Poster Session 3
Hugo Madar, MD, PhD (he/him/his)
Department of Obstetrics and Gynecology, Bordeaux University Hospital
Bordeaux, Aquitaine, France
Thomas Ferté, MD
Public Health Department, Clinical Epidemiology Unit, Bordeaux University Hospital, Bordeaux, France
Public Health Department, Clinical Epidemiology Unit, Bordeaux University Hospital, Aquitaine, France
Antoine Bénard, MD, PhD
Public Health Department, Clinical Epidemiology Unit, Bordeaux University Hospital, Bordeaux, France
Public Health Department, Clinical Epidemiology Unit, Bordeaux University Hospital, Bordeaux, Aquitaine, France
Kilian Trin, MD
Public Health Department, Clinical Epidemiology Unit, Bordeaux University Hospital, Bordeaux, France
Public Health Department, Clinical Epidemiology Unit, Bordeaux University Hospital, Aquitaine, France
Catherine Deneux-Tharaux, MD, PhD
Université Paris Cité, CRESS U1153, Obstetrical, Perinatal and Pediatric Epidemiology Research Team, EPOPé, INSERM, INRAE, Paris, France
Paris, Ile-de-France, France
Loïc Sentilhes, MD, PhD (he/him/his)
Head of the Department of Obstetrics and Gynecology
Department of Obstetrics and Gynecology, Bordeaux University Hospital, Bordeaux, France
Department of Obstetrics and Gynecology, Bordeaux University Hospital,, Aquitaine, France
Prophylactic tranexamic acid has been shown to reduce postpartum blood loss in cesarean deliveries, but its effectiveness may vary based on individual factors and cesarean management. Identifying the women who would benefit most could optimize treatment and minimize unnecessary exposure.
We reanalyzed data from the TRAAP2 trial to determine if the effect of tranexamic acid on postpartum hemorrhage (PPH) varies with women and management of the cesareans’ characteristics.
Study Design: We included all women in the modified intention-to-treat population of TRAAP2 who received tranexamic acid or placebo. The primary outcome was PPH, defined as blood loss over 1000 ml or a red-cell transfusion within two days post-delivery. Predictors for PPH included baseline characteristics and cesarean management details. Both logistic regression with lasso penalization (LRL) and random forest (RF) were used to estimate associations between PPH and predictors. Prediction accuracy was assessed using the area under the ROC curve (AUROC) and the precision-recall curve (AUPRC).
Results: The study included 4,367 women (2,190 in the tranexamic acid group and 2,177 in the placebo group). Lasso logistic regression outperformed random forest, with an AUROC of 0.72 (95% CI: 0.70-0.74) (Figure 1). The most important predictors of PPH were multiple pregnancy, number of cesareans, body mass index at the end of pregnancy, hemoglobin within 7 days before delivery, gestational age at delivery and cesarean during labor because of protracted labor. Model had poor calibration for benefit of tranexamic acid with an observed benefit almost constant across the quintile of predicted benefit with a discrimination for benefit of 0.04 (95%CI: -0.07 ; 0.15) (Figure 2).
Conclusion: Although tranexamic acid is effective for the prevention of calculated blood loss at population level, its individual efficacy is difficult to forecast accurately. This suggests that a targeted prophylactic approach, rather than universal use, may not be appropriate for preventing blood loss in cesarean deliveries.
