Duration and Appropriateness of Empirical Antibiotic Use in ICU: A Quality Improvement Project

Introduction: De-escalating empirical antibiotic therapy in critically ill patients to a targeted approach based on culture findings is strongly recommended within the framework of antimicrobial stewardship (AMS). We initiated a quality improvement project aimed at analyzing the impact of an AMS step-wise persuasive intervention targeting ICU physicians on the proportion of patients receiving prolonged empirical antibiotic therapy (>96 hours). Secondary outcomes included ICU mortality, duration of mechanical ventilation (MV), ICU and hospital length of stay (LOS), and severity of illness (SOFA score) at the time of prolonged use.


Methods: The study was conducted in three phases: three-month pre-intervention phase, one month of intervention, and three-month post-intervention phase. All adult ICU patients receiving empirical antibiotics were included. The step-wise intervention consisted of analysis of results of pre-intervention, presentation of results to clinicians of ICU team and stakeholders from department of Microbiology and Pharmacology to define actionable tasks, an introduction and persuasion of clinicians to use an algorithm prepared for de-escalation of empirical antibiotic. Real-time entry of antibiotic prescription in closed group online spreadsheet facilitated case-by-case discussion. Bayesian regression models were used to assess intervention impact, with sensitivity analyses across different prior assumptions.


Results: Under a neutral prior, the intervention yielded an odds ratio (OR) of 0.99 [95% CrI: 0.31–2.39] for prolonged empirical antibiotic use, with an absolute risk difference of −2.5% [−19.5%, 13.7%] and a 60% probability of benefit. Adjusted analyses revealed stronger effects (OR 0.27 [0.04–1.34]; ARD −9.3% [−21.0%, 2.4%]; 95% probability of benefit). The intervention led to a higher threshold for continuing empirical therapy, with a significant increase in median SOFA score at prolongation (4.00 vs 7.00; Δ = 2.75 [0.38, 4.94]; 99% probability). No significant improvements were observed in ICU mortality, duration of MV or LOS.


Conclusions: AMS step-wise persuasive intervention showed good posterior probability of reduction of prolonged empirical antibiotic use (60%) by reducing unnecessary prolongation and shifting physician behavior toward treating higher-severity patients.