A randomised trial published in The Lancet reports that benzylpenicillin significantly reduces acute kidney injury compared with flucloxacillin or cloxacillin in patients with penicillin-susceptible Staphylococcus aureus (PSSA) bacteraemia — though the trial’s primary non-inferiority endpoint on 90-day mortality was not formally met, according to results of the SNAP trial (ClinicalTrials.gov NCT05137119, DOI 10.1016/S0140-6736(26)00761-0, PMID 42309115).
The trial design
The SNAP (Staphylococcal Non-Antistaphylococcal Penicillin) trial was a multicentre randomised controlled trial comparing benzylpenicillin (penicillin G) with flucloxacillin in adults hospitalised with confirmed PSSA bacteraemia — organisms known to be fully susceptible to penicillinase-stable and narrow-spectrum penicillins. Standard-of-care in most of Europe and Australia for MSSA bacteraemia is flucloxacillin; benzylpenicillin is narrower-spectrum and has superior pharmacodynamic properties for susceptible organisms.
The primary endpoint was 90-day all-cause mortality, analysed as a non-inferiority test with a margin pre-specified in the protocol. The Bayesian posterior probability of non-inferiority reached 96.1%, which the investigators noted as providing strong evidence in favour of equivalence. However, the trial did not formally meet its pre-specified non-inferiority criterion: the primary Bayesian analysis found the upper bound of the 95% credible interval for the adjusted odds ratio (1.28) exceeded the pre-specified non-inferiority margin of OR < 1.20, so the primary endpoint was technically not met. The SNAP trial used a Bayesian — not frequentist — primary analysis throughout.
AKI reduced by half
The secondary endpoint of acute kidney injury told a different story. Patients in the benzylpenicillin arm experienced significantly fewer AKI events — approximately half the rate observed in the flucloxacillin or cloxacillin arm (cloxacillin was used in Canada, Israel, Singapore, and South Africa). This difference is biologically plausible: flucloxacillin carries a recognised nephrotoxicity risk, particularly with high-dose prolonged courses such as those used in deep-seated bacteraemia, whereas benzylpenicillin does not share this toxicity profile.
AKI in bacteraemia is not a soft endpoint. It prolongs hospital stay, worsens prognosis in patients already critically ill, and contributes to long-term renal decline. A halving of AKI events is a clinically meaningful secondary finding.
Implications for practice
The SNAP result puts antimicrobial stewardship in tension with clinical conservatism. Flucloxacillin has been the unchallenged first-line agent for MSSA bacteraemia in the UK and much of Europe for decades. A trial showing comparable 90-day mortality and substantially lower nephrotoxicity with the narrower-spectrum benzylpenicillin will prompt revision of clinical guidelines, though regulatory bodies and societies typically wait for formal non-inferiority confirmation before updating standards of care.
The 96.1% Bayesian posterior probability of non-inferiority is, to Bayesian practitioners, a strong statement. Whether that is sufficient for guideline change will be a matter for infectious disease societies; the trial provides the evidence base for that conversation.
The Lancet. 2026. DOI: 10.1016/S0140-6736(26)00761-0. PMID: 42309115.
Correction (2026-06-19): Three errors corrected by post-publication fact-check. (1) The trial used a Bayesian — not frequentist — primary analysis; the non-inferiority criterion (OR < 1.20) was not met because the upper 95% credible interval (1.28) exceeded the margin, not because a frequentist p-value threshold was missed. (2) The comparator arm was “flucloxacillin or cloxacillin” (cloxacillin was used in Canada, Israel, Singapore, and South Africa), not flucloxacillin alone; the article has been updated throughout. (3) The NCT identifier “NCT-SNAP” is not a valid registry number; the correct ClinicalTrials.gov identifier is NCT05137119.