Effectiveness and Accuracy of Tests for Preterm Delivery in Symptomatic Women: A Systematic Review

This study systematically reviewed evidence on the effectiveness and accuracy of predictive tests for preterm delivery among symptomatic women. The study included English-language systematic reviews (SRs) on any predictive test for preterm delivery among symptomatic women and primary studies for placental alpha-microglobulin-1. PubMed, Wiley Cochrane Library, the Centre for Reviews and Dissemination Database, the National Guidelines Clearinghouse, and the TRIP database were searched for SRs, PubMed and PubMed Central via the Wiley Cochrane Library were searched for primary studies. One reviewer performed study selection, with input from a second reviewer when needed. One reviewer appraised study quality and extracted: study characteristics (i.e., country, funding source, study design [primary studies] or synthesis method [SRs], study appraisal method [SRs]), population characteristics, index test(s) and cut-off points used, comparator(s) or reference standard(s), and outcomes. A second reviewed a random 10% sample. The authors synthesized the findings narratively. Of 451 unique records, the review included 22 (17 SRs, five primary studies). For effectiveness, there was evidence for use of transvaginal sonographic cervical length assessment (15-25 mm cut point) in reducing incidence of preterm delivery at <37 weeks (relative risk 0.64; 95% CI 0.44-0.94, one SR of three trials; n = 287) but lack of support for cervicovaginal fetal fibronectin. In terms of accuracy, one high-quality study within a best-evidence SR showed that cervical length measurement was useful to predict delivery within 48 hours (LR+ 6.43, 95% CI 5.17-8.00; LR- 0.03, 95% CI 0.00-0.42; n = 510) and 7 days (LR+ 8.61, 95% CI 6.65-11.14; LR- 0.03, 95% CI 0.00-0.18; n = 510). Accuracy of placental alpha-microglobulin-1 testing was not supported for most end points. In conclusion, some evidence supports the effectiveness of cervical length as a predictor of preterm delivery in symptomatic women. Evidence for most tests is limited in quality and quantity.