Objectives: To assess the role of cervical sonography and to compare various sonographic cervical parameters in their ability to predict spontaneous preterm birth in high-risk singleton gestations.
Design: A prospective cohort of 469 high-risk gestations were longitudinally evaluated between 15 and 24 weeks' gestation on 1265 occasions with transvaginal cervical sonography and transfundal pressure. The cervical parameters obtained were funnel width and length, cervical length, percent funneling and cervical index. The information obtained was used for patient management. Restriction of physical activities was initiated at cervical lengths of < or = 2.5 cm with cerclage as an option for cervical lengths of < or = 2.0 cm.
Results: Receiver operating characteristic curve analyses showed that a cervical length of < or = 2.5 cm between 15 and 24 weeks' gestation was equal to the other sonographic cervical parameters in its ability to predict spontaneous preterm birth. The sensitivities for delivery at < 28, < 30, < 32 and < 34 weeks' gestation were 94%, 91%, 83% and 76%, respectively, while the negative predictive values were 99%, 99%, 98% and 96%, respectively. The placement of a cerclage did not influence the positive and negative predictive values. In comparison to women with other risk factors, cervical length was best in the prediction of preterm birth in women with a prior mid-trimester loss; an optimal cut-off of < or = 1.5 cm had sensitivities for delivery at < 28, < 30, < 32 and < 34 weeks' gestation of 100%, 100% 92% and 81%, respectively. The rate of preterm delivery at < 34 weeks' gestation increased dramatically when the cervical length was < or = 1.5 cm. Cervical length was the only independent variable that entered the logistic regression model for the prediction of preterm delivery at < 34 weeks' gestation.
Conclusions: In high-risk singleton gestations a cervical length of < or = 2.5 cm was equal to other sonographic cervical parameters in its ability to predict spontaneous preterm birth and was better for the prediction of earlier forms of prematurity (at < 28 and < 30 weeks) than later forms (at < 32 and < 34 weeks). The optimal cervical lengths and their performance for predicting prematurity may be influenced by obstetric risk factors.