Limitations of Clinic-Based Studies on HIV and Fertility Preferences

The recent article by Snow et al.¹ contributes to a large body of literature on the volitional relationship between HIV infection and fertility preferences in sub-Saharan Africa. Although we agree with the finding that HIV suppresses fertility desires on average, we fear that their claims, particularly about the role of antiretroviral therapy (ART), extend well beyond the capabilities of the data and methods they employ. More generally, we worry about the habit of relying on clinic-based samples with noncomparable comparison groups to draw conclusions and make broad policy recommendations. This is a particularly curious problem when high quality population-based data sets (many longitudinal) exist that are better suited to address the questions at hand.

Clinic-based studies have uncovered important biomedical relationships between HIV and fertility, but they are limited in their ability to answer social and behavioral questions and to generalize to larger populations of HIV-positive (HIV+) and HIV-negative individuals. First, across Africa, clinic samples are disproportionately female, urban, wealthy, educated, pregnant, and ART savvy, and these differences cannot simply be “controlled away” statistically. Second, clinic-based studies are particularly vulnerable to social desirability bias.² Power relationships create a situation wherein respondents face pressure to give clinicians “right” answers, which threatens validity, as health care workers’ antinatalist attitudes toward HIV+ women are well documented in Africa.^3,4 To put it plainly, rural African women with HIV who want to have another child (and many of them do) are especially unlikely to disclose this preference to a nurse or to any person who resembles one.

What do population-based studies show about the relationship between HIV and fertility goals? Surveys and in-depth interviews conducted outside of hospital settings paint a more complex picture. First, the relationship varies across the reproductive life course.^5–7 Tsogolo la Thanzi, our population-based cohort of young women (aged 15–25 years) in rural Malawi, reveals no differences in the fertility desires of childless women by HIV status. Indeed, 100% of childless HIV+ women want them. Other studies from Mozambique and Malawi have found that men and women who perceive themselves to be at risk for infection want to speed up childbearing.^8,9 Additionally, high quality ethnographic studies highlight a considerable amount of reproductive ambivalence among HIV+ individuals—particularly those on ART.^5,10,11

As sociologists and demographers, we worry that studies like the one by Snow et al. oversimplify a relationship that is considerably more complicated. We recommend being cautious in making generalizations to larger populations.