Pregnancy incidence and correlates during the HVTN 503 Phambili HIV vaccine trial conducted among South African women

BACKGROUND: HIV prevention trials are increasingly being conducted in sub-Saharan Africa. Women at risk for HIV are also at risk of pregnancy. To maximize safety, women agree to avoid pregnancy during trials, yet pregnancies occur. Using data from the HVTN 503/"Phambili" vaccine trial, we report pregnancy incidence during and after the vaccination period and identify factors, measured at screening, associated with incident pregnancy. METHODS: To enrol in the trial, women agreed and were supported to avoid pregnancy until 1 month after their third and final vaccination ("vaccination period"), corresponding to the first 7 months of follow-up. Unsterilized women, pooled across study arms, were analyzed. Poisson regression compared pregnancy rates during and after the vaccination period. Cox proportional hazards regression identified associations with first pregnancy. RESULTS: Among 352 women (median age 23 yrs; median follow-up 1.5 yrs), pregnancy incidence was 9.6/100 women-years overall and 6.8/100 w-yrs and 11.3/100 w-yrs during and after the vaccination period, respectively [Rate Ratio = 0.60 (0.32-1.14), p = 0.10]. In multivariable analysis, pregnancy was reduced among women who: enrolled at sites providing contraception on-site [HR = 0.43, 95% CI (0.22-0.86)]; entered the trial as injectable contraceptive users [HR = 0.37 (0.21-0.67)] or as consistent condom users (trend) [HR = 0.54 (0.28-1.04)]. Compared with women with a single partner of HIV-unknown status, pregnancy rates were increased among women with: a single partner whose status was HIV-negative [HR = 2.34(1.16-4.73)] and; 2 partners both of HIV-unknown status [HR = 4.42(1.59-12.29)]. Women with 2 more of these risk factors: marijuana use, heavy drinking, or use of either during sex, had increased pregnancy incidence [HR = 2.66 (1.24-5.72)]. CONCLUSIONS: It is possible to screen South African women for pregnancy risk at trial entry. Providing injectable contraception for free on-site and supporting consistent condom use may reduce incident pregnancy. Screening should determine the substance use, partnering, and HIV status of both members of the couple for both pregnancy and HIV prevention. Trial Registration SA National Health Research Database DOH-27-0207-1539; Clinicaltrials.gov NCT0041372