Automatically-computed prehospital severity scores are equivalent to scores based on medic documentation.

BACKGROUND: Prehospital severity scores can be used in routine prehospital care, mass casualty care, and military triage. If computers could reliably calculate clinical scores, new clinical and research methodologies would be possible. One obstacle is that vital signs measured automatically can be unreliable. We hypothesized that Signal Quality Indices (SQI's), computer algorithms that differentiate between reliable and unreliable monitored physiologic data, could improve the predictive power of computer-calculated scores.
METHODS: In a retrospective analysis of trauma casualties transported by air ambulance, we computed the Triage Revised Trauma Score (RTS) from archived travel monitor data. We compared the areas-under-the-curve (AUC's) of receiver operating characteristic curves for prediction of mortality and red blood cell transfusion for 187 subjects with comparable quantities of good-quality and poor-quality data.
RESULTS: Vital signs deemed reliable by SQI's led to significantly more discriminatory severity scores than vital signs deemed unreliable. We also compared automatically-computed RTS (using the SQI's) versus RTS computed from vital signs documented by medics. For the subjects in whom the SQI algorithms identified 15 consecutive seconds of reliable vital signs data (n = 350), the automatically-computed scores' AUC's were the same as the medic-based scores' AUC's. Using the Prehospital Index in place of RTS led to very similar results, corroborating our findings.
CONCLUSIONS: SQI algorithms improve automatically-computed severity scores, and automatically-computed scores using SQI's are equivalent to medic-based scores.