Predicting the risk of unplanned cesarean delivery at admission to guide labor process management: a retrospective cohort study in Southwest China

BACKGROUND: Following the implementation of the three-child policy in China, an increasing number of pregnant people prefer vaginal delivery over cesarean delivery. However, concerns about unplanned cesarean delivery during labor have increased the demand for a quantitative risk assessment tool. Therefore, developing a prediction model tailored for pregnant Chinese people is essential for assisting clinical decision-making. OBJECTIVE: This study aimed to identify risk factors associated with unplanned cesarean delivery during a trial of labor and to develop a predictive model for determining this risk at admission. METHODS: This single-center retrospective cohort study analyzed 1,532 pregnant people with a term, singleton, cephalic pregnancy without a history of uterine surgery who were admitted for a trial of labor (April-December 2021). Of these, 138 underwent unplanned cesarean delivery and 1,394 achieved vaginal delivery. Predictors were selected via LASSO regression from an initial set of 13 candidate variables and used to build a multivariate logistic nomogram. The model was assessed by its AUC, sensitivity, specificity, accuracy, positive predictive value, negative predictive value and calibration (Hosmer-Lemeshow test). Clinical utility was evaluated via decision curve analysis, and internal validation was conducted via 1,000 bootstrap repetitions. RESULTS: Eight variables were identified as significant predictors of delivery mode: maternal height, maternal BMI, history of vaginal delivery, Bishop score at admission, premature rupture of membranes, gestational age, complications, and estimated fetal weight. The predictive model achieved an AUC of 0.811 (95% CI: 0.774-0.848). The Hosmer-Lemeshow test indicated P = 0.513. Internal validation yielded a mean AUC of 0.804 and a Brier score of 0.071, demonstrating good discrimination and calibration. CONCLUSION: A nomogram-based predictive model was developed to assess the risk of unplanned cesarean delivery during trial of labor. This model provides a quantitative tool to guide clinicians in risk assessment and optimize labor management, potentially reducing complications associated with unplanned cesarean deliveries.