Objectives (1) To identify correlates of incident Bacterial vaginosis (BV) among high-risk women and (2) to identify predictors of discordance between self-reported lack of semen exposure in the past 6 months and the detection of spermatozoa on Gram stain, which provides biological evidence of recent exposure.
Methods Analyses were based on among 871 HIV-infected and 439 HIV-uninfected women participating in HIV Epidemiology Research Study (HERS) which was conducted in 4 sites in the US Participants completed study visits conducted at baseline and at 6-month intervals thereafter. We conducted both cohort and case-crossover analyses, stratified by HIV infection status, to evaluate potential correlates of incident BV. We also used logistic regression to identify predictors of discordance between self-reported lack of exposure to semen and the detection of spermatozoa on Gram stain.
Results BV incidence was 21% among HIV-infected women and 19% among HIV-uninfected women. We found fewer correlates of incident BV when assessed with a case-crossover design than with a cohort design. Reporting frequent coitus (regardless of consistency of condom use) was correlated with incident BV in the cohort analyses but not in the case-crossover analyses. The sole correlate that emerged in both the cohort and case-crossover analyses among HIV-infected and -uninfected women was the detection of spermatozoa on Gram stain. Seven factors were associated with discordance between self-reported semen exposure and spermatozoa detection in the multivariable analysis. Discordance differed by study site and race/ethnicity and was more common among younger women. The following infections or conditions also were predictive of discordance: HIV (adjusted OR [aOR], 2.8; 95% CI, 1.7% to 4.6%), BV (aOR, 1.9; 95% CI, 1.5% to 2.5%), and human papillomavirus (aOR, 1.3; 95% CI, 1.0% to 1.8%). Finally, reporting current injection drug use (aOR, 0.6; 95% CI, 0.4% to 0.9%) was inversely related to discordance.
Conclusions The inconsistent association between condom use and BV found in prior studies could be the result of participant reporting bias. The present study found evidence of a relationship between semen exposure and incident BV. Also, given the number and range of correlates of discordance between self-reported and biological evidence of semen exposure, inaccuracies in the reporting of sexual behaviours cannot be assumed to be distributed randomly across a study population.