Janak JC, Clemens MS, Howard JT, Le TD, Cancio LC, Chung KK, Gurney JM, Sosnov JA, Stewart IJ. Burns 2018; 44(8): 1920-1929.
Affiliation
Uniformed Services University of the Health Sciences, 4301 Jones Bridge Rd, Bethesda, MD 20814, United States; David Grant USAF Medical Center, Clinical Investigation Facility, 101 Bodin Circle, Travis AFB, CA 94535, United States.
BACKGROUND: The injury severity score considers burn size and inhalation injury in estimating overall anatomical injury severity. Models that adjust for injury severity score in addition to total burn size and inhalation injury may therefore be double counting the risk from these individual burn characteristics, and obscuring (or overemphasizing) the contribution of risk from each source. The primary aim of this study was to compare differences in the estimated mortality risk of burn trauma using the traditional injury severity score (ISS) calculation and the non-burn injury severity score (NBISS) to examine how separating out the risk attributable to the burn injury versus other trauma changes the interpretation and clinical assessment.
METHODS: Among U.S. casualties sustaining burns during combat operations in Iraq and Afghanistan from March 2003 to October 2013, we performed a retrospective cohort study. Unadjusted, adjusted, and weighted Cox proportional hazards models were performed to estimate the risk of age, burn injury severity, and non-burn injury severity on mortality. Weighted hazard ratios and adjusted survival curves were performed using non-parametric inverse probability weighting.
RESULTS: Our final sample consisted of 902 service members with a mortality proportion of 5.7% (n=51). Adjusting for non-burn trauma with traditional ISS attenuated the risk of percent total body surface area burned (%TBSA) by 20% when modeled continuously [HR (95% CI): 1.27 (1.10-1.32) vs. 1.07 (0.99-1.15]. However, the adjusted model using NBISS only attenuated the associated mortality risk of burn size by 5% [HR (95% CI): 1.22 (1.12-1.34)] and had a similar model fit (AIC: 484.2 vs. 478.6). For the weighted Cox proportional hazards models, the risk from a large burn (%TBSA≥60) was also attenuated when adjusting for ISS [HR (95% CI): 2.80 (1.18-6.64)] compared to the model adjusting for NBISS [HR (95% CI): 5.63 (2.79-11.35)].
CONCLUSION: Our analysis comparing the use of traditional ISS and NBISS to measure comorbid non-burn trauma resulted in different interpretations for the effect of %TBSA on subsequent mortality. Our results suggest that the association of %TBSA with death can be obscured by the inclusion of traditional ISS. Therefore, we recommend using NBISS when constructing statistical models in this patient population.