Wolfe et al. respond

Gamper et al. raise an important concern regarding our sample. As we noted in our limitations section, although individuals were randomly selected from the 5 districts of Botswana with the highest number of HIV infected individuals, we did not sample from the more remote districts of Botswana, limiting the generalizability of our results. It is important to note, however, that the outcome of greatest interest in our study is the one showing an association between HIV stigma and perceived access to antiretroviral therapy; if, as the letter writers suggest, HIV stigma is greater among people who live further from health services and therefore have reduced access to antiretroviral therapy, we would expect data from these remote areas to support, rather than diminish, this outcome. The cultural context, highlighted by Gamper et al., is an important consideration when evaluating responses to questions about stigmatizing attitudes. Interestingly, in a population-based study conducted across Botswana in 2001, a majority of respondents readily admitted that they held stigmatizing attitudes toward people living with HIV/AIDS.1 This finding suggests that existing cultural barriers were inadequate to prevent the expression of stigmatizing attitudes prior to the introduction of the national treatment program. Our concern was that norms might have shifted during the rollout of the national program, increasing the risk of social desirability bias in our results. It was for this reason that we used a measure of anticipated stigma, which allowed respondents to focus explicitly on attitudes in their community rather than to put forward their own personal beliefs. We agree with Gamper et al. that it would have been interesting to assess awareness of, and attitudes toward, the national treatment program in the general population. Nevertheless, our measure of perceived access to treatment subsumes multiple related factors, including awareness of the treatment program, its geographic penetration, and its perceived effectiveness in reducing barriers to treatment among the general population. Similarly, although qualitative methods to identify barriers to treatment access among vulnerable groups were beyond the scope of this population-based study, we did obtain qualitative data from structured interviews with a convenience sample drawn from support groups for people living with HIV/AIDS from Gaborone, Serowe, and the surrounding villages and rural areas. In these interviews, key barriers to treatment access cited by respondents included inadequate access to food and gender inequality in relationships.2 We welcome the comments from Gamper et al., which highlight the importance of further exploring the etiology and repercussions of HIV stigma in the ever-evolving African context.