Validation of a predictive model for successful vaginal birth after cesarean section

Introduction: A strategy for reducing the number of cesarean sections is to allow vaginal delivery after cesarean section. Objective: To validate two predictive models, Metz and Grobman, for successful vaginal delivery after a cesarean section. Methods: Retrospective cohort study involving women with previous history of a previous segmental cesarean section, single pregnancy >= 37 weeks and cephalic presentation. The proportion of vaginal delivery in all pregnant women was determined, and it was compared with those (women) with successful delivery after cesarean section. Then, there were elaborated the models, and their predictive capacity was determined by curve-receiver-operator. Results: The proportion of successful delivery in pregnant women with a previous cesarean section and indication of vaginal delivery was 85.64%. The observed proportion of birth for each decile predicted in the Grobman model was less than 15%, except for the 91-100% decile, where it was 64.09%; the area under the curve was 0.95. For the Metz model, the actual successful delivery rate was lower than predicted in scores between 4 and 14, and within expected for a score between 15 and 23; the area under the curve was 0.94. Conclusions: The vaginal delivery rate after cesarean was lower than expected according to the predictive models of Grobman and Metz. The implementation of these models in a prospective way can lead to a higher rate of successful birth.