Hypothermia for prediction of death in severely injured blunt trauma patients.
Abstract:
Hypothermia is considered an independent predictor of death after trauma. The aim of this study was to assess these premises based on data from the TraumaRegistry DGU® (TR-DGU) using its outcome predication tool, the Revised Injury Severity Classification (RISC) score, in comparison with three previously published regression models by Shafi, Martin, and Wang. We hypothesized that body temperature on admission would improve accuracy of the RISC score. Data of 5,197 patients with documented body temperature on admission (T) and complete data for RISC score prognosis were selected from TR-DGU. Hypothermia was defined as T of 35°C or less. Patients were divided into hypothermia and normothermia group. Differences were assessed using Mann-Whitney U and chi-squared tests. Statistical significance was accepted at P < 0.01(*). Moreover, we performed multivariate logistic regression analyses using TR-DGU data on the four models (including RISC) with hospital mortality as dependant variable. Results are given as mean or odds ratio (OR) with 95% confidence intervals (95% CIs). Hypothermic patients were more severely injured (Injury Severity Score, 35.0 vs. 29.2 points*) and had higher rates of shock (38.3 vs. 16.8%*), organ failure (71.8 vs. 46%*), and sepsis (17.5 vs. 10.6%*). Survival was worse (29.2 vs. 13.7%*). Comparison of the above models revealed hypothermia as an independent risk factor (Martin: OR, 1.43 [95% CI, 2.21-1.42*]; and Wang: OR, 1.77 [95% CI, 2.21-1.42*]) only, although it would drop out from the model (RISC: OR, 1.12 [95% CI, 1.41-0.89; P = 0.33] and Shafi: OR, 1,.21 [95% CI, 1.60-0.92; P = 0.17]) as long as parameters to indicate hemorrhage and/or coagulopathy were included in sufficient number, a finding confirmed by a subsequent sensitivity analysis. We conclude that hypothermia is a result of injury severity and therefore unlikely to be an independent predictor of mortality. Our data suggest that hypothermia belongs closely to the hemorrhage/coagulopathy group of predictors.