Dear Editor,

Olson et al. report an important systemwide quality-improvement initiative to reduce intravenous therapy at discharge and overall antibiotic duration for uncomplicated gram-negative bloodstream infections (GN-BSI) across 22 hospitals. ^{Reference Olson, Adam and John1} The large sample size, multi-site setting and use of interrupted time-series (ITS) modeling represent notable strengths and make this work a valuable contribution to the stewardship literature.

Several methodological issues, however, merit clarification to aid interpretation and comparison with related studies. First, the institutional guideline and consensus recommendations emphasize a 7-day course for uncomplicated GN-BSI when predefined criteria are met. ^{Reference Heil, Bork and Abbo2} In contrast, the evaluative outcomes focus on proportions receiving <8 and <10 days of therapy, alongside median treatment duration that remains approximately 11 days in the final period. ^{Reference Olson, Adam and John1} Because “<10 days” is a relatively liberal threshold, these metrics may overstate the extent to which practice has converged on the 7-day standard. Future evaluations aligned more closely with the guideline target (for example, proportion achieving exactly 7 ± 1 days) would allow clearer assessment of fidelity to evidence-based duration.

Second, the study is a non-randomized before–after design with sequential interventions, and secular trends are an important threat to internal validity. The authors appropriately conduct an ITS sensitivity analysis with adjustment for case-mix variables, which is consistent with best practice for quasi-experimental stewardship studies. ^{Reference DiPietro, Dubrovskaya and Marsh3} Nonetheless, the ITS models were specified post hoc, and multiple analytic choices (period definitions, covariate selection, aggregation by month) introduce flexibility that can influence estimates of level and slope change. Explicit a priori analytic plans, as recommended in implementation-oriented stewardship research, ^{Reference Livorsi, Drainoni and Reisinger4} would further strengthen causal interpretation of system-level changes.

Finally, the authors observe substantial heterogeneity in response across hospital groups. Behavioral and organizational frameworks for antimicrobial stewardship highlight the importance of specifying intervention components, mechanisms of action, and local determinants of behavior when interpreting such variation. ^{Reference Livorsi, Drainoni and Reisinger4,Reference Thampi, Szymczak and Leis5} More detailed reporting of implementation strategies (e.g. intensity of audit and feedback, leadership engagement, and pharmacist capacity at each site) would help distinguish intervention effect from contextual effect and guide replication.

Olson et al. provide valuable descriptive data on stewardship practice and outcomes for GN-BSI. Clarifying alignment between guideline targets and outcome definitions, pre-specifying quasi-experimental analyses, and further detailing case definitions and implementation context, consistent with existing methodological guidance would enhance the interpretability and portability of their important findings.