RESUMEN
BACKGROUND: Traffic accidents are considered a public health problem and, according to the World Health Organization, currently is the eighth cause of death in the world. Specifically, pedestrians, cyclists and motorcyclists contribute half of the fatalities. Adequate clinical management in accordance with aggregation patterns of the body areas involved, as well as the characteristics of the accident, will help to reduce mortality and disability in this population. METHODS: Secondary data analysis of a cohort of patients involved in traffic accidents and admitted to the emergency room (ER) of a high complexity hospital in Medellín, Colombia. They were over 15 years of age, had two or more injuries in different areas of the body and had a hospital stay of more than 24 h after admission. A cluster analysis was performed, using Ward's method and the linfinity similarity measure, to obtain clusters of body areas most commonly affected depending on the type of vehicle and the type of victim. RESULTS: Among 2445 patients with traffic accidents, 34% (n = 836) were admitted into the Intensive Care Unit (ICU) and the overall hospital mortality rate was 8% (n = 201). More than 50% of the patients were motorcycle riders but mortality was higher in pedestrian-car accidents (16%, n = 34). The clusters show efficient performance to separate the population depending on the severity of their injuries. Pedestrians had the highest mortality after having accidents with cars and they also had the highest number of body parts clustered, mainly on head and abdomen areas. CONCLUSIONS: Exploring the cluster patterns of injuries and body areas affected in traffic accidents allow to establish anatomical groups defined by the type of accident and the type of vehicle. This classification system will accelerate and prioritize ER-care for these population groups, helping to provide better health care services and to rationalize available resources.
RESUMEN
BACKGROUND: Our purpose was to validate the performance of the ISS, NISS, RTS and TRISS scales as predictors of mortality in a population of trauma patients in a Latin American setting. MATERIALS AND METHODS: Subjects older than 15 years with diagnosis of trauma, lesions in two or more body areas according to the AIS and whose initial attention was at the hospital in the first 24 h were included. The main outcome was inpatient mortality. Secondary outcomes were admission to the intensive care unit, requirement of mechanical ventilation and length of stay. A logistic regression model for hospital mortality was fitted with each of the scales as an independent variable, and its predictive accuracy was evaluated through discrimination and calibration statistics. RESULTS: Between January 2007 and July 2015, 4085 subjects were enrolled in the study. 84.2% (n = 3442) were male, the mean age was 36 years (SD = 16), and the most common trauma mechanism was blunt type (80.1%; n = 3273). The medians of ISS, NISS, TRISS and RTS were: 14 (IQR = 10-21), 17 (IQR = 11-27), 4.21 (IQR = 2.95-5.05) and 7.84 (IQR = 6.90-7.84), respectively. Mortality was 9.3%, and the discrimination for ISS, NISS, TRISS and RTS was: AUC 0.85, 0.89, 0.86 and 0.92, respectively. No one scale had appropriate calibration. CONCLUSION: Determining the severity of trauma is an essential tool to guide treatment and establish the necessary resources for attention. In a Colombian population from a capital city, trauma scales have adequate performance for the prediction of mortality in patients with trauma.
