Cochran and Mays Respond.

Abstract

Including sexual orientation questions in general population surveys has relatively recent origin.1 But in a short time, their presence has profoundly altered what is known about health disparities affecting lesbians, gay men, bisexuals, and homosexually experienced heterosexuals. Because sexual orientation is a new survey construct, both methodological and conceptual issues related to its assessment bear close scrutiny. Hottes et al. caution us that antigay stigma may reduce disclosure rates and, thus, undercount men who have sex with men (MSM). Indeed, we noted this in our original article underscoring that this effect was especially likely in early waves of the General Social Survey-National Death Index (GSS-NDI).2 We agree that the estimated difference in HIV-related deaths between MSM and heterosexual men is at odds with well-founded expectations and may be attributable to underreporting. Prevalence of HIV-attributable deaths (2.6%; 95% confidence interval [CI] = 2.0%; 3.1% of all deaths), but not suicide-attributable deaths (1.9%; 95% CI = 1.4%, 2.3%), was higher in the GSS-NDI than that observed in the United States in 2006.3 Still, such deaths were rare, and our estimate could be unstable. Hottes et al. also suggest, without explanation, that MSM were undersampled (selection bias); we know of no mechanism by which this could have occurred. The higher suicide attempt rates among MSM, documented in numerous surveys of similar design to the GSS-NDI, do not inevitably lead, as Hottes et al. imply, to higher suicide mortality rates among MSM. Predictors of attempts and deaths differ.4 And the existing evidence, while sparse, contradicts the hypothesis of Hottes et al.: a similar, small study5 also failed to detect sexual orientation differences in suicide mortality in a sample where MSM, as compared with heterosexual men, had evidenced higher suicide attempt rates. Still, we heartily agree that the robustness of our results must be tested. Like many early surveys linking sexual orientation to health outcomes, alternate explanations should be entertained6 until a preponderance of evidence supports unambiguous conclusions. In this regard, the recent decision by the National Health Interview Survey (NHIS)7 to measure only sexual orientation identity (e.g., gay or bisexual) has important implications for our ability to track the health and mortality risks for MSM. Estimates suggest that more than half of MSM identify as heterosexual,8 and scientific evidence9,10 shows that their health issues differ substantially both from gay- or bisexual-identifying men and from exclusively heterosexual men. With this decision, health concerns for MSM will remain hidden in the NHIS, a major health surveillance system used to achieve our nation’s health goals.