Determine the associations between factors and sexual practices and the composition of the vaginal microbiome (VM) of women treated for bacterial vaginosis (BV).

Prospective cohort study.

The Melbourne Sexual Health Centre, Melbourne, Australia.

Seventy-five reproductive-age women diagnosed with clinical BV, treated with first-line antibiotics and followed for up to 6 months.

Women self-collected vaginal swabs and completed questionnaires at enrolment, the day following antibiotics and monthly for up to 6months until BV recurrence or no BV recurrence (n = 430 specimens). Bacterial composition was determined using 16S rRNA gene amplicon sequencing. The effects of ongoing factors on VM composition (utilising 291 monthly specimens) were assessed using generalised estimating equations population-averaged models, which accounted for repeated measures within individuals.

The relative abundance of vaginal bacterial taxa.

Women who reported ongoing sex with a regular sexual partner (RSP) had a VM comprised of increased relative abundance of non-optimal BV-associated bacteria (Adjusted co-efficient [Adjusted co-eff] = 11.91, 95% CI 3.39to20.43, P = 0.006) and a decreased relative abundance of optimal, Lactobacillus species (Adjusted co-eff = -12.76, 95% CI -23.03 to -2.49, P = 0.015). A history of BV was also associated with a decreased relative abundance of Lactobacillus spp. (Adjusted co-eff = -12.35, 95% CI -22.68, P = 0.019). The relative abundance of Gardnerella, Atopobium and Sneathia spp. increased following sex with an RSP.

Sex with an untreated RSP after BV treatment was associated with a VM comprised of non-optimal BV-associated bacteria. BV treatment approaches may need to include partner treatment if they are to achieve a sustained optimal VM associated with improved health outcomes.

Sex drives a return to a 'non-optimal' vaginal microbiota after antibiotics for bacterial vaginosis.