An ICU retrospective cohort study reported lower unadjusted mortality with ceftazidime–avibactam for carbapenem-resistant Klebsiella pneumoniae, but the association disappeared after multivariable adjustment. The analysis of 249 patients identified age, diabetes, lymphopenia, and hypoalbuminemia as independent predictors of death. The update is clinically relevant for antimicrobial stewardship because it highlights how confounding in real-world ICU populations can mask or exaggerate treatment comparisons. For infection teams, the practical takeaway is to treat efficacy signals cautiously when retrospective baseline differences are substantial. The study also reinforces that patient host factors may be as predictive as antibiotic choice in the sickest cohorts—particularly when fluid status, nutrition, and immune suppression vary across admissions.