We found during 1999–2006 that the percentage of S. aureus infections resistant to methicillin increased >90%, or ≈10% a year, in outpatients admitted to US hospitals. This increase was caused almost entirely by CA-MRSA strains, which increased >33% annually. Increases in the proportion of HA-MRSA isolates among outpatients were more variable, increasing ≈10% per year during 1999–2001 before the increase slowed; the proportion then decreased over the second half of the study period. This reduction in the growth of HA-MRSA isolates corresponds to a steep increase in the frequency of CA-MRSA skin and soft tissue infections among outpatients over an extremely short period, mostly during 2003–2005.
The frequency of CA-MRSA among inpatients increased nearly in conjunction with outpatient rates, overall and at each infection site. However, increases in blood and lung infections increased more among inpatients than in outpatients, which likely reflected the more severe status and increased likelihood of open wounds in inpatients. During this same period, rates of HA-MRSA decreased only ≈10%. Most of this decrease occurred during 2003–2005 and was mainly the result of a decrease in the frequency of HA-MRSA lung infections. This decrease was more likely the result of changes in empirical antimicrobial drug therapy for ventilator-associated pneumonia (24
) than a consequence of any changes in the epidemiology of MRSA.
Despite increases in the proportion of CA-MRSA strains among inpatients, the continuing high level of HA-MRSA suggests that in contrast to reports from local institutions (11
), CA-MRSA strains are adding to the problem of MRSA rather than replacing HA-MRSA strains. The fact that the frequency of HA-MRSA has decreased implies that some crowding out of HA-MRSA strains within the hospital may be occurring. However, lack of a decrease suggests that within the hospital, HA-MRSA strains may be more fit, and thus CA-MRSA strains are unable to replace them fully. The result is a coexistence of both strains in the hospital and maintenance of CA-MRSA because of the large influx of colonized and infected patients.
This finding is consistent with the biology of the 2 strains, which suggests differential fitness on the basis of the size of SCCmec
. In CA-MRSA strains, the predominant SCCmec
elements are types IV and V, which are smaller than the SCCmec
types typically found in HA-MRSA strains. These smaller genetic elements may increase the fitness of CA-MRSA strains outside hospital-related antimicrobial drug pressures, presumably by increasing mobility and growth potential (25
). However, their increased susceptibility to antibacterial agents in the hospital leaves them at a fitness disadvantage. The result is that although the community has effectively become a reservoir for the CA-MRSA strains that are continually introduced into the hospital population without genetic changes, they are unlikely to replace HA-MRSA strains in the hospital.
The large proportion of infections caused by CA-MRSA strains in hospitals with high frequencies of HA-MRSA has implications for drug-prescribing patterns within hospitals. Because CA-MRSA strains are generally susceptible to more antimicrobial drugs, persons with these infections may be able to be treated with less expensive antimicrobial drugs with fewer adverse outcomes. Moreover, appropriate therapy can reduce the likelihood of emergence of other resistant pathogens, such as vancomycin-resistant enterococci. Initial empiric therapy of infections with the suspected etiology of CA-MRSA must be tailored to antimicrobial drug susceptibility patterns within the local community and be based on efficacy studies that suggest specific effectiveness targets.
Kaplan suggested that empiric therapy should be modified if >10%–15% of CA-MRSA isolates become resistant to a specific empiric therapy (26
). Conversely, it may be appropriate to reintroduce a specific agent when susceptibility levels increase above a threshold. However, cycling strategies may not always be optimal (27
), and no efficacy studies have been conducted to establish this target. In addition, we urge caution in applying national results to the CA-MRSA antibiogram of a specific area. Although results showed an overall trend at the national level, specific results at individual testing centers tended to be more variable. Moreover, local health officials and hospitals should coordinate their efforts to identify susceptibility patterns at the community level, rather than at the hospital level, to optimize the gains from investments in infection control (28
The results of our study should be interpreted with caution because TSN provides information concerning only the site of isolate collection and not the infection. In addition, TSN only provides information on the collection location (i.e., outpatient or inpatient) and not case histories. Thus, some isolates may be difficult to classify in situations such as when an isolate was collected in the emergency department and then the patient was admitted or the patient was discharged and then returned as an outpatient. However, the effect of these situations is likely to be small because most isolates are from patients who can be classified as inpatients or outpatients.
A further limitation of the study is that although CA-MRSA isolate drug susceptibility patterns are technically genetically determined, the data enabled only phenotypic classification of isolates. In addition, as with any large time-series database, changes in surveillance or bias in the types of infections cultured over time, such as more severe or unusual infections, could alter the results. These findings suggest that more complicated bacteriology could alter the results. However, no general trend in the number of isolates collected was seen at individual testing centers, and resistance results from the TSN database were comparable to results of other national studies (1
). Furthermore, the striking increases over the study period suggest that the trends are likely robust to any bias.
In summary, we examined the frequency of CA-MRSA and HA-MRSA in inpatient and outpatient settings. Our results indicate that outpatients may be a major reservoir of CA-MRSA, which will continue to enter hospitals, exacerbating the problem of MRSA. However, although CA-MRSA isolates have undoubtedly spread within hospitals and are likely to continue to do so, without changes in the fitness of different strains, CA-MRSA strains are unlikely to displace HA-MRSA strains within the hospital.
Our findings have implications for local and national policies aimed at containing and preventing MRSA. More rapid diagnostic methods are urgently needed to better aid physicians in determining appropriate empiric therapy. Strategies for prevention of infection and treatment of patients with CA-MRSA within healthcare settings should be coordinated primarily at the local level in accordance with local susceptibility profiles. Lastly, infection control policies should take into account the role that outpatients likely play in the spread of MRSA and promote interventions that could prevent spread of MRSA from outpatient areas to inpatient areas.