Predictors of Methicillin Resistance in Healthcare-Associated Staphylococcus aureus Bloodstream Infections: The Role of Recent Antibiotic Use

Healthcare-associated bloodstream infections (HCA-Bsi) due to methicillin-resistant Staphylococcus aureus (MRsA) have been shown in most but not all studies, to be associated with increased mortality, longer length of hospital stay and higher cost to the health care system than HCA Bsis due to methicillin-susceptible Staphylococcus aureus (MssA) 1,2,3,4. A recent metaanalysis found an increased risk of 1.8 of acquiring MRsA in patients who had taken antibiotics, while previous antibiotic treatment has been associated in some studies with HCA Bsi by MRsA 5,6,7. A prediction model, aimed at identifying MRsA bacteremia, predicted that 80% of the nosocomial bacteremias would be resistant to methicillin if patients had experienced antimicrobials prior to the onset 8. The quantification of the association between MRsA infections and previous antibiotic therapy, especially regarding Bsis, may merit further exploration. Few studies have assessed quantitative antibiotic exposure, using defined daily doses (DDDs), but they have not focused especially on bloodstream infections 9,10. The aim of our study was to assess predictors of HCA Bsi by MRsA in comparison with cases of HCA Bsi by MssA, with a special focus on recent antibiotic use, both in a qualitative and a quantitative way. The Lakeside VA Medical Center (LsVAMC) was a 280-bed, tertiary-care, urban teaching hospital with approximately 5,000 admissions per year. Computerized medical records for all patients admitted and followed as outpatients to LsVAMC were available from 1991 and thereafter. From the electronic pharmacy notes we could access data on inpatient and outpatient intravenous antibiotic administration. Our study protocol was approved by the northwestern University institutional Review Board. The time-period for our study was between October 1997 and september 2001. We retrospectively identified two groups of inpatients, one group with a new positive MRsA blood culture and another group with a new positive MssA blood culture. Patients with a prior history of any type of positive MRsA or MssA culture up to 1991 were excluded. We considered the positive blood culture to be healthcare associated if 1) it was taken more than 48 hours after the current admission date and 2) if it was taken less than 48 hours after the current admission date but the patient had been hospitalized within the last month before the culture date. We used the computerized medical records of LsVAMC to obtain data pertinent to known risk factors for staphylococcal colonization and infection for our two study groups. We arbitrarily considered events up to 30 days before the positive culture date. We used the McCabe and Jackson classification system to assess the severity of underlying disease(s) and the Horn severity of illness index for the 2 groups on the day of the positive culture 11,12. Finally, we collected data about duration of hospital and MsiCU and CCU stay for the 2 study groups until their discharge or death and we compared mortality rates during the current hospital stay. We used Version 13 of sPss software for data processing, student’s t-test for comparison of continuous variables and Fisher’s exact test for categorical variables. Logistic regression analysis (backward stepwise) was used to assess independent predictors of MRsA bloodstream infection and Kaplan-Meier analysis for survival comparisons. A p value of 0.05 or less was considered clinically significant. Based on predefined inclusion and exclusion criteria, we identified 28 patients with MRsA HCA-Bsi and 32 patients with MssA HCA-Bsi eligible for further analysis. Characteristics significantly different between the 2 groups are shown together in table 1 while those not differing significantly are shown in table 2. Univariate analysis indicated significant differences in age, presence of chronic wounds and history of hemodialysis, duration of hospital stay and duration of MsiCU/CCU stay prior to the positive culture, intubation, surgical procedures and presence of indwelling urinary catheter for more than 24 hours. Regarding antibiotic indices, the MRsA group was characterized by a higher mean number of antibiotics used and mean number of antibiotic-days per patient, qualitative and quantitative use of penicillins, b-lactams and fluoroquinolones, qualitative use of aminoglycosides, and quantitative use of b-lactam/b-lactamase inhibitors and antibiotics as a whole. Factors significantly different between the 2 groups were used for logistic regression analysis. Two logistic regression models were carried out to assess the role of antimicrobial usage, using either qualitative or quantitative indices of antibiotic consumption (see table 1). From model i, a trend was noted for duration of prior hospital stay [OR 1.146, 95% Ci 0.98-1.341, p = 0.088), while, from model ii, mean antibiotic-days per patient emerged as the single independent predictor of HCA-Bsi by MRsA [OR 1.318, 95% Ci 1.027-1.692, p = 0.03]. no significant difference in mortality was noted by Kaplan-Meier analysis (data not shown). Patients of the MRsA group were older than the MssA patients and had clearly spent more days in-hospital and more days in the MsiCU/CCU before the culture date, features also found in other studies 10,13. Additionally, our MRsA group seemed to consist of a preponderance of “surgical” patients (higher rates of surgery and elective endotracheal intubation), while the MssA group of mostly “internal medicine” patients, with several patients receiving hemodialysis. Regarding antibiotic use, the MRsA group was characterized by significantly heavier antibiotic use overall and for individual classes by all types of analysis we performed (table 1). Most available studies of MRsA versus MssA Bsis are in agreement with our findings 8,10,13,14. Qualitative indices did not independently predict MRsA HCA Bsi in our study, in agreement with the study by McHugh et al, while Lodise et al did docu1 Evangelismos General Hospital, Athens, Greece, 2 Alexandra University Hospital, Athens, Greece, 3 University Hospital of Patras, Patra, Greece, 4 Loyola University Medical Center, Chicago, illinois.