Risk factors for positive admission surveillance cultures for methicillin-resistant Staphylococcus aureus and vancomycin-resistant enterococci in a neurocritical care unit

OBJECTIVE:Hospitals are under increasing pressure to perform active surveillance cultures for methicillin-resistant Staphylococcus aureus and vancomycin-resistant Enterococcus. This study aimed to identify patients at low and high risk for positive admission surveillance cultures for methicillin-resistant Staphylococcus aureus and vancomycin-resistant Enterococcus in a neurocritical care unit using readily ascertainable historical factors. DESIGN:Before/after study with nested case/control study. SETTING:Neurocritical care unit of an academic hospital. PATIENTS:During the intervention period (July 2007 to June 2008), after implementation of an admission surveillance culture screening program for methicillin-resistant Staphylococcus aureus and vancomycin-resistant Enterococcus, 2,059 patients were admitted to the neurocritical care unit for a total of 5,957 patient days. INTERVENTIONS:Cases had positive methicillin-resistant Staphylococcus aureus or vancomycin-resistant Enterococcus admission surveillance cultures within 48 hrs of hospital admission. Controls had negative cultures. MEASUREMENTS AND MAIN RESULTS:Admission surveillance cultures grew methicillin-resistant Staphylococcus aureus and vancomycin-resistant Enterococcus in 35 of 823 (4.3%) and 19 of 766 (2.5%) patients, respectively. Factors significantly associated with both methicillin-resistant Staphylococcus aureus and vancomycin-resistant Enterococcus colonization were intravenous antibiotics and hospitalization in the past year, immunocompromised health status, intravenous drug use, long-term hemodialysis, and known prior carrier status. Transfer from an outside hospital and residence in a long-term care facility in the past year were associated with vancomycin-resistant Enterococcus colonization. Classification and regression tree analysis was used to identify variables that best predicted positive methicillin-resistant Staphylococcus aureus and vancomycin-resistant Enterococcus surveillance cultures. A classification and regression tree model with six of these variables yielded an overall cross-validated predictive accuracy of 87.12% to detect methicillin-resistant Staphylococcus aureus colonization. For vancomycin-resistant Enterococcus, a four-variable classification and regression tree model (intravenous antibiotics, hospitalization and long-term patient care in the past year, and not being “admitted same day of procedure”) optimized the predictive accuracy (94.91%). There were no cases o