Early clindamycin for bacterial vaginosis in pregnancy (PREMEVA): a multicentre, double-blind, randomised controlled trial

Background Preterm delivery during pregnancy (<37 weeks' gestation) is a leading cause of perinatal mortality and morbidity. Treating bacterial vaginosis during pregnancy can reduce poor outcomes, such as preterm birth. We aimed to investigate whether treatment of bacterial vaginosis decreases late miscarriages or spontaneous very preterm birth. Methods PREMEVA was a double-blind randomised controlled trial done in 40 French centres. Women aged 18 years or older with bacterial vaginosis and low-risk pregnancy were eligible for inclusion and were randomly assigned (2:1) to three parallel groups: single-course or triple-course 300 mg clindamycin twice-daily for 4 days, or placebo. Women with high-risk pregnancy outcomes were eligible for inclusion in a high-risk subtrial and were randomly assigned (1:1) to either single-course or triple-course clindamycin. The primary outcome was a composite of late miscarriage (16-21 weeks) or spontaneous very preterm birth (22-32 weeks), which we assessed in all patients with delivery data (modified intention to treat). Adverse events were systematically reported. This study is registered with ClinicalTrials.gov, number NCT00642980. Findings Between April 1, 2006, and June 30, 2011, we screened 84530 pregnant women before 14 weeks' gestation. 5630 had bacterial vaginosis, of whom 3105 were randomly assigned to groups in the low-risk trial (n=943 to receive single-course clindamycin, n=968 to receive triple-course clindamycin, and n=958 to receive placebo) or high-risk subtrial (n=122 to receive single-course clindamycin and n=114 to receive triple-course clindamycin). In 2869 low-risk pregnancies, the primary outcome occurred in 22 (1.2%) of 1904 participants receiving clindamycin and 10 (1.0%) of 956 participants receiving placebo (relative risk [RR] 1.10, 95% CI 0.53-2.32; p=0.82). In 236 high-risk pregnancies, the primary outcome occurred in 5 (4.4%) participants in the triple-course clindamycin group and 8 (6.0%) participants in the single-course clindamycin group (RR 0.67, 95% CI 0. 23-2. 00; p=0.47). In the low-risk trial, adverse events were more common in the clindamycin groups than in the placebo group (58 [3.0%] of 1904 vs 12 [1.3%] of 956; p=0.0035). The most commonly reported adverse event was diarrhoea (30 [1.6%] in the clindamycin groups vs 4 [0.4%] in the placebo group; p=0.0071); abdominal pain was also observed in the clindamycin groups (9 [0.6%] participants) versus none in the placebo group (p=0.034). No severe adverse event was reported in any group. Adverse fetal and neonatal outcomes did not differ significantly between groups in the high-risk subtrial. Interpretation Systematic screening and subsequent treatment for bacterial vaginosis in women with low-risk pregnancies shows no evidence of risk reduction of late miscarriage or spontaneous very preterm birth. Use of antibiotics to prevent preterm delivery in this patient population should be reconsidered. Funding French Ministry of Health. Copyright (C) 2018 Elsevier Ltd. All rights reserved.