Advertisement

222: Development of a prediction model for cesarean-associated blood transfusion

      Objective

      To develop a model based on factors available at the time of cesarean delivery to predict risk of red blood cell transfusion in the intraoperative or postoperative period.

      Study Design

      We performed a secondary analysis of all women who underwent a cesarean delivery in a large, prospectively-assembled cesarean delivery registry. We divided this cohort randomly into a derivation group and a validation group. Using factors readily available to the clinician and also known to be associated with cesarean-associated red blood cell transfusion, we analyzed different strategies, including logistic regression, decision tree modeling, and random forest learning, in order to create the model with the best predictive ability in the derivation group. We then tested the model using the validation group.

      Results

      59,468 women were available for analysis and were split evenly into the derivation and validation groups. The overall rate of red blood cell transfusion was 2.7%. The prediction model is based on a multivariable logistic regression which includes race, number of prior cesarean deliveries, pre-delivery hematocrit, and the presence or absence of the following characteristics: multiple gestation, placenta previa, preeclampsia, placental abruption, chorioamnionitis, and use of general anesthesia. The area under the receiver operator characteristic curve for the derivation and validation groups is 0.82 (95% CI, 0.80 - 0.84) and 0.84 (95% CI, 0.82 - 0.85), respectively (p = 0.16) (Figure). Predicted probabilities of blood transfusion for four hypothetical patients are presented (Table).

      Conclusion

      A regression model that incorporates variables readily available to the clinician at the time of cesarean delivery can accurately predict the need for intra- or post-operative red blood cell transfusion. Use of this model, easily adaptable as a simple online calculator, may allow more appropriate allocation of blood bank resources at the time of cesarean delivery.
      Figure thumbnail fx1
      Figure thumbnail fx2