J. L. Gibson

Objective: To evaluate prospectively cervical length measurements and fetal fibronectin detection as predictors of spontaneous preterm delivery in an unselected population of twin pregnancies.Methods: Transvaginal ultrasound assessments of cervical length were performed serially at 18, 24, 28 and 32 weeks' gestation. Receiver–operating characteristic curves were generated at each time point to determine the optimal cut‐off for cervical length and rate of change in cervical length in the prediction of preterm delivery (< 35 weeks). A bedside assay for the detection of fetal fibronectin in the maternal vaginal secretions was performed prior to cervical length measurements from 24 weeks. Likelihood ratios (LRs) were used to assess the performance of each test.Results: The spontaneous preterm delivery rate was 16.5% in 91 studied twin pregnancies. A cervical length ≤ 25 mm at 18 weeks (LR+ 9.7, sensitivity 14.3%) and ≤ 22 mm at 24 weeks (LR+ 9.6, sensitivity 28.6%) were the best predictors of preterm delivery. A shortening of cervical length ≥ 2.5 mm per week between 18 and 28 weeks' gestation also predicted preterm delivery (LR+ 10.8, sensitivity 16.7%). There was no relationship between the detection of fetal fibronectin and preterm delivery.Conclusions: This study confirms the value of transvaginal ultrasound assessment of cervical length as a predictor of preterm delivery in twin pregnancies. However, the poor sensitivity of this test makes it unsuitable as a single predictor of preterm delivery. Fetal fibronectin testing does not identify twin pregnancies destined to deliver prematurely. Copyright © 2004 ISUOG. Published by John Wiley & Sons, Ltd.