Predicting and Estimating Burn Outcomes: An Institutional Analysis of 4622 Cases

Abstract

Abstract Advances in burn care have improved patient outcomes, and independently validated indices and predictors of burn outcomes warrant re-evaluation. The purpose of this study is to consolidate predictors of burn outcomes and determine the factors that significantly contribute to length-of-stay (LOS) and mortality. A retrospective review was conducted of all burn patients (n = 5778) admitted to a quaternary provincial burn unit from 1973 to 2017. Our inclusion criteria yielded 4622 independent cases. Multivariate linear and logistic regression models were generated, and area-under-receiver-operator-curve (AUROC) analysis was performed. Burn predictors included %TBSA, Baux (classic and revised) index, Abbreviated Burn Severity Index (ABSI), and Ryan score. Primary outcomes were mortality and LOS. Multivariate logistic regression for mortality showed the Baux index to be the best predictor for mortality (OR = 1.11, P < 0.001). The AUROC for Baux index was 0.95. With regard to LOS, ABSI was the best predictor for LOS (P < 0.001). ICU stay, ventilator use, alcoholism, and age are significantly associated with increased LOS. Interestingly, hypertension had a protective effect for LOS (P < 0.01) and trended towards a protective effect in mortality. Lethal score 50% (LS50) improved over the study period. The regressions show that burn mortality and LOS are best predicted with the Baux index and ABSI, respectively. Hypertension may have a protective effect on burn outcomes and may be attributed to increased perfusion to the periphery. These predictive scores are useful in determining institutional outcomes in burn surgery. Objective benchmarking of improvement in burn care outcomes can be established using LS50 trends.