Predicting Cesarean in the Second Stage of Labor

Abstract

To develop a prediction model for cesarean delivery (CD) in the second stage of labor using classification and regression tree (CART) analysis. Retrospective cohort of term women who reached 10-cm dilation. The primary outcome was CD at 10-cm dilation. Logistic regression and CART analysis were performed to identify factors that best predict second-stage CD. Only factors known at the time a patient reaches 10-cm dilation were used. Of 5,388 subjects who reached 10 cm, 88 (1.6%) underwent CD. The logistic regression model identified 4 risk factors for CD and produced an area under the receiver operator characteristic curve of 0.75 (95% confidence interval 0.70 to 0.81). CART analysis identified the most important variable in predicting second-stage CD was a station at or higher than 0 at complete dilatation, but correctly classified only 19.3% of CD. Second-stage cesarean cannot be reliably predicted based on antenatal and intrapartum characteristics by logistic regression or CART techniques.