The effect of gestational age and cervical length measurements in the prediction of spontaneous preterm birth in twin pregnancies: an individual patient level meta-analysis.

Abstract

To assess the effect of gestational age (GA) and cervical length (CL) measurements at transvaginal ultrasound (TVUS) in the prediction of preterm birth in twin pregnancy. Individual patient data (IPD) meta-analysis. International multicentre study. Asymptomatic twin pregnancy. MEDLINE and EMBASE searches were performed and IPD obtained from authors of relevant studies. Multinomial logistic regression analysis determined probabilities for birth at ≤28(+0) , 28(+1) to 32(+0) , 32(+1) to 36(+0) , and ≥36(+1) weeks as a function of GA at screening and CL measurements. Predicted probabilities for preterm birth at ≤28(+0) , 28(+1) to 32(+0) , and 32(+1) to 36(+0) . A total of 6188 CL measurements were performed on 4409 twin pregnancies in 12 studies. Both GA at screening and CL had a significant and non-linear effect on GA at birth. The best prediction of birth at ≤28(+0) weeks was provided by screening at ≤18(+0) weeks (P < 0.001), whereas the best prediction of birth between 28(+1) and 36(+0) weeks was provided by screening at ≥24(+0) weeks (P < 0.001). Negative prediction value of 100% for birth at ≤28(+0) weeks is achieved at CL 65 mm and 43 mm at ultrasound GA at ≤18(+0) weeks and at 22(+1) to 24(+0) weeks, respectively. In twin pregnancies, prediction of preterm birth depends on both CL and the GA at screening. When CL is <30 mm, screening at ≤18(+0) weeks is most predictive for birth at ≤28(+0) weeks. Later screening at >22(+0) weeks is most predictive of delivery at 28(+1) to 36(+0) weeks. In twins, we recommend CL screening in twins to commence from ≤18(+0) weeks. An individual patient meta-analysis assessing gestation and CL in the prediction of preterm birth in twins.