Background
Piperacillin/tazobactam or meropenem are often used to treat patients with severe bacterial infections. We aimed to compare the desirable and undesirable effects of empirical and/or definitive piperacillin/tazobactam versus carbapenems in patients with severe bacterial infections.

Methods
We searched PubMed, Embase, CENTRAL, Epistemonikos, and trial registers for randomised clinical trials of empirical and/or definitive piperacillin/tazobactam versus carbapenems in adult patients with severe bacterial infection (i.e., any bacterial infection requiring hospitalisation). The primary outcome was all-cause short-term mortality within 90 days. Secondary outcomes were all-cause long-term mortality, adverse events, quality of life, days alive without or duration of life support, secondary infections, selection of fungi or resistant bacteria, and days alive and out of hospital or hospital length of stay. We calculated relative risks (RRs) using random effects and fixed effect meta-analyses along with trial sequential analyses.

Results
We included 31 trials (n = 8790 patients) with overall high risk of bias. The RR for all-cause short-term mortality was 1.16 (95% confidence interval [CI]: 0.94–1.43, low certainty evidence), for adverse events 1.00 (98% CI: 0.96–1.04, moderate certainty evidence), for secondary infections 1.13 (98% CI: 0.76–1.68, very low certainty evidence), and for selection of fungi or resistant bacteria 1.61 (98% CI: 0.89–2.89, very low certainty evidence). There were no or limited data for the remaining outcomes.

Conclusions
Based on very low or low certainty evidence, piperacillin/tazobactam may be associated with less favourable outcomes in patients with severe bacterial infections as compared with carbapenems, but the information size for a robust conclusion has not been reached.