Predictive Characteristics of Methicillin-Resistant Staphylococcus aureus (MRSA) Nasal Swab for MRSA-positive Culture in Hospitalized Veterans

Providers often must decide whether to empirically treat hospitalized patients for MRSA. The results of routine MRSA nares swabs often are available prior to clinical culture results, so could conceivably help guide antibiotic selection. However, the reported predictive value of nares swabs is mixed. Therefore, we sought to define the predictive characteristics of MRSA nares swabs for the MRSA status of clinical Staphylococcus aureus(SA) isolates at the Minneapolis Veterans Affairs Medical Center (MVAMC). We retrospectively reviewed electronic health records (EHRs) of 599 MVAMC inpatients with a clinical SA isolate between 2013 and 2016. The SA isolates were from skin/soft tissue (n = 281), blood (n = 99), respiratory (n = 90), urine (n = 62), and bone/joint (n = 27). We recorded each isolate’s MRSA vs. MSSA status and the result of the temporally closest MRSA nares swab, then compared swab and culture results in relation to culture site and swab-to-culture interval. Overall, for identifying MRSA among patients with a clinical SA isolate, the MRSA nares swab’s sensitivity was 65.1%, specificity 96.2%, positive predictive value (PPV) 91.4%, and negative predictive value (NPV) 81.9%. The odds ratio (OR) of a positive MRSA nares swab for a MRSA-positive culture was 47.9 (95% confidence interval [CI] 25.7–89.2). Exclusion of the 70 nares swabs that were collected > 14 days before the clinical isolate increased the NPV to 84.0%, with a corresponding sensitivity 68.0%, specificity 83.9%, and PPV 90.3%. Test performance varied significantly by culture site (Table). A positive MRSA nares swab greatly increased the odds that a SA isolate was MRSA. However, sensitivity and NPV were lower than some prior studies. Our findings suggest that, for veterans with a severe infection that might be due to SA, a negative MRSA nares screen provides insufficient NPV to allow confident omission of empiric MRSA-active antibiotics. All authors: No reported disclosures.