Introduction:
     
     Patient and public involvement (PPI) is considered part of best-practice for health care delivery, research and policy. However, people with dementia are frequently excluded from PPI initiatives. While recent studies have investigated PPI of people with dementia in research, little is known about their involvement at the macro-level of care, that is, in health policy and guideline development. This scoping review maps the evidence on PPI of people with dementia at the macro-level of care, focusing on the methods, outcomes and mechanisms of involvement.
   

Methods:
     
     We systematically searched MEDLINE via PubMed, CINAHL, the Cochrane Library and GeroLit. Additionally, we performed forward and backward citation searching, manually tracked individual references, searched abstract books and yearbooks, and contacted authors of included reports to seek additional references. We analysed each method's mechanisms of involvement to assess whether measures were taken to maximise effective information transfer.
   

Results:
     
     We included 43 reports and identified 35 involvement methods, which we structured into six categories: 'focus groups and interviews', 'surveys and questionnaires', 'public events', 'meetings with decision-makers', 'serving as members of working groups', and 'multiple-step methods'. Most of the involvement took the form of consultations during the early stages of policy or guideline development. All involvement methods required verbal communication skills, almost all of the participants had mild dementia. We found that most reports did not clearly state the involvement outcomes. While a majority of methods had some facilitation in place to elicit information from participating people with dementia, only nine methods used a structured aggregation to synthesise participants' contributions into a joint statement.
   

Conclusion:
     
     We found limited evidence of dementia-adapted involvement. Future research might focus on exploring the mechanisms of involvement to adapt methods to specific target groups, such as people with impaired verbal communication or advanced dementia. We recommend using existing guidance to report PPI initiatives, as the reporting was often incomplete, which limits reproducibility.
   

Patient or public contribution:
     
     We discussed both our study protocol and our results with a working group of people with dementia, who provided valuable insight for our data interpretation. Our findings can serve such groups for future consultations.