INTRODUCTION: Infections may play an important role in mortality in burn patients, but this impact is not manifested in multivariate analyses that calculate the probability of death on admission. PATIENTS AND METHODS: In a cohort of 1,773 ICU patients, logistic regression analysis was used to determine the prognostic factors for death. Subsequently, the cohort was divided into 2 groups according to length of ICU stay: 1-10 days or more than 10 days. The groups contained a similar number of patients and the analysis was repeated in each one. RESULTS: Overall mortality in the 1,773 patients was 12.2%. In the stratified analysis mortality was 8.5% for the 890 patients who were in the ICU for 10 days or less and 15.8% for the 883 who stayed longer. The prognostic factors for death in the 2 groups were very different and had to be studied separately. In the group with a shorter ICU stay, the prognostic factors were similar to those of the entire cohort (eg, age, burned body surface area); in the group with a longer stay, these factors lost power and other factors, such as bacteraemia and pneumonia, gained significance. CONCLUSIONS: Estimation of the risk of death in ICU burn patients should be re-evaluated at 10 days after admission.
INTRODUCTION: Infections may play an important role in mortality in burn patients, but this impact is not manifested in multivariate analyses that calculate the probability of death on admission. PATIENTS AND METHODS: In a cohort of 1,773 ICU patients, logistic regression analysis was used to determine the prognostic factors for death. Subsequently, the cohort was divided into 2 groups according to length of ICU stay: 1-10 days or more than 10 days. The groups contained a similar number of patients and the analysis was repeated in each one. RESULTS: Overall mortality in the 1,773 patients was 12.2%. In the stratified analysis mortality was 8.5% for the 890 patients who were in the ICU for 10 days or less and 15.8% for the 883 who stayed longer. The prognostic factors for death in the 2 groups were very different and had to be studied separately. In the group with a shorter ICU stay, the prognostic factors were similar to those of the entire cohort (eg, age, burned body surface area); in the group with a longer stay, these factors lost power and other factors, such as bacteraemia and pneumonia, gained significance. CONCLUSIONS: Estimation of the risk of death in ICU burnpatients should be re-evaluated at 10 days after admission.