The diagnostic accuracy of cervico-vaginal fetal fibronectin in predicting preterm delivery: an overview.

Abstract

To determine the accuracy with which cervico-vaginal fetal fibronectin predicts preterm delivery using systematic quantitative overview of the available literature. Online searching of MEDLINE database (1966 to April 1996), scanning of bibliography of known primary and review articles and review of recent journal issues. Study selection, assessment of study quality and data extraction were performed in duplicate under masked conditions. Likelihood ratios were generated in subgroups of symptomatic and asymptomatic pregnant women by pooling data from different studies. An LR of > 10 or < 0.1 indicated conclusive changes in the pretest probability of preterm delivery while an LR of 5-10 or 0.2-0.1 indicated only moderate changes. Seven hundred and twenty-three symptomatic women with threatened preterm labour included in nine studies and 847 asymptomatic women (635 low risk and 212 high risk) included in six studies selected for meta-analyses. Likelihood ratios for positive and negative test results using delivery at < 37 and < 34 weeks of gestation, and within one week of testing as outcome measures. In symptomatic women a positive test predicted delivery < 37 weeks of gestation with a pooled likelihood ratio (LR) of 4.6 (95% CI 3.5-6.1) while a negative test had a pooled LR of 0.5 (95% CI 0.4-0.6). For delivery < 34 weeks of gestation, the pooled LR was 2.6 (95% CI 1.8-3.7) for a positive test and 0.2 (95% CI 0.1-0.5) for a negative test. For delivery within one week of testing, the pooled LR was 5.0 (95% CI 3.8-6.4) for a positive test and 0.2 (95% CI 0.1-0.4) for a negative test. In asymptomatic women at low risk of delivery < 37 weeks of gestation the pooled LR was 3.2 (95% CI 2.2-4.8) for a positive test and 0.8 (95% CI 0.7-0.9) for a negative test. In high risk asymptomatic women using delivery < 37 weeks of gestation as an outcome measure the pooled LR was 2.0 (95% CI 1.5-2.6) for a positive test and 0.4 (95% CI 0.2-0.8) for a negative test. For delivery < 34 weeks of gestation in high risk, asymptomatic women the pooled LR was 2.4 (95% CI 1.8-3.2) for a positive test and 0.6 (95% CI 0.4-0.9) for a negative test. The presence of fetal fibronectin in cervico-vaginal mucus has limited accuracy in predicting preterm delivery as the likelihood ratios for positive and negative test results generated only minimal to moderate changes in the pretest probability of preterm birth.