The incidence of and risk factors for MRSA bacteraemia in an HIV-infected cohort in the HAART era
Volume 9, Issue 10, Pages 858-862
Published Online: 27 Aug 2008
MD Burkey 1,2 LE Wilson 1 , RD Moore 1 , GM Lucas 1 , J Francis 3 and KA Gebo 1
1 Department of Medicine, Johns Hopkins University School of Medicine, Baltimore, MD, USA, 2 Department of Family Medicine, University of Pittsburgh Medical Center, Pittsburgh, PA, USA and 3 Department of Medicine, Yale University School of Medicine, New Haven, CT, USA
Correspondence: Dr Matthew D. Burkey, Department of Family Medicine, University of Pittsburgh Medical Center, 6470 Monitor St Pittsburgh, PA 15217, USA. Tel: +1 410-330-2154; fax: +1 410-955-7889; e-mail: email@example.com
*This work was presented in part at the 13th Conference on Retroviruses and Opportunistic Illnesses, Denver, CO, February 2006.
"There was a significant increase in the incidence of MRSA bacteraemia as well as all S. aureus bacteraemia over the study period from 2000 to 2004"...."To the best of our knowledge, this is the largest study of S. aureus bacteraemia to date in a US HIV-infected population. This study has several important findings. First, MRSA represented a large and growing proportion of S. aureus bacteraemias among our urban cohort of HIV-infected patients. Secondly, patients with MRSA bacteraemia were more likely to have low CD4 cell count, IDU exposure, and ESRD than were bacteraemia-free controls. Compared with patients with MSSA bacteraemia, those with MRSA bacteraemia were more likely to have ESRD and showed a trend towards a greater level of immunocompromise as measured by CD4 cell count."...."Initial antimicrobial therapy for presumed sepsis in HIV-infected patients may require agents active against MRSA, including vancomycin, linezolid and, in nonpulmonary infections, daptomycin, particularly in patients with risk factors for MRSA bacteraemia."
Objectives: To define the incidence and risk factors for methicillin resistant Staphylococcus aureus (MRSA) bacteraemia in an HIV-infected population.
Methods: From January 1, 2000 to December 31, 2004, we conducted a retrospective cohort study. We identified all cases of Staphylococcus aureus bacteraemia (SAB), including MRSA, among patients enrolled in the Johns Hopkins Hospital out-patient HIV clinic. A conditional logistic regression model was used to identify risk factors for MRSA bacteraemia compared with methicillin-sensitive SAB and no bacteraemia in unmatched (1:1) and matched (1:4) nested case-control analyses, respectively.
Results: Of 4607 patients followed for a total of 11 020 person-years (PY) of follow-up, 216 episodes of SAB occurred (incidence: 19.6 cases per 1000 PY), including 94 cases (43.5%) which were methicillin-resistant. The incidence of MRSA bacteraemia increased from 5.3 per 1000 PY in 2000-2001 to 11.9 per 1000 PY in 2003-2004 (P=0.001). Multivariate analysis demonstrated that independent predictors of MRSA bacteraemia (vs. no bacteraemia) were injection drug use (IDU), end-stage renal disease (ESRD) and CD4 count <200 cells/μL.
Conclusions: MRSA bacteraemia was an increasingly common diagnosis in our HIV-infected cohort, especially in patients with history of IDU, low CD4 cell count and ESRD.
HIV-infected patients appear to be at increased risk for Staphylococcus aureus and methicillin-resistant Staphylococcus aureus (MRSA) colonization and infection, including bloodstream infection [1-5]. Mirroring trends in the general population, rates of clinically significant MRSA infections have continued to increase over time in HIV-infected populations . Previous studies have also revealed distinct risk factor patterns for MRSA infections in patients with HIV, including low CD4 cell count, high HIV-RNA viral load, and absence of cotrimoxazole prophylaxis [3,6].
Most studies of MRSA infection in HIV-infected patients have primarily evaluated skin and soft tissue infections with only a small number of bloodstream infections represented [5-10]. Little is known about the epidemiology of and unique risk factors for MRSA bacteraemia in HIV-infected patients in the highly active antiretroviral therapy (HAART) era.
We performed a retrospective cohort analysis of HIV-infected patients receiving longitudinal primary care from the Johns Hopkins Hospital out-patient HIV clinic for the period 1 January 2000 to 31 December 2004 (risk factor analysis included data through 31 December 2005). The patient population at the clinic is largely urban, African American, male, and low-income and there are high rates of injecting drug use (IDU) and hepatitis C infection. A detailed description of the demographics, data collection methods, and database management of the clinic has previously been published .
For this study, microbiological data were abstracted from a hospital electronic database including all in-patient and out-patient blood culture results from the study period. S. aureus bacteraemia was defined as one positive blood culture for S. aureus. All isolates were screened for methicillin resistance using the agar dilution reference method .
The primary outcome of this study was an episode of MRSA bacteraemia. We used a nested, case-control analysis to assess sociodemographic and clinical factors associated with first episodes of MRSA bacteraemia. Cases were defined as an initial episode of MRSA bacteraemia that occurred during the study period. For the first analysis, up to four controls without S. aureus bacteraemia were randomly selected from the overall cohort for each case, matched on enrolment date, duration of follow-up and CD4 cell count at enrolment. A second case-control analysis compared cohort patients with methicillin-susceptible Staphylococcus aureus (MSSA) bacteraemia with those who had MRSA bacteraemia.
The following data were obtained from the database: age, sex, race, HIV transmission risk factors (IDU or non-IDU, based on enrolment assessment), CD4 cell count, plasma HIV-1 RNA, use of HAART (from prescription records), Pneumocystis carinii pneumonia (PCP) and Mycobacterium avium complex (MAC) prophylaxis at time of event, history of diabetes, and history of end-stage renal disease (ESRD).
Statistical analyses were performed using Stata 9.0 (Stata Corporation, College Station, TX). Univariate and multivariate regression analyses were used to assess risk factors for MRSA bacteraemia compared with (1) matched controls without S. aureus bacteraemia, and (2) unmatched MSSA bacteraemia controls. All reported P-values are two-tailed.
Between 1 January 2000 and 31 December 2004, 4607 patients were followed for a total of 11 020 person-years (PY) of follow-up. During the follow-up period, 216 total episodes of S. aureus bacteraemia occurred (incidence: 19.6 cases per 1000 PY) (Fig. 1). The incidence of S. aureus bacteraemia increased significantly between the 2-year periods 2000-2001 (incidence: 16.9 cases per 1000 PY) and 2003-2004 (23.6 cases per 1000 PY) (P=0.03).
Among the cases of S. aureus bacteraemia, 94 cases (43.5%) were methicillin-resistant (incidence: 8.5 per 1000 PY.) The proportion of methicillin-resistant strains increased during the follow-up period from 31.6% in 2000-2001 to 50.5% in 2003-2004 (P=0.06). The absolute incidence of MRSA bacteraemia also increased from 5.3 per 1000 PY in the 2-year period 2000-2001 to 11.9 per 1000 PY in 2003-2004 (P=0.001). Eighty-three of 89 first cases of MRSA bacteraemia were able to be matched based on the above criteria and were included in the analysis.
Among incident cases with MRSA bacteraemia, most patients were male (61.5%), were African American (91.6%) and had IDU as an HIV risk factor (74.7%). The mean age at event among incident cases was 42.3 years (range 27.2-63.6 years). These demographic characteristics are similar to those of the overall clinic population.
The last measured CD4 count prior to event was <200 cells/μL in 55 cases (66.9%), while only 14 cases (16.9%) had undetectable plasma HIV-1 RNA. Twenty-two cases (26.5%) were on HAART at the time of the event. Seven out of the 22 cases on HAART (32%) were virologically suppressed. Nineteen patients (22.9%) had ESRD and five (6.0%) had diabetes.
In the first case-control univariate analysis, factors associated with increased odds of MRSA bacteraemia compared with bacteraemia-free controls included African American race [odds ratio (OR)=4.54; 95% confidence interval (CI) 1.99-10.34], IDU (OR=4.59; 95% CI 2.57-8.20), and ESRD (OR=7.19; 95% CI 3.33-15.51) (Table 1). Compared with CD4 count >500 cells/μL at event, lower CD4 cell count at event was associated with increased odds of MRSA bacteraemia: CD4 count 201-500 cells/μL at event (OR=2.42; 95% CI 0.89-6.57), CD4 count 51-200 cells/μL (OR=10.09; 95% CI 3.51-29.00), and CD4 count <50 cells/μL (OR=18.56; 95% CI 6.55-52.61). Compared with plasma HIV-1 RNA <400 HIV-1 RNA copies/mL, detectable viral load at the event was associated with increased odds of MRSA bacteraemia (OR=3.05; 95% CI 1.63-5.73). All other variables evaluated, including sex, age, PCP or MAC prophylaxis, HAART and comorbid diagnosis of diabetes, were not associated with MRSA bacteraemia.
In univariate analysis stratified by sex, men who have sex with men (MSM) contact was associated with a reduced OR of MRSA bacteraemia (OR=0.23; 95% CI 0.10-0.51). Those who reported MSM contact were less likely to report IDU (_2<0.001). When corrected for IDU in a multivariate model, the association between MSM contact and MRSA bacteraemia was no longer statistically significant.
In those with an indication for PCP or MAC prophylaxis (CD4 count <200 cells/μL and CD4 count <50 cells/μL, respectively), neither form of opportunistic infection (OI) prophylaxis was associated with MRSA bacteraemia in univariate or multivariate analysis (see Table 1).
In adjusted multivariate regression analysis, IDU [adjusted odds ratio (AOR)=4.61; 95% CI 2.32-9.17], ESRD (AOR=7.78; 95% CI 2.92-20.72), and CD4 count 50-200 cells/μL (AOR=6.43; 95% CI 2.04-20.23) and CD4 count <50 cells/μL (AOR=21.52; 95% CI 6.74-68.69) compared with CD4 count >500 cells/μL were independent predictors of MRSA bacteraemia (Table 1). Race, plasma HIV-1 RNA, and use of HAART were not associated with MRSA bacteraemia after adjusting for CD4 cell count, IDU and ESRD.
The second case-control analysis compared MRSA bacteraemia cases to controls with MSSA bacteraemia. In univariate analysis, ESRD (OR=3.18; 95% CI 1.26-8.05) was associated with increased odds of MRSA bacteraemia. Compared with CD4 count >500 cells/μL, CD4 count 201-500 cells/μL (OR=3.79; 95% CI 1.14-12.55) and CD4 count 51-200 cells/μL (OR=3.50; 95% CI 1.08-11.37), but not CD4 count <50 cells/μL (OR=2.71; 95% CI 0.87-8.45), were associated with increased odds of MRSA bacteraemia. None of the other variables evaluated, including sex, age, race, IDU, plasma HIV-1 RNA, use of HAART, OI prophylaxis, and diabetes, was associated with MRSA bacteraemia (data not shown).
The association between MRSA and ESRD (AOR=2.89; 95% CI 1.12-7.49) remained significant when adjusted for CD4 cell count in a multivariate analysis.
To the best of our knowledge, this is the largest study of S. aureus bacteraemia to date in a US HIV-infected population. This study has several important findings. First, MRSA represented a large and growing proportion of S. aureus bacteraemias among our urban cohort of HIV-infected patients. Secondly, patients with MRSA bacteraemia were more likely to have low CD4 cell count, IDU exposure, and ESRD than were bacteraemia-free controls. Compared with patients with MSSA bacteraemia, those with MRSA bacteraemia were more likely to have ESRD and showed a trend towards a greater level of immunocompromise as measured by CD4 cell count.
There was a significant increase in the incidence of MRSA bacteraemia as well as all S. aureus bacteraemia over the study period from 2000 to 2004. This trend is similar to that observed for clinically significant MRSA infections in another urban HIV cohort over a similar study period (2000-2003) . However, our study focuses exclusively on MRSA bacteraemia and yields a large number of cases with this diagnosis. The increased incidence of all S. aureus bacteraemia was an unexpected finding that reflects trends observed in the general population over the past 20-40 years [13-15].
Our study builds on previous investigations of the epidemiology of MRSA infections in HIV-infected patients, adding evidence that low CD4 cell count, IDU and ESRD are important risk factors for severe MRSA infections [2,16-19]. IDU has been widely reported to be associated with S. aureus bacteraemia and intravascular infections in HIV-infected patients . This association may partly be explained by the higher rates of MRSA nasal colonization in IDUs [20,21]. Future studies will need to evaluate the association of MRSA colonization and bacteraemia in this population as nasal swabs were not available for our retrospective analysis.
Haemodialysis is also a recognized risk factor for bacteraemia, including MRSA, in the general population [22,23]. Other studies have demonstrated increased mortality and higher healthcare costs associated with MRSA bacteraemia compared with MSSA bacteraemia in haemodialysis patients . Our study adds to the literature by demonstrating that, in an HIV-infected population that is largely IDU, ESRD is an even stronger predictor of MRSA bacteraemia than IDU.
The finding of increased risk for MRSA infections in patients with advanced HIV disease as measured by CD4 cell count is consistent with previous research findings . The observed increased risk may reflect increased healthcare exposure in patients with more advanced HIV disease. Unlike previous studies, our results did not show a protective effect of PCP prophylaxis .
By comparing MRSA bacteraemia with MSSA bacteraemia cases, our study further delineates the risks for methicillin resistance among S. aureus bacteraemias in HIV-infected patients. We found that patients with MRSA bacteraemia were more likely than patients with MSSA bacteraemia to have ESRD and showed a trend towards greater immunosuppression as measured by CD4 cell count. Of note, IDU was not associated with MRSA bacteraemia when compared with MSSA bacteraemia after adjusting for ESRD and CD4 cell count.
There are several potential limitations to our study. First, our results are based on patients from a single institution with a high proportion of indigent patients and IDUs. While our institution is a tertiary referral centre and may not be reflective of all HIV clinics, most patients use our HIV clinic for primary care. Therefore, our findings may be generalizable to other urban HIV primary care sites. Secondly, we may have missed hospitalizations at outside hospitals, thereby underestimating the incidence of S. aureus bacteraemia. However, a recent analysis of Medicaid claims indicates that 96% of all hospitalizations among the cohort were collected in our database. Finally, although we evaluated ESRD as a risk factor for bacteraemia, we were unable to evaluate the presence of an indwelling catheter, as this variable was not regularly recorded in the medical record. Future studies will need to investigate the differences between indwelling catheters and permanent vascular access in this population.
Because of the retrospective nature of this study, we were unable to assess the molecular characteristics of isolates. Therefore, we were unable to conclude whether MRSA strains were community or nosocomially acquired. However, a small, randomly selected subset of blood specimens were available for Panton-Valentine leukocidin (PVL) analysis. Of those, 55% (five out of nine) were noted to be PVL positive. Further investigation of patterns of susceptibility to OI prophylactic medications may thus be warranted.
Data were not available on previous hospitalizations, previous administration of ß-lactams, or indwelling catheters, clinical factors demonstrated to predict methicillin resistance in other studies of clinically significant S. aureus infections [3,23]. Thus, while not conclusively demonstrating the source of MRSA bacteraemia, our study does indicate important sociodemographic (e.g. IDU), immunological (e.g. low CD4 cell count) and clinical (e.g. ESRD) characteristics of patients at risk for MRSA bacteraemia. These factors may serve as important clinical markers of MRSA likelihood in decisions regarding initial antimicrobial management.
In summary, we found that the incidence of MRSA bacteraemia increased significantly in HIV-infected patients from 2000 to 2004 in this large urban HIV out-patient clinic. IDU, ESRD, and low CD4 cell count were independent predictors for incident MRSA bacteraemia, while ESRD was more common in patients with MRSA bacteraemia than MSSA bacteraemia. Initial antimicrobial therapy for presumed sepsis in HIV-infected patients may require agents active against MRSA, including vancomycin, linezolid and, in nonpulmonary infections, daptomycin, particularly in patients with risk factors for MRSA bacteraemia.