Validation of a Prediction Model for Vaginal Birth after Cesarean Delivery Reveals Unexpected Success in a Diverse American Population

Objective To investigate the validity of a prediction model for success of vaginal birth after cesarean delivery (VBAC) in an ethnically diverse population. Methods We performed a retrospective cohort study of women admitted at a single academic institution for a trial of labor after cesarean from May 2007 to January 2015. Individual predicted success rates were calculated using the Maternal-Fetal Medicine Units Network prediction model. Participants were stratified into three probability-of-success groups: low (<35%), moderate (35-65%), and high (>65%). The actual versus predicted success rates were compared. Results In total, 568 women met inclusion criteria. Successful VBAC occurred in 402 (71%), compared with a predicted success rate of 66% (p = 0.016). Actual VBAC success rates were higher than predicted by the model in the low (57 vs. 29%; p < 0.001) and moderate (61 vs. 52%; p = 0.003) groups. In the high probability group, the observed and predicted VBAC rates were the same (79%). Conclusion When the predicted success rate was above 65%, the model was highly accurate. In contrast, for women with predicted success rates <35%, actual VBAC rates were nearly twofold higher in our population, suggesting that they should not be discouraged by a low prediction score.