Development and validation of a predictive model for severe postpartum hemorrhage in women undergoing vaginal delivery: A retrospective cohort study.

OBJECTIVE To develop a predictive tool to accurately screen women at high risk of severe postpartum hemorrhage (SPPH) undergoing vaginal delivery. METHODS We analyzed 28 150 mothers who underwent vaginal delivery after 28 weeks of pregnancy in the Third Affiliated Hospital of Guangzhou Medical University from January 2015 to August 2019. Two-thirds of the cohort were randomly allocated to a training set (n = 18 766) and the rest to a validation set (n = 9384). In the training set, we built a radiomic nomogram based on multivariate logistic analysis, and calibration and C-index were evaluated. The performance of the validated nomogram was then tested in the validation cohort. RESULTS Independent risk factors for SPPH in women undergoing vaginal delivery were previous cesarean section, history of PPH, in vitro fertilization, anemia, intrauterine death, prolonged labor, low-lying placenta, placental abruption, placenta accreta spectrum, and macrosomia. Good calibration was observed for the probability of SPPH in the validation cohort, and the C-index of the nomogram for the prediction of SPPH was 0.861 (95% confidence interval 0.820-0.902). CONCLUSION This model would be a useful tool to accurately screen for women at high-risk of SPPH undergoing vaginal delivery. It would be expected to be an effective tool to guide clinical practice and further reduce maternal mortality.