To estimate the prevalence of bacterial vaginosis by Nugent Gram stain criteria in a nonclinic national sample of young women entering recruit training; to examine clinical associations with bacterial vaginosis; and to evaluate the performance of a pH test card and Papanicolaou smear against Gram stain as screening tools for bacterial vaginosis.METHODS
A cross-sectional study of 1938 women was conducted. Self-collected vaginal swabs were applied to a colorimetric pH test card and a glass slide for Gram stain evaluation according to the Nugent criteria. Papanicolaou smears and samples for sexually transmitted diseases screening were collected during routine entry pelvic examinations.RESULTS
Bacterial vaginosis prevalence was 27%, with 28% in sexually experienced and 18% in non–sexually experienced women (P = .001). Bacterial vaginosis prevalence was 11% in Asian/Pacific Islanders, which was lower than in other nonwhite ethnic groups (P = .004). Clinically, bacterial vaginosis was directly related to multiple sexual partners (P = .026), self-report of vaginal discharge (P = .001), self-report of vaginal odor (P < .001), and concurrent Chlamydia trachomatis infection (P = .002), and inversely related to hormonal contraceptive use (P = .013). Vaginal discharge did not achieve statistical significance in multivariate analysis. Compared with the Nugent criteria, the sensitivities and specificities for bacterial vaginosis diagnosis were as follows: colorimetric pH test: 72% and 67%; Papanicolaou smear: 72% and 79%, respectively.CONCLUSION
Among these diverse young women, bacterial vaginosis occurs commonly in both sexually experienced and inexperienced young women and differs by race and ethnicity. The pH colorimetric test and Papanicolaou smear performed moderately well as screening tools for bacterial vaginosis. The inverse relationship of bacterial vaginosis with hormonal contraceptive use and its direct relationship with C trachomatis need further study.