Introduction Recent research indicates that around 8% of older people living in prison have signs or symptoms of dementia or mild cognitive impairment (MCI), yet the care they receive is not equivalent to care in the community and this means their needs may not be met. We co-developed an intervention specifically for older people living in prison with dementia/MCI (Dementia and Mild Cognitive Impairment in prison care pathway and training package–DECISION). To date, this has not been implemented or evaluated. This paper presents our protocol for a study to assess the feasibility and acceptability of DECISION. Methods This is a non-randomised, realist-informed mixed-methods feasibility study with integrated process evaluation, which will take place in two prisons in England. The intervention was codeveloped with experts with lived experience. Participants will include older people living in prison, staff working in prison and peer supporters. We will assess the feasibility and acceptability of the intervention (eg, numbers eligible; rates of recruitment and retention), and the evaluation design (eg, completion rates of standardised outcome measures). Methods will include semistructured, realist-informed interviews; an audit to assess implementation fidelity; focused ethnography; training questionnaires; and collection of resource use data. We will refine the DECISION programme theory using realist-informed methods to examine and refine how contexts and mechanisms interact to produce the intervention’s outcomes. Ethics and dissemination This study received a favourable ethical opinion from the Wales REC 3 Research Ethics Committee in January 2025 (reference number 24/WA/0323). HMPPS National Research Committee approval was also granted in January 2025 (reference number 2024-1451). Findings will be disseminated through a range of avenues, including stakeholder engagement events, open-access papers, conference presentations, evidence briefings for commissioners, providers and practitioners, and newsletters for service users.
Forsyth et al. (Sun,) studied this question.