Comparison of massive blood transfusion predictive models in the rural setting

Abstract

Background: Hemorrhage is the leading cause of preventable death in trauma patients, of which 3% require massive transfusion (MT). MT predictive models such as the Assessment of Blood Consumption (ABC), Trauma-Associated Severe Hemorrhage (TASH), and McLaughlin scores have been developed, but only included patients requiring blood transfusion during their hospital stay, excluding a large percentage of trauma patients. Our purpose was to validate these MT predictive models in our rural Level I trauma center patient population, using all major trauma victims, regardless of blood product requirements.

Methods: Review of all Level I trauma patients admitted in 2008 to 2009 was performed. ABC, TASH, and McLaughlin scores were calculated using 80% probability for the need for MT.

Results: Three hundred seventy-three patients were admitted; 13% had a penetrating mechanism and 52% were scene transports. MT patients had higher Injury Severity Score (median, 43 vs. 13; p < 0.001) and lower Trauma-Injury Severity Score (0.310 vs. 0.983; p < 0.001). Mortality was higher in MT patients (18.4% vs. 5.4%; p < 0.009). Thirty-eight (10%) required MT; 34 were predicted by ABC, one by TASH, and six by McLaughlin. ABC (area under the receiver operating characteristic [AUROC] = 0.86) was predictive of MT, whereas TASH (AUROC = 0.51) and McLaughlin (AUROC = 0.56) were not.

Conclusions: The ABC score correctly identified 89% of MT patients and was predictive of MT in major trauma patients at our rural Level I trauma center; the TASH and McLaughlin scores were not. The ABC score is simpler, faster, and more accurate. Based on this work, we strongly recommend adoption of the ABC score for MT prediction.