Predictors of methicillin-resistant Staphylococcus aureus colonization at hospital admission

Abstract

Background: The best strategy for active surveillance for methicillin-resistant Staphylococcus aureus (MRSA) remains unclear. We attempted to identify a risk factor score to predict MRSA colonization at hospital admission.

Methods: Data on 9 variables reported as risk factors for MRSA colonization were analyzed, and a risk factor score to predict MRSA colonization was generated using multivariable logistic regression and receiver operating characteristic curve analyses. This risk score was then prospectively validated.

Results: Four risk factors (nursing home residence, diabetes, hospitalization in the past year, and chronic skin condition/infection) were significantly associated with MRSA colonization (c-statistic = 0.846). A cut-off score of 8 or greater would result in screening 20% of admissions and would detect 71% of MRSA-colonized patients. In the prospective validation study, a cut-off score of 8 or greater required screening 21% of admissions and detected 54% of MRSA. Nursing home residence was the best predictor of MRSA colonization.

Conclusion: A similar risk factor-based screening strategy could be used to predict MRSA colonization in other institutions. Our data support routine screening of nursing home patients at hospital admission.

Keywords: MRSA; MRSA active surveillance; Prediction.