M. Subrize1, B. A. Asif1, M. Mukit1, H. C. Thode1, M. C. Henry1 1Stony Brook University Medical Center,Department Of Emergency Medicine,Stony Brook, NY, USA
Introduction:
In 2001, Suffolk County, New York implemented the use of the trauma prehospital care report (TPCR), which included the current American College of Surgeons (ACS) trauma triage criteria, a vehicle diagram to identify the location of damage, body diagrams to identify the location and type of bodily injury and the use of safety equipment. We sought to use the TPCR in conjunction with Trauma Registry data to examine which criteria were predictive of patient outcomes and resource use.
Methods:
We conducted a retrospective study of adult TPCR cases matched with the County’s regional Trauma Registry. All patients with a TPCR from 2003 to 2007 were eligible for inclusion. Outcomes included severe injury (Injury Severity Score > 16), need for major OR procedure (MOR), and death. Univariate analysis of triage criteria vs. outcomes was shown as relative risks. Multivariate analysis was performed using stepwise logistic regression analysis. Predictors were included if they were significant at the 0.05 level. Stepwise receiver operating characteristic (ROC) curves were also made using triage criteria as discrete (binary) predictors. At each step the sensitivity and specificity of each remaining criterion was calculated, and the criterion with the largest YI was accepted as the next predictor. This method was repeated until all remaining predictors had a YI of < 5.
Results:
17001 TPCR’s were analyzed. The most common cause of injury was due to motor vehicle collision. Nearly every trauma triage criterion was significantly associated with need for MOR, severe injury, and death. GCS <14 and flail chest were the most related to mortality (OR=12.4 CI 5.8-25.6; 11.4, 2.3-57.3 respectively).
GCS <14 and vehicle deformity were associated with every outcome and often the first criteria accepted into an outcome model. Mechanism criteria were included in 3 of the 6 outcome models. Speed over 40 mph was a useful indicator for ISS>15 or Major OR, but not death. The model with the highest sensitivity and specificity used mortality as the outcome (81% / 85%). Vehicle rollover, vehicle ejection, death of same vehicle occupant and pulse were not accepted into any models. Only death as an outcome had high sensitivity; all models had adequate specificity (78-85%).
Conclusion:
The current study explored different triage criteria and found vehicle deformity >20” and GCS <14 had the best predictive value in traumas causing death and were significant predictors in all other models as well. Vehicle rollover, vehicle ejection, death of same vehicle occupant and pulse had little impact on outcomes. The findings were discordant with other studies evaluating the 2011 trauma triage guidelines, and offer a glimpse into a unique environment, which includes a high percentage of automobile collisions.