PREVIEW-A&F: a mixed-methods eye-tracking methodology for prospective optimisation of electronic audit and feedback interventions

Background Electronic audit and feedback is a widely used implementation strategy for improving clinician practice, yet interventions demonstrate variable effectiveness. Current approaches to electronic audit and feedback development rely heavily on subjective usability feedback and rarely include prospective optimisation prior to implementation. This may contribute to repeated redesign cycles when engagement barriers are only identified after deployment, representing an underexamined contributor to slower implementation pathways. Methods We developed and tested a theory-informed methodology for prospectively evaluating engagement with electronic audit and feedback interventions. Central to the methodology was an a priori framework linking each element in the tool to its intended behavioural change technique (BCT), hypothesised behavioural mechanism, and required level of visual attention. To test the methodology, health professionals enrolled as investigators in an international multi-site penicillin allergy implementation study (international Network of Antibiotic Allergy Nations) reviewed their site-specific electronic audit and feedback report while wearing an eye-tracker and completing a subsequent think-aloud protocol. Gaze data and think-aloud data were independently analysed and then integrated to examine whether observed engagement aligned with the intervention’s intended behavioural design. Results Integration of gaze and think-aloud data identified convergence and divergence between intended and observed engagement with BCT-linked interface elements. Of 29 areas of interest, 14 were Convergent (intended and observed engagement aligned across both data streams); one Hidden Opportunity (visually attended as intended but BCT activation incomplete); seven Attentional Mismatch (fixation duration deviated from the intended level despite adequate processing); three Divergent (engagement failure across both data streams); and four Silent Engagement (no participant verbalised the element, so engagement could not be assessed from think-aloud data). Joint interpretation of the two data streams generated targeted redesign recommendations that would not have been identifiable using either method alone. Conclusions This proof-of-concept study establishes the feasibility of a theory-informed methodology for prospectively evaluating engagement with electronic audit and feedback interventions prior to broader implementation. Integration of objective measures of overt visual attention with qualitative evidence of verbalised processing depth allows interface-level barriers to be identified before implementation at scale, rather than after. Specifying intended BCTs a priori was critical to interpreting engagement against the intervention’s behavioural design and identifying opportunities for optimisation. Prospective optimisation may help reduce avoidable redesign cycles and support more efficient pathways to implementation readiness.