A systematic review and meta-analysis to identify behavioural content and active ingredients of antimicrobial stewardship education and training interventions in hospital-based care settings

BACKGROUND: The growing threat of antimicrobial resistance has led to efforts to improve the responsible use of antimicrobials (antimicrobial stewardship - AMS). AMS education and training is essential for providing healthcare professionals with the knowledge and skills required to change prescribing behaviours, but the design and delivery of education and training varies, and it is unclear what content, and methods make for more effective education and training. The aim of this systematic review was to apply behavioural science frameworks to specify the content of AMS education and training interventions in hospital settings to determine 'what works' and to evaluate their effectiveness and cost-effectiveness.

METHODS: We searched MEDLINE, EMBASE, and CENTRAL and hand searched studies included in a previous Cochrane review for studies published from January 2015 to February 2025. We applied behavioural science frameworks (Action, Actor, Context, Target and Time framework, Behaviour Change Wheel and Behaviour Change Technique Taxonomy) to code intervention descriptions and supplementary materials from published papers into target behaviours, modes of delivery and behaviour change strategies used. Meta-regressions were used to explore the (cost-)effectiveness of different target behaviours, modalities, and behaviour change strategies on reducing antibiotic consumption.

RESULTS: Of the 1845 studies identified, 64 were included in the review and 26 included in the meta-regression. Education/training was more effective in reducing antibiotic consumption when delivered face-to-face (β= - 2.65, 95% CI: - 5.23 to - 0.07, k = 21). In total, 29 behaviour change techniques were identified across interventions, with no individual behaviour change technique associated with reduced antibiotic consumption. Interventions using the broad intervention types of modelling (Providing an example for people to aspire to or imitate) (β= - 2.23 (95% CI: - 4.27 to - 0.18) and restriction (Using rules to reduce the opportunity to engage in the target behaviour or to increase the target behaviour by reducing the opportunity to engage in competing behaviours) (β = 2.95 (95% CI: 1.10 to 4.79) had significant effects on antibiotic consumption.

CONCLUSION: Our results suggest that AMS education and training interventions may be more effective when they focus on modelling and appropriate restriction, and when delivered in-person. However, more evidence is needed from well-designed studies that explicitly report intervention content, to enable firmer conclusions about the specific elements involved in effective AMS education and training.