Development of a Risk-Scoring System for Prediction of Blood Transfusion During Hospitalization for Delivery

Area of Science:

• Obstetrics and Gynecology
• Transfusion Medicine
• Health Informatics

Background:

• Blood transfusion during delivery is a significant concern in obstetric care.
• Predictive tools are needed to identify high-risk patients for timely intervention.
• Existing risk stratification methods for transfusion in pregnancy may lack precision.

Purpose of the Study:

• To develop and internally validate a practical, data-driven risk-scoring system.
• To predict the likelihood of blood transfusion during hospitalization for delivery.
• To create a tool for contemporary U.S. obstetric patients.

Main Methods:

• Secondary analysis of a multicenter cohort (n=21,780) of patients delivering between 2019-2020.
• Exclusion of patients with placenta accreta spectrum.
• Development of a multivariable logistic regression model with internal validation using k-fold cross-validation and stepwise backward elimination.

Main Results:

• 2.5% of patients received a blood transfusion.
• Highest risk factors identified: thrombocytopenia, placental abruption, or antepartum bleeding.
• The risk score ranged from 0-17, predicting transfusion risk from 1.0% to 84.4%.
• Area under the receiver operating curve (AUC) for prediction was 0.81.

Conclusions:

• A clinically applicable numeric risk score was developed to predict blood transfusion during delivery hospitalization.
• The risk-scoring system demonstrates good predictive performance in the validation subsample.
• External validation of this risk-scoring system is recommended for broader clinical application.