Final Evidence Summary
Bacterial Vaginosis in Pregnant Persons to Prevent Preterm Delivery: Screening
April 07, 2020
Recommendations made by the USPSTF are independent of the U.S. government. They should not be construed as an official position of the Agency for Healthcare Research and Quality or the U.S. Department of Health and Human Services.
By Leila C. Kahwati, MD, MPH; Rachel Clark, BA; Nancy Berkman, PhD; Rachel Urrutia, MD, MS; Sheila V. Patel, BS; Jennifer Zeng, MD, MPH; Meera Viswanathan, PhD
The information in this article is intended to help clinicians, employers, policymakers, and others make informed decisions about the provision of health care services. This article is intended as a reference and not as a substitute for clinical judgment.
This article may be used, in whole or in part, as the basis for the development of clinical practice guidelines and other quality enhancement tools, or as a basis for reimbursement and coverage policies. AHRQ or U.S. Department of Health and Human Services endorsement of such derivative products may not be stated or implied.
This article was published online in JAMA on April 7, 2020 (JAMA. 2020;323(13):1293-1309. doi:10.1001/jama.2020.0233).
Importance: Preterm delivery results in adverse outcomes; identifying and treating bacterial vaginosis may reduce its occurrence.
Objective: To update the evidence on screening and treatment of asymptomatic bacterial vaginosis in pregnancy for the US Preventive Services Task Force.
Data Sources: MEDLINE, Cochrane Library, and trial registries through May 29, 2019; bibliographies from retrieved articles, experts, and surveillance of the literature through December 31, 2019.
Study Selection: Fair- or good-quality English-language studies evaluating diagnostic accuracy of tests feasible within primary care; randomized clinical trials (RCTs); nonrandomized controlled intervention studies (for harms only); or meta-analyses of metronidazole or clindamycin.
Data Extraction and Synthesis: Two reviewers independently assessed titles/abstracts and full-text articles, extracted data, and assessed study quality; when at least 3 similar studies were available, meta-analyses were conducted.
Main Outcomes and Measures: Sensitivity, specificity, preterm delivery, maternal adverse effects, congenital birth defects, childhood cancer.
Results: Forty-four studies (48 publications) were included. No studies evaluated the benefits or harms of screening. Twenty-five studies (n = 15,785) evaluated the accuracy of screening tests; across individual studies and tests, sensitivity ranged from 0.36 to 1.0 and specificity ranged from 0.49 to 1.0. Among trials reporting findings from general obstetric populations (n = 7953), no significant association was observed between treatment and spontaneous delivery before 37 weeks (pooled absolute risk difference [ARD], –1.44% [95% CI, −3.31% to 0.43%]; 8 RCTs, n = 7571) or any delivery before 37 weeks (pooled ARD, 0.20% [95% CI, −1.13% to 1.53%]; 6 RCTs, n = 6307). Among 5 trials reporting findings among women with a prior preterm delivery, findings were inconsistent; 3 showed a significant beneficial effect, while 2 did not. Maternal adverse events from treatment were infrequent and minor (eg, candidiasis) but were slightly more common with active treatment compared with placebo across 8 RCTs. Two meta-analyses of observational studies reported no significant association between metronidazole exposure and congenital malformations (odds ratio, 0.96 [95% CI, 0.75 to 1.22]; odds ratio, 1.08 [95% CI, 0.90 to 1.29]). One cohort study reported no significantly increased incidence of childhood cancer among metronidazole exposed children (adjusted relative risk, 0.81 [95% CI, 0.41 to 1.59]). However, studies of in utero exposure had important limitations.
Conclusions and Relevance: Accuracy of screening tests for bacterial vaginosis varies. The evidence suggests no difference in the incidence of preterm delivery and related outcomes from treatment for asymptomatic bacterial vaginosis in a general obstetric population but was inconclusive for women with a prior preterm delivery. Maternal adverse events from treatment appear to be infrequent and minor, but the evidence about harms from in utero exposure was inconclusive.
Bacterial vaginosis is a common lower genital tract syndrome defined as a shift from normal hydrogen peroxide–producing lactobacilli to mixed anaerobes.1,2 Studies conducted between 1983 and 2006 estimate that only 25% to 50% of women with bacterial vaginosis report symptoms.3-5 Research has suggested bacterial vaginosis as a risk factor for preterm delivery; a 2007 meta-analysis of 32 studies estimated a pooled odds ratio for the risk of preterm delivery in the presence of asymptomatic bacterial vaginosis of 2.16 (95% CI, 1.56 to 3.00).6 The causal mechanism is not fully understood.7,8
Early identification and treatment of bacterial vaginosis may reduce the incidence of preterm delivery and its associated morbidity and mortality. This review was conducted to inform the US Preventive Services Task Force (USPSTF) for its update of the 2008 recommendation on screening and treatment of bacterial vaginosis in pregnancy to prevent preterm delivery.9 In 2008, the USPSTF recommended against screening for bacterial vaginosis in asymptomatic pregnant women at low risk for preterm delivery (D recommendation) and concluded that the evidence was insufficient for asymptomatic pregnant women at high risk for preterm delivery (I statement).
Scope of the Review
The analytic framework and key questions (KQs) that guided the review are shown in Figure 1. Detailed methods, evidence tables, sensitivity analyses, and contextual information are available in the full evidence report at https://www.uspreventiveservicestaskforce.org/uspstf/recommendation/bacterial-vaginosis-in-pregnancy-to-prevent-preterm-delivery-screening.
Data Sources and Searches
PubMed, the Cochrane Library, and EMBASE were searched for English-language articles published from January 1, 2006, through May 29, 2019. Because the previous reviews for the USPSTF did not include a systematic search for KQ2 (diagnostic test accuracy), a separate PubMed search from inception through December 31, 2005, was conducted to supplement the main search for this update. ClinicalTrials.gov, the Cochrane Clinical Trials Registry, and the World Health Organization International Clinical Trials Registry Platform were also searched. To supplement systematic electronic searches (eMethods in the Supplement), reference lists of pertinent articles and studies suggested by reviewers were searched. Ongoing surveillance was conducted through article alerts and targeted searches of journals to identify major studies published in the interim that may affect the conclusions or understanding of the evidence and the related USPSTF recommendation. The last surveillance was conducted on December 31, 2019.
Two investigators independently reviewed titles, abstracts, and full-text articles using prespecified inclusion criteria for each KQ (eMethods in the Supplement); disagreements about inclusion were resolved by a third reviewer. Briefly, for KQ1, KQ3, and KQ4, randomized clinical trials (RCTs) and relevant systematic reviews of RCTs, conducted in pregnant women or adolescents, were selected; for KQ1 and KQ3, participants had to be asymptomatic with respect to vaginal symptoms of bacterial vaginosis. For KQ1 and KQ3, studies that compared screening with no screening and reported health outcome benefits (eg, reduction in preterm delivery) or harms (eg, anxiety) were selected. For KQ2, studies that reported on diagnostic test accuracy for Amsel clinical criteria (vaginal pH >4.5, clue cells, discharge, amine odor)3 or laboratory-based tests in commercial use or feasible for use in primary care settings were selected. Participants were not required to be pregnant in studies selected for KQ2. For KQ4 and KQ5, trials that compared treatment with metronidazole or clindamycin vs placebo or no treatment in symptomatic or asymptomatic pregnant women with bacterial vaginosis and that reported health outcomes related to preterm delivery or other adverse pregnancy outcomes were selected. For KQ5, observational studies that reported on outcomes related to fetal exposure to metronidazole or clindamycin, such as carcinogenesis or congenital malformations, were eligible.
English-language studies that met all study selection criteria, were fair or good methodological quality, and were conducted in countries categorized as very highly developed by the 2017 United Nations Human Development Index were included.11 Studies included in the prior 2008 review were reassessed against the study selection and methodological quality criteria for this update.
Data Extraction and Quality Assessment
For each included study, 1 reviewer abstracted relevant study characteristics (ie, population, intervention, comparator) and data for eligible outcomes into a structured form. A second reviewer checked all data for completeness and accuracy. Some study authors were contacted to clarify data. Two senior reviewers independently assessed each study’s methodological quality using predefined criteria established by the USPSTF (eMethods in the Supplement) and others.12-16 Disagreements in study quality ratings were resolved through discussion or with an independent assessment from a third senior investigator. Studies reporting multiple outcomes may have been assigned different quality ratings for different outcomes
Data Synthesis and Analysis
For diagnostic test accuracy (KQ2), data related to sensitivity, specificity, and likelihood ratios were synthesized in tabular and narrative formats. When at least 3 studies using the same index test and test threshold were available, a quantitative synthesis was performed by fitting the bivariate model described by Reitsma et al17 with the metandi package in Stata version 15 (StataCorp) to generate a summary receiver operating characteristics curve and a pooled summary point estimate of sensitivity and specificity. For benefits of treatment (KQ4), findings were synthesized using both absolute risk differences (ARDs) and relative risk (RR) ratios. For harms of treatment (KQ5), odds ratios (ORs) were also used. When a quantitative synthesis was possible, a random-effects model with the inverse variance weighted method of DerSimonian and Laird with the metafor package in R version 2.0-0 (R Foundation for Statistical Computing) was used.18 Significance testing was based on the exclusion of the null value by the 95% CI around the pooled estimate.
The strength of evidence was assessed based on the Agency for Healthcare Quality and Research Methods Guide for Effectiveness and Comparative Effectiveness Reviews, which specifies the assessment of study limitations, directness, consistency, precision,and reporting bias for each intervention comparison and major outcome of interest.19 Two senior reviewers independently developed initial strength of evidence assessments for each relevant outcome and comparison across the KQs; disagreements were resolved through discussion and the independent assessment of a third senior reviewer.
Forty-four studies from 48 publications were included (Figure 2). Twenty-five studies of test accuracy (KQ2),20-44 13 RCTs evaluating the benefits of treatment with respect to preterm delivery and related pregnancy outcomes (KQ4),45-57 and 14 studies evaluating the harms of treatment (KQ5)45,48,49,51,52,55-63 were identified.
Benefits of Screening
Key Question 1. Does screening for bacterial vaginosis in asymptomatic pregnant adolescent and women reduce preterm delivery and related morbidity and mortality?
No studies were identified.
Accuracy of Screening
Key Question 2. What is the diagnostic accuracy of tests used to screen for bacterial vaginosis?
Twenty-five cross-sectional diagnostic test accuracy studies (n = 15,785) reported test accuracy for laboratory assays and for Amsel clinical criteria (complete or modified). Study characteristics are reported in eTable1 in the Supplement, and individual study methodological quality is described in eTables 2 through 7 in the Supplement. Six studies were assessed as good methodological quality;23,25,26,31,37,43 the others were assessed as fair quality generally because of unclear enrollment procedures and unclear information regarding blinding of index and reference test results. The reference standard assessed in nearly all studies was a Gram stain of vaginal secretions, most often interpreted using Nugent criteria, a scoring system based on quantity and morphotypes of organisms present.64,65 Two studies enrolled exclusively pregnant or asymptomatic women.23,40 Table 1 summarizes the accuracy of various tests; across individual studies and tests, sensitivity ranged from 0.36 to 1.0 and specificity ranged from 0.49 to 1.0.
Harms of Screening
Key Question 3. What are the harms of screening for bacterial vaginosis in asymptomatic pregnant adolescents and women?
No studies were identified.
Benefits of Treatment
Key Question 4. Does treatment of bacterial vaginosis during pregnancy reduce preterm delivery and related morbidity and mortality?
Thirteen RCTs (n = 8751) reported findings related to preterm delivery, other pregnancy outcomes, or clearance of bacterial vaginosis.45-57 Study characteristics are summarized in Table 2, with additional characteristics described in eTable 3 in the Supplement. Nine RCTs45,46,48,50,52,53,55-57 were assessed as good methodological quality, and 4 RCTs47,49,51,54 were assessed as fair quality, primarily because of concerns related to lack of information on allocation concealment and lack of information to assess adequacy of randomization,51 lack of treatment blinding,49,51 post hoc subgroup analyses,47,49 or lack of intent-to-treat analyses.54 Individual study methodological quality is described in eTable 4 in the Supplement. No studies reported subgroup findings by maternal or gestational age, race or ethnicity, HIV status, or other population characteristics specified by the KQs.
Four studies were conducted in the US;45,47,53,54 the others were conducted in Australia52 and various countries in Europe.46,48-51,55-57 Ten of the 13 studies (n = 7953)were conducted among general obstetric populations, meaning that patients were enrolled without regard to their risk for preterm delivery.45,46,48-53,55,56 The percentage of participants with a prior preterm delivery in these studies ranged from 0% to 10.9%. Two of these studies (n = 194) also reported findings among subgroups considered at high risk for preterm delivery because of a prior preterm delivery.45,52 Three of the 13 studies (n = 279)were conducted solely among participants considered at high risk for preterm delivery.47,54,57 All 3 defined high risk as a previous preterm delivery; however, 1 study also considered women with a prepregnancy weight less than 50 kg and no previous preterm delivery as high risk.47 Most studies identified asymptomatic patients during routine prenatal visits and enrolled participants during the second trimester, although criteria for enrollment varied. Three studies enrolled participants without regard to bacterial vaginosis status but reported findings for the subgroup of participants testing positive for bacterial vaginosis at study entry. Study findings are only reported in this article from the subgroups with bacterial vaginosis.47,49,57
Three studies evaluated the use of oral metronidazole,45,52,54 2 studies evaluated oral clindamycin,55,56 1 study evaluated oral metronidazole and erythromycin,47 and 7 evaluated intravaginal clindamycin.46,48-51,53,57 The dosages and durations of treatment varied across studies, and most, but not all, used a placebo control. Two studies repeated treatment if the test of a cure demonstrated persistent bacterial vaginosis,49,52 and 3 studies repeated dosing at a later follow-up point without regard to results from a test of cure for some or all participants.45,55,57 Twelve of the 13 RCTs45-56 reported findings related to preterm delivery prior to 37 weeks’ gestational age; 1 RCT57 only reported preterm delivery defined as delivery prior to 34 weeks. Detailed results are summarized in eTable 5 in the Supplement.
Preterm Delivery in General Obstetric Populations
Ten RCTs conducted among general obstetric populations reported preterm delivery outcomes (Figure 3). The absolute risk of delivery prior to 37 weeks’ gestational age in the placebo groups ranged from 3.1% to 15.7%. Among the 6 studies reporting all-cause preterm delivery, the pooled ARD comparing active treatment with control was 0.20% (95% CI, −1.13% to 1.53%; 6307 participants; I2 = 0%), and the pooled RR was 1.02 (95% CI, 0.86 to 1.20).45,46,51-53,55 No individual studies reported a significant difference between active treatment and control. Among the 8 studies reporting spontaneous preterm deliveries, the pooled ARD comparing active treatment with control was −1.44% (95% CI, −3.31% to 0.43%; 7571 participants; I2 = 61.9%), and the pooled RR was 0.78 (95% CI, 0.56 to 1.07).45,48-52,55,56 Two of the 8 studies reported statistically significant reductions in spontaneous preterm delivery for active treatment compared with control,50,56 while the other 6 reported no significant differences between active treatment and control.
One of the studies that reported a significant reduction in spontaneous preterm delivery enrolled participants (n = 409) with either bacterial vaginosis or intermediate flora;50 other population or intervention characteristics that might explain this inconsistency could not be identified.
Three RCTs reported the incidence of preterm delivery prior to 32 weeks’ completed gestation among a general obstetric population (eFigures 4 and 5 in the Supplement).45,51,55 The pooled ARD was −0.30% (95% CI, −0.97% to 0.38%; 5564 participants; I2 = 15.4%), and the pooled RR was 0.87 (95% CI, 0.54 to 1.42). All 3 studies observed no significant differences between active treatment and control. One RCT also reported no significant difference in preterm delivery at less than 34 weeks’ gestation (ARD, −0.04% [95% CI, −2.0% to 1.92%]; RR, 1.0 [95% CI, 0.7 to 1.5]).45
Other pregnancy-related outcomes in a general obstetric population for which a pooled summary estimate was calculatable are provided in eFigures 4 and 5 in the Supplement. No significant association between treatment and low birthweight, very low birthweight, or premature rupture of membranes was observed. Studies also reported outcomes for which pooled summary estimates could not be generated, including maternal peripartum infection,48 stillborn fetus,49 preterm labor,53 and neonatal mortality;55 authors observed no significant differences between active treatment and control for these outcomes.
Preterm Delivery in Women With Prior Preterm Delivery
Five RCTs reported preterm delivery outcomes in this subgroup; 3 reported incidence of preterm delivery at less than 37 weeks,45,47,52,54 1 reported incidence of preterm delivery at less than 34 weeks,57 and 1 reported incidence of preterm delivery at both less than 37 weeks and less than 34 weeks.39 Findings for this subgroup were not pooled because of heterogenous outcome measures.
In the 4 RCTs (n = 451) conducted among participants with a previous preterm delivery or that reported subgroup findings for such women, the incidence of preterm delivery at less than 37 weeks’ gestation in control groups ranged from 22.5% to 57.1% (Figure 4).45,47,52,54 Carey et al45 and Hauth et al47 reported all-cause preterm delivery, and Morales et al54 and McDonald et al52 reported spontaneous preterm delivery. Three of the 4 RCTs reported a statistically significant favorable treatment effect (ARDs ranging from –18.3% to –29.4%),47,52,54 while Carey et al45 (subgroup n = 160) observed no significant treatment effect (ARD, 7.50% [95% CI, −6.09% to 21.09%]). The inconsistency in findings could not be explained based on study or population characteristics (further details are reported in the eResults in the Supplement).
Two RCTs reported the incidence of preterm delivery at less than 34 weeks’ gestation among participants with a prior preterm delivery (Figure 4).54,57 In Morales et al (n = 80),54 4 participants (11%) in the placebo group and 2 participants (4.6%) in the oral metronidazole group had a spontaneous preterm delivery at less than 34 weeks (calculated ARD, −6.57% [95% CI, −18.5% to 5.40%]). Vermeulen and Bruinse57 reported the incidence of all-cause preterm delivery at less than 34 weeks’ gestation among a subgroup of 22 participants with bacterial vaginosis and observed 1 event in both the vaginal clindamycin and the placebo group.
With respect to other pregnancy outcomes, Morales et al (n = 80)54 reported a significant treatment effect on preterm labor (calculated ARD, −50.51% [95% CI, −69.41% to −31.60%]), premature rupture of membranes (calculated ARD, −28.79% [95% CI, −45.37% to −12.21%]), and birth weight less than 2500 g (calculated ARD, −19.7% [95% CI, −38.13% to −1.26%]). Vermeulen and Bruinse (n = 22)57 reported no significant treatment effect on neonatal sepsis.
Preterm Delivery Based on Bacterial Vaginosis Clearance Status
Some studies conducted among a general obstetric population reported preterm delivery outcomes for subgroups of participants who had documented clearance or persistence of bacterial vaginosis after treatment. Among a subgroup of participants who had follow-up Gram staining after initial testing and treatment, Carey et al (n = 1704)45 reported no significant difference in preterm delivery among women with clearance of bacterial vaginosis (incidence, 10.6%) vs those with persistence of bacterial vaginosis (incidence, 10.7%) (P = .95). Kekki et al48 also reported no significant difference in preterm delivery between active treatment and control among a subgroup (n = 121) of women with documented clearance of bacterial vaginosis 1 week after treatment (calculated ARD, 2.30% [95% CI, −1.45% to 6.06%]).
Harms of Treatment
Key Question 5. What are the harms of treatment of bacterial vaginosis in pregnant adolescents and women?
Fourteen studies reported on the harms of treatment. Eight RCTs reported on maternal adverse events,45,48,49,51,52,55-57 and 6 studies reported on adverse outcomes in children exposed to medication in utero.58-63 eTable 6 in the Supplement provides an assessment of individual study methodological quality.
Maternal Adverse Events
Among the 13 RCTs reporting on the benefits of treatment for bacterial vaginosis during pregnancy (KQ4), 8 (n = 7758) reported on maternal adverse events. These 8 RCTs included 4 trials of intravaginal clindamycin,48,49,51,57 2 trials of oral clindamycin,55,56 and 2 trials of oral metronidazole.45,52 Results from individual studies are presented in eTable 7 in the Supplement. Across this body of evidence, maternal adverse events from treatment with oral clindamycin or oral metronidazole generally occurred at a higher incidence compared with control treatment but were not severe (eg, gastrointestinal symptoms, candida infection). For example, in Carey et al (n = 1704; oral metronidazole),45 the ARD for gastrointestinal symptoms was 12.5%, and in Subtil et al (n = 2860; oral clindamycin),55 the ARD was 1.2%. Adverse events from intravaginal clindamycin were infrequent and mild (eg, vaginal itching).
Adverse Childhood Outcomes Associated With In Utero Exposure to Medication
Six studies (eTable 8 in the Supplement) reporting adverse childhood outcomes associated with in utero exposure to metronidazole were included.58-63 Three observational studies (n = 62,271)60-62 and 2 meta-analyses58,59 reported on outcomes related to congenital abnormalities and malformations, and 1 observational study (n = 328,846)63 reported on incidence of childhood cancer. One study was assessed as poor methodological quality because of confounding and because of a large amount of missing data;61 however, it was retained for continuity with the previous review. All other studies were assessed as fair methodological quality.
The studies included for this KQ did not provide information about the indication for metronidazole treatment; the setting of treatment (ie, inpatient vs outpatient); or the dose, duration, and route of treatment. Furthermore, the populations were not limited to women exposed to metronidazole specifically for the treatment of bacterial vaginosis in pregnancy, which may limit applicability; however, those studies were retained in this update for continuity with the previous review.
The 2 included meta-analyses found no significant association between metronidazole and congenital malformations (OR, 0.96 [95% CI, 0.75 to 1.22; N not reported]58 and OR, 1.08 [95% CI, 0.90 to 1.29; n = 199,451]59). Similarly, 2 of the 3 observational studies60-62 found no association between metronidazole and congenital abnormalities. The exception was reported by Czeizel and Rockenbauer.60 This fair-quality study (n = 47,963) found a significant association between congenital anomalies and exposure to metronidazole during the first month of gestation (OR, 2.24 [95% CI, 1.30 to 3.85]) but not for the second through third months or fourth through ninth months.60 The authors noted that because the first month of gestation is counted from the first day of the last menstrual period, several of these weeks of exposure may be before conception or during the all-or-none phase of fetal development; thus, this finding may be spurious or the result of recall bias or uncontrolled confounding.60
One cohort study among women enrolled in Tennessee Medicaid did not find an association between metronidazole exposure during pregnancy and diagnosis of first cancer before age 5 years among exposed children (n = 328,846; adjusted RR, 0.81 [95% CI, 0.41 to 1.59]).63
This evidence report reviewed studies on the diagnostic accuracy of screening tests for bacterial vaginosis and studies evaluating the benefits and harms of metronidazole or clindamycin treatment in pregnancy. Table 3 summarizes the evidence by KQ and provides an assessment of the strength of evidence. Compared with the 2008 review for the USPSTF on this topic, 2 RCTs were added and 2 RCTs were excluded. Despite this change in the evidence base, the overall conclusions about no benefit in a general obstetric population remain unchanged from the prior report.
For diagnostic accuracy (KQ2), the strength of evidence was assessed as low for adequate accuracy for all tests evaluated because of fair methodological quality and inconsistency. Most studies were conducted among symptomatic, nonpregnant women; thus, the applicability to asymptomatic pregnant women is not clear. For complete Amsel and modified Amsel clinical criteria, the sensitivities observed in the 2 studies23,40 conducted exclusively among pregnant women were lower than the pooled summary estimates, suggesting that the physiologic changes that occur in the vaginal environment during pregnancy may affect the sensitivity of 1 or more of the clinical criteria used to identify bacterial vaginosis. Furthermore, a lower sensitivity was not observed for the BD Affirm test in the 1 study conducted exclusively in pregnant women.41 Although no formal comparative assessment was conducted, the tests varied somewhat in accuracy. The laboratory-based tests (BD Affirm VPIII [Becton, Dickinson], BD Max, OSOM BVBLUE [Sekisui Diagnostics]) had higher sensitivities than those based on Amsel clinical criteria but lower specificities.
Among a general obstetric population, the strength of evidence was moderate for no benefit of treatment on all-cause preterm delivery because of imprecision and low for no benefit of treatment on spontaneous preterm delivery because of imprecision and inconsistency. With respect to precision, although most studies were powered for the outcome of preterm delivery, either a lower control group risk was observed than was expected or the treatment effect observed was smaller than expected, resulting in imprecise estimates. Regarding spontaneous preterm delivery, the strength of evidence was also influenced by methodological considerations. The consequences related to preterm delivery generally do not differ for medically indicated deliveries vs spontaneous deliveries, and treatment could result in a medical complication that results in delivery after randomization but before the outcome reporting window that would not be captured. In addition, because an indicated preterm delivery is a competing risk to a spontaneous preterm delivery, use of spontaneous delivery outcomes could introduce informative censoring.
Among women with a prior preterm delivery, the strength of evidence for preterm delivery at less than 37 weeks was insufficient because of inconsistency and imprecision. Furthermore, its applicability is limited to treatment with oral metronidazole. A source for the inconsistency in findings could not be identified. Findings from 3 of the 4 studies were based on subgroup analyses, some of which were post hoc. The 2 studies reporting preterm delivery at less than 34 weeks did not observe any significant differences between groups, but results were very imprecise.
Compared with placebo, the strength of evidence for serious maternal adverse events related to treatment was moderate for no difference for oral metronidazole and both oral and intravaginal clindamycin. Compared with placebo, the strength of evidence for minor adverse events was moderate for no difference for intravaginal clindamycin and was moderate for an increase in minor events for both oral metronidazole and oral clindamycin. These bodies of evidence were rated as moderate because of imprecision due to relatively infrequent events.
The strength of evidence for congenital malformations and incidence of cancer among children exposed to metronidazole in utero was insufficient. This evidence comprises observational studies with no more than fair methodological study quality, and despite large sample sizes, the incidence of these types of events was rare, resulting in imprecise estimates. This evidence applies to metronidazole exposure during pregnancy across a range of medical indications and is not specific to treatment for bacterial vaginosis.
This review has several limitations. First, no available evidence that directly evaluated the health benefits and harms of screening (KQ1 and KQ3) was identified. Second, for diagnostic test accuracy (KQ2), limited evidence was available for pregnant, asymptomatic populations. Most studies were of only fair methodological quality, and for most tests, moderate to substantial heterogeneity was observed. Most studies used Gram stain as a reference standard; however, in light of the advances in the molecular and microbiological understanding of bacterial vaginosis, this may be an imperfect standard.
Third, for benefits of treatment (KQ4) and adverse maternal events (KQ5), studies varied with respect to dose and duration of treatment, use of a test of cure, and methodological quality. The findings in women with a prior preterm delivery were inconsistent, and a source for this inconsistency could not be identified. Fourth, with respect to harms, trials were underpowered for maternal adverse events, and the comparative harms of treatment were not assessed.
Fifth, this review was limited to treatment with only metronidazole and clindamycin. Although other treatments for bacterial vaginosis are available, they have not been studied in pregnant women. Sixth, only observational studies were available to assess the harms to children related to in utero exposure to medications (KQ5), and all of these studies included women exposed to metronidazole for any indication, including but not limited to bacterial vaginosis. Given the infeasibility of conducting randomized studies large enough and over a long enough duration to provide definitive evidence on in utero exposure, it is unlikely that this body of evidence will become stronger. However, these medications have had widespread and longstanding use in clinical practice.
Accuracy of screening tests for bacterial vaginosis varies. The evidence suggests no difference in the incidence of preterm delivery and related outcomes from treatment for asymptomatic bacterial vaginosis in a general obstetric population but was inconclusive for women with a prior preterm delivery. Maternal adverse events from treatment appear to be in frequent and minor, but the evidence about harms from in utero exposure was inconclusive.
Source: This article was first published online in the Journal of the American Medical Association on April 7, 2020 (JAMA. 2020;323(13):1293-1309. doi:10.1001/jama.2020.0233).
Conflict of Interest Disclosures: None reported.
Funding/Support: This research was funded under contract HHSA-290-2015-00011-I, Task Order 11, from the Agency for Healthcare Research and Quality (AHRQ), US Department of Heath and Human Services, under a contract to support the US Preventive Services Task Force (USPSTF).
Role of the Funder/Sponsor: Investigators worked with USPSTF members and AHRQ staff to develop the scope, analytic framework, and key questions for this review. AHRQ had no role in study selection, quality assessment, or synthesis. AHRQ staff provided project oversight, reviewed the report to ensure that the analysis met methodological standards, and distributed the draft for peer review. Otherwise, AHRQ had no role in the conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript findings. The opinions expressed in this document are those of the authors and do not reflect the official position of AHRQ or the US Department of Health and Human Services.
Additional Contributions: We thank the following individuals for their contributions to this project: AHRQ staff, Tina Fan, MD, MPH, and Tracy Wolff, MD; former AHRQ staff, Quyen Ngo-Metzger, MD, MPH; current and former members of the USPSTF who contributed to topic deliberations; RTI International–University of North Carolina Evidence-based Practice Center staff, B. Lynn Whitener, DrPH; Carol Woodell, BSPH; Sharon Barrell, MA; and Loraine Monroe. USPSTF members, peer reviewers, and federal partner reviewers did not receive financial compensation for their contributions.
Additional Information: A draft version of the full evidence report underwent external peer review from 4 content experts (Mark Klebanoff, MD, MPH, Nationwide Children’s Hospital; John Thorp, MD, University of North Carolina; Valerie J. King, MD, MPH, Oregon Health & Science University; and Julie van Schalkwyk, MD, University of British Columbia) and 2 additional federal partner reviewers (National Institutes of Health; Centers for Disease Control and Prevention). None of the reviewers received compensation for their role in reviewing the report. Comments from reviewers were presented to the USPSTF during its deliberation of the evidence and were considered in preparing the final evidence report.
1. Srinivasan S, Fredricks DN. The human vaginal bacterial biota and bacterial vaginosis. Interdiscip Perspect Infect Dis. 2008;2008:750479. doi:10.1155/2008/750479
2. Livengood CH. Bacterial vaginosis: an overview for 2009. Rev Obstet Gynecol. 2009;2(1):28-37.
3. Amsel R, Totten PA, Spiegel CA, Chen KCS, Eschenbach D, Holmes KK. Nonspecific vaginitis: diagnostic criteria and microbial and epidemiologic associations. Am J Med. 1983;74(1):14-22. doi:10.1016/0002-9343(83)91112-9
4. ACOG Committee on Practice Bulletins—Gynecology. ACOG Practice Bulletin: clinical management guidelines for obstetrician-gynecologists, number 72, May 2006: vaginitis. Obstet Gynecol. 2006;107(5):1195-1206.
5. Klebanoff MA, Schwebke JR, Zhang J, Nansel TR, Yu KF, Andrews WW. Vulvovaginal symptoms in women with bacterial vaginosis. Obstet Gynecol. 2004;104(2):267-272. doi:10.1097/01.AOG.0000134783.98382.b0
6. Leitich H, Kiss H. Asymptomatic bacterial vaginosis and intermediate flora as risk factors for adverse pregnancy outcome. Best Pract Res Clin Obstet Gynaecol. 2007;21(3):375-390. doi:10.1016/j.bpobgyn.2006.12.005
7. Hillier SL, Nugent RP, Eschenbach DA, et al; Vaginal Infections and Prematurity Study Group. Association between bacterial vaginosis and preterm delivery of a low-birth-weight infant. N Engl J Med. 1995;333(26):1737-1742. doi:10.1056/NEJM199512283332604
8. Goldenberg RL, Culhane JF, Iams JD, Romero R. Epidemiology and causes of preterm birth. Lancet. 2008;371(9606):75-84. doi:10.1016/S0140-6736(08)60074-4
9. Calonge N, Petitti DB, DeWitt TG, et al; US Preventive Services Task Force. Screening for bacterial vaginosis in pregnancy to prevent preterm delivery: U.S. Preventive Services Task Force recommendation statement. Ann Intern Med. 2008;148(3):214-219. doi:10.7326/0003-4819-148-3-200802050-00007
10. Procedure Manual. US Preventive Services Task Force (USPSTF). Published 2015. Accessed November 20, 2017. https://www.uspreventiveservicestaskforce.org/uspstf/procedure-manual
11. Human Development Index and its components. United Nations Development Programme. Published 2016. Accessed October 15, 2017. http://hdr.undp.org/en/composite/HDI
12. Higgins JP, Green S. Cochrane Handbook for Systematic Reviews of Interventions. Cochrane Collaboration. Published 2011. Accessed August 24, 2016. https://training.cochrane.org/handbook
13. Sterne JA, Hernán MA, Reeves BC, et al. ROBINS-I: a tool for assessing risk of bias in non-randomised studies of interventions. BMJ. 2016;355:i4919. doi:10.1136/bmj.i4919
14. Critical appraisal notes and checklists: checklist 4: case control studies. Scottish Intercollegiate Guidelines Network (SIGN). Published 2012. Accessed August 24, 2016. https://www.sign.ac.uk/checklists-and-notes.html
15. Whiting P, Savović J, Higgins JP, et al; ROBIS Group. ROBIS: a new tool to assess risk of bias in systematic reviews was developed. J Clin Epidemiol. 2016;69:225-234. doi:10.1016/j.jclinepi.2015.06.005
16. Whiting PF, Rutjes AW, Westwood ME, et al; QUADAS-2 Group. QUADAS-2: a revised tool for the quality assessment of diagnostic accuracy studies. Ann Intern Med. 2011;155(8):529-536. doi:10.7326/0003-4819-155-8-201110180-00009
17. Reitsma JB, Glas AS, Rutjes AW, Scholten RJ, Bossuyt PM, Zwinderman AH. Bivariate analysis of sensitivity and specificity produces informative summary measures in diagnostic reviews. J Clin Epidemiol. 2005;58(10):982-990. doi:10.1016/j.jclinepi.2005.02.022
18. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials. 1986;7(3):177-188. doi:10.1016/0197-2456(86)90046-2
19. Agency for Healthcare Research and Quality. Methods Guide for Effectiveness and Comparative Effectiveness Reviews. Agency for Healthcare Research and Quality; 2014. AHRQ publication 10(14)-EHC063-EF.
20. Bradshaw CS, Morton AN, Garland SM, Horvath LB, Kuzevska I, Fairley CK. Evaluation of a point-of-care test, BVBlue, and clinical and laboratory criteria for diagnosis of bacterial vaginosis. J Clin Microbiol. 2005;43(3):1304-1308. doi:10.1128/JCM.43.3.1304-1308.2005
21. Chen HM, Chang TH, Lin FM, et al. Vaginal microbiome variances in sample groups categorized by clinical criteria of bacterial vaginosis. BMC Genomics. 2018;19(suppl 10):876. doi:10.1186/s12864-018-5284-7
22. Gallo MF, Jamieson DJ, Cu-Uvin S, Rompalo A, Klein RS, Sobel JD. Accuracy of clinical diagnosis of bacterial vaginosis by human immunodeficiency virus infection status. Sex Transm Dis. 2011;38(4):270-274.
23. Gratacós E, Figueras F, Barranco M, et al. Prevalence of bacterial vaginosis and correlation of clinical to Gram stain diagnostic criteria in low risk pregnant women. Eur J Epidemiol. 1999;15(10):913-916. doi:10.1023/A:1007673531595
24. Gutman RE, Peipert JF, Weitzen S, Blume J. Evaluation of clinical methods for diagnosing bacterial vaginosis. Obstet Gynecol. 2005;105(3):551-556. doi:10.1097/01.AOG.0000145752.97999.67
25. Hellberg D, Nilsson S, Mårdh PA. The diagnosis of bacterial vaginosis and vaginal flora changes. Arch Gynecol Obstet. 2001;265(1):11-15. doi:10.1007/s004040000109
26. Hay PE, Taylor-Robinson D, Lamont RF. Diagnosis of bacterial vaginosis in a gynaecology clinic. Br J Obstet Gynaecol. 1992;99(1):63-66. doi:10.1111/j.1471-0528.1992.tb14395.x
27. University of Pittsburgh. A validation study of genzyme diagnostics OSOM Trichomonas rapid test and BVBlue test [NCT00682851]. ClinicalTrials.gov. Updated 2016. Accessed January 28, 2020. https://clinicaltrials.gov/ct2/show/NCT00682851
28. Hilmarsdóttir I, Hauksdóttir GS, Jóhannesdóttir JD, Daníelsdóttir T, Thorsteinsdóttir H, Olafsson JH. Evaluation of a rapid Gram stain interpretation method for diagnosis of bacterial vaginosis. J Clin Microbiol. 2006;44(3):1139-1140. doi:10.1128/JCM.44.3.1139-1140.2006
29. Landers DV, Wiesenfeld HC, Heine RP, Krohn MA, Hillier SL. Predictive value of the clinical diagnosis of lower genital tract infection in women. Am J Obstet Gynecol. 2004;190(4):1004-1010. doi:10.1016/j.ajog.2004.02.015
30. Myziuk L, Romanowski B, Johnson SC. BVBlue test for diagnosis of bacterial vaginosis. J Clin Microbiol. 2003;41(5):1925-1928. doi:10.1128/JCM.41.5.1925-1928.2003
31. Platz-Christensen JJ, Larsson PG, Sundström E, Wiqvist N. Detection of bacterial vaginosis in wet mount, Papanicolaou stained vaginal smears and in gram stained smears. Acta Obstet Gynecol Scand. 1995;74(1):67-70. doi:10.3109/00016349509009947
32. Sha BE, Gawel SH, Hershow RC, et al. Analysis of standard methods for diagnosing vaginitis: HIV infection does not complicate the diagnosis of vaginitis. J Low Genit Tract Dis. 2007;11(4):240-250. doi:10.1097/LGT.0b013e318033dfed
33. Schwebke JR, Hillier SL, Sobel JD, McGregor JA, Sweet RL. Validity of the vaginal gram stain for the diagnosis of bacterial vaginosis. Obstet Gynecol. 1996;88(4, pt 1):573-576. doi:10.1016/0029-7844(96)00233-5
34. Schwebke JR, Gaydos CA, Nyirjesy P, Paradis S, Kodsi S, Cooper CK. Diagnostic performance of a molecular test versus clinician assessment of vaginitis. J Clin Microbiol. 2018;56(6):e00252-e18. doi:10.1128/JCM.00252-18
35. Singh RH, Zenilman JM, Brown KM, Madden T, Gaydos C, Ghanem KG. The role of physical examination in diagnosing common causes of vaginitis: a prospective study. Sex Transm Infect. 2013;89(3):185-190. doi:10.1136/sextrans-2012-050550
36. Cartwright CP, Lembke BD, Ramachandran K, et al. Comparison of nucleic acid amplification assays with BD affirm VPIII for diagnosis of vaginitis in symptomatic women. J Clin Microbiol. 2013;51(11):3694-3699. doi:10.1128/JCM.01537-13
37. Byun SW, Park YJ, Hur SY. Affirm VPIII microbial identification test can be used to detect Gardnerella vaginalis, Candida albicans and Trichomonas vaginalis microbial infections in Korean women. J Obstet Gynaecol Res. 2016;42(4):422-426. doi:10. 1111/jog.12913
38. Rouse AG, Gil KM, Davis K. Diagnosis of bacterial vaginosis in the pregnant patient in an acute care setting. Arch Gynecol Obstet. 2009;279(4):545-549. doi:10.1007/s00404-008-0766-5
39. Rebbapragada A, Howe K, Wachihi C, et al. Bacterial vaginosis in HIV-infected women induces reversible alterations in the cervical immune environment. J Acquir Immune Defic Syndr. 2008;49(5):520-522. doi:10.1097/QAI.0b013e318189a7ca
40. Mastrobattista JM, Bishop KD, Newton ER. Wet smear compared with gram stain diagnosis of bacterial vaginosis in asymptomatic pregnant women. Obstet Gynecol. 2000;96(4):504-506.
41. Witt A, Petricevic L, Kaufmann U, Gregor H, Kiss H. DNA hybridization test: rapid diagnostic tool for excluding bacterial vaginosis in pregnant women with symptoms suggestive of infection. J Clin Microbiol. 2002;40(8):3057-3059. doi:10.1128/JCM.40.8.3057-3059.2002
42. Sonnex C. The amine test: a simple, rapid, inexpensive method for diagnosing bacterial vaginosis. Br J Obstet Gynaecol. 1995;102(2):160-161. doi:10.1111/j.1471-0528.1995.tb09071.x
43. Schmidt H, Hansen JG. A wet smear criterion for bacterial vaginosis. Scand J Prim Health Care. 1994;12(4):233-238. doi:10.3109/02813439409029246
44. Briselden AM, Hillier SL. Evaluation of affirm VP Microbial Identification Test for Gardnerella vaginalis and Trichomonas vaginalis. J Clin Microbiol. 1994;32(1):148-152. doi:10.1128/JCM.32.1.148-152.1994
45. Carey JC, Klebanoff MA, Hauth JC, et al; National Institute of Child Health and Human Development Network of Maternal-Fetal Medicine Units. Metronidazole to prevent preterm delivery in pregnant women with asymptomatic bacterial vaginosis. N Engl J Med. 2000;342(8):534-540. doi:10.1056/NEJM200002243420802
46. Guaschino S, Ricci E, Franchi M, et al. Treatment of asymptomatic bacterial vaginosis to prevent pre-term delivery: a randomised trial. Eur J Obstet Gynecol Reprod Biol. 2003;110(2):149-152. doi:10.1016/S0301-2115(03)00107-6
47. Hauth JC, Goldenberg RL, Andrews WW, DuBard MB, Copper RL. Reduced incidence of preterm delivery with metronidazole and erythromycin in women with bacterial vaginosis. N Engl J Med. 1995;333(26):1732-1736. doi:10.1056/NEJM199512283332603
48. Kekki M, Kurki T, Pelkonen J, Kurkinen-Räty M, Cacciatore B, Paavonen J. Vaginal clindamycin in preventing preterm birth and peripartal infections in asymptomatic women with bacterial vaginosis: a randomized, controlled trial. Obstet Gynecol. 2001;97(5, pt 1):643-648.
49. Kiss H, Petricevic L, Husslein P. Prospective randomised controlled trial of an infection screening programme to reduce the rate of preterm delivery. BMJ. 2004;329(7462):371. doi:10.1136/bmj.38169.519653.EB
50. Lamont RF, Duncan SL, Mandal D, Bassett P. Intravaginal clindamycin to reduce preterm birth in women with abnormal genital tract flora. Obstet Gynecol. 2003;101(3):516-522.
51. Larsson PG, Fåhraeus L, Carlsson B, Jakobsson T, Forsum U; Premature Study Group of the Southeast Health Care Region of Sweden. Late miscarriage and preterm birth after treatment with clindamycin: a randomised consent design study according to Zelen. BJOG. 2006;113(6):629-637. doi:10.1111/j.1471-0528.2006.00946.x
52. McDonald HM, O’Loughlin JA, Vigneswaran R, et al. Impact of metronidazole therapy on preterm birth in women with bacterial vaginosis flora (Gardnerella vaginalis): a randomised, placebo controlled trial. Br J Obstet Gynaecol. 1997;104(12):1391-1397. doi:10.1111/j.1471-0528.1997.tb11009.x
53. McGregor JA, French JI, Jones W, et al. Bacterial vaginosis is associated with prematurity and vaginal fluid mucinase and sialidase: results of a controlled trial of topical clindamycin cream. Am J Obstet Gynecol. 1994;170(4):1048-1059. doi:10.1016/S0002-9378(94)70098-2
54. MoralesWJ, Schorr S, Albritton J. Effect of metronidazole in patients with preterm birth in preceding pregnancy and bacterial vaginosis: a placebo-controlled, double-blind study. Am J Obstet Gynecol. 1994;171(2):345-347. doi:10.1016/S0002-9378(94)70033-8
55. Subtil D, Brabant G, Tilloy E, et al. Early clindamycin for bacterial vaginosis in pregnancy (PREMEVA): a multicentre, double-blind, randomised controlled trial. Lancet. 2018;392(10160):2171-2179. doi:10.1016/S0140-6736(18)31617-9
56. Ugwumadu A, Manyonda I, Reid F, Hay P. Effect of early oral clindamycin on late miscarriage and preterm delivery in asymptomatic women with abnormal vaginal flora and bacterial vaginosis: a randomised controlled trial. Lancet. 2003;361(9362):983-988. doi:10.1016/S0140-6736(03)12823-1
57. Vermeulen GM, Bruinse HW. Prophylactic administration of clindamycin 2% vaginal cream to reduce the incidence of spontaneous preterm birth in women with an increased recurrence risk: a randomised placebo-controlled double-blind trial. Br J Obstet Gynaecol. 1999;106(7):652-657. doi:10.1111/j.1471-0528.1999.tb08363.x
58. Burtin P, Taddio A, Ariburnu O, Einarson TR, Koren G. Safety ofmetronidazole in pregnancy: a meta-analysis. Am J Obstet Gynecol. 1995;172(2, pt 1):525-529. doi:10.1016/0002-9378(95)90567-7
59. Caro-Patón T, Carvajal A, Martin de Diego I, Martin-Arias LH, Alvarez Requejo A, Rodríguez Pinilla E. Is metronidazole teratogenic? a meta-analysis. Br J Clin Pharmacol. 1997;44(2):179-182. doi:10.1046/j.1365-2125.1997.00660.x
60. Czeizel AE, Rockenbauer M. A population based case-control teratologic study of oral metronidazole treatment during pregnancy. Br J Obstet Gynaecol. 1998;105(3):322-327. doi:10.1111/j.1471-0528.1998.tb10094.x
61. Diav-Citrin O, Shechtman S, Gotteiner T, Arnon J, Ornoy A. Pregnancy outcome after gestational exposure to metronidazole: a prospective controlled cohort study. Teratology. 2001;63(5):186-192. doi:10.1002/tera.1033
62. Sørensen HT, Larsen H, Jensen ES, et al. Safety of metronidazole during pregnancy: a cohort study of risk of congenital abnormalities, preterm delivery and low birth weight in 124 women. J Antimicrob Chemother. 1999;44(6):854-856. doi:10.1093/jac/44.6.854
63. Thapa PB, Whitlock JA, Brockman Worrell KG, et al. Prenatal exposure to metronidazole and risk of childhood cancer: a retrospective cohort study of children younger than 5 years. Cancer. 1998;83(7):1461-1468. doi:10.1002/(SICI)1097-0142(19981001)83:7<1461::AID-CNCR25>3.0.CO;2-1
64. Nugent RP, Krohn MA, Hillier SL. Reliability of diagnosing bacterial vaginosis is improved by a standardized method of Gram stain interpretation. J Clin Microbiol. 1991;29(2):297-301. doi:10.1128/JCM.29.2.297-301.1991
65. Ison CA, Hay PE. Validation of a simplified grading of Gram-stained vaginal smears for use in genitourinary medicine clinics. Sex Transm Infect. 2002;78(6):413-415. doi:10.1136/sti.78.6.413
66. Andrews WW, Hauth JC, Cliver SP, Conner MG, Goldenberg RL, Goepfert AR. Association of asymptomatic bacterial vaginosis with endometrial microbial colonization and plasma cell endometritis in nonpregnant women. Am J Obstet Gynecol. 2006;195(6):1611-1616. doi:10.1016/j.ajog.2006.04.010
67. Lowe NK, Neal JL, Ryan-Wenger NA. Accuracy of the clinical diagnosis of vaginitis compared with a DNA probe laboratory standard. Obstet Gynecol. 2009;113(1):89-95. doi:10.1097/AOG.0b013e3181909f63
68. Kurkinen-RätyM, Vuopala S, Koskela M, et al. A randomised controlled trial of vaginal clindamycin for early pregnancy bacterial vaginosis. BJOG. 2000;107(11):1427-1432. doi:10.1111/j.1471-0528.2000.tb11660.x
69. Gaydos CA, Beqaj S, Schwebke JR, et al. Clinical validation of a test for the diagnosis of vaginitis. Obstet Gynecol. 2017;130(1):181-189. doi:10.1097/AOG.000000000000209
Evidence reviews for the US Preventive Services Task Force (USPSTF) use an analytic framework to visually display key questions addressed by the review to allow the USPSTF to evaluate the effectiveness and safety of a preventive service. The questions are depicted by linkages that relate interventions to outcomes. A dashed line indicates an outcome that precedes subsequent outcomes. Refer to the USPSTF Procedure Manual for further details.10
|Test||No. of studies/
No. of participants
|BD Affirm, pooled||5/2936||0.87 (0.80-0.92)||0.81 (0.73-0.88)||4.6 (3.1-6.8)||0.16 (0.11-0.26)|
|BV Blue, range||3/864||0.61-0.92||0.86-0.99||6.3-41.3||0.09-0.14|
|Complete Amsel clinical criteriaa, pooled||15/7171||0.76 (0.63-0.85)||0.95 (0.89-0.98)||14.1 (6.8-29.2)||0.26 (0.17-0.39)|
|Modified Amsel clinical criteriab, pooled||5/2674||0.67 (0.54-0.78)||0.96 (0.93-0.98)||17.3 (10.4-28.8)||0.34 (0.24-0.48)|
Abbreviation: LR, likelihood ratio.
a Clinical diagnosis is based on the presence of at least 3 of 4 criteria: vaginal pH greater than 4.5, at least 20% of epithelial cells are clue cells on microscopy, amine odor when potassium hydroxide is added to vaginal secretions, thin homogenous discharge.
b Similar to complete Amsel criteria except the requirement for thin homogenous discharge is waived. Studies vary with respect to whether all 3 remaining criteria were required or whether 2 of 3 remaining criteria were required
|Source||Country||Study Quality||Interventions||No. (%)||Outcomes reported|
|With bacterial symptoms||Nulliparous||Nonwhite||With prior PTD|
|Carey et al,45 2000
Andrews et al,66 2006
|US||Good||Placebo (987 randomized)||0||407 (41.2)||841 (85.2)||110 (11.1||All-cause and spontaneous PTD <37, 35, and 32 wk
Birth weight <2500 and 1500 g
Subgroup findings for women with prior PTD; treatment for chlamydia and bacterial vaginosis clearance
|Oral metronidazole (1000 mg dose 4 times on days 0, 2,14, and 16) (966 randomized)||0||436 (45.1)||822 (85.1)||103 (10.7)|
|Guaschino et al,46 2003||Italy||Fair||No treatment (57 randomized)||0||35 (61.4)||NR||3 (5.3)||All-cause PTD <37 wk
Birth weight <2500 g
Preterm or term PROM
|Intravaginal clindamycin (2% cream once daily for 7 d) (55 randomized)||0||39 (70.9)||NR||5 (9.1)|
|Hauth et al,47 1995||US||
|Placebo (87)||30 (16)a||150 (79)a||56 (65.1)b||All-cause PTD <37 wk
Subgroup findings among women with prior PTD
|Oral metronidazole (750 mg daily for 7 d) and erythromycin (999 mg daily for 14 d) (176 randomized)||NR||84 (19)a||309 (71)a||121 (70.3)b|
|Kekki et al,48 2001
Kurkinen-Räty et al,68, 2000
|Finland||Good||Placebo (188 randomized)||0||Mean parity 1.9||NR||0||Spontaneous PTD <37 wk
Maternal peripartum infection
Subgroup findings among participants with clearance of bacterial vaginosis and participants with intermediate flora
|Intravaginal clindamycin (2% cream once daily for 7 d) (187 randomized)||0||Mean parity 1.7||NR||0|
|Kiss et al,49 2004||Austria||Fair||No treatment (179 randomized)c||0||NR (47.8)||NR (2)||Between 33 and 36 wk: 45 (2.1)
Between 23 and 32 wk: 24 (1.1)
|Spontaneous PTD <37 wk|
|Intravaginal clindamycin (2% cream once daily for 6 d) and treatment with oral clindamycin (300 mg twice a day) if still positive at 24 to 27 wk gestation (177 randomized)c||0||NR (47.9)|
|Lamont et al,50 2003||UK||Good||Placebo (201 randomized)||0||112 (56)||63 (31)||11 (8)||Spontaneous PTD <37 wk
Birth weight <2500 and 1500 g
|Intravaginal clindamycin (2% cream once daily for 3 d) (208 randomized)||0||111 (53)||58 (28)||10 (7)|
|Larsson et al,51 2006||Sweden||Fair||No treatment (411 randomized)||0||187 (45.5)||NR||Among parous women: 13 (6.0)||All-cause PTD <37 wk
Spontaneous PTD <37 and <32 completed
|Intravaginal clindamycin (2% cream once daily for 7 d) (408 randomized)||0||186 (45.5)||NR||Among parous women: 20 (9.2)|
|McDonald et al,52 1997||Australia||Good||Placebo (440 randomized||0||144 (32.7)||53 (12.3)||24 (5.5)||All-cause and spontaneous PTD <37 wk
Subgroup findings for women with prior PTD
|Oral metronidazole (800 mg daily for 2 d) repeated at 28 weeks for women with persistence (439 randomized)||0||139 (31.7)||47 (10.8)||22 (5.0)|
|McGregor et al,53 1994||US||Good||Placebo (69 analyzed)||0||Mean parity, 1.0 (range 0-6)||87 (61.2)||15 (10.9)||All-cause PTD <37 wk
Birth weight <2500 g
|Intravaginal clindamycin (2% cream once daily for 7 d) (60 analyzed)||0|
|Morales et al,54 1994||US||Fair||Placebo (36 analyzed)||NR||2.2 (1.1)||18 (50)||80 (100)||Spontaneous PTD <37 and 34 wk
Birth weight <2500
|Oral metronidazole (750 mg daily for 7 d) (44 analyzed)||NR||2.4 (1.2)||20 (45)|
|Subtil et al,55 2018||France||Good||Placebo (956)||NR||NR||NR||NR||All-cause and spontaneous PTD <37 wk
Spontaneous PTD <32 wk
|Oral clindamycin (600 mg daily for 4 d or 3 courses [600 mg daily] for 4 d, each 1 mo apart) (1904 randomized)|
|Ugwumadu et al,56 2003||UK||Good||Placebo (245)d||NR||0.8 (1.0)||93 (39)||22 (9)||Spontaneous PTD <37 wk
Subgroup findings among participants with intermediate flora
|Oral clindamycin (600 mg daily for 5 d) (249 randomized)d||NR||0.8 (1.1)||86 (36)||24 (10)|
|Vermeulen and Bruinse,57 1999||The Netherlands||Good||Placebo (11)e||NR||1.4 (0.9)||NR||11 (100)||All-cause PTD <34 wk
|Intravaginal clindamycin (2% cream once daily for 7 d) at 26 wk and again at 32 wk (11 randomized)e||NR||1.6 (0.9)||NR||11 (100)|
Abbreviations: NR, not reported; PROM, premature rupture of membranes; PTD, preterm delivery.
a For total study population including those with and without bacterial vaginosis.
b For subgroup with bacterial vaginosis.
c This study randomized a total of 4429 participants to vaginal smear screening, but only a subset of participants tested positive for bacterial vaginosis and received treatment; only data for the bacterial vaginosis–positive subset of the study population was abstracted.
d Represents the full randomized population; findings reported only for the subgroup of women with bacterial vaginosis (203 in placebo group; 207 in intervention group).
e Represents the number of women with bacterial vaginosis who were randomized to placebo and active treatment; total number randomized was 168 (85 placebo; 83 active treatment).
Mixed-methods test of moderators for all-cause vs spontaneous preterm delivery: P = .054 for absolute risk difference and P = .002 for risk ratio. OMindicates oralmetronidazole; PTD, preterm delivery; VC, intravaginal clindamycin.
a Includes spontaneous late abortion (16 weeks).
Mixed-methods test of moderators for all-cause vs spontaneous preterm delivery at less than 37 weeks’ gestation (P = .14). For Hauth et al,47 data from the subgroup of participants with bacterial vaginosis and history of prior preterm delivery were used. For Carey et al,45 data from the subgroup of participants with a history of prior preterm delivery were used. For McDonald et al,52 data from the subgroup of participants with bacterial vaginosis and history of prior preterm delivery were used. For Vermeulen and Bruinse,57 data from the subgroup of participants with bacterial vaginosis were used. OC indicates oral clindamycin; OM, oral metronidazole; VC, intravaginal clindamycin.
No. of studies (No. of participants)
|Summary of findings||Consistency/precision||Other limitations||EPC assessment of strength of evidence||Applicability|
|KQ1: Benefits of screening|
|KQ2: Diagnostic test accuracy|
|BD Affirm VPIII
5 Cross-sectional studies36,37,41,44,67 (2936)
|Pooled sensitivity, 0.87 (95% CI, 0.80 to 0.92)
Pooled specificity, 0.81 (95% CI, 0.73 to 0.88)
Pooled LR+, 4.6 (95% CI, 3.1 to 6.8)
Pooled LR–, 0.16 (95% CI, 0.11 to 0.26)
|Inconsistenta; preciseb||4 of 5 studies with fair methodological quality (unclear enrollment procedures, unclear masking of test results, spectrum bias)||Low for adequate accuracy||Only 1 study conducted in pregnant women; all studies conducted in symptomatic women|
1 Cross-sectional study34,69 (1338)
|Sensitivity, 0.93 (95% CI, 0.91 to 0.94)
Specificity, 0.92 (95% CI, 0.90 to 0.94)
LR+, 10.9 (95% CI, 8.3 to 14.5)
LR–, 0.08 (95% CI, 0.06 to 0.10)
|Unknown consistency; precisec||Excluded participants with intermediate flora from analysis||Lowd for adequate accuracy||Symptomatic women|
3 Cross-sectional studies20,27,30 (864)
|Sensitivity range across studies, 0.61 to 0.92
Specificity range across studies, 0.86 to 0.9
|Inconsistente (more inconsistent for sensitivity than specificity); precisef (more precise for specificity than sensitivity)||All studies with fair methodological quality (unclear enrollment, unclear masking of results, spectrum bias)||Low for adequate accuracy||Symptomatic, nonpregnant women|
|Complete Amsel criteria
15 Cross-sectional studies20-24,26-35 (7171)
|Based on 14 of the 15 studies:
Pooled sensitivity, 0.76 (95% CI, 0.63 to 0.85)
Pooled specificity, 0.95 (95% CI, 0.89 to 0.98)
Pooled LR+, 14.1 (95% CI, 6.8 to 29.2)
Pooled LR–, 0.26 (95% CI, 0.17 to 0.39)
|Inconsistentg; preciseh (more precise for specificity than sensitivity)||12 of 15 studies with fair methodological quality (unclear enrollment, unclear masking of test results, spectrum bias), heterogeneity in application of clinical criteria and unit of analysis (patients vs visits)||Low for adequate accuracy||Only 1 study conducted exclusively in pregnant women; most studies conducted in symptomatic women|
|Modified Amsel criteria
5 Cross-sectional studies23,33-35,40 (2674)
|Based on 4 of the 5 studies:
Pooled sensitivity, 0.67 (95% CI, 0.54 to 0.78)
Pooled specificity, 0.96 (95% CI, 0.93 to 0.98)
Pooled LR+, 17.3 (95% CI, 10.4 to 28.8)
Pooled LR–, 0.34 (95% CI, 0.24 to 0.48)
|Inconsistenti (more inconsistent for sensitivity than specificity); precisej (more precise for specificity than sensitivity)||4 of 5 studies with fair methodological quality (unclear enrollment, unclear masking of test results, spectrum bias)||Low for adequate accuracy||2 studies conducted exclusively in asymptomatic, pregnant women|
|KQ3: Harms of screening|
|KQ4: Benefits of treatment|
|6 RCTs45,46,51-53,55 (6307)||
All-cause preterm delivery <37 wk in general obstetric population:
Pooled ARD, 0.20% (95% CI, −1.13% to 1.53%)
Pooled RR, 1.02 (95% CI, 0.86 to 1.20)
|Consistent; imprecisek||All but 1 study of good methodological quality; no reporting bias detected||Moderate for no benefit of treatment||Applies to treatment of asymptomatic patients with oral or vaginal clindamycin or oral metronidazole; history of prior PTD in this population ranged from 0%-10.9%|
|8 RCTs45,48-52,55,56 (7571)||Spontaneous preterm delivery <37 wk in general obstetric population:
Pooled ARD, −1.44% (95% CI, −3.31% to 0.43%)
Pooled RR, 0.78 (95% CI, 0.56 to 1.07)
|Inconsistentl; imprecisem||All but 2 studies of good methodological quality; no reporting bias detected||Low for no benefit of treatment||Same as previous row|
|3 RCTs45,51,55 (5564)||Preterm delivery <32 wk in general obstetric population:
Pooled ARD, −0.30% (95% CI, −0.97% to 0.38%)
Pooled RR, 0.87 (95% CI, 0.54 to 1.42)
|Consistent; precisen||1 study of fair methodological quality; outcome was spontaneous PTD in 2 studies and all-cause PTD in the other study; no reporting bias detected||High for no benefit of treatment||Same as previous row|
|5 RCTs45,46,50,53,55 (5377)||Birth weight <2500 g in general obstetric population:
Pooled ARD, 0.39% (95% CI, −1.74% to 2.53%)
Pooled RR, 1.03 (95% CI, 0.83 to 1.2
|Consistent; impreciseo||All studies of good methodological quality; no reporting bias detected||Moderate for no benefit of treatment||Same as previous row|
|3 RCTs45,50,55 (5149)||
Birth weight <1500 g in general obstetric population:
Pooled ARD, 0.06% (95% CI, −0.99% to 1.12%)
Pooled RR, 1.05 (95% CI, 0.50 to 2.18
|Consistent; precisep||All studies of good methodological quality; no reporting bias detected||High for no benefit of treatment||Same as previous row|
|4 RCTs46,52,53,55 (3568)||Preterm PROM or PROM in general obstetric population:
Pooled ARD, 0.10% (95% CI, −1.32% to 1.52%)
Pooled RR, 1.11 (95% CI, 0.72 to 1.72)
|Consistent; preciseq||All studies of good methodological quality; no reporting bias detected; 1 study reported PROM while others reported preterm PROM||Moderate for no benefit of treatment||Same as previous row|
|4 RCTs45,47,52,54 (451)||Preterm delivery <37 wk (all-cause or spontaneous) in women with prior preterm delivery:
ARDs range from −29.4% to 7.5%
RRs range from 0.17 to 1.33
Results statistically significant in 3 of the 4 studies favoring treatment
|Inconsistentr; imprecises||2 studies of fair methodological quality; findings from 3 studies were from subgroup analyses, and it is not clear that they were preplanned
Unable to definitively identify source(s) of inconsistency
|Insufficient||Applies to treatment of asymptomatic patients with a prior PTD with oral metronidazole|
|2 RCTs54, 123 (102)||Preterm delivery <34 wk in women with prior preterm delivery:
ARD 0% in 1 study and −6.57% (95% CI, −18.5% to 5.4%) in other study
|Consistent; impreciset||Both studies with fair study quality; results from 1 study were from subgroup analysis||Insufficient||Applies to treatment of asymptomatic patients with a prior PTD with vaginal clindamycin or oral metronidazole|
|KQ5: Harms of treatment (maternal harms)|
4 RCTs48,49,51,57 (1718)
|Heterogenous outcomes reported||Consistent; impreciseu||Although RCTs were mostly of good methodological quality, adverse event outcome measurement and reporting were not well described and studies were not powered for adverse events||Moderate for no difference in serious AEs or minor harms (intravaginal clindamycin)||Applies to treatment of asymptomatic pregnant women with bacterial vaginosis|
2 RCTs55,56 (3345)
|Serious AEs not observed in either group in 1 study;55 not reported in the other study56
Higher incidence of adverse effects with active treatment in 1 study (ARD, 1.79% [95% CI, 0.75% to 2.84%])55
Higher incidence of stopping medication with active treatment in both studies, but findings were statistically significant in only 1 study (ARD, 3.33% [95% CI, 0.38% to 6.27%]55; ARD, 3.65% [95% CI, −0.27% to 7.56%]56
|Consistent; imprecisev||Moderate for no difference in serious AEs but more minor harms (oral clindamycin and metronidazole)|
2 RCTs45,52 (2776)
|Higher incidence of adverse effects or AEs with active treatment in both studies, but findings were statistically significant in only 1 study (ARD, 12.51% [95% CI, 9.33% to 15.69%]45; ARD, 2.56% [95% CI, −0.36% to 5.47%]52)||Consistent; imprecisew|
|KQ5: Harms of treatment (harms to children from in utero exposure to medication)|
|3 Observational studies60-62 (62,271)
2 Meta-analyses of observational studies58,63 (>199,541)
|Congenital malformations among children exposed to metronidazole in utero:
ORs and RR, estimates from individual studies range from 0.44 to 2.24; CIs range from 0.11 to 4.23
Congenital malformations among children exposed to metronidazole in utero:
Pooled OR, 0.96 (95% CI, 0.75 to 1.22)58
Pooled OR, 1.08 (95% CI, 0.90 to 1.29)59
|Consistent; imprecisex||Studies of poor to fair methodological quality, did not address confounding, variation in outcome definition, potential for recall bias in case-control study
Older analyses that did not use current methods for conducting and reporting analyses, included studies were not assessed for risk of bias
|Insufficient||Applies to metronidazole exposure across a range of indications (not specific to women with bacterial vaginosis)|
|1 Observational study63
(328,846 participants with 1,172,696 person-years)
|Cancer incidence before age 5 y among children exposed to metronidazole: adjusted RR, 0.81 (95% CI, 0.41 to 1.59)||Consistency unknown; imprecisey||Fair methodologic quality; baseline imbalances between groups and potential for residual confounding||Insufficient||Applies to metronidazole exposure across a range of indications; not specific to women with bacterial vaginosis|
Abbreviations: AE, adverse event; ARD, absolute risk difference; EPC, Evidence-based Practice Center; KQ, key question; LR, likelihood ratio; OR, odds ratio; PROM, premature rupture of membranes; PTD, preterm delivery; RCT, randomized clinical trial; RR, relative risk.
a The 95% prediction region covers nearly one-third of the receiver operating characteristic (ROC) space (eFigure 1 in the Supplement), suggesting at least moderate inconsistency in estimates across studies not easily be explained by differences in study populations or settings.
b The 95% confidence region is relatively small and the CI around the area under the curve fairly narrow, suggesting precise estimates (eFigure 1 in the Supplement).
c Based on the upper and lower CIs for sensitivity and specificity, the LR+ would range from 10.67-11.11 and the LR– from 0.078-0.82, resulting in minimal variation in posttest probabilities, suggesting precise estimates.
d Overall strength of evidence downgraded for study limitations and a single-study body of evidence with unknown consistency.
e Range of estimates across the 3 studies inconsistent for sensitivity but reasonably consistent for specificity. In particular, 1 study had markedly lower sensitivity (0.61) than the others (0.88 and 0.917). This study was only reported in ClinicalTrials.gov, and very little information about the study setting and population was available to understand why this result was inconsistent with the other 2 studies.
f The LR+ and LR– at the upper and lower limits of the sensitivity and specificity CIs for each study are reasonably similar and result in only small differences in posttest probabilities.
g The 95% prediction region covers more than one-third of the ROC space (eFigure 2 in the Supplement), suggesting at least moderate inconsistency in estimates of sensitivity and specificity not easily be explained by differences in study populations or settings.
h Confidence region is quite small; thus, estimate was judged as precise, although more precise for specificity than for sensitivity (eFigure 2 in the Supplement).
i Although the prediction region covers only one-fifth of the summary ROC space, the shape of the region suggests that future studies could lie in the space of relative poor sensitivity and high specificity or equally likely the space of relatively poor specificity and high sensitivity; visual inspection of the plot also suggests inconsistency (eFigure 3 in the Supplement).
j The 95% confidence region suggests reasonable precision for estimates of specificity but some imprecision in estimates of sensitivity (eFigure 3 in the Supplement).
k Optimal information size (OIS) criteria not met, sample size of 7116 required to detect a 20% RR reduction based on 9% control group risk, α = .05, power = 0.80, 2-tailed test. Further, the width of the CI around the RR could not exclude a clinically meaningful benefit or harm; despite the narrow range of the CI around the ARD, the population burden from even a small increase or decrease in PTD could be clinically meaningful.
l Although CIs are mostly overlapping, there is some inconsistency in both the direction and magnitude of effect, as 2 studies observed a statistically significant effect of –5.80% and –9.96%, vs the other studies that are much closer to a null effect (ARDs ranging from –2.25% to 1.09%); I2 = 61.9% for the ARD.
m OIS criteria not met; sample size of 9920 required to detect a 20% RR reduction based on 7% control group risk (average risk across studies), α = .05, power = 0.80, 2-tailed test. Further, CIs for both the ARD and RR span a range that could be considered a clinically meaningful benefit or no difference.
n Low baseline risk (<5%) and sample sizes greater than 2000 in each group; thus, OIS is met. Because of infrequent events, more emphasis was placed on ARD than RR when evaluating precision.
o OIS criteria not met; sample size of 7116 required to detect a 20% RR reduction based on a 9% control group risk (average across these studies), α = .05, power = 0.8, 2-tailed test.
p Low baseline risk (<5%) and sample sizes greater than 2000 in each group; thus, OIS is met. Because of infrequent events, more emphasis placed on ARD than RR when evaluating precision.
q OIS criteria not met; sample size of 24,798 required to detect a 20% RR reduction based on a 3% control group risk (average across these studies), α = .05, power = 0.8, 2-tailed test.
r Three studies have statistically significant moderate treatment effect sizes; while 1 study shows an increase in preterm delivery from treatment but is not statistically significant (source of inconsistency unexplained).
s OIS criteria not met; sample size of 1248 required to detect a 20% RR reduction based on 38% control group risk, α = .05, power = 0.80, 2-tailed test.
t OIS criteria not met; sample size of 1874 required to detect a 20% RR reduction based on 29% control group risk, α = .05, power = 0.80, 2-tailed test.
u OIS criteria not met; infrequent events reported.
v OIS criteria not met; sample size of 45,236 required to detect a 20% relative risk increase based on a 2% control group risk, α = .05, power = 0.8, 2-tailed test.
w OIS study included 155,504 participants;58 the other study included 199,451 participants;59 3 studies overlapped between the 2 analyses.
x OIS criteria met. Sample size of 17,128 required to detect an elevated RR of 1.20 with α = .05, power = 0.80, 2-sided test. However, the null effect cannot be excluded, and CIs from both the individual studies and the meta-analyses span a clinically meaningful range of benefit and harms; thus, the estimate was considered imprecise.
y OIS criteria not met; sample size of more than 6 million required to detect a 20% RR increase based on a 0.0142% control group risk, α = .05, power = 0.80, 2-tailed test.