Cervical length measurement for the prediction of preterm birth in symptomatic women with a twin pregnancy: a systematic review and meta-analysis.

Objective. The aim of this study was to assess whether cervical length measurement (CL) could predict preterm birth (PTB) in symptomatic women with a twin pregnancy. Methods. We searched MEDLINE and EMBASE to identify studies investigating the accuracy of CL measurement in predicting PTB in symptomatic women with a twin pregnancy. We extracted data to construct two-by-two tables and used bivariate meta-analysis to generate point estimates of sensitivity and specificity. Results. Five studies (N = 226) were included. Variation in definition of PTB and cut-off points for CL was strong. One study investigated delivery within seven days, demonstrating a sensitivity of 1.0 (95% CI: 0.83-1.0) and a specificity of 0.31 (95% CI 0.2-0.43) for a CL cutoff at 25 mm. Three studies reported on predicting PTB < 37 weeks at a CL cutoff of 30 mm, with sROC point estimates of 0.76 (95% CI: 0.66 to 0.84) and 0.37 (95% CI: 0.21 to 0.56) for sensitivity and specificity, respectively. For preterm birth <34 weeks, no pooled estimates could be estimated since only 2 studies with large heterogeneity were identified. Conclusions. There is limited evidence on the accuracy of cervical length measurement testing the prediction of preterm birth in symptomatic women with a twin pregnancy, especially on the most important outcome, that is, delivery within 7 days.