Predictive model o risk of primary postpartum hemorrhage at the first level of care

Abstract

Introduction: Early detection of postpartum hemorrhage risk factors by healthcare professionals during pregnancy and the puerperium can enable them to take action to prevent it.

Objective: To develop and validate a risk prediction model for primary postpartum hemorrhage in women who delivered vaginally.

Method: A cross-sectional case-control study was conducted among 84 postpartum women at the Charef Polyclinic in Wilaya Djelfa from January to December 2024. Twenty-one of the women were diagnosed with postpartum hemorrhage (the case group), while 63 were not (the control group). Percentages were used as a summary measure of qualitative variables, and the cross-product ratio (OR), confidence interval (CI), and chi-squared test were determined. A predictive model was developed using multivariate analysis with binary logistic regression, and its predictive capacity was assessed using the area under the curve.

Results: The predictors included in the model were episiotomy [OR 3,336;( 2,319-4,799); p=0,000], cervical laceration [OR 3,120; (2,211-4,402); p=0,000], clot retention [OR 2,917; (2,003-4,247); p= 0,000], fetal macrosomia [OR 2,892; (2,028-4,125); p=0,000], prolonged labor [OR 1,921; (1.351-2,732); p=0,000], and primiparity [OR 2,572; (1,798-3,680); p 0,000]. The area under the curve was 0.804, and the Hosmer-Lemeshow test was greater than 0.05 (p=0.23).

Conclusions: The risk factors selected in the study were associated with birth trauma, so reducing their practice would be an effective measure for preventing further postpartum blood loss. Furthermore, the predictive model obtained demonstrated very good discrimination and calibration capabilities.