The Missing Men: HIV Treatment Scale-Up and Life Expectancy in Sub-Saharan Africa

Delivery of effective HIV antiretroviral therapy (ART) to the more than 6 million persons with HIV in South Africa is well underway, with early data on the impact of this massive public health effort demonstrating a reversal of the previous decade’s precipitous decline in population life expectancy [1]. Although South Africa’s age and sex disparities in HIV acquisition have traditionally been described as disadvantaging young women [2], accumulating evidence now suggests a reverse disparity: although HIV care is available to both men and women and is nominally free of charge, women are more likely to be tested for HIV, engage in pre-treatment care, initiate treatment earlier, stay on treatment, and survive [3–6]. To adopt the classic Eisenberg and Power [7] analogy of health care as current flowing through an electric circuit, the voltage drops along the entire circuit of HIV care, from HIV infection to AIDS-free survival, are larger for men compared with women (Fig 1). There are simply too many missing men. In recent years, studies from South Africa [8], as well as Rwanda [9] and Uganda [10], have begun to demonstrate the cumulative impact of these voltage drops, which, in total, result in an approximately 10-year life expectancy gap between men and women initiating ART at 20 years of age (Fig 2). However, the findings of these studies should be interpreted in light of important limitations. First, they were based solely on data obtained from persons enrolled in HIV treatment programs. Poverty, food insecurity, HIV stigma, and geographic barriers still exert outsize influences on HIV testing, treatment, and retention in these settings [11–14], so it is unlikely that these enrollees are representative of the entire population of persons with HIV. Second, mortality had to be estimated among those lost to follow up [9,10]. Because persons in HIV treatment programs are much more likely to be lost to care than confirmed as dead [15,16], and because the vast majority of HIV-related mortality events go unreported [17], the mortality estimates in these studies are likely to be biased. Third, and perhaps most notably, none of these studies directly observed non-HIV mortality. Thus, while they were able to document trends in mortality among persons with HIV, they were unable to assess the extent to which these changes were related to HIV care or to unrelated secular trends in health and health behavior. In this context, the research article by Jacob Bor and colleagues [18] that appears this week in PLOS Medicine provides new evidence of a widening gender gap in life expectancy, using data obtained from a general population sample in rural South Africa from 2001–2011, covering a period of coincident ART scale-up. By surveilling all persons in the region—whether HIV-negative, HIV-positive in care, or HIV-positive but not in care—and by using verbal autopsies to categorize mortality events that were recorded by the surveillance teams, this