Clinical Outcomes for Patients with Bacteremia Caused by Vancomycin-Resistant Enterococcus in a Level 1 Trauma Center

MAJOR ARTICLE Clinical Outcomes for Patients with Bacteremia Caused by Vancomycin-Resistant Enterococcus in a Level 1 Trauma Center Anti-Infective Research Laboratory, Detroit Receiving Hospital, Wayne State University, Detroit, Michigan To assess the degree of morbidity and mortality attributable to vancomycin resistance in enterococcal bacteremia (EB), a matched case-control study was conducted. Patients with bacteremia due to vancomycinresistant enterococcus (VRE) were matched to control patients with bacteremia due to vancomycin-susceptible enterococcus. During 1996–2000, 65 patients with cases of clinically significant VRE bacteremia were identified, and 53 of these patients were successfully matched. In this group of patients, VRE bacteremia was found to be an independent predictor of crude mortality (odds ratio [OR], 4.0; 95% confidence interval [CI], 1.2–13.3) and the infection-related mortality rate (OR, 5.2; 95% CI, 1.4–20.0). It was also an independent predictor of the rate of clinical failure at day 7 after the onset of EB (OR, 4.6; 95% CI, 1.2–17.3) and overall clinical failure (OR, 4.3; 95% CI, 1.3–14.5) and was associated with a longer mean length of hospitalization after the onset of EB, compared with that for control patients (22.7  1.88 vs. 15.9  1.7 , P p .006). These observations indicate that vancomycin resistance in EB independently affects outcomes. Infections due to vancomycin-resistant enterococcus (VRE) have become the focus of increased attention during the past decade. The first enterococcal strains with glycopeptide resistance were reported from France in 1986 [1]. The Centers for Disease Control and Prevention (CDC) reported that, by 1993, there had been a 20-fold increase in the percentage of nosocomial isolates of enterococci resistant to vancomycin, with greater increases in isolates recovered from intensive Received 21 August 2001; revised 7 November 2001; electronically published 4 March 2002. Presented in part: 40th Interscience Conference on Antimicrobial Agents and Chemotherapy, Toronto, Ontario, Canada, September 2000. a Present affiliation: Division of Clinical Pharmacology, Albany Medical College, Albany, New York. Reprints or correspondence: Dr. Peggy S. McKinnon, Dept. of Pharmacy Services and Anti-Infective Research Laboratory, Detroit Receiving Hospital, Wayne State University, 4201 St. Antoine Blvd., 1-B UHC, Detroit, MI, 48201 (pmckinno@dmc .org). Clinical Infectious Diseases 2002; 34:922–9  2002 by the Infectious Diseases Society of America. All rights reserved. 1058-4838/2002/3407-0006$03.00 922 • CID 2002:34 (1 April) • Lodise et al. care units (ICUs) [2]. In 1998, the SCOPE study demonstrated an overall vancomycin resistance rate of ∼14%, and reported that the rate of vancomycin resistance among strains of Enterococcus faecium in the northeast section of the United States was 63% [3]. The lack of viable treatment options for VRE infection has contributed to the growing concern about this organism. The development of vancomycin resistance in this organism further diminishes the treatment options, which were already limited because of other intrinsic and acquired forms of resistance in enterococcus. Despite the dearth of therapeutic options, the degree of morbidity and mortality attributable to vancomycin resistance in enterococcal infections remains controversial. Although most studies that have compared VRE infections and vancomycin-susceptible enterococcus (VSE) infections have demonstrated that a higher crude mortality rate is associated with the former, it is unclear whether infection with a vancomycinresistant strain is an independent predictor of mortality [4–15]. Thomas P. Lodise, Peggy S. McKinnon, Vincent H. Tam,a and Michael J. Rybak PATIENTS AND METHODS Setting We performed the study at Detroit Receiving Hospital (DRH), which is a 279-bed level 1 trauma center in Detroit, Michigan. DRH is part of the Detroit Medical Center and is a major teaching facility for the health sciences at Wayne State University. Medical specialties include trauma, critical care, surgery, cardiology, neurology, and internal medicine. The hospital is the regional burn center and has a spinal-cord injury unit. Study Population The study focused on episodes of EB identified from 1996 through 2000. Only episodes of EB that met the CDC criteria for bloodstream infection were included in the analysis [17]. When a patient had 11 episode of EB, only the first episode was considered. Study Design To evaluate the effect of vancomycin resistance on the clinical and microbiologic outcomes associated with EB, a retrospective matched case-control study was performed. Patients with VRE bacteremia were designated “case patients” and patients with VSE bacteremia were designated “control patients.” Case and control patients were matched in 1:1 ratio with use of a stepwise method similar to one described elsewhere [18–21]. Matching criteria included the following: age; Acute Physiology and Chronic Health Evaluation (APACHE) II score [22] at the onset of EB; admitting service (i.e., medicine vs. surgery); hospital unit at the onset of EB (ICU vs. non-ICU); and length of hospital stay prior to the onset of EB (stratified into 4 time categories, as follows: ⭐3, 4–8, 9–28, and 128 days). Matching Process To select the most suitable control patient for every patient with VRE bacteremia, a 15-point scoring system was used, similar to that described elsewhere [18–21]. The matching score was calculated in the following manner: for age, 4 points were given if the age difference was 5 years, and 2 points were given if the age difference was 10 years; for APACHE II score at the onset of EB, 4 points were given if the difference in the APACHE II scores was 4, and 2 points were given if the difference in the APACHE II scores was 8; for the admitting service (i.e., medicine vs. surgery), 2 points were given in case of concordance; for the hospital unit at the onset of EB, 2 points were given in case of concordance; and, for the length of hospital stay prior to the onset of EB, 3 points were given if the duration was in the same time category. A score of 12 was the minimum accepted score for a control patient. Control patients were selected without knowledge of patients’ outcome status. Data Collection Clinical data. Data for each patient were extracted from medical records. Data obtained included the following: age, sex, admitting service (i.e., medicine vs. surgery), ward of admission (i.e., ICU vs. non-ICU), comorbidities, receipt of prior antibiotic therapy, origin of infection (i.e., community-acquired or hospital-acquired), length of hospitalization prior to the onset of EB (total and in the ICU), hospital unit at the onset of EB (ICU vs. non-ICU), presence of a central venous or urinary catheter, receipt of total parenteral nutrition or mechanical ventilation at the onset of EB, source of bacteremia, and severity of illness at the (...truncated)