Predicting nosocomial bloodstream infections using surrogate markers of injury severity: Clinical and methodological perspectives

Document Type


Publication Date


Publication Title

Nursing Research





First Page


Last Page



analysis of variance, article, blood transfusion, Catheterization, Central Venous, central venous catheterization, Chest Tubes, classification, critical illness, cross infection, emergency health service, human, Humans, immunosuppressive agent, Immunosuppressive Agents, Infection control, injury, injury scale, Injury severity, Injury Severity Score, length of stay, Linear Models, Logistic Models, metabolism, methodology, Mid-Atlantic Region, Nosocomial bloodstream infection, outcome assessment, Outcome Assessment (Health Care), prediction and forecasting, Predictive Value of Tests, Prospective Studies, prospective study, risk assessment, sensitivity and specificity, sepsis, serum albumin, standard, statistical model, statistics, Surrogate markers, Trauma Centers, Trauma patients, tube, United States, validation study, Wounds and Injuries


Background: Injury severity indices are numerical scores that are utilized to predict nosocomial bloodstream infections (BSI) in critically ill patients. However, surrogate markers of injury severity (SMIS) may be more clinically meaningful than these commonly used numerical injury severity indices with respect to the control and prevention of nosocomial BSI. Objective: The purpose of this study was to demonstrate the clinical and research implications of using the SMIS in predicting nosocomial BSI. Method: A prospective nonexperimental cohort study was conducted on 361 critically ill trauma patients. Three logistic regression models were examined for their clinical relevance and statistical parsimony. The first model included the Injury Severity Score (ISS) and 5 other independent predictors, and excluded the SMIS. The second model included all study variables. The third model excluded the ISS. Results: The analysis suggested that number of blood units transfused, number of central venous catheters inserted, and use of chest tube(s) were the SMIS. The ISS was found to be an independent predictor of nosocomial BSI only when the SMIS were not included in the model. The model that included the SMIS and excluded the ISS explained the highest variance in nosocomial BSI and had the best negative predictive value (93%). Discussion: Clinicians can use knowledge of SMIS to develop interventions that minimize the risk of nosocomial BSI. Hence, the SMIS can serve not only as a prediction tool but also as a way to enhance control and prevention strategies for BSI.