Epidemiologic aspects and results of applying the TRISS methodology in a Spanish trauma intensive care unit (TICU).

OBJECTIVE: Analysis of epidemiologic aspects in a trauma intensive care unit (TICU) and assessment of predicted outcomes. DESIGN: Prospective study. Samples collected over a 2-year period. SETTING: A Spanish TICU at a tertiary care centre. PATIENTS: A group of 404 trauma patients. INTERVENTIONS: TRISS methodology was applied. MAIN RESULTS: Mean age was 35.8 +/- 17 years. Mortality was 19.6% over a median ISS = 17. Blunt trauma was more frequent than penetrating trauma (90.1% versus 9.9%). Car accident was the major aetiological factor (32.4%) and the highest mortality was among struck pedestrians (26.4%). The cranial region showed the highest incidence of lesion (57.9%) and the neurological complications on stage were the commonest reported on the discharge forms (49.7%). Mechanical ventilatory support (MVS) was applied in 53.2% of patients, with a relative mortality of 35.8%. Survivors differed significantly from nonsurvivors in terms of age, Glasgow Coma Scale rating, RTS, ISS, TRISS, stage and number of complications reported. The risk factors found to be associated with mortality were injury to cranial and abdominal/pelvic regions and age over 65. The TRISS total accuracy was 0.88 (sensitivity = 0.67; specificity = 0.93; area under the ROC curve = 0.85 +/- 0.03). Forward stepwise logistic regression analysis selected age, ISS and RTS as the best predictors of survival. When our TRISS results were compared with those anticipated on the basis the MTOS, an injury severity mismatch appeared (z = 0.02; M = 0.78). CONCLUSIONS: We found a 19.6% mortality in the TICU. Cranial and abdominal/pelvic injury and age over 65 were the main risk factors on admittance. Clinically, we finally agreed with the majority of TRISS outcome predictions. However, we could not statistically validate the apparent clinical goodness of the TRISS methodology.

Robust linear regression taking into account errors in the predictor and response variables.

We developed a robust regression technique that is a generalization of the least median of squares (LMS) technique to the field in which the errors in both the predictor and the response variables are taken into account. This simple generalization is limited in the sense that the resulting straight line is found by using only two points from the initial data set. In this way a simulation step is added by using the Monte Carlo method to generate the best robust regression line. We call this new technique 'bivariate least median of squares' (BLMS), following the notation of the LMS method. We checked the robustness of the new regression technique by calculating its breakdown point, which was 50%. This confirms the robustness of the BLMS regression line. In order to show its applicability to the chemical field we tested it on simulated data sets and real data sets with outliers. The BLMS robust regression line was not affected by many types of outlying points in the data sets.