Development of a regression model for the prediction of a success of external cephalic version

Background: The success of external cephalic version was affected by a series of factors. Development of a regression model incorporating multiple important factors will help prediction of success of external cephalic version. Here, we describe the development of a logistic regression model for predicting the success of external cephalic version for breech presentation in pregnancy. Methods: We retrospectively reviewed the records of 100 pregnant women who were pregnant with breech presentation during 37-40 weeks of gestation, and who were admitted in Fujian Provincial Maternity and Children’s Hospital affiliated Hospital of Fujian Medical University from Jan 2016 to Jan 2017. They were without other indications for cesarean section and ready to receive external cephalic version. The clinical data of the subjects were subjected to univariate analyses for selecting variables to be included in the logistic regression model. AUC of the regression model was calculated for estimate of the accuracy of the model. Independent variables with P≤0.1 in the multivariate analysis were selected for establishment of a scoring system of risk factors. External cephalic version success rate was compared between groups that were divided according to a scoring system. ROC curve analysis was employed to convert the quantitative variables into bi-categorical variables to be included as independent variables in the multivariate logistic regression model. Results: The success rate of external cephalic version was 62% (62/100), and 48 cases (80%) underwent vaginal delivery at the end. The following variables were included in the regression model: parity, the sum of uterine height and presentation height, amniotic fluid volume, spontaneous transposition times during pregnancy and fetal head holding. The AUC of the model was 0.929 (95% CI: 0.878-0.981). The comparison between groups divided by scoring revealed groups with higher scores had higher success ECV rate. Conclusion: Our regression model showed a good performance for predicting ECV success for breech presentation in single pregnancy.