A widely used hospital antibiotic is tied to a higher risk of dying, a review of 110 trials found

A systematic review and Bayesian meta-analysis published in JAMA Network Open has tied cefepime, a broad-spectrum antibiotic infused intravenously in hospitals worldwide, to a higher risk of death than other antibiotics in its class. The analysis, led by Zahra N. Sohani of the Division of Infectious Diseases and Medical Microbiology at Hopital Maisonneuve-Rosemont in Montreal, pooled results from 110 randomized clinical trials covering more than 22,000 patients who received either cefepime or another beta-lactam antibiotic.

Across those trials, 778 deaths occurred among 11,726 patients given cefepime, a rate of 6.6%, compared with 6.2% among patients given a comparator beta-lactam. Run through a Bayesian model, that gap translated into a 94.4% probability that cefepime carries higher all-cause mortality than the alternatives it was tested against.

A meta-analysis built from 110 randomized trials

Sohani’s team pulled together outcomes from 110 separate randomized trials rather than relying on any single study, an approach designed to smooth out the noise that comes from small trials run in different hospitals, countries and patient populations. Cefepime is a fourth-generation cephalosporin, prized in intensive care and oncology wards for covering a wide range of gram-negative bacteria, including some resistant strains that narrower antibiotics miss.

Cephalosporins belong to the broader beta-lactam family, the most widely prescribed class of antibiotics in hospitals, alongside penicillins and carbapenems. All three work the same basic way, by disrupting the construction of bacterial cell walls, which is why researchers treated other beta-lactams as the natural comparison group rather than antibiotics from an unrelated class.

Rather than testing for a single yes-or-no significance threshold, the researchers used a Bayesian meta-analysis, a method that produces a probability instead of a pass-fail result. Across all 110 trials, that probability came out to 94.4%. When the analysis was narrowed to the 73 trials that had gone through full peer review and publication, the probability climbed further, to 98.6%.

Six deaths for every hundred patients treated

The raw numbers behind that probability are modest on their face: 6.6% of cefepime patients died during the trials, against 6.2% of patients on other beta-lactams, a difference of four-tenths of a percentage point. Spread across more than 11,000 patients per group, though, that gap adds up to dozens of additional deaths, which is what gave the Bayesian model enough signal to flag it with high confidence.

The association was not uniform across every patient. It showed up more strongly in adults than in children, and it was most pronounced among patients being treated for febrile neutropenia, a dangerous combination of fever and low white blood cell counts that often follows cancer chemotherapy. Those are exactly the patients for whom cefepime is frequently a first-line choice, which sharpens the stakes of the finding rather than softening them.

A famously narrow dosing window

Researchers involved in the analysis pointed to cefepime’s pharmacology, not necessarily the drug’s basic chemistry, as a likely explanation. Cefepime has a comparatively narrow therapeutic window, meaning the margin between a dose that clears an infection and one that turns toxic is smaller than for many other antibiotics. That narrow margin cuts both ways: underdosing can leave a serious infection undertreated, while overdosing raises the risk of neurotoxicity, including confusion and seizures, particularly in patients whose kidneys are already struggling to clear the drug.

That dosing-based explanation reframes the finding as a question of how cefepime is administered rather than whether it should be used at all. The study itself did not test whether tighter kidney-function-based dosing or closer neurotoxicity monitoring would close the mortality gap, leaving that as an open question for future trials rather than a settled answer.

Kidney function complicates the picture further because it changes with age, illness and hydration, all of which shift how quickly a patient clears the drug. A dose calculated for one point in a hospital stay may no longer fit days later if kidney function has declined, which is part of why intensive-care and oncology patients, already among the sickest people in a hospital, may be more exposed to the risk the meta-analysis flagged.

Hospitals face a decision before guidelines catch up

Cefepime remains widely used for exactly what it treats well: hospital-acquired pneumonia, febrile neutropenia and other serious gram-negative infections where few narrower-spectrum antibiotics offer comparable coverage. An abrupt shift away from it, absent a proven substitute for those same infections, could leave some patients undertreated while institutions wait for clearer guidance.

Other researchers who reviewed the findings alongside publication argued that the mortality signal likely reflects a dosing problem rather than a reason to abandon the drug outright, echoing the pharmacology-based explanation Sohani’s team raised. For now, prescribers are left weighing a statistically real, Bayesian-confirmed risk against the absence of an equally effective, clearly safer alternative for the infections cefepime is chosen to treat.

This article was produced with the assistance of AI and reviewed by Morning Overview editors prior to publication.

More from Morning Overview