To the Editors:
In their recent article in JAIDS Peterson et al1 assessed the relationships between social discrimination (perceived racism and homophobia) and prevalent HIV infection using baseline data from a cohort study of black and white men who have sex with men (MSM) in Atlanta, GA. They conducted logistic regression analyses stratified by race and found that perceived racism predicted unprotected (condomless) anal intercourse, but not prevalent HIV infection, among black MSM. Homophobia was not a significant predictor of prevalent HIV infection or condomless intercourse among white or black men. The authors conclude that their findings do not support the hypothesis that perceived social discrimination explains the racial disparity in HIV prevalence between black and white MSM.
We appreciate the contributions made by Peterson et al to addressing health disparities, and join them in their recommendation for future research to explore manifestations of institutional and network-level social discrimination, and their impacts on HIV infection. However, there are several methodological and theoretical considerations that should temper interpretations of these results.
First, by conducting separate within-group analyses of black and white men, the authors identified sources of variation in HIV prevalence within each group. However, such stratified analyses may not help explain the disparity in HIV prevalence between black and white men. Causes of variation within groups are not necessarily the same as causes of intergroup variation. This is particularly notable when sources of intergroup variation are largely constant or invariant in 1 group.2 We believe that the authors' use of within-group analyses to answer a question about between-group differences may have contributed to a form of conceptual bias that Schwartz and Carpenter2 term “type III error,” which occurs when there is a mismatch between the research question that an analysis is able to answer and the intended research question. In this case, a priori hypotheses were well matched to analysis methods, but type III error may have been introduced in the interpretation of results. Specifically, their study compared the effect of racism between black men who have experienced less perceived racism and black men who have experienced more. This analysis did not assess the interaction between being black and experiencing greater levels of racism, an interaction implied by the interpretation concerning the racial disparity in HIV prevalence. By only measuring perceived racism among black participants and stratifying analyses based on race, the authors are not able to evaluate the synergistic effects of race and perceived racism on HIV prevalence.
Second, although the authors correctly note that the cross-sectional nature of the data precludes causal inference, they do approach the data with a causal question—whether or not perceived discrimination increases risk of HIV infection for black MSM. Using the measures available at baseline, associations between potentially long-standing HIV infection and past-year perceived racial discrimination and homophobia were assessed. Perceived social discrimination scores at the time of survey completion may not temporally precede prevalent HIV infection. Given that perceived racism was associated with condomless anal intercourse (a proximate risk factor for HIV infection) among black MSM, we think there would be a value in a longitudinal evaluation of the association between perceived racism and incident HIV infection among black men who were HIV negative at baseline.
Understanding underlying mechanisms for disparities in HIV prevalence and incidence remains an important focus in HIV research, given the disproportionate HIV disease burden being shouldered by MSM worldwide, especially racial/ethnic minority men. However, this area of research remains underdeveloped. For example, measures of perceived discrimination are subject to substantial measurement error because of differences in recognition, attribution, recall, and reporting that may themselves be related to health status and/or behavior.3,4 In addition, the ways in which interpersonal and structural racism are enacted and perceived and their impact on HIV risk may vary between groups and geographical locales. It is also crucial to explicate the theoretical bases of hypothesized relationships between specific measures of social discrimination and health outcomes. Peterson et al chose HIV prevalence as an outcome in recognition of the growing consensus that differences in sexual or injection risk do not explain the disparity in HIV infection between black and white MSM.5 As they note in their discussion, intraracial sexual partnering could mediate a relationship between perceived racism and HIV prevalence among black MSM. However, they do not provide any other potential pathways by which perceived racism could lead to increased HIV prevalence, and we cannot think of any that are not mediated by sexual or injection risk behaviors. Thus, it is not clear whether the use of prevalent HIV infection as an outcome represents a substantial change from previous research.
In addition, we believe it is premature to conclude that social discrimination does not explain disparities in HIV infection, or that investigations concerning interpersonal discrimination should be abandoned in favor of studies focusing on other units of analysis. It may be the case that perceived racism is most relevant to understanding variations in HIV sexual risk and vulnerability among black men, whereas network and social-structural level forms of racism better account for the racial disparity in HIV prevalence. Thus, longitudinal studies measuring both interpersonal and structural forms of discrimination and social oppression, and their interactions, may be necessary to better understand the role of racial discrimination in the HIV epidemic among black MSM.
Peterson et al inspire us to reinvigorate research in the area of social predictors of HIV disparities. We must pursue investigations designed to understand the causes of HIV disparities among MSM. This requires us to refine our measures of discrimination, theorize social oppression as co-occurring across multiple levels, and design longitudinal studies with MSM inclusive of large numbers of ethnic minority men to better attribute causes of HIV prevalence and incidence.
1. Peterson JL, Bakeman R, Sullivan P, et al.. Social discrimination and resiliency are not associated with differences in prevalent HIV infection in black and white men who have sex with men. J Acquir Immune Defic Syndr. 2014;66:538–543.
2. Schwartz S, Carpenter KM. The right answer for the wrong question: consequences of type III error for public health research. Am J Public Health. 1999;89:1175–1180.
3. Schwartz S, Meyer IH. Mental health disparities research: the impact of within and between group analyses on tests of social stress hypotheses. Soc Sci Med. 2010;70:1111–1118.
4. Krieger N, Carney D, Lancaster K, et al.. Combining explicit and implicit measures of racial discrimination in health research. Am J Public Health. 2010;100:1485–1492.
5. Millett GA, Flores SA, Peterson JL, et al.. Explaining disparities in HIV infection among black and white men who have sex with men: a meta-analysis of HIV risk behaviors. AIDS. 2007;21:2083–2091.