Prediction of preterm delivery by fetal fibronectin: a meta-analysis.

OBJECTIVE To assess the cervicovaginal fetal fibronectin test to predict preterm delivery. DATA SOURCES We searched MEDLINE, Current Contents, Index Medicus, and proceedings of meetings for studies published between 1991 and June 1997. METHODS OF STUDY SELECTION Inclusion criteria were prospective cohort study; test performed between 20 and 36 weeks' gestation; fetal fibronectin measured by a previously described assay, with a cutoff level set at 50 ng/mL; test results not disclosed to women or physicians; and fewer than 20% of study participants excluded from the analysis. TABULATION, INTEGRATION, AND RESULTS Twenty-nine relevant studies were stratified according to the prevalence of preterm delivery, the number of tests performed, and delivery before 34, 35, or 37 weeks. Sensitivities, specificities, and likelihood ratios were calculated in each study. The summary estimates of the likelihood ratio for tests yielding positive results or tests yielding negative results along with their 95% confidence intervals (CIs) were computed in each stratum according to a random-effects model. All summary likelihood ratios for a test yielding positive results indicated a significant association with preterm delivery. The strongest association was found between a single test with positive results and delivery before 37 weeks in a low-risk population (likelihood ratio 7.5; 95% CI 4.6, 12.3). This association also was found in high-risk women (likelihood ratio 3.5; 95% CI 2.6, 4.6). In high-risk women, a test yielding negative results was associated with a reduction in risk of preterm delivery (likelihood ratio 0.4; 95% CI 0.3, 0.5). CONCLUSION Fetal fibronectin in cervicovaginal secretions is associated with preterm delivery in both high-risk and low-risk women.