Comparison of outcomes in patients with methicillin-susceptible Staphylococcus aureus (MSSA) bacteremia who are treated with β-lactam vs vancomycin empiric therapy: a retrospective cohort study

We performed a retrospective cohort study of adult inpatients diagnosed with their first episode of MSSA bacteremia at two tertiary care hospitals in Vancouver, Canada, between January 2007 and December 2014, inclusive. Consecutive patients were included if they had MSSA bacteremia and either cloxacillin or cefazolin was prescribed for definitive therapy (penicillin was an acceptable alternative if the isolate was proven to be susceptible). Patients were excluded if there was missing data for 28-day mortality, no empiric therapy was administered, death occurred within 24 h following diagnosis of bacteremia, or polymicrobial bacteremia. Patients were stratified based on empiric treatment with β-lactams or vancomycin. The β-lactam group received one or more of cloxacillin, cefazolin, β-lactam/β-lactamase inhibitor combination, a third generation cephalosporin or a carbapenem, with or without vancomycin. In both groups, other antimicrobials may have been prescribed during empiric and definitive therapy.

Bacteremia was defined as the isolation of MSSA from one or more blood culture bottles. Bacteremia identified within 72 h of hospital admission was considered community-onset, while bacteremia diagnosed after more than 72 h of hospital admission was deemed hospital-onset. Immunocompromised state was present if any of the following were described: neutropenia (≤ 1.5 × 10⁹/L), congenital immune deficiencies, or use of immunosuppressants (TNF-α inhibitors, prednisone ≥ 10 mg/day or its equivalent, cancer chemotherapy, methotrexate, cyclophosphamide, mycophenolate mofetil, calcineurin inhibitors, mTOR inhibitors, azathioprine and any other drug generally considered to significantly weaken the immune system). Definite infective endocarditis was diagnosed using the modified Duke criteria [19]. The source of bacteremia was either stated explicitly or inferred as the most likely source based on available clinical data and microbiological results. Metastatic complications included infections that occurred distant from the presumed primary source such as septic emboli, mycotic aneurysms, osteoarticular infections, and distant abscesses. Surgical source control included only procedures performed in the operating theatre. Empiric therapy began with the first dose of empiric antibiotics and ended with the start of definitive therapy. Definitive therapy began when antimicrobial susceptibilities were known and one of the following treatments was prescribed: 1) cloxacillin or cefazolin 2) discontinuation of other empiric antibiotics for patients already on cloxacillin or cefazolin empirically, or 3) continuation of empiric cloxacillin or cefazolin. Definitive therapy ended when cloxacillin or cefazolin was stopped. Time to receipt of antibiotics was measured from the time of obtaining the first positive blood culture to the time of the first dose of antibiotic. If a patient was already on antibiotics at the time of the first positive blood culture, the time to receipt of antibiotics was zero. We calculated the hours of empiric β-lactam exposure based on start and stop dates, start and stop times, and dosing frequency.

Our primary outcome was 28-day all-cause in-hospital mortality. Secondary outcomes were 90-day all-cause in-hospital mortality, recurrent infection at 6 months, duration of bacteremia and hospital length-of-stay (LOS). Time to mortality was measured from the date of the first positive blood culture to the date of death. Recurrent infection occurred when MSSA bacteremia was diagnosed again following completion of a treatment course for the previous episode of MSSA bacteremia. Duration of bacteremia was the time difference between the first positive blood culture and the first negative blood culture. Patients without follow-up blood cultures were excluded from the analysis for duration of bacteremia. Hospital LOS was measured from the date of admission to the date of discharge.

Patients with MSSA bacteremia were extracted from the medical microbiology laboratory information systems and medical records were reviewed. A single reviewer collected data on patient demographics and comorbidities, blood culture results and antimicrobial therapy from electronic and paper databases.

Our predicted mortality difference between the β-lactam and vancomycin group was 15 % based on a previous study [17]. We estimated a sample size of 200 for each group to capture a 15 % difference in mortality with 80 % power at a two-tailed alpha level of 0.05. Baseline categorical variables were described as counts and percentages, and differences between groups were assessed with chi-square or Fisher’s exact tests. Continuous variables were presented as means and standard deviations, or medians and interquartile range. Differences between groups were assessed using parametric t-tests or non-parametric Mann–Whitney-U tests, as appropriate. Logistic regression methods were used to model the odds ratio of death and recurrent infection in the β-lactam compared to vancomycin group. Linear regression model was conducted for hospital LOS and duration of bacteremia. The two outcomes were log-transformed in the analysis to improve normality of the distribution of residuals. All models were adjusted for pre-specified confounding variables including age, sex, age-adjusted Charlson-comorbidity index [20], Pitt bacteremia score, infectious diseases consultation, infective endocarditis and time to receipt of empiric antibiotics. These factors have been shown to affect mortality in patients with SAB [1]. The duration of bacteremia was further adjusted for surgical source control. All analyses were performed using the SAS 9.4 software.