Cervical length at 16-22 weeks' gestation and risk for preterm delivery

Objective: To determine whether cervical length, as measured by transvagina l sonogram in asymptomatic gravidas at 16-22 weeks, is associated with risk of spontaneous preterm delivery. Methods: In a prospective observational study in an unselected urban tertia ry care population, cervical length was measured by transvaginal ultrasound during routine anatomic surveys in 760 gravidas at 16 0/7 to 22 6/7 weeks. The predictor variable was cervical length and the outcome variable was ge stational age at delivery. Care providers were not masked to the results. S pontaneous preterm delivery was analyzed as before 37, 35, and 32 weeks. Sh ortened cervical length was defined by the tenth, fifth, and two and a half percentiles for our population. Yates-corrected chi (2) was used to evalua te the significance on univariate analysis of the relative risk (RR) and 95 % confidence intervals (CI). Multiple logistic regression analysis was used to control for background variables in evaluating the probability of prete rm delivery at less than 35 weeks. Sensitivity, specificity, positive and n egative predictive values also were calculated. Results: Cervical lengths were normally distributed (mean 38.5 +/- 8.0 mm a t 19.9 +/- 1.5 weeks) independent of gestational age at measurement, and th e tenth, fifth, and two and a half percentiles were 30, 27, and 22 mm, resp ectively. Eighty-five women delivered before 37 weeks, 51 before 35 weeks, and 27 before 32 weeks. Relative risks (95% CI) for spontaneous preterm del ivery before 37 weeks were 3.8 (2.6, 5.6), 5.4 (3.3, 9.0), and 6.3 (3.0, 13 .0) for the tenth, fifth, and two and a half percentiles, respectively; RRs for before 35 weeks were 4.5 (2.9, 6.9), 7.5 (4.5, 12.5), and 7.8 (3.6, 16 .7); and for before 32 weeks were 5.2 (3.3, 8.3), 9.7 (5.8, 16.1), and 8.4 (3.6, 19.9), respectively. Multiple logistic regression analysis confirmed that cervical length was a significant predictor of preterm birth before 35 weeks, and that paras had a 43% greater risk compared with nulliparas. Sen sitivity ranged from 13-44%, specificity 90-99%, positive predictive value 15-47%, and negative predictive value 80-98%. Conclusion: Transvaginal measurement of cervical length during routine ultr asound at 16-22 weeks' gestation in asymptomatic gravidas might help identi fy women at risk for spontaneous preterm delivery. (Obstet Gynecol 2000;96: 972-8. (C) 2000 by The American College of Obstetricians and Gynecologists .).