Purpose: To review the evidence about screening for bacterial vaginosis during pregnancy to prevent preterm delivery.
Data Sources: MEDLINE, the Cochrane Library, and trial registries through May 29, 2019; bibliographies from retrieved articles, outside experts, and surveillance of the literature through December 31, 2019.
Study Selection: Two investigators independently selected studies using a priori inclusion and exclusion criteria. We selected studies that evaluated the diagnostic accuracy of commercially available tests or tests feasible within primary care settings for bacterial vaginosis. We also selected controlled trials of treatment with metronidazole or clindamycin for bacterial vaginosis during pregnancy that reported preterm delivery or maternal adverse effect outcomes, and we selected observational studies that evaluated harms to children from in utero exposure to the medications. We excluded studies with poor methodological quality and studies conducted in developing countries.
Data Extraction and Analysis: One investigator extracted data and a second checked accuracy. Two reviewers independently rated methodological quality for all included studies using predefined criteria. When at least three similar studies were available, meta-analyses were conducted.
Data Synthesis: We included 44 studies. We did not identify any studies directly evaluating health benefits or harms of screening. Twenty-five studies evaluated the accuracy of screening tests; most were conducted in nonpregnant, symptomatic women. The sensitivity and specificity varied by test: BD Affirm (pooled sensitivity, 0.87 [95% confidence interval {CI}, 0.80 to 0.92], pooled specificity, 0.81 [95% CI, 0.73 to 0.88]; 5 studies; 2,936 participants), BD Max (sensitivity, 0.93 [95% CI, 0.91 to 0.94]; specificity, 0.92 [95% CI, 0.90 to 0.94]; 1 study; 1,338 participants), BVBLUE (sensitivity range, 0.61 to 0.92; specificity range, 0.86 to 0.99; 3 studies; 864 participants), Amsel’s clinical criteria (pooled sensitivity, 0.76 [95% CI, 0.63 to 0.85]; pooled specificity, 0.95 [95% CI, 0.89 to 0.98]; 14 studies; 5,790 participants), and modified Amsel’s clinical criteria (pooled sensitivity, 0.67 [95% CI, 0.54 to 0.78]; pooled specificity, 0.96 [95% CI, 0.93 to 0.98]; 4 studies; 2,477 participants).
Thirteen randomized, controlled trials (RCT) compared either oral metronidazole or oral or intravaginal clindamycin with either a placebo control or with no treatment for asymptomatic bacterial vaginosis in pregnancy. Among a general obstetric population, six RCTs reported no difference in any delivery before 37 weeks gestation (pooled absolute risk difference [ARD], 0.20% [95% CI, −1.13% to 1.53%]; 6,307 participants), and eight RCTs reported no difference in spontaneous delivery before 37 weeks (pooled ARD, −1.44% [95% CI, −3.31% to 0.43%]). No treatment effects were observed for other pregnancy outcomes, including delivery before 32 weeks gestation, low birth weight, premature rupture of membranes, and several others. In the four RCTs reporting preterm delivery before 37 weeks among women with a prior preterm delivery, three reported a significant reduction for treatment compared with control, and one reported no difference. In two RCTs reporting preterm delivery before 34 weeks among women with a prior preterm delivery, both reported no difference between treatment and control groups.
Fourteen studies reported on harms of treatment. Among eight RCTs reporting maternal adverse effects, events were infrequent and minor (e.g., candidiasis, gastrointestinal upset) but were slightly more common for oral clindamycin and metronidazole compared with placebo. Six observational studies reported on adverse effects on children exposed to oral metronidazole in utero. Two meta-analyses of observational studies reported no difference in congenital malformations in exposed children (odds ratio [OR], 0.96 [95% CI, 0.75 to 1.22]; OR, 1.08 [95% CI, 0.90 to 1.29]). Findings from three additional studies published subsequent to these metanalyses observed similar results. One cohort study reported no increased incidence of childhood cancer among exposed children (adjusted relative risk, 0.81 [95% CI, 0.41 to 1.59]).
Limitations: Only English-language studies were included. No direct evidence for the benefits or harms of screening was identified. The evidence on diagnostic accuracy may have limited applicability to pregnant, asymptomatic populations. We did not assess comparative accuracy of tests or comparative effectiveness or harms of treatments. Studies of treatment were generally underpowered for harm outcomes. We did not evaluate treatments other than metronidazole and clindamycin.
Conclusions: We identified no direct evidence that compared screening with no screening and that reported health outcomes. Diagnostic test accuracy studies were mostly conducted in nonpregnant, symptomatic women; across individual studies the sensitivity of the various tests ranged from 0.36 to 1.0 and the specificity ranged from 0.49 to 1.0. RCTs conducted in general obstetric populations reported no difference in the incidence of preterm delivery and related outcomes for treatment with metronidazole or clindamycin compared with placebo. The evidence is inconclusive for treatment in women with a prior preterm delivery. Maternal adverse events from treatment with metronidazole or clindamycin are infrequent and minor. The observational study evidence about harms to children from in utero exposure to medication is inconclusive because of study limitations and imprecision.