Table 4 A summary of the aim, methods and results for each intervention development phase

Phase 1: To describe the response to a national audit

Interviews, observations and documentary analysis, iteratively presenting the findings to both a co-design group and an advisory group

Data collection was manual, took an average of 37 min per record and involved variable interpretation both of the notes and the standards being measured. Feedback took 15 months and was typically seen by two or three people. The clinical lead was responsible for developing an organisational action plan. They did this drawing upon national, rather than local, results. The actions were constrained by what the clinical lead could personally deliver. There was little evidence that they selected actions aligned to an analysis of influences upon performance. The action plan was reviewed, amended and approved at directorate- and organisational-level committees. In reaching their decision, the committees discussed the motivation of the audit provider, the validity of the data, relative performance, triangulation with other data and risks to organisational priorities.

Phase 2: To identify and specify enhancements

Co-design methods involving facilitated discussion about evidence, theory and the use of a specifying framework. Research team developed a logic model to review theoretical coherence.

The co-design group identified the opportunity to enhance data collection, feedback and action planning. They prioritised action planning and defined the outcome sought.

The research team identified theoretically-coherent target behaviours that aligned to the outcome: target low baseline performance; address recipient priorities; develop trust and credibility; present meaningful comparisons; present loss-framed data; identify and address barriers to performance; develop a conceptual model; involve stakeholders; consider the opportunity cost.

The co-design group considered these proposals, rejecting the use of loss-framing. The co-design group specified how the selected behaviours could be delivered. Their specification was used to group the target behaviours into a series of ‘steps’ (Table 1).

Phase 3: To develop a strategy to implement the enhancements

Co-design methods involving facilitated discussion using theory-informed toolkit

The key normalisation process theory mechanisms anticipated to influence implementation were coherence and cognitive participation. Individual and collective specification, initiation, legitimation were important ingredients. There were differences in the ingredients between the steps. We selected a strategy including an educational workshop and virtual educational outreach. Figure 2 represents the content of the intervention after phase 6; at phase 3, the target behaviours did not include collaboration nor feedback.

Phase 4: To test and refine the intervention

Delivery of the intervention followed by interviews exploring fidelity, feasibility, acceptability and appropriateness [9, 23, 27]

All the BCTs in the manual were delivered. There was evidence for fidelity of receipt, fidelity of enactment and of acceptability of the intervention. Potential enhancements to the intervention included addressing the time commitment, creating opportunities to learn from others, further supporting the analysis of influences and developing content so those developing improvement actions gain feedback about the impact of their actions. We revised the content, re-named the intervention, amended activities to address time commitment, incorporate a more structured analysis of influences, support collaboration and to develop local feedback mechanisms.

Phase 5: To adapt the intervention to a second national audit

Re-design and consultation to reflect differences in clinical topic, context, contractual requirements and recent evidence.

Delivery changed in response to: the new clinical topic, by changing examples based on dementia standards to those within the diabetes audit and incorporating delivery through a credible source (NDA Clinical lead); context, pandemic related restrictions meant that it was delivered virtually, requiring additional content about the use of Microsoft Teams and Google JamBoard; the NDA contractual requirements, to address a pre-determined target for improvement thereby narrowing the recipients’ selection of priorities (step 1); recent evidence [26], that there may be benefits from highlighting the intervention deliverers’ recent and credible experience in quality improvement and having multi-disciplinary quality improvement team delivery that includes opinion leaders.

Phase 6: To test and refine the intervention

Coding the BCTs within the manual, delivery of the intervention followed by coding the BCTs delivered and interviews exploring fidelity, appropriateness and acceptability. Re-design and consultation to reflect interview findings.

All BCTs identified in the written protocol were delivered by facilitators. Participants reported positive attitudes towards the intervention and that the intervention was appropriate.