Multifactorial probit analysis of mortality in burned patients

Abstract

Burn mortality statistics may be misleading unless they account properly for the many factors which may influence outcome. In reviewing such factors in our patients, we identified age, total burn area, third degree burn area, prior bronchopulmonary disease, abnormal Pao2, and airway edema as the factors present on admission which best distinguished survivors from nonsurvivors. Using multifactorial probit analysis, we then calculated the contribution of each to the probability of fatal outcome. The resultant six-factor model significantly improved estimation of the probability of fatal outcome when compared to probit analysis based only on the traditional factors of age and total burn area. It also revealed a spectrum of mortality probabilities varying with the additional factors present. Although crucial in comparing different approaches to burn care, consideration of such prognostic factors will not eliminate the need for randomized treatment trials, because other factors, some of which are obscure, may also influence mortality rates in burned patients.