PREDICTING SURVIVAL IN BURNED PATIENTS

Data pertaining to 562 consecutive admissions for burn treatment were analysed to identify factors related to survival. Besides socioeconomic, demographic and burn-related variables, three indices to measure burn severity, were proposed and evaluated with the help of multiple regression and discriminant analyses. The results of multiple regression analysis showed that one of the proposed indices, total burned surface (TBS), based on presence or absence of burn injury on 11 different body sites, turned out to be the best single predictor of survival TBS alone accounted for 64.5 per cent of the total variance in survival variable. Combined use of TBS with type and severity of burn, age, sex, etc. did wt appreciably raise the value of R2. The results of the discriminant analysis yielded a cut-off point of 20 TBS score which provided maximum separation between survivors and fatalities, Using this cut-off point (20 per cent) the TBS index provided the correct prediction of the eventual survival status in about 93 per cent of 562 patients. This cut-off point score of 20 was cross-validated on an independent sample of 924 Cases. The prediction in 79 per cent of patients could be made correctly.