Longitudinal assessment of bacterial vaginosis prior to and during incident pregnancy: an observational study in Kenyan adolescent girls and young women

OBJECTIVE: To determine bacterial vaginosis (BV) status at multiple time points among adolescent girls and young women (AGYW) and assess the impact of pregnancy on their BV status. DESIGN: Longitudinal cohort study. SETTING: Thika, Kenya. PARTICIPANTS: AGYW aged 16–20 years enrolled prior to first sex or reporting only a single lifetime partner. MAIN OUTCOME MEASURES: The primary outcome was relative risk (RR) of BV during pregnancy compared with before pregnancy by analysing longitudinal trends in BV over time. BV risk was estimated using Poisson regression models. RESULTS: A total of 121 AGYW became pregnant in the parent cohort and had BV results before, during or after pregnancy. Point prevalence of BV was 11.0% at visits >12 months pre-pregnancy, 13.0% at 3–12 months pre-pregnancy, 22.1% at <3 months pre-pregnancy and 13.4% during pregnancy. Compared with visits during pregnancy, RR of BV was 1.65 (95% CI: 1.00 to 2.71; p=0.05) at visits <3 months pre-pregnancy, 0.97 (95% CI: 0.62 to 1.52; p=0.90) at visits 3–12 months pre-pregnancy and 0.82 (95% CI: 0.44 to 1.53; p=0.53) at visits 12 months pre-pregnancy. An adjusted analysis including age, income, residence, date of first sex, recent sexual activity and positive sexually transmitted infection test resulted in small changes in risk estimates, with adjusted RR of BV of 1.66 (95% CI: 1.04 to 2.67; p=0.04) at visits <3 months pre-pregnancy compared with visits during pregnancy. CONCLUSIONS: BV risk during pregnancy was lower than during the immediate pre-pregnancy period. Hormonal changes in pregnancy may reduce BV.