Accurate monitoring of antimicrobial consumption is essential for developing effective stewardship interventions. Hospitals are mandated to report antibiotic treatment duration using the CDC's Standardized Antimicrobial Administration Ratio (SAAR). We aimed to create a metric for measuring antimicrobial spectrum to complement SAAR, benchmarked against local and individual resistance.

This retrospective cohort study utilised data from nine acute care hospitals within BJC Healthcare from January 2018 to December 2023. Adult patients meeting CDC sepsis event criteria were included, focussing on infections caused by Gram-negative bacilli (GNB). Data on patient characteristics, microbiology results, treatments, and timing were collected. We defined the antibiotic effective spectrum (AES) as the ratio of susceptible GNB to all GNB cultured. Mathematical optimisation identified the optimal antibiotic as the antibiotic with the narrowest effective spectrum active against the isolated GNB. We calculated the empirical (before culture results are available) and directed Optimal Spectrum Antibiotic Ratio (OSAR) as the ratio between the AES chosen clinically to the optimal AES, reported it for each hospital yearly, and compared it to SAAR and the Antibiotic Spectrum Index (ASI). Sensitivity analyses focused on empirical undertreatment (OSAR <1, where the empirical choice did not cover the cultured GNB).

We included 10,277 GNB cultured from clinical specimens in 7997 distinct patients. The AES ranged from 0.15 for ampicillin to 0.94 for amikacin. Empirical OSAR varied between 1.08 (95% CI 1.05–1.10) and 1.89 (95% CI 1.51–2.28), with undertreatment present in 1396 (13.6%) of cases overall and wide variations across hospitals. Directed OSAR was lower, ranging between 0.74 (95% CI 0.73–0.75) and 1.45 (95% CI 1.41–1.49). OSAR, SAAR and ASI had different trends across hospitals over the six study years.

The OSAR yielded different conclusions about antibiotic utilisation, complementing duration-based measures like SAAR. It determines the appropriateness of antibiotic spectrum based on local and individual resistance.

The project was funded by the CDC (5U54CK000609).

Antimicrobial utilisation

Antibiotic spectrum

Gram-negative bacilli

Mathematical optimisation

© 2026 The Authors. Published by Elsevier B.V.