People with memory problems can struggle with everyday activities and may stop doing things they want to do. They are more prone to accidents and have a higher risk of falling. Occupational therapists can advise how to do daily activities more easily and safely. Physiotherapists can teach exercises which increase activity and improve balance, and may help maintain memory.

There is little research on how to make these interventions work for people with memory problems. The investigators have developed two activity and exercise programmes suitable for people with memory problems. The investigators will study them in a feasibility trial. One programme involves high-intensity supervision (50 visits over one year), the other moderate-intensity supervision (11 visits over three months). The investigators will compare these with standard falls prevention assessment and advice (1-3 therapist visits). The investigators will encourage participants to exercise by themselves or with family members over the year, and once the programme ends.

People with early dementia or memory problems will be eligible for this study. If possible, the investigators will also recruit a family member. Participants will be recruited from memory clinics or the 'Join Dementia Research' register. The intervention will be delivered over a maximum of 1 year in their own homes. Researchers will visit to collect information at baseline and at 12 months. The investigators will measure ability in activities of daily living, activity, quality of life, memory and health service use. Participants will complete weekly falls diaries. Intervention persistence will be measured for 24 months.

The investigators will conduct interviews and discussion groups to help develop the programmes, and understand how they work in practice ('process evaluation'). The investigators will also do initial work on health economic modelling, dissemination and implementation.

Study findings will be used to refine the intervention, and inform a planned definitive randomised controlled trial.

This is a feasibility randomised controlled trial which aims to answer practicability and feasibility questions required to ensure a large scale trial is successful. The specific questions are:

1. Can the investigators develop and implement a successful and safe rehabilitation support worker (RSW) training programme?
2. Can the investigators recruit participants at a sufficient rate across multiple sites? Will potential participants consent? Do randomisation systems work?
3. Can the investigators deliver the intervention across sites? Can the investigators tailor the intervention? Do the components work together?
4. Can the intervention be undertaken at home?
5. Is the intervention acceptable, tolerated and adhered to? How many withdraw?
6. What level of supervision intensity is required for the main trial that will enable engagement at a level likely to be effective at preventing falls? Or whether, in practice, level of supervision can be matched to individual participant characteristics.
7. What proportion continue to adhere during 24 months follow-up?
8. Are there unexpected or adverse consequences?
9. Can the investigators collect trial health status and falls data at baseline and follow-up? Does blinding work? Is the assessment schedule too burdensome?
10. Are our sample size assumptions correct? The feasibility trial will act as an external pilot trial. The investigators have chosen a randomised controlled trial design to minimise researcher bias during the allocation to groups and to enable differences at follow up to be attributed to the therapy rather than the characteristics of the participants involved. A control arm has been included of standard falls prevention, as requested by the NIHR.

Participants will either be recruited through Memory Clinics (the initial discussion about the study will be by the nurse or doctor in the clinic), or through the 'Join Dementia Research' dementia research volunteer register. If the patient is interested in taking part in the research, a researcher will visit them in their own home at a convenient time, assess their capacity to consent (The investigators will only recruit patients who have capacity to give informed consent), ensure they meet the eligibility criteria and then take informed consent. The investigators will collect information about participants in an interview conducted in their own home.

Baseline assessment

Those agreeing to take part will be screened for suitability and a baseline assessment undertaken at home or in a clinic setting. This will include:

1. demographic and contact details, carer demographic and contact details
2. Medical and falls history, including previous fractures, recent hospitalisation, and drugs taken
3. falls risk factors, including vision and lying and standing blood pressure
4. CANTAB neuro-psychological assessment
5. outcome variables (DAD activities of daily living scale, Nottingham Extended ADL Scale, IPAQ activity questionnaire, EQ5D and DEMQol quality of life scales (participant and proxy), short FES-I falls efficacy (fear of falling) scale, Hospital Anxiety and depression scale (HADS), muscle strength, Berg balance scale, Timed Up and Go test, SHARE frailty instrument) .
6. Motivation and adherence factors - Self-Reported Habit Index, BREQ-3 regulations in exercise questionnaire, Basic Psychological Need Satisfaction and Frustration Scale, Participants perceptions of healthcare providers motivation communication style, healthcare providers (clinicians) perceptions of their own motivational communication style.
7. Clinician's work motivation (Work Extrinsic and Intrinsic Motivation Scale) and Basic Psychological Need Satisfaction and Frustration Scale.
8. Carer strain (caregiver strain index).
9. Participant and carer service use (CSRI).

Following the baseline assessment, participants will be individually randomised, stratified by site, co-resident carer and history of previous falls, using an allocation algorithm accessed by a secure web portal to the system held at the clinical trials unit NWORTH, in Bangor University. The randomisation system will be maintained by a statistician independent of the analysis and research teams to ensure blinding of analysis. Clinical researchers will inform the participant about the treatment and follow-up plan and arrange the clinical assessments (allowing masking of the RAs to treatment arm). Access to the study website will be password protected and only accessible by authorised individuals.

Due to the nature of the intervention, blinding of intervention is impossible for participants and therapists administering it. Outcome assessment and statistical analysis will be blind to allocation.

Once randomised, the participant will start one of the activity programmes (see below).

The intervention will be for 12-months. Follow up will be at 12 months, with persistence (and hospitalisation) for 24 months. Participants will be asked to keep a detailed daily diary including activities and exercises undertaken. Participants will be telephoned monthly to prompt return of this diary. Health and social care services received will be collected monthly and data on habit formation will be collected alternate months throughout the trial period by telephone. Data on participant's perceptions of motivation communication will be collected by telephone at months 1, 3, 6. The investigators will pilot and refine the measures to be used in the full assessment battery, including cognition and quality of life. The investigators will measure withdrawal, outcome measures, simple fitness, strength and balance variables, and acceptability of the intervention and research interview schedule. The investigators will consider the impact of patient characteristics (whether they live alone, severity and type of cognitive impairment, co-morbidity) on likelihood of completing self-directed or carer-supported exercise.

The investigators will ascertain falls by diary or calendar, supervised by the carer or prompted by weekly SMS text messages or telephone calls, and data from GPs, hospitals and ambulance service, blind to allocation. The investigators will ensure that falls ascertainment is appropriate for a population with cognitive impairment, and is not biased across intervention arms (e.g. by therapists or RSWs prompting recall of falls).

Researchers will visit participants at home after 12 months (+/- 2 weeks) to complete the Disability Assessment for Dementia (DAD) ADL scale,, the Nottingham Extended ADL Scale, IPAQ activity questionnaire, and 3-item CANTAB cognitive tests, short falls efficacy scale (fear of falling, FES-I), DEMQoL and EQ5D quality of life questionnaires, muscle strength (Lafayette dynamometer), single- and dual-task Timed Up and Go (TUG) test, Hospital Anxiety and Dementia Scale (HADS), SHARE frailty instrument and the Berg Balance Scale, resting and post-exercise pulse rate. The investigators will ask participants to wear pedometers in weeks 1, 26 and 50 as an objective measure of activity. The investigators will ascertain health and social care use, in particular injurious falls, and hospital admissions using electronic healthcare records and the Client Service Receipt Inventory (CSRI). The investigators will use an interviewer-delivered questionnaire (which will be piloted) to ask about the acceptability and ease of adherence with the programme. Researchers collecting follow up data will be blind to allocation.

Participants will be sent a brief questionnaire at 18 and 24 months asking about persistence with exercise and advice. The nominated carer will be notified and asked to supervise completion, or support to complete it offered by telephone or home visit. Hospital visits up to 24 months will be ascertained from hospital electronic administration records.

Inclusion Criteria:

1. age 65 or over (no maximum)
2. a diagnosis of dementia or an assessment of MCI (of any subtype), attendance at a memory assessment service, or on the Join Dementia Research register, with a cognitive score in the range defined, with no other apparent cause for cognitive impairment
3. Able to walk without human help
4. Able to communicate in English.
5. Able to see, hear and have dexterity sufficient to perform neuropsychological tests
6. Capacity to give consent to participate, and agreeing to do so
7. Montreal Cognitive Assessment (MoCA) 15-25 or standardised Mini-mental state examination (sMMSE) 18-26 or Addenbrookes Cognitive Examination (ACE-III) 60-94 .

Carer participants will be spouses, family members or others in a caring relationship who see the patient participant most weeks, or speaks by telephone most weeks, are willing to take part and can communicate in English.

Exclusion Criteria:

1. Co-morbidity preventing participation (e.g. severe breathlessness, pain, psychosis, Parkinson's or other severe neurological disease)
2. Life expectancy of less than 1 year
3. Likely to be unable to undertake the intervention regularly (e.g. planned elective surgery, planning to move away or commitments elsewhere).
4. Unable to communicate in English.