What and how do different stakeholders contribute to intervention development? A mixed methods study. [version 2; peer review: 2 approved]

Background: UK Medical Research Council guidelines recommend end-user involvement in intervention development. There is limited evidence on the contributions of different end-users to this process. The aim of this Study Within A Trial (SWAT) was to identify and compare contributions from two groups of end-users - people with diabetes’ (PWD) and healthcare professionals’ (HCPs), during consensus meetings to inform an intervention to improve retinopathy screening uptake. Methods: A mixed method, explanatory sequential design comprising a survey and three semi-structured consensus meetings was used. PWD were randomly assigned to a PWD only or combined meeting. HCPs attended a HCP only or combined meeting, based on availability. In the survey, participants rated intervention proposals on acceptability and feasibility. Survey results informed the meeting topic guide. Transcripts were analysed deductively to compare feedback on intervention proposals, suggestions for new content, and contributions to the final intervention. Results: Overall, 13 PWD and 17 HCPs completed the survey, and 16 PWD and 15 HCPs attended meetings. For 31 of the 39 intervention proposals in the survey, there were differences (≥10%) between the proportion of HCPs and PWD who rated proposals as acceptable and/or feasible. End-user groups shared and unique concerns about proposals; both were concerned about informing but not scaring people when communicating risk, while concerns about resources were mostly unique to HCPs and concerns about privacy were mostly unique to PWD. Fewer suggestions for new intervention content from the combined meeting were integrated into the final intervention as they were not feasible for implementation in general practice. Participants contributed four new behaviour change techniques not present in the original proposals: goal setting (outcome), restructuring the physical environment, material incentive (behaviour) and punishment. Conclusions: Preferences for intervention content may differ across end-user groups, with feedback varying depending on whether end-users are involved simultaneously or separately.