The evidence regarding the prevention of adverse outcomes in pregnant women with bacterial vaginosis (BV) is conflicting. Some evidence suggests that treatment of high-risk women (those who have already had a pre-term birth) and treatment of women before 20 weeks' gestation may have a beneficial effect on some adverse outcomes, but further evidence is needed to confirm this.

A Cochrane systematic review (search date: to May 2006) considered the use of antibiotics to prevent adverse outcomes associated with BV in pregnancy [McDonald et al, 2007]. The review included 15 RCTs of good quality (n = 5888); five trials used oral metronidazole alone, one used oral metronidazole plus erythromycin, one used oral clindamycin, one used ampicillin, one used vaginal metronidazole gel, and six used intravaginal clindamycin:

• The risk of pre-term birth less than 37 weeks' gestation was not significantly decreased with any antibiotic compared with no treatment or placebo (odds ratio [OR] 0.91, 95% CI 0.78 to 1.06; 15 trials of 5888 women). There was also no evidence of an effect on pre-term pre-labour rupture of membranes (OR 0.88, 95% CI 0.61 to 1.28; four trials of 2579 women), nor on the incidence of low-birthweight deliveries (OR 0.95, 95% CI 0.77 to 1.17; seven trials of 4107 women).
• As there were sufficient trials in the review, the comparisons were stratified, to explore the effect of the intervention on the outcomes, by the following factors: oral versus vaginal antibiotics, women with a previous pre-term birth, clindamycin compared with placebo treatment; and treatment before 20 weeks' gestation:
• Oral versus intravaginal antibiotics: the route of administration appeared to make no difference to the outcomes.
• Women with a previous pre-term birth: the use of antibiotics was associated with a significant decrease in the risk of pre-term pre-labour rupture of membranes (OR 0.14, 95% CI 0.05 to 0.38; two trials of 114 women) and low birthweight (OR 0.31, 95% CI 0.13 to 0.75; two trials of 114 women) but not of subsequent pre-term birth less than 37 weeks' gestation (OR 0.83, 95% CI 0.59 to 1.17; five trials of 622 women). All five studies compared oral metronidazole (various doses) with placebo.
• Clindamycin (oral or intravaginal) versus placebo: there was no significant difference in pre-term birth less than 37 weeks' gestation (OR 0.80, 95% CI 0.60 to 1.05; six trials of 2406 women), pre-term pre-labour rupture of membranes (OR 2.52, 95% CI 0.69 to 9.25; one trial of 100 women), or low birthweight (OR 1.03, 95% CI 0.73 to 1.45; four trials of 1648 women).
• Treatment before 20 weeks' gestation: antibiotic treatment was associated with a significant decrease in risk of pre-term birth less than 37 weeks' gestation (OR 0.72, 95% CI 0.55 to 0.95; five trials of 2287 women) and pre-term pre-labour rupture of membranes (OR 0.14, 95% CI 0.04 to 0.44; one trial of 80 women), but not low birthweight (OR 0.79, 95% CI 0.44 to 1.41; two trials of 489 women).
• Two earlier systematic reviews produced similar findings [Riggs and Klebanoff, 2004; Okun et al, 2005].
• One RCT has investigated oral metronidazole for the prevention of pre-term birth in women designated to be at high risk for pre-term delivery on the basis of clinical history and a positive test for cervicovaginal fetal fibronectin [Shennan et al, 2006]. The study was stopped early owing to difficulty in recruitment.
• Among the 100 women randomized to treatment, there were significantly more pre-term deliveries in those treated with metronidazole (33/53) than those who received placebo (18/46) (RR 1.6, 95% CI 1.05 to 2.4). However, only 13 of 96 women (14%) initially tested positive for BV. Further, appropriately powered studies are needed to confirm or refute these results in women with BV.