DASC-LOT Framework and S3 Metric: A Novel Evaluation and Benchmarking Method to Assess Initiation, Duration, and Spectrum of Antimicrobial Usage at Inpatient Hospital Settings
Michihiko Goto, Shungo Imai, Hyunkeun Cho, James A Merchant, Matthew Bidwell Goetz, Eli N Perencevich, Bruce Alexander, Qianyi Shi, Andrea Holcombe, Daniel J LivorsiAbstract
Background
Inpatient antimicrobial stewardship programs (ASPs) promote avoiding unnecessary initiation, excessively long duration, and overly broad-spectrum selection of antimicrobials, but currently available metrics do not adequately reflect or distinguish these three components. We aimed to develop a novel framework that captures them while providing information specific to each, based on Days of Antimicrobial Spectrum Coverage (DASC) and length of therapy (LOT).
Methods
We developed a three-level conditional modeling framework to extract hospital-level variability, risk-adjusted for three components (initiation, duration, and spectrum). This was applied to data from 118 Veterans Health Administration (VHA) hospitals, with models built on 2022–2023 data and validated with 2024 data. Patient demographics, intensive care status, admitting specialty, 86 comorbidities, and 224 procedure categories were considered candidate variables for risk adjustment. Overall hospital performance was evaluated using a composite metric that integrated three components (“S3 [start, stop, select] Metric”) and was visualized in a radar chart for each hospital.
Results
The cohort included 727,958 unique patients with 9,363,922 days present. Hospital-level usage ranged widely (DASC per 1,000 DP: 1,311-5,275 [interquartile range (IQR): 2,738-3,563]; LOT per 1,000 DP: 132.2-517.6 [IQR: 301.0-367.5]). Risk-adjustment models included 116 variables for initiation, 77 for duration, and 91 for spectrum components. S3 metrics ranged from 0.703 to 1.572 (IQR: 0.906–1.086). A three-component evaluation could provide specific information on each hospital's usage patterns.
Conclusions
We propose a DASC-LOT framework and S3 metric to assess ASP practices in initiation, duration, and spectrum separately while providing overall composite benchmarking.