Predicting risk of postpartum haemorrhage: a systematic review

Background Postpartum haemorrhage (PPH) causes substantial morbidity and mortality worldwide. A reliable prognostic tool for PPH has potential to aid prevention efforts. Objective Systematically to identify and appraise prognostic modelling studies for prediction of PPH. Search strategy MEDLINE, Embase, CINAHL and the Cochrane Library were searched using a combination of terms and synonyms including 'prediction tool', 'risk score' and 'postpartum haemorrhage'. Selection criteria Any observational or experimental study developing a prognostic model for women's risk of PPH. English language publications. Data collection and analysis Predesigned data extraction form to record: data source; participant criteria; outcome; candidate predictors; actual predictors; sample size; missing data; model development; model performance; model evaluation; interpretation. Main results Of 2146 citations screened, 14 studies were eligible for inclusion. Studies addressed populations of women who experienced placenta praevia, placenta accreta spectrum, vaginal birth, caesarean birth (CS) and the general obstetric population. All studies were at high risk of bias due to low sample size, no internal validation, suboptimal or no external validation or no reporting or handling of missing data. Five studies raised applicability concerns. Three externally validated and three internally validated studies show potential for robust external validation. Conclusion Of 14 prognostic models for PPH risk, three have some potential for clinical use: in CS, in placenta accreta spectrum disorders with MRI placental Evaluation and in placenta praevia. Future research requires robust internal and external validation of existing tools and development of a model for use in the general obstetric population. Tweetable abstract Current PPH prediction tools need external validation: one for CS, one for placenta praevia and one for placenta accreta. Tools are needed for labouring women.

Automated summary

From 2146 citations, 14 studies were included. While these studies addressed various populations of women, they all had high risk of bias due to lack of internal or external validation, as well as issues in handling missing data. Only three studies show potential for robust validation, indicating the need for further research in this area.