Prevalence of STIs, pregnancy, contraception and reported sexual behaviour at baseline and scheduled follow-up visits among 1020 potential trial participants.

1. For each variable, we modelled the change in prevalence between visits using a random-effects logistic regression model adjusting for visit as a continuous variable and including a random intercept. This modelling approach adjusted for the expected additional correlation between multiple responses by the same participant. We calculated formal tests of significance using likelihood ratio tests, and the p-values from these tests are presented in the final column of the table.2.T. vaginalis sample N = 220. Measurements made at recruitment (visit 1) and at 12 months (visit 5).3. Pregnancy status was an initial study exclusion criterion. Prevalence at Visit 01 was therefore zero among eligible participants.