Antibiotic Stewardship and Length of Stay in Osteoarticular Infections: Predictive Modeling and Multivariate Analysis of Mismatch and Optimization Delay.

BACKGROUND

Osteoarticular infections (OAIs) require prompt, targeted antimicrobial therapy. Delays in antibiotic optimization - whether due to empirical mismatches, diagnostic uncertainty, or systemic inefficiencies - prolong hospitalization and increase costs.

METHODS

Using the MIMIC-IV database, we retrospectively identified adult OAI admissions. Mismatch was defined when empirical antibiotics were inactive against the isolated pathogen. Optimization delay was calculated from empirical initiation to the first susceptible antibiotic. Multivariate log-linear regression assessed the impact of mismatch, delay, age, gender, and comorbidities on length of stay (LOS). Cost modeling applied a standardized €600/day estimate from European tertiary centers, used as a reference value for relative savings.

RESULTS

Among 3,093 admissions (mean age 59 years), mismatch occurred in 13% with a mean delay of 8.8 days. Mismatch increased LOS by ∼64%, and each day of delay increased LOS by 2.4% (p<0.001). Younger mismatch patients had disproportionately longer stays. Congestive heart failure (CHF), chronic kidney disease (CKD), and immune deficiency prolonged LOS. Simulated 2-3-day reductions shortened LOS by 1.6-2.4 days, saving 600-900 per patient and freeing 2-3 beds per mismatch scenario.

CONCLUSIONS

Reducing optimization delays yields measurable clinical, economic, and operational benefits. Rapid diagnostics and predictive modeling could support earlier optimization. Excluding culture-negative OAIs may limit generalizability.