Objective: 1. To reexamine the predictive value of the variables usually used in admission scores in burned patients (age, total body surface area burned (TBSAB), full thickness burn (FTB), inhalation injury (IHT) and sex). 2. To evaluate whether risk factors (alcohol abuse (AA), nicotine abuse (NA)) or preexisting diseases influence outcome significantly.

Design: Retrospective study of prospectively collected data.

Patients: 498 burned patients admitted to the burn ICU within a 5 years period. The mean TBSAB was 29% and the mean age 38 years. 42% of the patients suffered burns greater than 30% and the incidence of IHT amounted to 43%.

Methods: Univariate analyses were used to determine the independent relation of the variables to mortality. The relative weight of the variables was estimated using the step-wise logistic regression model. An additional analysis of subgroups was performed using classification and regression trees (CART).

Results: The univariate analyses identified the following variables to have significant influence on mortality: age, TBSAB, FTB, IHT, sex, AA, NA, the combination of AA and NA, preexisting neurological diseases and cardiovascular diseases. The step-wise logistic regression analysis identified age and TBSAB to have the most important influence on the outcome. Of minor weight was IHT followed by FTB and sex. The weight of IHT was found to be 1.7 fold higher than the impact of FTB and sex. A significant influence of IHT was found in all patients, but especially in patients with a medium risk of death (20%-45%) regarding age and TBSAB. In this group AA and NA additionally caused a significant impact on mortality. In patients with a higher or lower probability of survival AA and NA did not influence the outcome. The CART analysis identified TBSAB to be the most discriminative variable followed by age. In the group up to 20% TBSAB age was the only additionally significant variable regarding the outcome. In the group with a TBSAB between 20% and 60% age, sex and AA became important variables. In patients up to 72 years with a medium risk of mortality (20%-70%) IHT, AA, combined AA and NA, sex, preexisting neurological diseases and cardiovascular diseases significantly influenced outcome. In older patients IHT was the only additional variable of importance.

Conclusions: The study demonstrates that besides the "classical" variables of bum scores as age, TBSAB and IHT other variables such as sex, AA, NA and preexisting diseases have significant influence on the outcome. These variables especially gain important predictive value in patients with a medium risk of mortality.