Impact of Empiric Carbapenem Beta-Lactam (CBL) versus Non-Carbapenem Beta-Lactam (N-CBL) Treatment on Mortality and Microbiological Response Rates for Third-Generation Cephalosporin-Resistant (3GCR) Enterobacteriaceae Bloodstream Infections (BSI)

Background: Carbapenem β-lactams (CBLs) have been the class of choice for third generation cephalosporin-resistant (3GCR) Enterobacteriaceae infections, but emergence of carbapenemase-producing bacteria highlight the need to limit their use to prevent resistance. Current literature shows mixed evidence regarding the effectiveness of carbapenem-sparing β-lactam regimens. This study aimed to compare 30-day mortality and microbiological response for patients with 3GCR bloodstream infections (BSI) when treated empirically with CBLs or non-carbapenem β-lactams (N-CBLs; cefepime [FEP] or piperacillin/tazobactam [TZP]).

Methods: This retrospective study included patients with 3GCR Enterobacteriaceae BSI from 1/2009-9/2015 at two Boston teaching hospitals. Patients were included if 1) they had ≥1 blood culture positive for Escherichia coli, Enterobacter spp., and/or Klebsiella spp., 2) their culture was resistant to 3GC and susceptible to both CBLs and N-CBLs (either FEP or TZP), and 3) they received ≥ 48 hours of CBL or N-CBL therapy within 72 hours of the index culture. Patients with polymicrobial BSI or a BSI within the previous 30 days were excluded. The primary outcome was 30-day all-cause hospital mortality. Secondary outcomes included microbiological response, duration of BSI, and BSI length of stay (LOS; index date to discharge).

Results: Of 125 included patients, 90 received CBL and 35 received N-CBL (15 FEP; 20 TZP). Baseline characteristics of the groups were similar; median patient age and APACHE II scores were 66.2 years and 14.6, respectively. After 48 hours of treatment, a switch in therapy from empiric N-CBL to CBL was common (54.3%). There was no difference in 30-day mortality between CBL and N-CBL (8.9% vs. 14.3%; p =0.51) groups, respectively. Microbiological response rates were similar between groups (approx. 94% eradication) as well as duration of BSI (approx. 2 ± 1 days) and BSI LOS (11 ± 16 days [CBL] vs. 12 ± 22 [N-CBL]; p = 0.79).

Conclusion: No differences in 30-day mortality nor microbiological response were revealed among this cohort of 125 patients with 3GCR BSI treated empirically with CBL or N-CBL (FEP or TZP). While larger studies should be conducted to validate findings, N-CBLs may be an effective CBL-sparing strategy.