Predictors of nosocomial pneumonia in critically ill trauma patients

Abstract

Nosocomial pneumonia is the most common pulmonary complication in trauma patients and the leading cause of death from nosocomial infections. The purposes of this study were to identify underlying dimensions of variables that increase the risk of nosocomial pneumonia and to identify predictors of nosocomial pneumonia in critically ill trauma patients. A prospective cohort study of 246 consecutive critically ill trauma patients evaluated 22 risk factors for nosocomial pneumonia. The incidence of nosocomial pneumonia was 18.3%. A three factor structure identified with exploratory factor analysis using principle components extraction with varimax rotation was verified with confirmatory factor analysis using LISREL 8 (Goodness of Fit Index.97). Four indicators (days of intubation, integrity of endotracheal tube cuff, presence of a nasogastric tube, and receiving continuous enteral feedings) explained 37% of the variance in the treatment construct. Four indicators (Glasgow Coma Score, Abbreviated Injury Score (AIS) of the head, intracranial pressure monitoring, and Injury Severity Score (ISS)) explained 50% of the variance in the head injury construct. ISS and AIS of the chest explained 25% of the variance in the chest injury construct. Treatment construct score > 2.4 (p =.0002), head injury construct (p =.0045), age (p =.015), and no prior use of antibiotics (p<.0001) were identified by forward stepwise logistic regression as significant multivariate predictors of nosocomial pneumonia (Model chi-square = 53.908, df 4, p<.00009). Although the regression model correctly classified 84.6% of subjects, only 33.3% of patients with pneumonia were correctly classified. Study results support the existence of a 3 factor structure underlying risk factors for nosocomial pneumonia in critically ill trauma patients. Further research is indicated to confirm the identified 3 factor structure and regression model in other samples of trauma patients.