807: A Novel prediction model for external cephalic version success

Abstract

Breech presentation accounts for 3% to 4% of term pregnancies. Breech vaginal delivery has become less accepted following the publication of the Term Breech Trial in 2000. Successful external cephalic version (ECV) can reduce the rates of non-cephalic delivery and cesarean section (CS), but does not lack complications. Currently, maternal consultation on ECV success is limited to quoted success rates, varying from 35-83% in different studies. Clinical predictive factors have been identified for successful ECV, but have yet to be integrated in a predictive model, potentially assisting in the precise counselling of gravidas with non-cephalic presentation. A case-control study of patients who underwent ECV from January 2011 to July 2018, at a single tertiary center. Successful ECV was compared in a single and multivariate analysis using demographic and clinical independent variables to failed ECV group. A model predicting ECV success algorithm was assembled based on probability calculations extracted from the multivariate analysis. 701 women underwent ECV during the study period. ECV success rate was 60.34%. Characteristics of success & failure groups are shown on Table 1. Complication rate was 7.7% (n=54), of which 77.7% (n=42) were transient changes in fetal heart rate, 0.28% (n=2) suspected abruption and one case of neonatal anemia following ECV. Emergency post procedural CS rate was 1.2%. There were no cases of fetal demise. Positive predictors of successful ECV were Parity ≥1 (RR=4.55, CI 3.19-6.49), AFI≥200mm (RR=2.4, CI 1.33-4.46). Anterior placentation and maternal weight ≥ 70 kg were found to be negative predictors (RR=0.34, CI 0.236-0.479 and RR=0.59, CI 0.41-0.85 respectively). Subsequently a prediction algorithm was built on probability calculations based on the 4 predictors of ECV success; parity (0 or ≥1), AFI (<100, 100-200, >200), placentation (anterior, non-anterior) and maternal weight (<70kg, ≥70 kg). Use of the predictive model enabled excellent prediction of patient specific ECV success rates from 7% up to 95%. An algorithm for prediction of successful ECV based on clinical maternal parameters is presented.