Shachaf Shiber, D. Yahav, T. Avni
The Journal of antimicrobial chemotherapy
BACKGROUND Data on the relative efficacy of β-lactam/β-lactamase inhibitors (BL/BLIs) versus carbapenems are scant. METHODS This is a systematic review and meta-analysis of randomized controlled trials (RCTs) comparing any BL/BLI versus any carbapenem for the treatment of sepsis. The primary outcome was all-cause mortality. A broad search was conducted with no restrictions on language, publication status or date. Two reviewers independently applied the inclusion criteria and extracted the data. Assessment of risk of bias was performed using the domain-based approach. Subgroup analyses were used to investigate heterogeneity and focus on patient groups more likely to harbour ESBL-positive bacteria. Risk ratios (RRs) with 95% CIs were calculated and pooled. RESULTS Thirty-one RCTs were included. There was no difference between BL/BLIs and carbapenems in terms of mortality (RR 0.98, 95% CI 0.79-1.20), without heterogeneity. No differences were observed with regard to clinical or microbiological failure and bacterial superinfections. The results were not affected by risk of bias. No differences were detected in the subgroups of patients with nosocomial infections, Gram-negative infections and neutropenic fever. Adverse events requiring discontinuation were more common with BL/BLIs, on account of an increased incidence of diarrhoea. However, Clostridium difficile-associated diarrhoea (RR 0.29, 95% CI 0.10-0.87) was more frequent with carbapenems and seizures were more frequent with imipenem (RR 0.21, 95% CI 0.05-0.93). CONCLUSIONS No differences in efficacy between BL/BLIs and carbapenems exist in RCTs including patient populations with a certain, albeit unknown, rate of ESBL-positive bacteria causing infections.