ICU Stewardship Linked to Lower Broad‑Spectrum Use, Improved Gram‑negative Susceptibility, and Cost Savings

Investigators in Brazil reported an interrupted time-series evaluation of an Antimicrobial Stewardship Program (ASP) introduced in January 2021 in the adult ICU of a high-complexity public hospital, examining outcomes from January 2019 through December 2023. The analysis tracked three domains around program implementation: antibiotic consumption expressed as DDD/1000 patient-days, temporal susceptibility patterns assessed with Joinpoint regression, and direct antimicrobial acquisition costs. Using this quasi-experimental before/after framework, the authors described associations between the ASP period and changes in use, ecology, and spending within a single center.

The ASP was described as a structured bundle that included local guideline development and dissemination, prospective audit and feedback on antimicrobial prescriptions, and formulary restriction with mandatory pre-authorization for selected broad-spectrum agents, alongside multidisciplinary meetings and education. For utilization, investigators extracted ICU dispensing data from the hospital information system and calculated consumption as DDD/1000 patient-days using the WHO ATC/DDD Index (2023), with pharmacy dispensing as the numerator and ICU patient-days as the denominator. For susceptibility surveillance, they assembled clinical-culture isolate data interpreted under BrCAST criteria, excluded identifiable screening/surveillance samples when possible, and analyzed isolates without patient-level deduplication to reflect unit-level ecology. Joinpoint regression was used to identify inflection points in monthly trends, complemented by pre/post comparisons aligned to January 2021.

For consumption, the authors reported statistically significant post-ASP reductions for ceftriaxone (Mann–Whitney p=0.0011; Joinpoint decline −1.61%/month from March 2020 to November 2023), meropenem (t-test p=0.0034; −1.58%/month from March 2020 to November 2023), and piperacillin–tazobactam (t-test p=0.003; −0.66%/month across the observation window). Vancomycin consumption was also lower after the intervention (pre 75.99 vs post 48.28 DDD/1000 patient-days; p<0.001), with an early Joinpoint segment showing −5.46%/month from February 2020 to March 2021. In contrast, polymyxin B did not show a meaningful reduction (pre 15.03 vs post 20.23 DDD/1000 patient-days; p=0.1011) and was modeled as rising 1.99%/month without a detected inflection. At the aggregate ATC J01 level, mean monthly consumption decreased from 410.24 to 276.30 DDD/1000 patient-days, a 32.65% relative reduction that the authors reported as statistically significant.

For susceptibility, the aggregated Gram-negative trend increased after 2021, with an overall monthly percent change of +1.89% (p<0.05) in the Joinpoint model. Species-level analyses identified post-intervention improvements for Klebsiella pneumoniae (June–September 2021, +70.1%/month; p<0.05), Escherichia coli (June 2021–September 2023, +7.71%/month; p<0.05), and Enterobacter cloacae (August 2020–July 2023, +7.33%/month; p<0.05). Other organisms showed limited change in the authors’ models, including Pseudomonas aeruginosa with an essentially flat trajectory (−0.11%/month, non-significant) and Acinetobacter baumannii with persistently low susceptibility and no significant improvement (−0.93%/month, non-significant). Among Gram-positive organisms highlighted by the authors, Enterococcus faecium was described as highly variable with no sustained upward trend (−1.77%/month, non-significant), and Staphylococcus epidermidis remained low without a meaningful response (−6.37%/month, non-significant). Overall, the reported susceptibility signal was heterogeneous across pathogens despite the aggregated Gram-negative rise.

The retrospective economic analysis combined observed utilization changes with acquisition cost estimates to report projected annual direct antimicrobial savings of USD 174,431.42. The largest per-agent contributors were vancomycin (USD 60,796.45), meropenem (USD 51,691.65), piperacillin–tazobactam (USD 45,251.67), and ceftriaxone (USD 22,727.43), while polymyxin B was reported with a negative annual balance (USD −6,035.79). In Discussion and Conclusions, the authors interpreted the overall pattern as an association between the ASP period and reduced broad-spectrum use, favorable ecological trends in several Gram-negative pathogens, and lower direct antimicrobial costs. They also noted design- and data-related caveats, including the single-center observational nature, isolate-level surveillance without patient deduplication, and limited ability in a retrospective dataset to uniformly exclude screening samples, framing the findings as observed associations within this ICU setting.

Key Takeaways:

• The authors reported post-ASP decreases in multiple commonly used broad-spectrum agents; polymyxin B use was higher post-ASP and the pre/post difference was not statistically significant.
• Aggregated Gram-negative susceptibility rose in the Joinpoint analysis, with species-level improvements reported for K. pneumoniae, E. coli, and E. cloacae alongside organisms showing limited change.
• Estimated direct antimicrobial cost savings were reported and appeared concentrated in a small set of high-impact agents in the authors’ accounting.