Butler-Laporte G, et al. - Researchers investigated if in cases of S. aureus bacteremia, a score based on patient level factors and methicillin resistant Staphylococcus aureus (MRSA) colonization could predict the risk of MRSA infection and inform the need for empiric coverage. For MRSA infection, researchers designed a predictive score based on patient level characteristics, and MRSA colonization as measured by screening done 30 days before infection (30-Day criteria), or at any time before infection (Ever-Positive criteria) using modern machine learning statistical methods (LASSO regression and random forests). Findings suggested colonization status to be the only independent and reliable predictor of MRSA infection in cases of S. aureus bacteremia. An appropriate clinical approach seemed to be the one that is based on a patient’s known MRSA colonization status and on local susceptibility patterns.