Background: Community-associated (CA) methicillin-resistant Staphylococcus aureus (MRSA) infection has become common worldwide. Some researchers have argued that empirical therapy for MRSA should be given only to patients with suspected CA S. aureus infections who have risk factors for acquisition of MRSA. However, there are no prospective data examining this approach.

Methods: We prospectively enrolled consecutive patients who were hospitalized with S. aureus infection, administered a detailed questionnaire, and collected clinical and microbiological information.

Results: Of the 280 consenting patients, 180 were adults with CA S. aureus infection. Among these subjects, 108 (60%) had MRSA infection, and 78 (40%) had methicillin-susceptible S. aureus (MSSA) infection. MRSA infection was associated with younger age (P<.0001); skin/soft-tissue infection (P=.015); snorting/smoking illegal drugs (P=.01); recent incarceration (P=.03); lower comorbidity index (P=.01); more frequent visits to bars, raves, and/or clubs (P=.03); and higher frequency of laundering clothes in hot water (P=.05). However, the sensitivity, specificity, and predictive values for these factors for discriminating CA-MRSA infection from CA-MSSA infection were relatively poor. Post-hoc modeling revealed that, even in a 10% (i.e., low) MRSA prevalence population, patients lacking the 3 strongest MRSA risk factors would still have a 7% posttest probability of MRSA. Most MRSA strains belonged to the ST-8/USA300 genotype, contained SCCmec type IV, and shared virulence factors commonly found in the ST1:USA400 clone. MSSA strains were genotypically heterogeneous.

Conclusions: We found that clinical and epidemiological risk factors in persons hospitalized for CA S. aureus infection cannot reliably distinguish between MRSA and MSSA. Our findings have important implications for the choice of empirical antibiotic therapy for suspected S. aureus infections and for infection control.