Antimicrobial resistance (AMR) is a major global health threat, contributing to an estimated 254,000 deaths annually among children under five. A key driver of AMR is the inappropriate use of antimicrobials in human medicine. Antimicrobial stewardship (AMS) programs aim to optimize antimicrobial use by promoting the appropriate choice of drug, dose, and duration. Yet achieving optimal prescribing is particularly challenging in complex hospital environments. Clinical decision support systems (CDSS) offer real-time, guideline-based recommendations and have emerged as promising tools to support clinicians.

Despite their potential, many AMS interventions are supported by weak evidence, often relying on non-randomized studies with limited methodological rigor. There is a pressing need for well-designed research to assess the true impact of these tools and to better understand the barriers and facilitators influencing their implementation.

In the context of this PhD work, I have approached this problem through:

• A qualitative study conducted in three hospitals across France and Switzerland (two linguistic regions) identified local barriers and facilitators to CDSS adoption for antimicrobial prescribing.
• The development of a CDSS (COMPASS) embedded in electronic health record systems across three hospital networks, providing real-time, locally adapted guidance on antimicrobial selection, dosing, and duration.
• A multicenter cluster randomized controlled trial across 24 hospital wards to assess the impact of the CDSS on antimicrobial prescribing. In addition to the CDSS, clinicians in intervention wards received regular feedback on their prescriptions’ appropriateness.

The intervention did not reduce overall antimicrobial use (primary outcome) and led to only limited improvements in prescribing behavior—specifically, an increased rate of switch to oral therapy. These findings contrast with earlier, less rigorous studies and highlight the importance of conducting randomized trials in AMS research. Factors contributing to the limited impact included low baseline antibiotic use and insufficient uptake of the tool. Nevertheless, this work generated valuable insights into CDSS development and implementation, detailed in a published perspective article.

The COVID-19 pandemic has reinforced the importance of evidence-based clinical practice and the ongoing need to uphold AMS principles, even in crisis conditions. Looking forward, the digitalization of healthcare in low- and middle-income countries (LMICs) creates new opportunities for CDSS implementation in settings with high AMR burden. However, both integrated systems and mobile applications face similar challenges—chiefly adoption, integration into clinical workflows, and sustained use.

Emerging technologies like artificial intelligence (AI)-driven CDSS may improve antimicrobial decision-making by harnessing large datasets for tailored recommendations. Still, these innovations bring new risks and will likely face similar implementation challenges. Strengthening AMS requires continued investment in implementation science to adapt tools to local realities and support clinician engagement.

Ultimately, AMS programs must be grounded in robust evidence and supported at all levels—global, national, and institutional—to sustainably improve prescribing practices and tackle AMR.