Review reports slightly higher mortality with cefepime than other antibiotics
A large analysis of randomized trials found a possible difference in all-cause mortality, but the overall estimate remained uncertain.
High evidenceReviewInterpret with caution
Medical disclaimer: This article summarizes research findings and is for informational purposes only. It is not medical advice.
Editorial illustration — not from the study.
The analysis examined randomized trials comparing cefepime with other beta-lactam antibiotics for infections, including situations such as febrile neutropenia and gram-negative infections. The researchers combined the results using a Bayesian random-effects model and focused on death within 30 days, or the closest reported time point.
In the full analysis, 6.6% of patients receiving cefepime died compared with 6.2% receiving another beta-lactam. The estimated odds ratio was 1.10, with a 95% credible interval of 0.98 to 1.24. A separate analysis limited to published peer-reviewed trials estimated higher odds of death with cefepime, but the findings still require cautious interpretation and do not by themselves establish that cefepime caused the deaths.
The question reviewed
Researchers conducted a systematic review and Bayesian random-effects meta-analysis of randomized clinical trials available through May 25, 2026. The trials included children and adults and compared cefepime alone with another beta-lactam alone, or compared cefepime combined with another drug against another beta-lactam combined with that same drug. Prophylactic cefepime and cefepime combinations containing a beta-lactamase inhibitor were excluded. The main outcome was all-cause mortality at 30 days or the closest reported time point. In total, 110 trials involving 22,608 patients were included; 11,726 received cefepime and 10,882 received a comparator beta-lactam.
What the review concluded
Across all 110 trials, 778 of 11,726 patients receiving cefepime died, compared with 674 of 10,882 patients receiving another beta-lactam. The pooled odds ratio for mortality was 1.10, with a 95% credible interval of 0.98 to 1.24; the model estimated a 94.4% probability that the odds ratio was above 1. In the analysis of 73 published peer-reviewed trials involving 15,411 patients, the estimated odds ratio was 1.17, with a 95% credible interval of 1.02 to 1.34, and the model estimated a 98.6% probability of higher mortality with cefepime. The researchers reported that results generally favored higher odds of mortality with cefepime across comparator antibiotics, clinical indications, and doses of 2 grams or more every 12 hours. These are pooled comparisons from trials and do not, by themselves, prove that cefepime caused the deaths.
Who this is relevant to
These findings are most directly relevant to children and adults enrolled in randomized trials comparing cefepime with other beta-lactam antibiotics for the types of infections studied. They may not apply equally to people with different infections, illness severity, treatments, or clinical circumstances. The results are population-level comparisons and should not be interpreted as predicting what will happen to any particular patient.
What this could mean
Cefepime is used for certain serious infections, so possible differences in mortality are relevant to discussions about its safety. The findings suggest that cefepime safety may warrant careful consideration in clinical guidance, but they do not show that the drug is responsible for an individual death or establish that the same result applies to every patient, infection, dose, or comparator antibiotic.
Limitations & evidence assessment
The overall credible interval ranged from a possible small reduction to a possible increase in mortality, so the main pooled estimate did not clearly exclude no difference. The review combined trials involving different ages, infections, treatments, comparator antibiotics, and study conditions; the abstract does not report enough detail to judge how these differences affected the pooled result. The abstract also does not provide the overall risk-of-bias findings, the number of participants lost to follow-up, or the duration of follow-up beyond the use of 30 days or the closest reported time point. The separate analysis of published peer-reviewed trials produced a stronger association, but restricting analyses to published studies can introduce publication-related bias. As with any meta-analysis, the conclusions depend on the quality and comparability of the included trials and on the statistical model used.
Why this evidence level: Meta-analysis pooling multiple studies sits at the top of common evidence hierarchies.
Evidence levels are editorial estimates derived from study metadata — they are not clinical appraisals.
// Source
JAMA Network Open · 2026 · DOI: 10.1001/jamanetworkopen.2026.33017
This systematic review and meta-analysis examined whether ultra-low-dose computed tomography (CT) can detect pulmonary, or lung, nodules while using less radiation than standard-dose CT. Across 42 studies involving 15,792 patients, pooled sensitivity for nodule detection was 91.9%, but results varied greatly between studies. The review also found that ultra-low-dose CT used an average effective dose of 0.22 millisieverts (mSv), while detection was less reliable for some small or hazy nodules and in patients with a high body mass index.
In an online randomized experiment involving 3,519 Chilean parents, researchers tested whether adding an “ULTRAPROCESSED” warning to Chile’s existing nutrient labels changed views of potato chips. The added label improved identification of ultraprocessed products and lowered reported purchase intentions and perceived healthfulness, but the effects were smaller when products already had more nutrient warnings. The study measured stated responses, not actual purchases or health outcomes.
A systematic review and network meta-analysis examined nonhormonal prescription medicines for hot flashes and related symptoms among adult women with breast cancer or a history of breast cancer. Across 19 study or substudy records involving 2,224 participants, five medicines showed statistically greater symptom improvement than placebo, but the evidence for comparing safety between medicines was limited.