Objective: Various types of diagnostic and monitoring techniques are available in the prehospital environment. It is unclear how increasing complexity of diagnostic equipment improves the ability to predict the need for a life-saving intervention (LSI). In this study, we determined whether the addition of diagnostic equipment improved the predictive power of vital signs and scores obtained only by physical examination.
Methods: Institutional review board approval was obtained for an analysis of 793 prehospital trauma patient records collected during helicopter transport by Emergency Medical Services personnel. Exclusion of severe head injuries and patients with incomplete data resulted in 381 patients available for analysis. Data sets were classified on the basis of the instrumentation requirements for capturing the given measurements and were defined by three groups: Group 1, vital signs obtained with no equipment (radial, femoral, and carotid pulse character; capillary refill; motor and verbal components of the Glasgow Coma Scale [GCS]); Group 2, Group 1 plus eye component of the GCS and pulse oximetry (Spo(2)); and Group 3, Group 2 plus fully automated noninvasive blood pressure measurements, heart rate, end-tidal carbon dioxide, and respiratory rate. LSIs performed during transport and in the hospital were recorded. Data were analyzed using a multivariate logistic regression model to determine which vital signs were the best predictors of LSI.
Results: Radial pulse character and GCS verbal and motor components had the best predictive power for the need of a prehospital LSI in Group 1 (receiver operating characteristic [ROC] curve, 0.97). Radial pulse character together with the eye component of the GCS and the motor component of the GCS provided the best prediction of a need for a prehospital LSI for Group 2 (ROC curve, 0.97). Addition of all supplementary vital signs measured by an automated monitor (Group 3) resulted in an ROC curve of 0.97. Given an abnormal radial pulse character (weak or absent) and abnormal GCS verbal and motor components, the probability of needing an LSI was greater than 88%.
Conclusion: In this cohort of patients, predicting the need for an LSI could have been achieved from GCS motor and verbal components and radial pulse character without automated monitors. These data show that simple and rapidly acquired manual measurements could be used to effectively triage non-head-injured trauma casualties. Similar results were obtained from manual measurements compared with those recorded from automated medical instrumentation that may be unavailable or difficult to use in the field.