Prediction of preterm premature rupture of membranes by transvaginal ultrasound features and risk factors in a high-risk population.

OBJECTIVE: To determine the transvaginal ultrasound features and risk factors that predict preterm delivery (PTD) preceded by either preterm premature rupture of membranes (PPROM) or preterm labor (PTL) in a high-risk population. STUDY DESIGN: Singleton gestations at high-risk for preterm delivery were screened prospectively with transvaginal ultrasound between 14 and 24 weeks. Univariate and multivariate regression analyses were performed for historical obstetric risk factors and transvaginal ultrasound features (cervical length (CL) and cervical funneling (CF)) for the prediction of PTD preceded by PPROM or PTL at < 32 and < 35 weeks' gestation, respectively. Indicated preterm deliveries were excluded. RESULTS: Of 321 patients screened, 71 (22%) delivered at < 35 weeks, 40 (56%) after PPROM and 31 (44%) after PTL. Preterm delivery at < 32 weeks was seen in 47 patients (15%), with 20 (43%) occurring after PTL and 27 (57%) after PPROM. Univariate analysis showed the following to be significant predictors of PPROM at < 35 weeks: a past history of PTD between 25 and 30 weeks (P < 0.008), cerclage in the current pregnancy (P < 0.0001), bacterial vaginosis (P < 0.011), CL < 25 mm (P < 0.0001) and CF > 25% (P < 0.0001). The following were found to be significant predictors of PTL at < 35 weeks: black race (P < 0.02), a cerclage in the current pregnancy (P < 0.01), CL < 25 mm (P < 0.001) and CF > 25% (P < 0.0001). The significant predictors of PPROM at < 32 weeks were cerclage in the current pregnancy (P < 0.006) and CL < 25 mm (P < 0.0001); for PTL at < 32 weeks they were a past history of spontaneous miscarriage between 14 and 24 weeks (P < 0.02), black race (P < 0.021), cerclage in the current pregnancy (P < 0.001) and CL < 25 mm (P < 0.001). Multivariate regression analysis revealed the significant predictors of PPROM at < 35 weeks to be: a history of PTD between 25 and 30 weeks with odds ratio (OR) 4.8 (95% confidence interval (CI), 1.9-11.5) and CL < 25 mm with OR 7.9 (95% CI, 3.6-17.5). The significant predictors of PTL at < 35 weeks on multivariate regression were black race with OR 2.2 (95% CI, 1.0-4.8) and unemployment with OR 1.2 (95% CI, 1.0-1.5). For deliveries < 32 weeks, CL < 25 mm (OR, 10.1; 95% CI, 3.2-32) was the only significant predictor of PPROM. Significant predictors of PTL were unemployment (OR, 1.3; 95% CI, 1.1-1.7), a history of spontaneous miscarriage between 14 and 24 weeks (OR, 6.2; 95% CI, 1.4-25), black race (OR, 8.6; 95% CI, 1.5-47) and CL < 25 mm (OR, 4.4; 95% CI, 1-19). The sensitivity, specificity and positive and negative predictive values of CL < 25 mm for PPROM at < 35 weeks were 73%, 69%, 25% and 95%, respectively; for PTL at < 35 weeks they were 58%, 66%, 15% and 94%; for PPROM at < 32 weeks they were 85%, 68%, 20%, and 98%; for PTL at < 32 weeks they were 70%, 66%, 12% and 97%. CONCLUSION: There are differences in both transvaginal ultrasound findings and risk factors that predict PPROM or PTL. Of the pathways leading to PTD, CL was a more significant predictor of PPROM compared with PTL, while sociodemographic factors were more predictive of PTL.