MRSA Risk 6-Fold Higher in HIV+: Community-Associated Methicillin-Resistant Staphylococcus aureus and HIV: Intersecting Epidemics
Clinical Infectious Diseases April 1 2010;50:979-987|
Kyle J. Popovich,1,2 Robert A. Weinstein,1,2 Alla Aroutcheva,1,2 Thomas Rice,2 and Bala Hota1,2
1Rush University Medical Center and 2Stroger Hospital Of Cook County, Chicago, Illinois
"In conclusion, the CA-MRSA and HIV infection epidemics have intersected, impacting community and hospital settings. The CA-MRSA burden in Cook County is >6-fold higher among HIV-infected patients than it is among HIV-negative patients. Control of MRSA in the community needs to involve effective use of preventive strategies in communities and networks at highest risk.....CA-MRSA caused increasing numbers of SSTIs (soft-tissue infections )among HIV-infected patients in community and hospital settings".... in comparison, in 2008, 10.5% of tested individuals with incident tuberculosis cases in the United States had HIV infection , making the number of tuberculosis cases in 2008 among HIV-positive patients about 1354. ..... Outbreaks of CA-MRSA infection have been observed in populations whose common attribute appears to be close person-to-person contact-young children in daycare facilities , prisoners , aboriginal populations [8, 9], athletes , military personnel , and men who have sex with men (MSM) ....HIV infection is a suggested risk factor for CA-MRSA infection; however, variables in addition to HIV-related immunosuppression, such as housing type and location of residence, may contribute to the risk of MRSA acquisition.....in addition to affecting patients without health care exposures, the CA-MRSA infection epidemic has impacted certain populations with high-risk behaviors or who have alternative housing (eg, MSM , illicit drug users , and prisoners ). Hota et al  observed that the CA-MRSA epidemic disproportionately affected the urban poor who resided in public housing. We also observed that, among HIV-infected patients hospitalized with CA-MRSA SSTIs, alternative housing was a risk factor for CA-MRSA acquisition."
Background. Single-site studies have suggested a link between human immunodeficiency virus (HIV) and community-associated methicillin-resistant Staphylococcus aureus (CA-MRSA).
Methods. Population-level incidence of HIV-infected patients with CA-MRSA versus community-associated methicillin-susceptible S. aureus (CA-MSSA) infection was assessed in the Cook County Health and Hospitals System (CCHHS), a multi-hospital and ambulatory care center. Rates in zip codes, including those with a high density of individuals with prior incarceration (ie, high-risk zip codes), were calculated. We did a nested case-control analysis of hospitalized HIV-infected patients with S. aureus skin and soft-tissue infections (SSTIs).
Results. In CCHHS, the incidence of CA-MRSA SSTIs was 6-fold higher among HIV-infected patients than it was among HIV-negative patients (996 per 100,000 HIV-infected patients vs 157 per 100,000 other patients; p<.001). The incidence of CA-MRSA SSTIs among HIV-infected patients significantly increased from 2000-2003 (period 1) to 2004-2007 (period 2) (from 411 to 1474 cases per 100,000 HIV-infected patients; relative risk [RR], 3.6; p<.001), with cases in period 1 clustering in an area 6.3 km in diameter (p=.035) that overlapped high-risk zip codes. By period 2, CA-MRSA SSTIs among HIV-infected patients were spread throughout Cook County. USA300 was identified as the predominant strain by pulsed-field gel electrophoresis (accounting for 86% of isolates). Among hospitalized HIV-infected patients, the incidence of CA-MRSA increased significantly from period 1 to period 2 (from 190 to 779 cases per 100,000 HIV-infected patients; RR, 4.1; p<.001). Risks for CA-MRSA by multivariate analysis were residence in alternative housing (eg, shelters), residence in high-risk zip codes, younger age, and infection in period 2.
Conclusions. HIV-infected patients are at markedly increased risk for CA-MRSA infection. This risk may be amplified by overlapping community networks of high-risk patients that may be targets for prevention efforts.
Methicillin-resistant Staphylococcus aureus (MRSA) infections were once restricted to hospitals and long-term care facilities. However, clonal transmission of MRSA-in the United States, USA300 by pulsed-field gel electrophoresis (PFGE) -has emerged in the community, causing infections in people without prior health care . The clinical impact of community-associated MRSA (CA-MRSA) ranges from minor skin and soft-tissue infections (SSTIs)  to invasive disease, such as necrotizing pneumonia , fasciitis , and bacteremia . Outbreaks of CA-MRSA infection have been observed in populations whose common attribute appears to be close person-to-person contact-young children in daycare facilities , prisoners , aboriginal populations [8, 9], athletes , military personnel , and men who have sex with men (MSM) .
Risk factors different from those typically associated with health care-associated MRSA , such as poor hygiene , illicit drug use , incarceration , and crowded living situations  have been linked to CA-MRSA. Furthermore, risk of MRSA acquisition in the community may be unevenly distributed geographically , with community-based social networks possibly serving as sources for acquisition and transmission .
HIV-infected patients have increased rates of S. aureus colonization [18, 19] and infection , in particular SSTIs [21, 22]. HIV-infected patients also appear to be at increased risk for persistent S. aureus nasal colonization  and for recurrent S. aureus infections at low CD4+ cell counts . Because of increased risk for S. aureus and MRSA colonization among HIV-infected patients [19, 25, 26], we assessed the impact that CA-MRSA has had on this population. We performed a community and hospital study to evaluate the epidemiology of CA-MRSA in HIV-infected patients. In the community study, we estimated the incidence of CA-MRSA SSTIs in the Cook County Health and Hospitals System (CCHHS) by HIV status and determined whether geographic clustering was present. In the hospital study, we examined the impact that CA-MRSA SSTIs have had on hospitalization rates for HIV-infected patients.
The CA-MRSA and HIV infection epidemics appear to have intersected. We found that the risk of CA-MRSA SSTIs in CCHHS was >6-fold higher for HIV-infected patients than it was for HIV-negative patients. CA-MRSA caused increasing numbers of SSTIs among HIV-infected patients in community and hospital settings. From period 1 (2000-2003) to period 2 (2004-2007), there was a nearly 4-fold increase in CA-MRSA SSTIs among HIV-infected patients in CCHHS. Among hospitalized HIV-infected patients, the incidence of CA-MRSA SSTIs increased >4-fold from period 1 to period 2. This increase in CA-MRSA disease was in addition to, not in replacement of, CA-MSSA SSTIs.
We have identified risk factors and potential transmission networks contributing to the propagation of the CA-MRSA infection epidemic in our HIV-infected patients. HIV infection is a suggested risk factor for CA-MRSA infection; however, variables in addition to HIV-related immunosuppression, such as housing type and location of residence, may contribute to the risk of MRSA acquisition. Shet et al  observed increased MRSA colonization among HIV-infected patients without significant immune dysfunction, compared with that among HIV-negative patients, which suggested that factors other than CD4+ count may contribute to increased colonization burden. Others have found that, in addition to affecting patients without health care exposures, the CA-MRSA infection epidemic has impacted certain populations with high-risk behaviors or who have alternative housing (eg, MSM , illicit drug users , and prisoners ). Hota et al  observed that the CA-MRSA epidemic disproportionately affected the urban poor who resided in public housing. We also observed that, among HIV-infected patients hospitalized with CA-MRSA SSTIs, alternative housing was a risk factor for CA-MRSA acquisition.
CA-MRSA infection outbreaks have been described in jails and prisons [7, 36, 37]. Lowy et al  observed a 25.5% rate of S. aureus nasal colonization among prisoners in 2 New York state prisons; 10.5% of isolates were MRSA and carried SCCmec IV, the genetic element predominantly associated with CA-MRSA . David et al  found that 63.5% of SSTIs at the Cook County Jail from March 2004 through August 2005 were due to MRSA; 92.3% of genotyped MRSA isolates belonged to sequence type 8, a genotype often associated with CA-MRSA . In a study of S. aureus clinical cultures from the San Francisco County jail system, Tattevin et al  observed that 82% of MRSA infections were due to USA300 strains. The downstream effects (when inmates return to the community) of the apparent MRSA colonization burden in jails and prisons are unknown. We observed that high-risk zip codes in Cook County (ie, those with high numbers of residents who were previously incarcerated) were associated with increased risk of CA-MRSA acquisition among HIV-infected patients. The initial focus of CA-MRSA infection in high-risk zip codes suggests that an early epidemic-driver was transmission of strains among networks of high-risk individuals.
Diep et al  also observed that CA-MRSA disease may not be evenly distributed geographically. They found that, over a 3-year period in San Francisco, there was an increased incidence of infection due to USA300 strains in 8 contiguous zip codes that had higher proportions of MSM, which suggested spread of this strain in particular community networks. We observed that, over an 8-year period, CA-MRSA SSTIs evolved geographically from zip codes having high individual-level prior incarceration rates, suggesting possible community networks of strain transmission to the broader Cook County.
The exact mode of transmission of CA-MRSA strains within and from community networks is unknown. Diep et al  hypothesized that, in San Francisco, particular high-risk behaviors likely contributed to high levels of disease. In Cook County, we observed patient and community factors that were associated with increased levels of infection, particularly residence in alternative housing and high-risk zip codes, which suggested a dynamic interplay among individual factors, housing type, location of residence, and HIV status.
Our findings and the findings of others [7, 10, 11, 16, 17] suggest that particular populations are disproportionately affected by the CA-MRSA infection epidemic. Based on the CDC's 2006 HIV prevalence estimates  and projecting our data, adjusted for age, sex, and race, we estimate that there may be at least 12,790 CA-MRSA SSTIs each year in HIV-infected patients. Although our extrapolation is crude (eg, based on our urban safety-net population), the CA-MRSA disease burden for HIV infection is significant; in comparison, in 2008, 10.5% of tested individuals with incident tuberculosis cases in the United States had HIV infection , making the number of tuberculosis cases in 2008 among HIV-positive patients about 1354. Given the potential excess use of parenteral and/or expensive antibacterials active against MRSA or need for hospitalization, excess cost and health care system burden may be substantial. Effective practices for controlling the intersection of the CA-MRSA and HIV infection epidemics are unclear. Particular community settings that appear to be epicenters for the spread of CA-MRSA (eg, prisons and high-risk zip codes) may represent targets for prevention efforts.
Our study has limitations. First, it is retrospective, with epidemiologic definitions used for cohort identification. However, we used PFGE and a published phenotypic prediction rule that was based on antibiotic susceptibility to validate our data . Second, differences between case patients and control subjects may not be generalizable to non-urban populations, although the notion of social networks as an essential component of CA-MRSA transmission may apply elsewhere. Third, we noted increased rates of MSSA infection over time-this could be the result of CA-MRSA strains that lost SCCmec IV, which we did not assess, or the result of more frequent culturing of individuals with SSTIs. However, the frequency of this occurrence is unclear. Fourth, our denominator of HIV-negative patients may include some HIV-infected patients whose illness was not diagnosed. Fifth, we do not know if culturing rates among HIV-positive and HIV-negative patients changed before or during the study period. However, based on several studies that documented the emergence of CA-MRSA in communities and emergency departments across the United States, awareness of this epidemic has extended beyond HIV-infected patients. Therefore, a drastic difference in culturing practices between HIV-infected and HIV-negative patients would be unlikely to explain our findings. Finally, a large percentage of patients with CA-MRSA and CA-MSSA SSTIs in population-level analysis were hospitalized. It is unclear whether this is a marker of illness severity, whether it suggests a lower threshold in our emergency department for hospitalizing HIV-infected patients, or whether it reflects an underreporting of minor SSTIs in communities. However, if the latter were the case, our incidence estimates would be an underestimation of total CA-MRSA disease burden among HIV-infected patients in CCHHS. Furthermore, it is unclear whether the lack of change from period 1 to period 2 in the proportion of case patients who were hospitalized, despite an increase in the incidence of CA-MRSA SSTIs, represents case patients with less severe illness or earlier appropriate treatment.
In conclusion, the CA-MRSA and HIV infection epidemics have intersected, impacting community and hospital settings. The CA-MRSA burden in Cook County is >6-fold higher among HIV-infected patients than it is among HIV-negative patients. Control of MRSA in the community needs to involve effective use of preventive strategies in communities and networks at highest risk.
Geographic clustering analysis.
There are 187 zip codes in Cook County, and 9 are identified as high-risk zip codes . During the period from January 2000 through December 2003 (period 1), 19 (10%) of the zip codes in Cook County had a RR >1 of CA-MRSA versus CA-MSSA SSTIs among HIV-infected patients. A geographic cluster 6.3 km in diameter (p=.035) was identified that represented 13 (18%) of 73 cases. This cluster overlapped high-risk zip codes  (Figure 1A).
By the period from January 2004 through December 2007 (period 2), 40 (21%) of the zip codes in Cook County were associated with a RR >1 of CA-MRSA versus CA-MSSA infection among HIV-infected patients, which was a significant increase in comparison with period 1 (21% vs 10%; p=.03), which suggested that dissemination occurred directly or indirectly from high-risk to other zip codes (Figure 1B). Areas affected by CA-MRSA were spread throughout Cook County by period 2. Although no significant geographic clusters were identified in period 2, several areas with increased risk of CA-MRSA SSTIs continued to overlap with high-risk zip codes.
The incidence of CA-MRSA SSTIs among HIV-infected patients increased significantly from period 1 to period 2 in high-risk zip codes (from 671 to 1799 cases per 100,000 HIV-infected patients; p<.001) and other zip codes (from 302 to 1351 cases per 100,000 HIV-infected patients; p<.001) (Table 1).
Incidence of CA-MRSA SSTIs among HIV-negative and among HIV-positive individuals increased significantly from period 1 to period 2 in CCHHS (from 61 cases to 253 cases per 100,000 HIV-negative patients and from 411 cases to 1474 cases per 100,000 HIV-infected patients, respectively) (Table 2). The overall incidence of CA-MRSA SSTIs was >6-fold higher among HIV-positive individuals than it was among HIV-negative individuals (RR, 6.3; 95% confidence interval [CI], 5.6-7.1; p<.001).
Among HIV-positive individuals, there was a significant increase in SSTIs due to CA-MSSA from period 1 to period 2 (from 404 cases to 650 cases per 100,000 HIV-infected patients; p=.004), although this increase was less than that for CA-MRSA (Table 2). Proportions of HIV-infected patients who were hospitalized during period 1 and period 2 with CA-MRSA (46% and 49%, respectively) or CA-MSSA (51% and 50%, respectively) SSTIs were similar.
Genotypic analysis showed that CA-MRSA SSTIs in HIV-infected patients were largely due to USA300 strains. Among the 186 isolates available for PFGE analysis, USA300 strains were the predominant cause of SSTIs in outpatient (57 [88%] of 65), emergency room (17 [85%] of 20), and inpatient settings (86 [85%] of 101).
There were 162 CA-MRSA and 91 CA-MSSA SSTIs in hospitalized HIV-infected patients during the period 2000-2007. Rates increased from period 1 to period 2; the incidence of CA-MRSA SSTIs increased (from 190 cases to 779 cases per 100,000 HIV-infected patients; p<.001) to a much greater extent than did the incidence of CA-MSSA SSTIs (from 218 cases to 346 cases per 100,000 HIV-infected patients; p=.04) (Table 3).
Among CA-MRSA SSTIs, there was a statistically significant increase observed for both community-onset disease and community-onset, health care-associated disease (from 120 cases to 479 cases per 100,000 HIV-infected patients and from 70 cases to 300 cases per 100,000 HIV-infected patients, respectively; p<.001).
PFGE was performed on 92% of the 2004-2006 CA-MRSA SSTI isolates obtained from hospitalized HIV-infected patients. The phenotypic prediction rule, based on antibiotic susceptibility, was 87% accurate in predicting CA-MRSA PFGE pattern (USA300 or USA400 ). Of patients with CA-MRSA SSTIs, 100 (62%) had no prior health care exposures, and 62 (38%) had an MRSA infection in the previous 6 months or health care exposure in the previous year (eg, hemodialysis, long-term care facility residence, surgery, or hospitalization). Among isolates obtained from patients without prior health care exposures, 91% had the CA-MRSA phenotype, and among genotyped isolates, 91% were identified as USA300 or USA400 by PFGE. Among isolates obtained from patients with prior health care exposure, 74% had the CA-MRSA phenotype, and among genotyped isolates, 86% were identified as USA300 or USA400. Within the subset of patients with prior health care exposures, 40% of isolates had the CA-MRSA phenotype in period 1, whereas 81% of isolates had this antibiotic susceptibility pattern in period 2, which suggested that, during the study period, there was a possible replacement of traditional MRSA strains with genotypically defined CA-MRSA (ie, USA300) strains among individuals with health care exposures .
By univariate analysis, significant risk factors for CA-MRSA SSTIs (case patients) versus CA-MSSA SSTIs (control subjects) among HIV-infected patients (Table 4) were younger age, prior MRSA infection, infection in period 2, and residence in a high-risk zip code and/or residence in alternative housing. Degree of HIV-associated immunosuppression was not significantly different between groups; however, mean CD4+ cell count (194 cells/mm3) and mean log10viral load (3.5 copies/mL) in hospitalized patients were not reflective of well-controlled disease.
On multivariate analysis (Table 5), significant risk factors for CA-MRSA SSTIs were residence in alternative housing (odds ratio [OR], 4.3; 95% CI, 1.2-15.4; p=.03), residence in high-risk zip codes (OR, 1.9; 95% CI, 1.1-3.4; p=.03), younger age (OR, 0.96; 95% CI, 0.93-0.997; p=.03), and infection in period 2 (OR, 2.8; 95% CI, 1.5-5.3; p=.002). There were no significant differences in clinical outcomes between HIV-infected patients with CA-MRSA SSTIs and HIV-infected patients with CA-MSSA SSTIs.
Using electronic data, we retrospectively studied HIV-infected patients with community-associated S. aureus SSTIs who received medical care during the period 2000-2007 in the multicenter, regional CCHHS in Chicago, Illinois. CCHHS is the safety-net system in Cook County, which is the second most populous county in the United States. CCHHS has 3 major hospitals: Stroger (formerly Cook County) Hospital (CCH), a 464-bed hospital and the major safety-net hospital in the region; Provident Hospital, a 140-bed community-level hospital; and Oak Forest Hospital (OFH), which during the study was a 670-bed long-term care facility. In addition, CCHHS has an extensive ambulatory system; the Ruth M. Rothstein CORE Center is an affiliated HIV clinic with a catchment population of 4490. The case definition used for community-associated S. aureus disease included patients with S. aureus SSTIs cultured in an outpatient clinic, emergency department, or hospital 72 h into hospitalization. Given the evolving epidemiology  for community-associated disease, and because a large proportion of HIV-infected patients frequently have prior antibiotic and health care exposures [28, 29], patients with health care exposures in the previous year were not excluded from our case definition. We used 2006 Centers for Disease Control and Prevention (CDC) HIV prevalence estimates  to project national estimates of CA-MRSA SSTIs among HIV-infected patients.
Geographic clustering analysis.
HIV-infected patients with community-associated S. aureus SSTIs with care in CCHHS were used in population-level analysis. The relative risk (RR) of CA-MRSA versus community-associated methicillin-susceptible S. aureus (CA-MSSA) SSTIs was estimated for each zip code in Cook County by time among HIV-infected individuals. Arcview 9.0 (ESRI) was used to geocode and visualize addresses. Individuals were mapped according to their zip code of residence, and spatial clustering was assessed using individuals with MSSA SSTIs as control subjects (Bernoulli method of spatial scan statistic; SaTScan software, version 5.1). In addition, zip codes that included HIV-infected patients at increased risk of CA-MRSA (vs CA-MSSA) were identified in period 1 (2000-2003) and period 2 (2004-2007) by calculating RRs by zip code. CA-MSSA was the comparator, because it can be a community-onset infection for which risk factors similar to those for CA-MRSA exist, but secular trends may differ, therefore allowing assessment of geographic clustering.
Using a published research report from the Urban Institute Justice Policy Center (Washington, DC) , zip codes in Cook County with high individual levels of prior incarceration were identified (so-called high-risk zip codes). High-risk zip codes were included in geographic analysis to assess the relation of these and other zip codes to risk for CA-MRSA SSTIs among HIV-infected individuals.
Population analysis.Incidences of CA-MRSA and CA-MSSA SSTIs were estimated for HIV-positive and HIV-negative patients who received medical care in CCHHS. The denominator for HIV-infected patients was the annual number of HIV-infected patients who received care in CCHHS; the denominator for HIV-negative patients was calculated using the proportion of residents in each zip code that sought care at CCHHS for SSTIs based on discharge diagnosis data reported to the Illinois Hospital Association. These proportions were multiplied with census measures of the total population in each zip code, creating an estimate of the at-risk population in each zip code. Estimates were summed across zip codes to generate a catchment population for SSTI in Cook County for CCHHS . We reduced this total by the estimated seroprevalence of HIV in the general population (0.4%)  to generate an estimate of HIV-negative individuals in the catchment population. Denominators for calculated rates in high-risk and non-high-risk zip codes were HIV-infected patients in high-risk and other zip codes, respectively.
Genotypic analysis with PFGE  was performed on isolates available from January 2004 through December 2006 with results interpreted as described by McDougal et al . Genotypic results were not used for the community-associated definition.
With use of electronic and clinical data, we retrospectively studied HIV-infected patients hospitalized at CCH with S. aureus SSTIs cultured within 48 h before to 72 h after hospitalization from January 2000 through December 2007. Case patients or control subjects were HIV-infected patients whose SSTI culture grew MRSA or MSSA, respectively. Patients were included irrespective of health care exposures in the previous year.
We estimated the incidence of community-associated S. aureus SSTIs among hospitalized HIV-infected patients at CCH, with the denominator being the annual number of HIV-infected patients who received care in the CORE Center catchment area. Demographic, housing, zip code, risk factor, and outcome data were obtained, and a case-control analysis was performed. Alternative housing was defined as current residence in subsidized housing, shelters, or substance-abuse centers. Residence in high-risk zip codes was defined as an address in a zip code with a high density of prior incarceration and not having residence in alternative housing. We applied a published phenotypic prediction rule that was based on antibiotic susceptibility  to test performance of this rule in comparison with PFGE and to assess correlation of our epidemiologic definition for community-associated disease with USA300 types in this population.
SPSS software, version 10 (SPSS), was used for statistical analysis. χ2 analysis was used to examine categorical variables; continuous variables were analyzed with the independent-samples t test; and logistic regression was used for multivariate analysis. Statistically important ( ) factors on univariate analysis were included in multivariate analysis. Variables were removed using backward elimination of covariates with P values >.15. The study was reviewed by our Institutional Review Board; need for informed consent was waived.