Association between labor induction and severe postpartum hemorrhage: a propensity score–weighted cohort study using longitudinal bleeding trajectory analysis

Background Postpartum hemorrhage (PPH) remains a major obstetric complication, and the association between labor induction and severe PPH remains uncertain, partly because of confounding by indication and limited evaluation of longitudinal bleeding patterns. Methods This retrospective cohort study included 1,108 consecutive women who underwent singleton vaginal delivery at ≥ 28 weeks of gestation at Fujian Maternity and Child Health Hospital between January 2022 and December 2025. Participants were classified according to labor onset into labor induction ( n = 376) and spontaneous labor ( n = 732) groups. Labor induction was defined as the use of pharmacologic or mechanical methods to initiate labor before spontaneous onset; recorded methods included balloon catheter, amniotomy, intravenous oxytocin, dinoprostone, and misoprostol, and individual women could receive more than one method. The primary outcome was a composite severe PPH endpoint defined by blood loss ≥ 1000 mL, blood transfusion, or advanced hemostatic intervention within 24 h postpartum. Inverse probability of treatment weighting (IPTW) followed by doubly robust weighted logistic regression was used to account for measured confounding. Longitudinal cumulative blood loss during the first 24 h was analyzed using LMMs. Sensitivity analysis was restricted to term deliveries (≥ 37 weeks). Results Severe PPH occurred in 399 women. After IPTW, most covariates achieved adequate balance, although gestational age remained mildly imbalanced (absolute SMD = 0.20) and was additionally adjusted for in the doubly robust model. Labor induction was not significantly associated with severe PPH (OR = 0.92, 95% CI 0.71–1.20, P = 0.555). In the longitudinal analysis, cumulative blood loss increased significantly across postpartum time points ( F = 152.902, P < 0.001), whereas neither the main effect of labor induction ( F = 1.98, P = 0.159) nor the labor induction-by-time interaction ( F = 0.25, P = 0.991) was statistically significant. These findings remained consistent in the sensitivity analysis restricted to term deliveries (≥ 37 weeks), in which labor induction was not significantly associated with severe PPH (adjusted OR = 1.08, 95% CI 0.80–1.46; P = 0.613) and the labor induction-by-time interaction was not significant ( P = 0.159). Conclusion In this cohort of women undergoing vaginal delivery, labor induction was not independently associated with the composite severe PPH outcome or with a different longitudinal pattern of cumulative postpartum blood loss. Prospective studies are warranted to confirm these findings and to determine whether associations differ across specific induction methods.