Development of the Susceptibility-Spectrum Discrepancy Index (S2DI): A novel metric for antimicrobial stewardship in hospitalised patients
Tsuzuki, S.; Koizumi, R.; Asai, Y.; Hashimoto, Y.; Inoue, N.; Ohmagari, N.
Show abstract
Objectives: Optimising parenteral antimicrobial use is central to antimicrobial resistance (AMR) control, yet its appropriateness is difficult to assess. We aimed to develop a quantitative indicator to evaluate the appropriateness of parenteral antimicrobial therapy in hospitalised patients with bloodstream infections. Methods: We developed the Susceptibility-Spectrum Discrepancy Index (S2DI), reflecting the discrepancy between antimicrobial susceptibility of blood culture isolates and the spectrum width of prescribed agents. Using a database from 67 National Hospital Organization hospitals in Japan, we identified patients with Staphylococcus aureus or Escherichia coli bacteraemia from 2017 to 2023. An expert panel of 10 infectious disease physicians independently ranked antimicrobial susceptibility (A) and spectrum width of commonly used agents (B). S2DI was defined as B minus A on day 7 after treatment initiation, with values closer to zero indicating more appropriate therapy. S2DI was calculated for individual cases, aggregated at the hospital level, and analysed using linear mixed-effects models with hospital-level random effects. Results: A total of 4,505 S. aureus and 9,563 E. coli bacteraemia cases were included. Median S2DI was 1 (IQR 0-1) for S. aureus and 2 (IQR 0-3) for E. coli. For both pathogens, later calendar years were significantly associated with more favourable S2DI, suggesting gradual improvement in antimicrobial use. In E. coli bacteraemia, female sex and younger age were also associated with more appropriate therapy. Conclusions: Although variation across hospitals persists, appropriateness of parenteral antimicrobial use has improved over time. S2DI is a simple metric that may support optimisation of antimicrobial use.
Matching journals
The top 8 journals account for 50% of the predicted probability mass.