Massive Transfusion Protocol Predictive Modeling in the Modern Electronic Medical Record

Area of Science:

• Medical Informatics
• Trauma Surgery
• Emergency Medicine

Background:

• The Emory model was selected for integration into the hospital's EMR as a real-time clinical decision support tool for predicting massive transfusion protocol (MTP) activation.
• The model's continuous variable output facilitates periodic re-calibration to optimize sensitivity and specificity.

Purpose of the Study:

• To integrate a predictive model for MTP activation and delivery into the electronic medical record (EMR).
• To externally validate the model using prospectively gathered data and assess its accuracy and precision over time.

Main Methods:

• Prospectively collected data from trauma activations were used, inputting heart rate, systolic blood pressure, base excess (BE), and mechanism of injury into the EMR-integrated model.
• MTP delivery was defined as 6 units of packed red blood cells/6 hours (MTP1) or 10 units in 24 hours (MTP2).
• Receiver Operating Characteristic (ROC) and Precision-Recall (PR) curves were constructed at 6, 12, and 20 months to evaluate model performance.

Main Results:

• Data from 1162 patients were analyzed.
• Areas under the ROC curves for MTP activation, MTP1, and MTP2 delivery at 20 months were 0.831, 0.879, and 0.905, respectively (all P < 0.001).
• Areas under the PR curves at 20 months reached 0.371 for MTP activation, 0.339 for MTP1, and 0.355 for MTP2 delivery.

Conclusions:

• A predictive model for MTP activation and delivery was successfully integrated into the EMR and externally validated.
• The model demonstrated improved performance over time, with the continuous probability output allowing for optimization of sensitivity and specificity.