Risk prediction models for postpartum hemorrhage after vaginal delivery:a Meta-analysis

LIU Huan; CHEN Lifen; FU Yanyu; HUANG Shasha; LI Shiliang

doi:10.12102/j.issn.2095-8668.2025.23.004

Abstract：Objective:To systematically evaluate the risk prediction models for postpartum hemorrhage after vaginal delivery and provide support and reference for the optimization of existing models and the construction of new models.Methods:The computer was used to search CBM,CNKI,WanFang Data,VIP,PubMed,EMbase,Cochrane Library,and Web of Science.The retrieval time was from the inception to April 20,2025.The included studies were evaluated using the PROBAST risk bias assessment tool,and a Meta-analysis was conducted using RevMan 5.4 software.Results:A total of 13 articles were included,including 13 predictive models involving 29 617 patients.The AUC values of the models ranged from 0.630 to 0.951,with 12 models having an AUC≥0.7.The overall risk of bias assessment for all models was high risk,and the applicability evaluation was low risk.The results of the Meta-analysis showed that gestational hypertensive disorders(OR=3.39,95%CI 2.68-4.29),neonatal weight(OR=2.81,95%CI 2.04-3.86),placenta previa(OR=3.54,95%CI 2.65-4.72),age(OR=1.19,95%CI 1.03-1.38),placental residue/adhesion/implantation(OR=3.64,95%CI 1.93-6.86),third stage of labor time(OR=1.06,95%CI 1.04-1.07),uterine atony(OR=3.73,95%CI 2.11-6.61),multiple pregnancies(OR=4.12,95%CI 2.61-6.50)were predictive factors for postpartum hemorrhage after vaginal delivery(P<0.05).Conclusion:Existing evidence shows that the overall predictive performance of the risk prediction models for postpartum hemorrhage after vaginal delivery is good.However,since most of the models in this study have not been fully externally validated and the calibration methods have not been unified,continuous improvement is still needed in future model construction to explore more clinically applicable predictive models.