Epidemiologic studies of the potential link between the use of hormonal contraception by HIV-1-negative women and increased risk of HIV-1 acquisition have yielded disparate results, with some studies suggesting an association while others have not.2 Methodologic considerations have likely contributed to some of these differences and have hampered the ability to directly compare these data. For instance, many studies infrequently assessed both hormonal contraceptive use and HIV-1 infection status, making it difficult to determine whether hormonal contraceptives were used at the time of HIV-1 acquisition. Of the fifteen prospective studies reviewed in 2007, only three assessed HIV-1 infection status at least monthly,6–8 and many assessed contraceptive use only at baseline. Furthermore, differences in sexual behaviors between users and non-users of hormonal contraception may confound results, particularly in the cross-sectional studies.
Two of the largest, high-quality prospective studies to evaluate associations between hormonal contraceptive use and HIV-1 transmission highlight the difficulties in summarizing these data. In a prospective study of 1272 female sex workers in Mombasa, Kenya with monthly follow-up, both oral contraceptive pills (hazard ratio [HR], 1.5; 95% confidence interval [CI], 1.0–2.1) and DMPA (HR, 1.8; 95% CI, 1.4–2.4) were associated with greater risk of HIV-1 in analyses controlling for demographic factors, sexual behavior, condom use, and sexually transmitted infections.8 In a later reanalysis, further adjustment for HSV-2 status, which was not controlled for in the original paper, did not alter these results, nor was there evidence of effect modification by HSV-2 serostatus.9 In the second study, which enrolled 4439 women from family planning clinics in Uganda and Zimbabwe, women were followed at 3-monthly intervals.10 Neither oral contraceptive pills (HR, 0.99; 95% CI 0.69–1.42) nor DMPA (HR, 1.25; 95% CI, 0.89–1.78) was associated with HIV-1 acquisition overall in the original analysis.10 However, in the subset of women who were seronegative for HSV-2 at study enrollment, both oral contraceptive pills (HR, 2.85; 95% CI, 1.39–5.81) and DMPA (HR, 3.97; 95% CI, 1.98–8) was associated with HIV-1 acquisition.10 A reanalysis of the data from this second study has recently been performed, using a different analytic methodology to better account for time-dependent confounding. In this reanalysis, a modest, but statistically significant, increased HIV-1 risk was seen for women using DMPA (HR, 1.48; 95% CI, 1.02–2.15) but not oral contraceptives (HR, 1.19, 95% CI, 0.8–1.76), and the results for both oral contraceptives and DMPA use in HSV-2 seronegative women were strengthened (oral contraceptives: HR, 2.06, 95% CI 0.75–4.92; DMPA: HR, 4.49, 95% CI, 1.98–10.17).11 In addition, younger women (≤24 years of age) using either DMPA (HR, 2.76, 95% CI 1.62–4.72) or oral contraceptives (HR, 2.02, 95% CI, 1.15–3.55) were also at increased risk of HIV-1 acquisition.11
It should be noted that very few studies had sufficient statistical power to detect the 50–80% increased risk of HIV-1 infection suggested by the Mombasa study of high-risk sex workers.2,8 In fact, studies of high-risk populations, such as sex workers, have generally shown that hormonal contraception use is associated with elevated risk of HIV-1 acquisition.7,8,12,13 In several studies, the risk of HIV-1 acquisition in high-risk women may be higher in users of DMPA than in users of oral contraceptive pills.7,8,14 Additional evidence that hormonal contraceptives may increase risk of HIV-1 acquisition in high-risk women comes from the observation that women who use hormonal contraceptives are more likely to become simultaneously infected with more than one variant of HIV-1 (Odds Ratio [OR], 2.7; 95% CI, 1.3–5.6).15 Therefore, messages about the potential risks associated with hormonal contraceptives may have to be tailored to the population.
To more clearly establish the role of hormonal contraceptives in altering the risk of HIV-1 acquisition, future studies must be carefully designed to account for potential confounders and bias including sexual habits, condom use, and other behaviors that might affect HIV-1 acquisition risk. Such studies must be sufficiently powered to detect as modest as a 50% increase in HIV-1 infection incidence and should carefully assess hormonal contraceptive use, HSV-2 status, surrogate markers of risk, and HIV-1 infection status at multiple, frequent timepoints during the follow-up period. Ideal groups to study would include adolescents, who are at high risk of HIV-1 acquisition yet there is little available data, and discordant couples, in whom precise measurement of transmission rates from infected male partners to uninfected female partners can be determined.