depression scale without diagnostic cut-offs or for whom a depression score was not available.
Because multiple drug use is common and combining a range of substances used may create a more reliable predictor than individual drug use indicators,32 we computed a categorical indicator encompassing any use of the following substances: cocaine/stimulants, heroin/opiates, ecstasy, hallucinogens, or sedatives. Alcohol use was not included in the indicator as drinking to intoxication was assessed only at a single site, so it was not possible to differentiate problem drinking. The time frame of substance use varied by study and ranged from the last 30 to 365 days. An additional category of “missing” was included for participants without substance use data.
Descriptive analyses were performed to identify means, standard deviations, and bivariate correlations of the demographic, depression, substance use, and adherence variables. We used logistic regression in combination with robust standard error correction to evaluate medication adherence as a continuous variable between 0 and 1. This approach has been previously recommended for the analysis of continuous percentage data.33,34 (The magnitude and statistical significance of the results were comparable under a linear regression model.) The unadjusted association between adherence and the demographic, depression, and substance use variables was examined with logistic regression. Additional logistic regression models were used to examine subgroup associations between race/ethnicity and adherence at each level of the other demographic, depression, and substance use variables.
Because the data originated from multiple studies, we first conducted preliminary analyses to identify any site-specific associations between the medication adherence outcome and each of the demographic, depression, and substance variables. For each predictor variable, the adherence outcome was regressed on site, the predictor, and the predictor × site interaction in separate models. An omnibus Wald χ2 test of the predictor × site interactions evaluated whether the relation between the predictor and medication adherence varied by site. The interaction effect with site was significant for race/ethnicity, age, depression, and substance use. Consequently, we controlled for the main effect of site and its interactions with those 4 variables in the final model. Site effects were modeled using a fixed effect approach because of the small number of clusters and because the sites were not sampled randomly from all possible locations in the United States. Based on the final model, adjusted mean adherence was calculated for African American, Latino, and white participants with all other variables held at their mean levels. In the case of categorical covariates such as gender, the mean level represented the average proportion of individuals in the respective categories. Consequently, the adjusted mean reflected a weighted average of adherence across the different subgroups.
Next, a sequential logistic regression was used to determine if the association of race/ethnicity with ART adherence would persist after controlling for relevant (1) demographic and (2) substance use and depression confounders. Omnibus Wald χ2 tests evaluated whether the variables in each step of the model were collectively associated with medication adherence. In step 1, race/ethnicity (predictor 1), age in years (2), gender (3), annual household income (4), and education (5) were entered into the model to account for demographic associations with medication adherence. In step 2, substance use (6) and depression (7) were entered to assess the additional effect of mental health symptoms on adherence after controlling for demographic factors, site, and the interaction of site with all predictors that varied by site. As site was not of substantive interest in the final results, we report the average effect across sites all predictions, weighted by the relative sample size among the cities included in the analysis.
The final analytic sample of 1809 participants ranged in age from 18 to 72 years and was mostly (67%) male. Participants were 53% African American (n = 951), 14% Latino (n = 250), and 34% white (n = 608). Although most were well educated (66% had a high school or general education degree and 11% had a college degree), more than half (59%) reported an annual household income of less than $10,000. Forty-one percent were heterosexual, 33% were gay or lesbian, and 14% were bisexual. With respect to baseline medication, 31% of participants were on a once-daily regimen, 61% were on a twice-daily regimen, and 8% were on a thrice-daily regimen (Table 1). Regimens were based on a protease inhibitor (51%), a nucleoside reverse transcriptase inhibitor (29%), a nonnucleoside reverse transcriptase inhibitor (16%), a nucleotide reverse transcriptase inhibitor (3%), a fusion inhibitor (<1%), or another medication (<1%).
To assess for differences between participants with complete data (60%), those who were missing substance use only (2%), depression only (15%), or both substance use and depression data (23%), χ2 tests and 1-way analyses of variance were conducted on categorical and continuous demographic characteristics, respectively. There was no racial/ethnic difference between individuals with complete data and those without. Individuals with complete data were more likely to have at least a high school diploma [χ2 (2) = 20.39, P < 0.001], possess a higher annual household income [χ2 (4) = 21.11, P < 0.001], be female [χ2(1) = 11.22, P = 0.001], and be of older age [F (1,1807) = 18.21, P < 0.001].
Descriptive and Bivariate Analyses
Means, standard deviations, and bivariate correlations of the adherence, demographic, substance use, and depression variables are presented in Table 1. The mean standardized depression score was 0.77 (SD = 1.65). Among the subgroup of participants from studies that used a depression scale with established diagnostic cutoffs (n = 1115), 60% (n = 673) had none or minimal signs of depression, 25% (n = 282) had mild to moderate depression, and 14% (n = 160) had severe depressive symptomatology. Among the participants with substance use data (n = 1364), 65% reported using no substances in the 30–365 days preceding their baseline assessment, 26% reported using one substance, and 10% reported using 2 or more substances. African American [35%; χ2 (1) = 6.38, P = 0.01] and Latino [48%; χ2 (1) = 21.93, P < 0.0001] participants were more likely to report use of at least 1 substance compared with white participants (28%).
Mean adherence levels by race/ethnicity and other prospective predictors are reported in Table 2. African American (65%), Latino (64%), and white (72%) evidenced an omnibus difference in mean adherence, χ2 (2) = 22.96, P < 0.001. With respect to bivariate analyses, all the predictor variables had statistically significant associations with adherence (P < 0.05).
Sequential Logistic Regression
The sequential logistic regression model results are presented in Table 3. The demographic variables in step 1 were collectively associated with medication adherence [omnibus χ2 (36) = 172.32, P < 0.0001]. When controlling for only demographic variables, race/ethnicity was a significant predictor of ARV medication adherence [omnibus test: χ2 (21) = 119.09, P < 0.0001]. The substance use and depression variables introduced in step 2 improved upon the prediction of medication adherence [χ2 (24) = 61.94, P < 0.0001]. When controlling for demographics, substance use, and depression, race/ethnicity remained a significant predictor of medication adherence [omnibus test: χ2 (21) = 122.10, P < 0.0001].
The relationship between race/ethnicity and adherence was comparable in the model controlling for only demographic variables (step 1) and the model controlling for demographics in addition to substance use and depression (step 2). In the full model controlling for all variables, the odds of 100% medication adherence were 40% lower [odds ratio (OR) = 0.60, 95% CI: = 0.52 to 0.70, P < 0.001] among African Americans compared with white participants and 28% lower (OR = 0.72, 95% CI: = 0.53 to 0.98, P = 0.04) compared with Latino participants. The odds of 100% medication adherence were comparable among Latinos compared with white participants (OR = 0.84, 95% CI: = 0.61 to 1.15, P = 0.27). Adjusted mean adherence in the full model was 61% among African Americans (95% CI: = 59% to 64%), 69% among Latinos (95% CI = 63% to 75%), and 72% among white participants (95% CI: = 69% to 75%). Mean adherence among African Americans was significantly lower than adherence among Latinos (Estimate = −8%, P = 0.04) and whites (Estimate = −11%, P < 0.001). The difference in mean adherence between Latinos and whites was not statistically significantly (Estimate = 3%, P = 0.27).
In this analysis of pooled data from 13 studies in the United States, racial/ethnic disparities in adherence persisted even after controlling for key demographic variables (ie, sex, age, income, education, and site), depression, and substance use. The underlying mechanisms leading to these racial/ethnic disparities remain unclear, although we can speculate based on the experience of individuals of minority race/ethnicity in the United States.
The marginalization experienced by racial/ethnic minorities is the United States can be understood as involving overt discrimination as well as “microaggressions”, which are “brief and commonplace daily verbal, behavioral, or environmental indignities, whether intentional or unintentional, that communicate hostile, derogatory, or negative racial slights and insults toward people of color”.35 As discrimination is related to worse health outcomes and maladaptive health behaviors,36 it may be implicated in worse ART adherence.37 For example, among 152 HIV-positive African American men who have sex with men, Bogart et al38 found that 40% reported racial discrimination in the last 6 months, which was significantly associated in longitudinal analyses with lower adherence. Racial/ethnic minorities who experience discrimination may find it more difficult to trust their providers, a distrust which may also negatively impact adherence. For example, Saha et al39 found that compared with white patients (n = 201), African Americans (n = 1104) expressed lower levels of trust in their provider, which was associated with worse adherence.
Levels of trust in providers may have a basis in experience. A study of 1886 participants (54% white, 28% black, 14% Hispanic) from the HIV Cost and Services Utilization Study21 found that 40% reported having a discriminatory health care experience since their HIV diagnosis and that 24% failed to “completely” or “almost completely” trust their health care providers. In a structural equation model in the same study, discrimination predicted distrust, weaker treatment benefit beliefs, and, in turn, poorer adherence. Distrust seemed to operate on medication adherence by increasing treatment-related psychological distress and weakening treatment benefit beliefs.
Related recent work from Bogart et al38 has pointed to conspiracy beliefs as possible explanations. Among 177 blacks, they identified 2 distinct belief categories: genocidal beliefs (eg, HIV was created by humans) and treatment-related beliefs (eg, people on ART are the government's guinea pigs). Although both genocidal and treatment-related conspiracy beliefs were related to electronically monitored adherence in univariate analyses, in multivariate tests only the latter were linked to a lower likelihood of optimal adherence at 1 month. Most of this work has been done with African Americans; more work with Latinos is needed.
Health literacy is another possible explanation for racial/ethnic disparities in ART adherence. People with low literacy may not be able to read or identify their medications, which makes it difficult for them to adhere to their prescribed regimens.40 HIV-positive African Americans with low levels of educational literacy have been found to be more nonadherent to ART.41 In addition, Osborn et al42 found that when health literacy was included as a mediator in their analyses, the association of African American race and worse 4-day self-reported ART adherence was reduced to nonsignificance. More specifically, Waldrop-Valverde et al22 investigated numeracy or the ability to understand and use numbers in daily life. They found that poor management of a simulated ARV regimen among African American men and women with HIV was mediated by lower numeracy.
The current research points to many possible directions for future intervention. As Bogart et al43 have pointed out, adherence interventions designed to address culturally specific roots of nonadherence may help to overcome medical mistrust. These might involve culturally tailored components delivered ideally by peer advocates or community-level interventions with recognized popular opinion leaders. Also, future work will need to disaggregate the racial/ethnic categories summarized for convenience here into the broad groupings of “African Americans” and “Latinos.” Earlier work based on the Multi-site AIDS Cohort Study, for example, indicated that blacks of Caribbean descent had worse adherence than those from other regions.12 There may be important Latino subgroup differences as well. We know of no work that has begun to investigate these differences.
Limitations of the present study include the reliance on self-report measures of depression and substance use and their lack of uniformity in measurement across studies. For example, a consistent duration for substance-use assessment across all studies (instead of the actual range of 30–365 days) would have been preferable but would have necessitated our limiting the analyses to the few studies that used the same duration for measurement. Our current analysis represents a conservative estimate of controlling for any substance use at all but may mask the adherence-reducing effects of more intensive substance use. This could potentially affect our main findings if this higher severity of substance use was more common in one racial/ethnic group than another. Our approach to combining the different depression measures, though recommended per Anastasi,30 is limited by the quality of the normative estimates available (Wagner et al31) and the degree to which the different scales' items tapped different aspects of depression. The threat to validity is thought to be minimal; however, as the norms are based on very large population samples and there is good overlap of items on these scales. Furthermore, in the prior study we demonstrated that the findings observed using this approach approximated findings obtained from using a more conservative approach of converting raw scores using standard cut-offs for mild/moderate/severe depression across all measures. As such, we are confident in this recoded variable's ability to control for the effects of depression in the analyses presented here.
Another limitation is that data on all variables were not available from each study, although participants with missing data on depression and substance use were retained in the analyses by creating a category for missing data (ie, a missing indicator approach). A broader assessment of mental health would have been helpful, perhaps validated by clinician rating. Importantly, the MACH14 data set had a limited array of variables available for investigation. Factors we could not include here—such as self-efficacy, social support, housing status, food insecurity, or insurance status—may have accounted for the race/ethnicity and adherence association.
In conclusion, this study of pooled data from 13 different US studies confirmed prior reports of racial/ethnic disparities in ART adherence, even when adherence was more objectively measured and analyses adjusted for key demographic variables, substance use, and depression. The findings may help to explain the worse health outcomes for racial/ethnic minorities with HIV in the United States and suggest a need for interventions targeting their adherence to achieve the US National HIV/AIDS Strategy goal to reduce HIV-related health disparities. There is a need for more research on the mechanisms of these racial/ethnic disparities to inform intervention development and better address these inequities.
The authors would like to thank all the patients who participated in each of the individual studies.
1. Hogg RS, Heath K, Bangsberg D, et al.. Intermittent use of triple-combination therapy is predictive of mortality at baseline and after 1 year of follow-up. AIDS. 2002;16:1051–1058.
2. Palella FJ, Delaney KM, Moorman AC, et al.. Declining morbidity and mortality among patients with advanced human immunodeficiency virus infection. N Engl J Med. 1998;338:853–860.
3. Hartzell JD, Spooner K, Howard R, et al.. Race and mental health diagnosis are risk factors for highly active antiretroviral therapy failure in a military cohort despite equal access to care. J Acquir Immune Defic Syndr. 2007;44:411–416.
4. Halkitis PN, Palamar J, Mukherjee P. Analysis of HIV medication adherence in relation to person and treatment characteristics using hierarchical linear modeling. AIDS Patient Care STDS. 2008;22:323–335.
5. Halkitis PN, Parsons JT, Wolitski RJ, et al.. Characteristics of HIV antiretroviral treatments, access and adherence in an ethnically diverse sample of men who have sex with men. AIDS Care. 2003;15:89–102.
6. Kleeberger CA, Buechner J, Palella F, et al.. Changes in adherence to highly active antiretroviral therapy medications in the multicenter AIDS cohort study. AIDS. 2004;18:683–688.
7. Kleeberger CA, Phair JP, Strathdee SA, et al.. Determinants of heterogeneous adherence to HIV-antiretroviral therapies in the multicenter AIDS cohort study. J Acquir Immune Defic Syndr. 2001;26:82–92.
8. Lazo M, Gange SJ, Wilson TE, et al.. Patterns and predictors of changes in adherence to highly active antiretroviral therapy: longitudinal study of men and women. Clin Infect Dis. 2007;45:1377–1385.
9. Pence BW, Ostermann J, Kumar V, et al.. The influence of psychosocial characteristics and race/ethnicity on the use, duration, and success of antiretroviral therapy. J Acquir Immune Defic Syndr. 2008;47:194–201.
10. Sullivan PS, Campsmith ML, Nakamura GV, et al.. Patient and regimen characteristics associated with self-reported nonadherence to antiretroviral therapy. PLoS One. 2007;2:e552.
11. Silverberg MJ, Leyden W, Quesenberry CP Jr, et al.. Race/ethnicity and risk of AIDS and death among HIV-infected patients with access to care. J Gen Intern Med. 2009;24:1065–1072.
12. Oh DL, Sarafian F, Silvestre A, et al.. Evaluation of adherence and factors affecting adherence to combination antiretroviral therapy among White, Hispanic, and Black men in the MACS Cohort. J Acquir Immune Defic Syndr. 2009;52:290–293.
13. Springer SA, Chen S, Altice F. Depression and symptomatic response among HIV-infected drug users enrolled in a randomized controlled trial of directly administered antiretroviral therapy. AIDS Care. 2009;21:976–983.
14. Arnsten JH, Demas PA, Grant RW, et al.. Impact of active drug use on antiretroviral therapy adherence and viral suppression in HIV-infected drug users. J Gen Intern Med. 2002;17:377–381.
15. Harzke AJ, Williams ML, Nilsson-Schönnesson L, et al.. Psychosocial factors associated with adherence to antiretroviral medications in a sample of HIV-positive African American drug users. AIDS Care. 2004;16:458–470.
16. Carrico AW, Riley ED, Johnson MO, et al.. Psychiatric risk factors for HIV disease progression: the role of inconsistent patterns of antiretroviral therapy utilization. J Acquir Immune Defic Syndr. 2011;56:146–150.
17. Bing EG, Burnam MA, Longshore D, et al.. Psychiatric disorders and drug use among human immunodeficiency virus-infected adults in the United States. Arch Gen Psychiatry. 2001;58:721–728.
18. Cook JA, Cohen MH, Burke J, et al.. Effects of depressive symptoms and mental health quality of life on use of highly active antiretroviral therapy among HIV-seropositive women. J Acquir Immune Defic Syndr. 2002;30:401–409.
19. Lewis-Fernández R, Das AK, Alfonso C, et al.. Depression in US Hispanics: diagnostic and management considerations in family practice. J Am Board Fam Pract. 2005;18:282–296.
20. Gonzalez JS, Hendriksen ES, Collins EM, et al.. Latinos and HIV/AIDS: examining factors related to disparity and identifying opportunities for psychosocial intervention research. AIDS Behav. 2009;13:582–602.
21. Thrasher AD, Earp JAL, Golin CE, et al.. Discrimination, distrust, and racial/ethnic disparities in antiretroviral therapy adherence among a national sample of HIV-infected patients. J Acquir Immune Defic Syndr. 2008;49:84–93.
22. Waldrop-Valverde D, Osborn CY, Rodriguez A, et al.. Numeracy skills explain racial differences in HIV medication management. AIDS Behav. 2010;14:799–806.
23. Liu H, Wilson IB, Goggin K, et al.. MACH14: a multi-site collaboration on ART adherence among 14 institutions. AIDS Behav. 2012; in press.
24. Cohen J, Cohen, Patricia, et al.. Applied Multiple Regression/Correlation Analysis for the Behavioral Sciences. Mahwah: Psychology Press; 2003.
25. Beck AT, Steer RA, Brown GK. Beck Depression Inventory-II (BDI-II). San Antonio, TX: Psychological Corporation; 1996.
26. Beck AT, Ward CH, Mendelson M, et al.. An inventory for measuring depression. Arch Gen Psychiatry. 1961;4:561–571.
27. Radloff LS. The CES-D scale: A self-report depression scale for research in the general population. Appl Psychol Meas. 1977;1:385.
28. Derogatis LR, Melisaratos N. The brief symptom inventory: an introductory report. Psychol Med. 1983;13:595–605.
29. Rogers WH, Wilson IB, Bungay KM, et al.. Assessing the performance of a new depression screener for primary care (PC-SAD). J Clin Epidemiol. 2002;55:164–175.
30. Anastasi A. Psychological Testing. 6th ed. New York: Macmillan; 1988.
31. Wagner GJ, Goggin K, Remien RH, et al.. A closer look at depression and its relationship to HIV antiretroviral adherence. Ann Behav Med. 2011;42:352–360.
32. Peretti-Watel P, Spire B, Lert F, et al.. Drug use patterns and adherence to treatment among HIV-positive patients: Evidence from a large sample of French outpatients (ANRS-EN12-VESPA 2003). Drug Alcohol Depend. 2006;82(suppl 1):S71–79.
33. Papke LE, Wooldridge JM. Econometric methods for fractional response variables with an application to 401 (k) plan participation rates. J Appl Econ. 1996;11:619–632.
34. Huh D, Flaherty BP, Simoni JM. Optimizing the analysis of adherence interventions using logistic generalized estimating equations. AIDS Behav. 2012;16:422–431.
35. Sue DW, Capodilupo CM, Torino GC, et al.. Racial microaggressions in everyday life: Implications for clinical practice. Am Psychol. 2007;62:271–286.
36. Myers HF. Ethnicity- and socio-economic status-related stresses in context: an integrative review and conceptual model. J Behav Med. 2008;32:9–19.
37. Bird ST, Bogart LM, Delahanty DL. Health-related correlates of perceived discrimination in HIV care. AIDS Patient Care STDS. 2004;18:19–26.
38. Bogart LM, Wagner GJ, Galvan FH, et al.. Longitudinal relationships between antiretroviral treatment adherence and discrimination due to HIV-serostatus, race, and sexual orientation among African-American men with HIV. Ann Behav Med. 2010;40:184–190.
39. Saha S, Sanders DS, Korthuis PT, et al.. The role of cultural distance between patient and provider in explaining racial/ethnic disparities in HIV care. Patient Educ Couns. 2011;85:e278–e284.
40. Kripalani S, Weiss BD. Teaching about health literacy and clear communication. J Gen Intern Med. 2006;21:888–890.
41. Kalichman SC, Ramachandran B, Catz S. Adherence to combination antiretroviral therapies in HIV patients of low health literacy. J Gen Intern Med. 1999;14:267–273.
42. Osborn CY, Paasche-Orlow MK, Davis TC, et al.. Health literacy: an overlooked factor in understanding HIV health disparities. Am J Prev Med. 2007;33:374–378.
43. Bogart LM, Wagner G, Galvan FH, Banks D. Conspiracy beliefs about HIV are related to antiretroviral treatment nonadherence among African Americans with HIV. J Acquir Immune Defic Syndr. 2009;53:648–655.