K. He1, P. Zhang1,2, S. C. Wang1,2 2International Center Of Automotive Medicine,Ann Arbor, MI, USA 1University Of Michigan,Department Of Surgery,Ann Arbor, MI, USA
Introduction:
Early identification of severely injured patients in Motor Vehicle Collisions (MVC) is crucial. Mortality in this population is reduced by one quarter if these patients are directed to a level I trauma center versus a non-trauma center. The Centers for Disease Control and Prevention (CDC) Guidelines for Field Triage of Injured Patients recommends occupants at 20% or greater risk of Injury Severity Score (ISS) 15+ be urgently transported to a trauma center. With the increasing availability of vehicle telemetry technology, there is a great potential for advanced automatic collision notification (AACN) systems to improve trauma outcomes by detecting patients at risk for severe injury and facilitating early transport to trauma centers. We compared first responder field triage to a real-world field test of our updated injury severity prediction (ISPv2) algorithm using crash outcomes data from General Motors vehicles equipped with OnStar.
Methods:
We performed a literature search to determine the sensitivity of first responder identification of ISS 15+ MVC occupants. We used National Automotive Sampling System Crashworthiness Data System (NASS_CDS) data from 1999-2013 to construct a functional logistic regression model predicting the probability that one or more occupants in a non-rollover MVC would have ISS 15+ injuries. Variables included principal direction of force, change in velocity, multiple vs. single impacts, presence of older occupants (≥55 years old), presence of female occupants, belt use, and vehicle type. We validated our model using 2008-2011 crash data from Michigan vehicles with AACN capabilities identified from OnStar records. We confirmed telemetry crash data sent from the vehicles using police crash reports. We obtained medical records and imaging data for patients transported from the scene for evaluation and treatment. ISS was assumed to be ≤15 for MVC occupants not transported for medical assessment. We used our ISPv2 algorithm and transmitted telemetry data to predict the probability that an occupant had ISS 15+ injuries and compared our prediction to the observed injuries for each occupant and each vehicle.
Results:
Recent studies have found field triage to be 50-66% sensitive in identifying ISS 15+ occupants. Our study population included 924 occupants in 836 crash events. The median age was 41 years old, 57% were female, 21% were right-sided passengers, and 1.2% experienced an ISS 15+ injury. Our ISPv2 model was 72.7% sensitive (ISPv2 ≥0.2 when ISS 15+) and 93% specific (ISPv2<0.2 when ISS ≤15) for identifying seriously injured MVC patients.
Conclusion:
Our second generation ISP algorithm was highly specific and more sensitive than current field triage in identifying MVC patients at risk for ISS 15+ injuries. This real-world field study shows telemetry data transmitted before dispatch of emergency medical systems is superior in selecting patients who require urgent transfer to trauma centers.