Assessment of coagulation markers and decision curve analysis for predicting transfusions and hemostatic interventions in severe postpartum hemorrhage: A nationwide observational study in Japan

Abstract Background Postpartum hemorrhage (PPH) remains a major cause of direct obstetric mortality. We aimed to predict pregnant women with severe PPH, defined as cases requiring hemostatic interventions—such as transcatheter arterial embolization, hysterectomy, bimanual uterine compression, or intrauterine balloon tamponade—or blood transfusions. Methods This retrospective observational study utilized a large-scale database covering approximately 600 hospitals in Japan from 2014 to 2023. Women of ≥18 years of age, emergency hospitalized for placental abruption, amniotic fluid embolism, or peripartum hemorrhage were included. Laboratory test (platelets, fibrinogen, fibrinogen/fibrin degradation product [FDP], and prothrombin time-internationalized ratio [PT-INR]) results at admission and perinatal parameters were collected. The primary outcome was receiving hemostatic intervention or blood transfusion. Risk factors were assessed by logistic regression and non-linear regression analyses. Decision curve analysis (DCA) was used to compare Japanese obstetric scores with platelet adjustment and three existing obstetric disseminated intravascular coagulation (DIC) scores. Results Among 902 obstetric inpatients meeting the inclusion criteria, 343 received hemostatic interventions or blood transfusions; 559 did not. Multivariate logistic regression revealed fibrinogen level (adjusted odds ratio [OR]: 0.994, 95% CI: 0.991–0.998), but not platelet count (adjusted OR: 0.980, 95% CI: 0.933–1.029) as a risk factor for the primary outcome. Non-linear regression using restricted cubic splines demonstrated that the predictive risk increased steeply when platelet count and fibrinogen fell below 10 × 10⁴/μL and 200 mg/dL, respectively. DCA revealed that the platelet-adjusted obstetric DIC score developed in this study showed the highest net benefit across most threshold probabilities when compared with the pregnancy-modified ISTH DIC score, the JAAM-2 DIC score, and the new Japanese obstetric DIC score, although the advantage was modest. Conclusion Platelet counts and fibrinogen levels may be useful for identifying women with severe PPH who are at risk of requiring hemostatic interventions and transfusions.