Comparison of predictive blood transfusion scoring systems in trauma patients and application to pre-hospital medicine.

Hemorrhage leads to 30% to 40% of trauma deaths, with up to 50% of these deaths transpiring before hospital arrival. There is a growing amount of experience and research with prehospital blood administration, but few tools exist to identify the need and impact of a prehospital blood program in a community. Validating a blood use prediction tool locally will allow us to apply that validation to prehospital patients in other communities. Multiple algorithmic scoring tools that predicted the use of blood products were assessed using data from Baylor Scott and White Memorial Hospital, a level I trauma center. A total of 100 men and 51 women were included in the study, 99 of whom received a blood transfusion within 2 hours of hospital arrival. Comparing the scoring systems using our internal data, we found that three scoring systems were approximately equal at determining the need for blood products: Criteria A for the Zhu et al scoring system had a specificity and positive predictive value (PPV) of 92% while maintaining a sensitivity and negative predictive value (NPV) of 48%. Similarly, the EBTNS scoring system with a cutoff of ≥6 resulted in a specificity of 90%, PPV of 91%, sensitivity of 56%, and NPV of 52%. Lastly, the ABC scoring system with a cutoff of ≥2 had a specificity of 94%, PPV of 91%, sensitivity of 38%, and NPV of 56%. These scoring tools can be used in the prehospital setting to predict the need for blood in geographic areas in order to help with asset utilization.