Inappropriate Empirical Antibiotic Therapy and Mortality in Critical Illness: A Retrospective Cohort Study with Propensity Score Analysis in an Indonesian ICU

Abstract

Introduction: Inappropriate empirical antibiotic therapy (IEAT) is a critical driver of mortality in sepsis, particularly in regions with high antimicrobial resistance (AMR) like Southeast Asia. This study aimed to quantify the association between IEAT and 28-day mortality in a critically ill Indonesian patient cohort, employing advanced statistical methods to control for confounding.

Methods: We conducted a retrospective cohort study of 280 adult patients who received empirical antibiotics and had positive cultures upon admission to a tertiary ICU in Indonesia (January 2022–December 2023). The primary exposure was the appropriateness of the initial antibiotic regimen (IEAT vs. AEAT) based on in-vitro susceptibility. We used multivariate logistic regression and a 1:1 propensity score-matched (PSM) analysis to adjust for baseline differences in patient severity, including APACHE II score and the presence of septic shock.

Results: In the full cohort, 108 patients (38.6%) received IEAT. The 28-day mortality was profoundly higher in the IEAT group than the AEAT group (77.8% vs. 8.1%; p < 0.001). After multivariate adjustment, IEAT remained a powerful predictor of mortality (Adjusted Odds Ratio [aOR]: 38.72; 95% CI: 18.91–79.30; p < 0.001). In the PSM cohort of 200 patients with balanced baseline characteristics, the association remained strong and significant (OR: 25.15, 95% CI: 11.54–54.81; p < 0.001). Local prescribing patterns revealed that levofloxacin monotherapy, the most common regimen, had an inappropriateness rate of 76.4%.

Conclusion: Inappropriate empirical antibiotic therapy is strongly associated with a substantially increased risk of death in critically ill Indonesian patients. This association persists after rigorous adjustment for confounding. These findings highlight the urgent need for robust antimicrobial stewardship programs, guided by dynamic local surveillance, to combat the lethal impact of AMR.