Key points are not available for this paper at this time.
OBJECTIVE: To identify the independent predictors of nosocomial bloodstream infections (BSIs) among critically ill adult trauma patients. DESIGN: A prospective, cohort design was used to study patients who met predetermined inclusion criteria. Basic descriptive and univariate statistical analyses were performed to identify unadjusted predictors. A forward stepwise multivariate logistic regression analysis was then conducted to identify independent predictors of nosocomial BSI. SETTING: Level I university-affiliated shock trauma center. PATIENTS: Three hundred sixty-one critically ill adult trauma patients, 55 of whom developed nosocomial BSIs (15.2%). RESULTS: Data analysis of 45 variables indicated that only 9 were independent predictors of nosocomial BSI: presence of a chest tube, use of immunosuppressive agents, presence of microbial resistance, length of stay, presence of preexisting infection, percentage change of serum albumin levels, patient disposition, transfusion of 10 or more units of blood, and number of central venous catheters (CVCs) for patients who had 4 or more. The classification index of the final regression model at a cut-off point of 0.5 had a specificity of 97.4%, a sensitivity of 60%, a positive predictive value of 76.7%, a negative predictive value of 93%, and an overall precision of 91%. CONCLUSION: In this study, only 9 variables were independent predictors of nosocomial BSI. Our findings are specific to critically ill adult trauma patients and should be interpreted within the context of this particular population.
El‐Masri et al. (Sun,) studied this question.