Replacement of HA-MRSA by CA-MRSA Infections at an Academic Medical Center in the Midwestern United States, 2004-5 to 2008

We noted anecdotally that infections designated as health care-associated (HA-) MRSA by epidemiologic criteria seemed to be decreasing in incidence at the University of Chicago Medical Center (UCMC) after 2004. We compared MRSA patients seen at any site of clinical care at UCMC and the isolates that caused their infections in 2004-5 (n = 545) with those in 2008 (n = 135). The percent of patients with MRSA infections cultured > 2 days after hospital admission decreased from 19.5% in 2004-5 to 7.4% in 2008 (p = 0.001). The percent in 2004-5 compared with 2008 who had a hospitalization (49.1% to 26.7%, p = 0.001) or surgery (43.0% to 14.1%, p<0.001) in the previous year decreased. In 2008 a greater percent of patients was seen in the emergency department (23.1% vs. 39.3%) and a smaller percent both in intensive care units (15.6% vs. 6.7%) and in other inpatient units (40.7% vs. 32.6%) (p<0.001). The percent of patients with CA-MRSA infections by the CDC epidemiologic criteria increased from 36.5% in 2004-5 to 62.2% in 2008 (p<0.001). The percent of MRSA isolates sharing genetic characteristics of USA100 decreased from 27.9% (152/545) to 12.6% (17/135), while the percent with CA-MRSA (USA300) characteristics increased from 53.2% (290/545) to 66.7% (90/135). The percent of infections that were invasive did not change significantly. Our data suggest that HA-MRSA infections, both by epidemiologic and microbiologic criteria, relative to CA-MRSA, decreased between 2004-5 and 2008 at UCMC.


Introduction
The epidemiology of methicillin-resistant Staphylococcus aureus (MRSA) changed dramatically in the first decade of the twentyfirst century with the emergence of genotypically novel community-associated (CA-) MRSA strains [1]. CA-MRSA strains were first noted in the late 1990s as causes of skin and soft tissue infections (SSTIs) as well as severe invasive syndromes in people with no known previous exposure to a health care setting. Subsequently, a single genetic background, identified as USA300 by pulsed-field gel electrophoresis (PFGE), became the most common cause of SSTIs in U.S. emergency departments (EDs) in 2004 [2] and 2008 [3]. USA300 was the most common genetic background of MRSA causing bacteremia in 23 U.S. hospitals in 2009-2010 [4]. In surveillance at 12 laboratories in Minnesota, CA-MRSA infections defined by an epidemiologic criterion accounted for 33% of all MRSA infections in 2005, an increase from 11% in 2000 [5].
USA300 has progressively increased as a proportion of MRSA isolates causing infections cared for at large medical centers. As a consequence, the overall number of MRSA infections has increased [6]. This has resulted from 4 epidemiologic trends. First, the incidence of outpatient and ED MRSA SSTIs [7,8] and the incidence of CA-MRSA infections resulting in hospitalization [9] has increased in the past decade. Second, USA300 has increasingly been the cause of nosocomial infections and infections among people with healthcare exposure [10]. Third, there is evidence that there were fewer invasive MRSA infections in the U.S. among patients with recent health care exposure [11]. Fourth, during the past decade, among all MRSA infections in several U.S. populations, there has been an increase in the percent that are classified epidemiologically as CA-MRSA [5,[12][13][14].
We noted anecdotally that infections designated as health careassociated (HA-) MRSA by epidemiologic criteria seemed to be decreasing in incidence at the University of Chicago Medical Center (UCMC) after 2004. Previously, we assessed the clinical and molecular epidemiology of MRSA infections at UCMC from July 1, 2004 to June 30, 2005 [15]. At that time, we found that CA-MRSA infections, whether defined by epidemiologic or microbiologic criteria, were very common among inpatients and outpatients in the clinic and the ED among both adults and children. Here, we compare the results of that study with the clinical and microbiologic characteristics of MRSA patients and their MRSA isolates at UCMC in 2008. Our hypothesis was that between 2004-5 and 2008, there was a major shift in the overall molecular epidemiology of MRSA infections at our center, with a greater percent of infections caused by MRSA strains bearing SCCmec type IV, a greater percent of isolates carrying the genes for the Panton-Valentine leukocidin (PVL), and a higher percent of strains having the ST8 multilocus sequence type (MLST), characteristics, when taken together, that identify USA300 [16].

Ethics Statement
The study was approved by the University of Chicago Biological Sciences Division Institutional Review Board (IRB). The IRB approved the following consent process. In-person consent was not feasible because the study was performed without specific funding, and patients, in the majority of cases, were no longer inpatients at our institution at the time of enrollment. Telephone consent was obtained utilizing an IRB-approved text read to potential subjects; for children ,7 years of age, consent was obtained from a parent or guardian. For children 8 to 18 years of age, in addition to consent from a parent or guardian, telephone assent was obtained from the patient. Patients who refused consent were not enrolled. Consent was documented on a paper consent form for each enrolled subject except in cases when the subject (or a parent or guardian for a child) could not be reached despite at least 5 attempts. In cases when telephone consent could not be obtained, a waiver of consent was granted by the IRB, and note was made of this in the study records.
UCMC is comprised of pediatric and adult hospitals serving the population on the south side of Chicago as well as tertiary care patients from elsewhere in the midwestern United States, with 26,288 admissions, 82,828 ED visits, and 388,277 outpatient visits in fiscal year 2008. UCMC follows the guidelines of the Health Care Infection Control Practices Advisory Committee of the Centers for Disease Control and Prevention [17]. Also, in March 2008 UCMC began a nasal screening for MRSA carriage, using a PCR-based test, of all patients admitted to an intensive care unit, as mandated by Illinois state law. Carriers identified in this way were placed on contact precautions.

Bacterial Isolates
We began prospectively collecting and storing MRSA isolates obtained by the UCMC Clinical Microbiology Laboratory on November 1, 2003. MRSA was isolated from 839 unique patients cultured as part of an evaluation for an infection in January 1 -December 31, 2008. A random 20% sample of these patients and their isolates was selected (n = 168). The first MRSA isolate obtained from each selected subject was chosen for analysis except when the first isolate represented asymptomatic colonization, in which case the first isolate from the patient's clinical infection was used.

Medical Record Review
Medical records were reviewed for each enrolled subject to assess age, gender, race, type of insurance, location of care when the MRSA culture was obtained, clinical MRSA syndrome, the presence or absence of an indwelling vascular catheter, comorbid medical conditions, and any previous MRSA infection or colonization. As in our previous study [15], each MRSA clinical syndrome was assigned to a syndrome category (SSTI, osteomyelitis or septic arthritis, bacteremia and endocarditis, urinary tract infection [UTI], pneumonia or other). All infections were separately classified as invasive if the isolate was obtained from a normally sterile site, and all other infections were considered noninvasive. Invasive infections included pneumonia, bacteremia, endocarditis, sepsis, osteomyelitis, septic arthritis, pyomyositis, peritonitis, necrotizing fasciitis and other deep-seated skin and soft tissue infections, and cholecystitis. Subjects with isolates repre-senting asymptomatic colonization were excluded from further analysis. Data collected for the 2008 subjects were identical to those collected for the 2004-5 patient cohort previously described at UCMC [15] except that, in 2008, data on previous surgery were collected for the year prior to the MRSA infection. For the 2004-5 data, history of surgery was assessed only for 6 months prior.

Microbiological Methods
For each MRSA isolate, antimicrobial susceptibilities to erythromycin, ciprofloxacin, rifampin, gentamicin, trimethoprimsulfamethoxazole, and vancomyin were determined by Vitek (Vitek 2, bioMérieux Vitek, Inc., Durham, NC). However, the Dzone test was performed on all isolates found to be clindamycin susceptible and erythromycin resistant on single-agent testing, according to NCCLS Guidelines [18].

Genotyping
For each studied isolate, multilocus sequence typing (MLST) was performed as described [19]. Detection of the Panton Valentine leukocidin (PVL) genes was performed by PCR as described [20]. The SCCmec type in the isolates was determined by a panel of PCR assays that allowed for the determination of the mec and ccr complex [21,22].

Criteria for CA-and HA-MRSA
Eight criteria commonly used to distinguish CA-from HA-MRSA infections were applied to each patient and his or her MRSA isolate. Clinical criteria included the (1) the 48-hour criterion, (2) CDC epidemiologic definitions for CA-MRSA and HA-MRSA [11], and (3) SSTI as the clinical syndrome. (1) By the 48-hour criterion, an infection was considered to be CA-MRSA if it was cultured as an outpatient or ,2 days after admission to the hospital, and other infections were considered to be HA-MRSA.
(2) A MRSA infection was considered to be HA-MRSA by the CDC epidemiologic definitions if, in the year prior to culture, the subject had surgery, hospitalization, hemodialysis or a stay in a long-term care facility, if an indwelling vascular catheter was in place at the time of culture, or if the subject was an inpatient hospitalized for .2 days at the time of culture. Otherwise, the subject was considered to have a CA-MRSA infection [11]. The CDC has a third category, the health-care-associated communityonset (HACO) group, and we include this in the group of HA-MRSA infections by the CDC criteria. Patients and their MRSA isolates were also separately classified as CA-MRSA by isolate characteristics, including (4) the presence of the PVL genes, (5) the presence of SCCmec type IV, (6) the ST8 MLST genotype, (7) clindamycin susceptibility, and (8) non-multidrug resistance [non-MDR, defined as resistance to $2 tested non-b-lactam antibiotic drugs]), as done in our previous study [15].

Statistical Analysis
For each clinical and microbial characteristic, subjects and their isolates in 2004-5 [15] were compared with those in 2008. Data were compared by the chi-square, Fisher Exact, or t-test, as appropriate (Stata 11, College Station, TX). The percent pairwise concordance for every possible pair of the 8 criteria for CA-MRSA was tabulated.

Results
Among 168 patient-isolates designated for study, 149 were enrolled. Upon chart review, 135 (90.6%) were determined to have had an infection and 14 (9.4%) only had MRSA colonization. Further analyses were limited to the 135 enrolled subjects with clinical infections. Comparing patients with MRSA infections in 2004-5 and 2008, there was no significant difference in the distribution of clinical syndromes, pediatric vs. adult age groups, or race, but the subjects differed significantly by distribution of types of insurance coverage, with the percent of uninsured and privately insured higher in the 2008 patient cohort and the percent publicly insured lower (p = 0.001) ( Table 1). Although this change was statistically significant, the magnitude of the change was not great, and it is difficult to determine the importance of the finding.

Characteristics of infections: Invasive and non-invasive
The percent of infections that were invasive did not change significantly.

Discussion
Our data suggest that in 2008, the proportion of MRSA infections that were HA-MRSA, both by epidemiologic and microbiologic criteria, decreased compared with 2004-5 at UCMC. Most strikingly, there was a significant decrease in the  percent of MRSA infections cultured from inpatients more than 2 days after admission, in the percent that had surgery or a hospital stay in the previous year, and in isolates having the ST5 background, the most common background of U.S. HA-MRSA strains (USA100). At the same time, the percent of all MRSA infections that were invasive did not change during the two periods studied. The shift in molecular epidemiology of MRSA infections, with more USA300-like and fewer USA100-like isolates causing infection in 2008 compared with 2004-5, suggests that both of these genetic backgrounds are associated with invasive disease. This pattern, i.e., a shift in genetic backgrounds without a significant shift in the percent of infections that were invasive, also suggests that the development of an invasive infection may be more likely driven by patient characteristics and exposures than by characteristics of the bacteria causing infections although such patient characteristics were not assessed in our chart reviews.
Patients with MRSA infections over time were increasingly seen in the ED and less often in the inpatient setting at our center. This shift in site of care for MRSA infections occurred with no significant change in the overall distribution of the types of clinical syndromes caused by MRSA comparing 2004 with 2008. While there was a non-significant increase in the percent of SSTIs among patients in the 2008 cohort compared with the 2004-5 cohort, the percent that had bacteremia, endocarditis, or sepsis or other invasive infections was similar. This suggests that while the focus of onset of MRSA infections moved to the community in 2008, the shift was not simply because of an increase in the percent of SSTIs among MRSA infections.
There were 2 dominant genetic backgrounds at UCMC in 2004-5 and 2008. Comparing these 2 time periods, there was a shift in their relative prevalence, with a lower percent of ST5 and a greater percent of ST8 isolates, i.e., from USA100 to USA300. The most common HA-MRSA genetic background, characteristically ST5 with SCCmec II and lacking the PVL genes, decreased from about one-third to less than one-sixth of all MRSA isolates that we sampled at our medical center. The USA300-like isolates, i.e., ST8 carrying the SCCmec type IV element and PVL+, caused both CA-and HA-MRSA infections by the CDC epidemiologic criteria, as noted in a few other U.S. centers [10,23]. Few CA-MRSA infections were caused by USA-100-like isolates.
We found that as USA300 has become an increasingly prevalent genetic background among MRSA infections, USA100 became less so. In 2008, at our center less than one-quarter of HA-MRSA infections defined epidemiologically were caused by USA100-like isolates. This trend has been shown in few previous studies. While CDC Active Bacterial Core surveillance documented a decrease in invasive HA-and HACO-MRSA infections between 2005-2008, it did not report on non-invasive MRSA infections [11], as we have done.
A limitation to our study is our decision to use a 20% sample of all unique patients with a MRSA infection in 2008 at our medical center. It is possible that this number of patients is not an adequate representation of all patients in 2008. However, our patient sample was randomly selected, and with this sample, we demonstrated a significant change in the percent of infections caused by CAcompared with HA-MRSA infections at our medical center, no matter which of the many criteria are used to distinguish them.
As HA-MRSA strains become increasingly infrequent in the U.S., it is possible that we will enter an era in which USA300 is nearly the exclusive cause of MRSA infections in inpatient and outpatient settings. This is concerning because isolates of this background have been shown to be easily transmissible and virulent, potentially exacerbating infection control efforts. Further surveillance is warranted for genotypic change and for changes in antimicrobial susceptibility among MRSA isolates causing clinical infections at U.S. medical centers.