A personalised music intervention for people with dementia in care homes is being tested against standard care in a trial involving 406 residents across 58 UK care homes. This matters because dementia often causes agitation, aggression, and distress—behaviours that are difficult to manage with medication, which carries side effects. Personalised music, tailored to an individual’s life history and preferences, offers a drug-free alternative, but robust evidence from large trials is lacking. The PUMA study aims to fill that gap by training care staff as "Dementia Care Champions" to embed personalised music into each resident’s care plan using a structured model called FITS. If the intervention reduces distress—measured by the Neuropsychiatric Inventory-Nursing Home version over 12 months—care homes could adopt a low-cost, scalable tool that improves residents’ quality of life without relying on antipsychotics. The study also tracks wellbeing, resource use, and cost-effectiveness, giving commissioners and care providers the data needed to decide whether to roll out the approach nationally. A process evaluation will reveal what helps or hinders implementation in real-world settings.
View original technical description
Research question: In people with dementia living in care homes, does incorporating personalised music into care plans using the FITS (Focussed Intervention Training and Support) model reduce distress compared to standard practice without a systematic music intervention? Background: Recent research highlights the need for robust trials evaluating personalised music interventions for people with dementia in care homes. Aims and objectives: To determine if using the FITS model to incorporate personalised music into care plans reduces changed behaviours in residents with dementia measured by the Neuropsychiatric Inventory-Nursing Home version (NPI-NH). Secondary objectives include reporting resident experience using the Music in Dementia Assessment Scale (MiDAS), conducting a process evaluation, and determining cost-effectiveness. Methods: Multi-centre, parallel cluster randomised controlled trial with process and economic evaluations. 58 diverse UK care homes will be randomised in a 1:1 ratio, aiming to recruit an average of 7 participants per home for a total of 406 residents with dementia and associated changes in behaviour. The intervention involves training care staff as "Dementia Care Champions" to use the FITS approach to tailor personalised music within individual care plans. Comparator is usual care. The study will use a three-level mixed-effects linear model adjusted for covariates and clustering. Randomisation will use stratified blocks within strata based on care home characteristics. Outcome assessors will be independent but not blind to allocation. Primary outcome is the average NPI-NH score over 12 months. Secondary outcomes include quality of life (QUALID), wellbeing (EQ-HWB-Proxy), and resource use. A mixed-methods process evaluation guided by Normalization Process Theory will examine implementation, mechanisms and context. 10 of 28 intervention care homes will be purposively sampled for case studies involving surveys with the 28 Dementia Care Champions, interviews with 20 staff and 20 resident/family dyads, and longitudinal analysis of MiDAS scores. An economic evaluation will determine cost per quality-adjusted life year using the EQ-HWB-Proxy, with costs based on intervention delivery and health/social care resource use. Cost-effectiveness analysis will use a similar multi-level model to the primary analysis. Timelines: 42 month study - 8 months setup, 16 months recruitment, 12 months follow-up, 6 months for analysis and dissemination. Impact and dissemination: Results will guide personalised music interventions in dementia care. Dissemination through presentations, publications, training and engagement with care providers and dementia charities.
Plain English summaries and category classifications on this site are generated by AI and may not perfectly reflect the original research.
Is something wrong? Let us know