Effectiveness and Costs of Molecular Screening and Treatment for Bacterial Vaginosis to Prevent Preterm Birth: The AuTop Randomized Clinical Trial

ABSTRACT One of the risk factors for preterm birth (PTB) is bacterial vaginosis (BV), a common, often asymptomatic, vaginal dysbiosis. The earlier BV is diagnosed based on gestational age, the higher the risk of PTB. The effectiveness of a screen-and-treat strategy for BV during pregnancy remains a source of debate. One meta-analyses, including 5 studies and 2346 patients, showed a benefit to screen and treat using clindamycin. Another, with 21 studies and 7847 patients, did not recommend BV screening but observed reductions in preterm delivery by 50% and miscarriages by 80%. In another systematic review, with 48 studies, there was varying accuracy across conventional screening tests for BV and suggested no or inconclusive efficacy in the treatment of asymptomatic BV in the general obstetric population and in those with a history of preterm delivery. Based on these studies, French and international organizations recommend against screening for BV with conventional diagnosis tools in low-risk populations. However, molecular diagnostic tools have been shown to be more accurate in identifying vaginal microbiota than conventional tools, such as clinical diagnosis based on Amsel or Nugent criteria. Molecular tools have been shown to provide an objective, reproducible, quantitative diagnosis of BV, identifying emergent pathogen species, such as Atopobium vaginae (now known as Fannyhessea vaginae). To date, no randomized studies have been conducted to assess the impact of molecular tools on a screen-and-treat intervention to BV. The aim of this study was to assess whether a screen-and-treat intervention using a molecular diagnostic tool is cost-effective in reducing the rate of PTB. The AuTop Trial was a prospective, open-label superiority trial conducted in 19 French maternity hospitals from March 9, 2015, to December 18, 2017. Included were adult women in early pregnancy (<20 weeks of gestation), with no history of PTB or late abortion and no major risk factors for prematurity. Excluded were patients who had extrauterine pregnancy or nonprogressive pregnancy, or were treated with antibiotics a week before inclusion in the study. Women were randomly assigned 1:1 to undergo molecular screening and treatment (intervention group) or receive usual care (control group). The intervention group was asked to self-collect vaginal swabs and return them within 15 days of collection for each month until 28 weeks of gestation. If BV was detected, treatment with azithromycin (1 g repeated after 48 hours) or amoxicillin (2 g/d for 7 days) was provided within 48 hours. A molecular biology-based rapid diagnostic tool designed for point-of-care testing was developed using real-time polymerase chain reaction assays to quantify DNA levels of A. vaginae. The tool's specificity, sensitivity, positive predictive value, and negative predicative value of the tool were 99%, 95%, 95%, and 99%, respectively, compared with other diagnostic techniques. The control group received usual care with no systematic screening of BV. A total of 6671 women were randomly assigned 1:1 to the intervention group (n = 3333) or control group (n = 3338). At inclusion, a total of 3438 were nulliparous pregnancies, with 1671 in the intervention group and 1767 in the control group. In the intention-to-treat analysis, no reduction in the rate of PTB was observed between the intervention and control group (3.8% vs 4.6%, respectively; risk ratio, 0.83; 95% confidence interval [CI], 0.66–1.05; P = 0.12). The average cost of intervention per woman was €203.60 (US $218.00). No significant differences were observed in the total cost in the intervention group versus control group (€3344.30 [US $3580.50] vs €3272.90 [US $3504.10]). In addition, no evidence of superiority was observed for the intervention strategy compared with usual care. However, in the nulliparous subgroup, the number of PTBs was significantly lower in the intervention group (3.6%; 95% CI, 2.9–4.6) than the control group (5.9%, 95% CI, 4.8–7.2), and the cost was nonsignificant. In conclusion, the molecular diagnostic tool for screening and treating BV did not significantly reduce the risk of PTBs. Although no significant benefit was found in the overall population, the screen-and-treat strategy using a molecular tool was more effective than usual care in reducing PTBs among nulliparous pregnant women.