Computerized decision support for antimicrobial prescribing: what every antibiotic steward should know

To examine the potential role of computerized clinical decision support systems (CDSS) in antimicrobial stewardship (AMS) and to identify significant challenges concerning their effectiveness and implementation.

Narrative review.

This review is based on existing literature regarding CDSS in AMS across various healthcare environments, such as hospitals and primary care facilities. The systems evaluated include both stand-alone tools and those integrated into electronic health records (EHR), featuring expert rule-based logic and new machine learning (ML) models. CDSS capabilities include prescribing guidance, alerts, resistance prediction, and de-escalation protocols.

CDSS are intended to aid in antimicrobial prescribing by integrating clinical guidelines with data specific to each patient. Despite their theoretical potential, their effectiveness is often hindered by inconsistent incorporation into clinical practices, low user engagement, and inadequate design. Many systems are reactive, not well-suited to user needs, or lack transparency in their recommendations. Evaluating these systems is challenging due to varied outcomes, poor methodological quality of studies, and the complexity of attributing causality in intricate care settings. Barriers to implementation include alert fatigue, perceived time constraints, poor fit with existing workflows, and resistance to change. Instances like the COMPASS trial demonstrate the disconnect between design and practical application, underscoring the necessity for user-focused development, clear reasoning, and a balanced approach between mandatory and advisory elements.

CDSS have the potential to improve antimicrobial use, but widespread impact is hindered by evaluation and implementation challenges. Realizing their value requires better integration, usability, and rigorous research frameworks tailored to complex healthcare settings.