Background: Vancomycin is often used as empiric therapy for methicillin-resistant Staphylococcus aureus (MRSA), but can be associated with clinically important adverse events including renal failure. MRSA colonization swabs are primarily used for infection control; their use as a diagnostic test to inform the decision to add empiric vancomycin therapy has not been well elucidated.

Methods: We performed a Medline and Embase systematic review for peer-reviewed studies reporting the diagnostic accuracy of using MRSA colonization status to predict MRSA infections. Meta-analysis was performed using Cochrane guidelines. Grey literature was excluded.

Findings: 29 studies were included involving 24225 patients. In cases where the pathogen is not known to be S. aureus, specificities were greater than 85% for bacteremia, lower respiratory tract infections, skin and soft tissue infections (SSTI), and all infections pooled together. Sensitivities ranged between 54.0% and 77.5%. In cases where the pathogen is known to be S. aureus, we found studies on bacteremia and SSTI and arrived at pooled estimates of sensitivities ranging between 56.6% and 56.9%, and of specificities greater than 91%. Most importantly, for most infections in settings where the prevalence of MRSA as a causative organism is below 15%, the negative predictive value of a negative MRSA colonization swab exceeds 90%.

Interpretations: In settings of low-moderate MRSA prevalence, negative MRSA screening swabs may prevent unnecessary vancomycin use. More research is needed to assess if this strategy can mitigate the cost of screening in areas with a low MRSA colonization rate.