Background
Previous studies investigating the prevalence of postpartum haemorrhage have shown inconsistent findings. One reason for the inconsistency is the method of measuring blood loss, with subjective estimation of blood loss to diagnose postpartum haemorrhage known to be inaccurate. We aimed to synthesise the prevalence of postpartum haemorrhage and severe postpartum haemorrhage according to the method of blood loss assessment and mode of birth.
Methods
In this systematic review and meta-analysis, we searched PubMed, Web of Science, Embase, Google Scholar, Cochrane, ClinicalTrials.gov, and ISRCTN between Jan 1, 1960, and Aug 11, 2025 for studies of postpartum haemorrhage available in English with no restrictions on participants' age, ethnicity, or geographical setting. At least two authors independently assessed heterogeneity (τ2), publication bias (Doi plots) and certainty of evidence (GRADE approach). A random-effects model was used to meta-analyse the prevalence of postpartum haemorrhage (defined as blood loss ≥500 mL) and severe postpartum haemorrhage (defined as blood loss ≥1000 mL). The outcomes were synthesised pooled prevalence of postpartum haemorrhage and severe postpartum haemorrhage according to blood loss assessment method (subjective and objective) and mode of birth (vaginal and caesarean). This study is registered with PROSPERO (CRD42023418635).
Findings
Of 24 969 identified records, the final quantitative synthesis comprised 81 eligible studies (n=42 709 185 women), conducted in 40 countries. The pooled prevalence of objectively diagnosed postpartum haemorrhage (≥500 mL) after vaginal birth was 12·6% (95% CI 10·1–15·2, τ2=0·02 [n=329 782]), compared with 3·9% (3·0–4·9, τ2=0·02 [n=24 330 539]) for subjective diagnosis. For severe postpartum haemorrhage after vaginal birth, the pooled prevalence was 3·3% (2·6–4·1, τ2=0·01 [n=3 061 456]) using objective methods, compared with 2·3% (1·3–3·6, τ2=0·02 [n=2 465 141]) using subjective methods. The pooled prevalence of postpartum haemorrhage after caesarean birth was 8·2% (3·7–14·3, τ2=0·04 [n=120 635]) using subjective methods; there were no data for objective methods. For severe postpartum haemorrhage after caesarean birth, the pooled prevalence was 8·5% (5·8–11·6, τ2=0·01 [n=480 306]) for objective methods compared with 3·7% (2·7–4·9, τ2=0·02 [n=11 921 326]) for subjective methods. There was evidence of publication bias for all study outcomes. All outcomes had moderate-certainty evidence, except for subjectively diagnosed postpartum haemorrhage after caesarean birth which had low-certainty evidence.
Interpretation
Postpartum haemorrhage and severe postpartum haemorrhage rates were substantially higher when using objective blood loss assessment methods for diagnosis, suggesting underestimation of blood loss with subjective assessment methods. Awareness of the correct prevalence of postpartum haemorrhage can help establish its health-care burden and improve outcomes.
