Background and Importance Clinical-decision support systems (CDSS) are commonly used in clinical practice to generate antimicrobial stewardship (ASP)-alerts, which could help implement evidence-based recommendations.

Aim and Objectives To analyse use, effectiveness, and positive predictive value (PPV) of a bundle of ASP alerts generated by CDSS in a first-level hospital.

Material and Methods Observational, retrospective study. Inclusion criteria: ASP alerts generated between 1 November 2021 and 31 August 2022. The bundle of alerts included (1) >7 days of intravenous antimicrobial therapy (IAT), (2) transition from IAT to oral therapy, (3) antimicrobial dosage adjustment in renal impairment, (4) therapeutic antibiotic monitoring (TAM) and (5) duration of restricted antimicrobials (RA) (carbapenems, daptomycin, piperacillin/tazobactam, linezolid, tigecycline, ceftazidime/avibactam, echinocandins and voriconazole) >72 hours. Total number of generated alerts, number of patients with at least one alert during their hospital stay, type of alert and antimicrobial related that triggered the alert were recorded and analysed.

Effectiveness was calculated as a proportion between alerts requiring intervention and total number of alerts. PPV was calculated as a proportion between accepted interventions and total number of alerts. Both proportions were expressed as percentages (%).

Results A total of 2,546 alerts (on 927 patients) generated during the time of study. Most frequent antimicrobials that triggered the alerts were: 28.6% piperacillin/tazobactam (727/2,546), 13.6% meropenem (346/2,546), 7.5% linezolid (190/2,546), 6.7% levofloxacin (171/2,546) and 6.2% ceftriaxone (158/2,546). The type of ASP-alert generated was: >7 days of AIT (32.0%), duration of RA >72 hours (31.6%), antimicrobial dosage adjustment in renal impairment (19.2%), transition from IAT to oral therapy (13.2%) and TAM (4.0%).

The effectiveness was 14.5%, with a PPV of 9.6%. By type, effectiveness was 9.5% (type 1), 21.1% (type 2), 11.0% (type 3), 19.6% (type 4) and 18.1% (type 5). PPV for these alerts was 6.2% (type 1), 19.9% (type 2), 9.2% (type 3), 11.8% (type 4) and 8.7% (type 5).

Conclusion and Relevance The most frequently triggered ASP-alerts were duration of IAT and RA, and antimicrobial dosage adjustment in renal impairment. However, those alerts with a higher PPV were transitions from IAT to oral therapy and TAM. Further studies are needed to determine ASP-alerts with a higher effectiveness to optimise their use and to avoid alert fatigue.

Conflict of Interest No conflict of interest