Is vaginal washing associated with increased risk of HIV-1 acquisition?

McClelland and colleagues found an increased risk of HIV-acquisition for vaginal washing among 1270 Kenyan female sex workers participating in a 10-year open cohort study [adjusted hazard ratio (HR) for washing with water alone, 2.64, 95% confidence interval (CI), 1.00–6.97; and for soap 3.84, 95% CI, 1.51–9.77] [1]. We also studied vaginal washing as a risk factor for HIV acquisition in a prospective cohort of 4531 women attending family planning clinics in Zimbabwe and Uganda, but did not find a statistically significant association (univariate HR for washing with water and/or soap 0.96, 95% CI, 0.72–1.27; unpublished data). Vaginal drying or tightening was associated with HIV-1 acquisition in univariate (HR, 1.49; 95% CI, 1.03–2.15) but not in multivariate models (HR, 1.09; 95% CI, 0.71–1.67). Our results are consistent with results of a South African study recently published by Meyer et al. [2]. We hypothesize that several important differences between the study cohorts might explain the difference in findings. Incident HIV-1 infection occurred in 222 women in the Kenyan cohort (7.7 per 100 person-years), and in 213 women in the Zimbabwe/Uganda cohort (2.8 per 100 person-years). The Kenyan sex workers almost universally engaged in vaginal washing; only 6% reported no washing at all. Most used soaps (71%) and 23% used water alone. In our study, two-thirds of women in Zimbabwe and Uganda engaged in vaginal washing at baseline; almost all of them used water alone and 13% (Zimbabwe) to 46% (Uganda) also used soaps. In both cohorts, the most common vaginal washing method was with a finger. The use of traditional substances was rare in Kenya (1%) and Uganda (2%) but relatively common in Zimbabwe (12%; mostly for vaginal drying or tightening). McClelland et al. did not report on the frequency of vaginal washing, the proportion of sex workers with large numbers of clients (as the median number of sexual contacts was only two per week, most appeared to have few), nor the timing of washing in relation to sex. In our cohort, most women had one partner, and cleansed the vagina on average 1 (Zimbabwe) to 1.7 times per day (Uganda) independently of sexual activity. The published literature suggests that sex workers engage in vaginal practices more frequently, and are more likely to wash to prepare for sex, than women in the general population (reviewed in [3]). These and other cohort differences, together with differences in exposure to HIV, might explain why an association was found among Kenyan sex workers but not among Zimbabwean and Ugandan women from the general population. Several differences in study methodologies are also apparent. In the main analysis of the Kenyan study, vaginal practices were ascertained at study enrollment only, and were not included as time-dependent variables. While the authors argue that they were more interested in ‘usual’ practices, which might be best captured in baseline reports, in fact, data on actual practices in the period prior to HIV seroconversion are probably more informative, especially because behaviors change over time. Second, the Kenyan analyses did not control for baseline herpes simplex type 2 virus, which was a strong predictor of HIV acquisition in our study as well as other African studies [4]. Furthermore, whereas we used PCR testing for both Neisseria gonorrhoeae and Chlamydia trachomatis, in Kenya the first was diagnosed using culture and the second was not assessed. Finally, in our study, we hypothesized that vaginal practices might increase HIV acquisition by causing epithelial disruption of the cervico-vaginal mucosa, or by causing genital inflammation or disturbances in the vaginal flora (bacterial vaginosis or candidiasis), which in turn could increase a woman's risk for HIV. As McClelland et al. point out, these direct and indirect pathways have been evaluated in cross-sectional studies, but the results have been equivocal [5–7]. It therefore would be helpful if additional data on the various pathways between vaginal washing and HIV acquisition from the Kenyan study are presented. Our results and those of Meyer et al. indicate that interventions solely aimed at modifying vaginal practices, as proposed by McClelland et al., would likely not result in important gains in HIV prevention in general populations in Africa. The reasons why women engage in vaginal practices have been studied extensively in recent years, and many of these reasons are closely linked to reproductive tract infections (RTIs; articulated by women as the removal of dirty secretions and bad smells) and achieving a desired amount of lubrication during sex [3]. Rather than focusing on vaginal practices per se, it would be better to address vaginal practices in the context of RTI screening and treatment programs, and when promoting vaginal products (e.g., condoms, diaphragms, and in the future hopefully also vaginal microbicides).