Reply to: ‘Is vaginal washing associated with increased risk of HIV-1 acquisition?’

Abstract

In their letter, van de Wijgert and colleagues [1] report that they found no association between vaginal washing and HIV-1 acquisition among women attending family planning clinics in Zimbabwe and Uganda. These results differ from our study in a cohort of female sex workers in Kenya [2], among whom vaginal washing was associated with a significantly increased risk of HIV-1. We agree that differences in the populations and their background HIV-1 risk may partly explain the variation between the two studies. An additional factor that may help to explain the difference in study results is the frequency and timing of vaginal washing. Although we did not collect this information when the Mombasa cohort was initiated in 1993, we have collected detailed data about vaginal washing practices since 2002 from cohort participants enrolling in a trial of periodic presumptive treatment for vaginal infections. Of the 310 women enrolled, 300 (97%) reported that they perform vaginal washing, with a median frequency of three (interquartile range 2–3) times per day. Vaginal washing was performed after sex in 299 (100%), when bathing in 294 (98%), after urinating in 83 (28%), and before sex in only 16 (5%) women. The association between the frequency of vaginal washing and HIV-1 risk should be a topic for future investigation. Although methodological differences between the Kenya and Zimbabwe/Uganda studies have been noted, we feel that these are unlikely to explain the differences in results. We used the baseline report of vaginal washing as our measure of exposure because we felt that this better captured the women's usual behavior over time. More importantly, we felt that this measure was less subject to bias as a result of washing in response to genital symptoms (including symptomatic genital tract infections) or a perception of partners' risk. Nonetheless, we also evaluated the association between vaginal washing as a time-dependent covariate and HIV-1 seroconversion. In this analysis, vaginal washing with water [hazard ratio (HR) 3.63, 95% confidence interval (CI) 1.43–9.20, P = 0.007] and with soap (HR 3.82, 95% CI 1.56–9.33, P = 0.003) were significantly associated with an increased risk of HIV-1 acquisition. Differences in laboratory testing were noted between our study and the study reported by van de Wijgert and colleagues [1]. We did not include serological testing for herpes simplex virus (HSV) 2 or testing for Chlamydia trachomatis. However, we did adjust our analyses for the presence of genital ulcerations, mucopurulent cervical discharge, and microscopic evidence of cervicitis, the likely mechanisms by which herpes and Chlamydia increase the risk of HIV-1. It is important to note that neither HSV-2 serostatus nor incident chlamydial infection would confound our results unless these exposures were also associated with baseline vaginal washing. Although women may wash in response to episodes of HSV-2 or Chlamydia, it seems unlikely that practices reported at enrollment would be closely related to subsequent genital infections. We thus feel that confounding by HSV-2 and Chlamydia does not explain the increased risk of HIV-1 acquisition associated with vaginal washing. We recently evaluated the association between vaginal washing and Lactobacillus colonization (by culture). Compared with women who reported no vaginal washing, there was a step-wise decrease in the detection of vaginal Lactobacillus during follow-up among women who washed with water (HR 0.58, 95% CI 0.41–0.82, P = 0.002) and with soap (HR 0.46, 95% CI 0.36–0.61, P < 0.001). Similar findings were observed for the detection of hydrogen peroxide producing lactobacilli (water only, HR 0.53, 95% CI 0.30–0.95, P = 0.03; soap HR 0.39, 95% CI 0.27–0.55, P < 0.001). Vaginal colonization with Lactobacillus, and particularly with hydrogen peroxide-producing Lactobacillus species, has been associated with a significantly lower risk of HIV-1 acquisition [3], supporting the hypothesis that vaginal washing may increase HIV-1 susceptibility by eliminating vaginal lactobacilli. In summary, our recent study demonstrates that vaginal washing may substantially increase women's risk of acquiring HIV-1. This adds to evidence that vaginal washing, which has no confirmed medical value, increases the risk of pelvic inflammatory disease, other genital tract infections, and adverse obstetrical outcomes [4]. We feel that intervention strategies aimed at modifying intravaginal practices should be evaluated as a possible female-controlled HIV-1 prevention strategy. This should not be interpreted as indicating that vaginal washing should be addressed to the exclusion of other strategies. We agree that vaginal washing interventions could be most effective as part of a comprehensive programme including the treatment and prevention of genital tract infections, barrier methods, and other HIV-1 risk-reduction strategies. Combined interventions of this type are a focus of our current research.