Antimicrobial stewardship supports rational antibiotic use. However, balancing access to antibiotic treatment while controlling resistance is challenging. This research used a threshold logistic modeling approach to identify targets for antibiotic usage associated with carbapenem-resistant Acinetobacter baumannii, carbapenem-resistant Klebsiella pneumonia, and extended-spectrum β-lactamases-producing Escherichia coli incidence in hospitals.

This study utilizes an ecological population-level design. Monthly pathogen cases and antibiotic use were retrospectively determined for inpatients between January 2015 and December 2023. The hospital pharmacy and microbiology information systems were used to obtain this data. Thresholds were identified by applying nonlinear modeling and logistic regression.

Incidence rates of 0.199, 0.175, and 0.146 cases/100 occupied bed-days (OBD) for carbapenem-resistant A baumannii, carbapenem-resistant K pneumonia, and extended-spectrum β-lactamases-producing E coli, respectively, were determined as the cutoff values for high (critical) incidence rates. Thresholds for aminoglycosides (0.59 defined daily dose [DDD]/100 OBD), carbapenems (6.31 DDD/100 OBD), piperacillin-tazobactam (3.71 DDD/100 OBD), third-generation cephalosporins (3.71 DDD/100 OBD), and fluoroquinolones (1.91 DDD/100 OBD), were identified. Exceeding these thresholds would accelerate the gram-negative pathogens' incidence rate above the critical incidence levels.

Threshold logistic models can help inform and implement effective antimicrobial stewardship interventions to manage resistance within hospital settings.