Estrategia

Baseline data on the participants (persons with dementia)

Potential participants were visited at home by a registered research nurse or physiotherapist. Informed consent was obtained (seeConsent), then eligibility was checked, including carrying out the sMMSE. Once the participant was confirmed as being eligible, descriptive data were collected, including age, gender, marital status, ethnicity, educational attainment and employment status.

The next stage was to collect the data for the primary and secondary outcomes, as given inTable 3. The type of dementia (vascular, Alzheimer’s or mixed) was ascertained after the baseline visit by reference to the participant’s medical notes.

The primary outcome was the global cognition score of the Alzheimer’s Disease Assessment Scale–Cognitive Subscale (ADAS-Cog).70_{The funder’s brief was clear that the primary purpose of the trial and of the exercise}

programme should be to determine whether or not exercise can modify cognitive functioning. Cognitive deficits are central to dementia and widely understood as the most important treatment target.77_The

ADAS-Cog was collected directly from the participant, takes about 30–40 minutes to administer and has established sensitivity to change. It is widely considered the gold standard primary outcome in treatment trials for dementia, with relatively well-established treatment effect sizes.78_{We also collected the maze and}

number of cancellation optional items of the ADAS-Cog as additional items to be reported separately. Initially, the primary outcome measure was to be the sMMSE, which, although widely used in clinical evaluation, is a very global, relatively insensitive measure of cognitive function. The ADAS-Cog is

acknowledged to be more sensitive in detecting cognitive change in individuals with MMD than a variety of other measures, including the sMMSE.79_{The use of a valid but more sensitive primary outcome measure}

(the ADAS-Cog) made it possible to reduce the sample size of the study.

Secondary outcomes were chosen to reflect the broad impact of dementia on function, behaviour and quality of life. When possible, we chose instruments that were dementia specific, well validated and not excessively burdensome. Secondary outcomes are also detailed inTable 3. In all instances we used the published guidance about who the primary respondent for the questionnaire should be. We used the Bristol Activities of Daily Living Scale (BADLS).71_{This carer-rated instrument of participant ability is dementia}

specific, sensitive to change and widely used in clinical trials. We also collected data using the Quality of Life in Alzheimer’s Disease (QoL-AD) scale.73_{This is a 13-item dementia-specific scale that can be}

completed by a carer or participant; both were collected for the DAPA trial but the participant response was considered the primary data source. We also collected the EuroQol-5 Dimensions, three-level version (EQ-5D-3L),72_{a 5-dimension generic (i.e. not dementia-specific) measure of health-related quality of life}

(HRQoL). The EQ-5D-3L was reported by both the participant and carer, with the participant being the primary data source, and it allows a calculation of health utilities for application in economic evaluations. We used the Neuropsychiatric Inventory (NPI),74_{which includes important predictors of carer breakdown}

such as depression and agitation. We collected cost data using the Client Services Receipt Inventory TABLE 3 Outcome measures

Domain Measure Description Completed by

Primary

Cognition ADAS-Cog70 _{Takes 30–40 minutes to complete. Includes} 11 tasks targeting three domains (memory, language and praxis). Scores range from 0 to 70 points, with higher scores indicating greater cognitive impairment. A 4-point difference is considered clinically important79

Participant

Secondary

Function BADLS71

Takes 15 minutes to complete. Includes 20 daily activities. Scores range from 0 to 60 points, with higher scores indicating greater impairment

Carer (rating participant)

HRQoL EQ-5D-3L72 _{Takes a few minutes to complete. Includes} health state classification system with five dimensions and a VAS thermometer. Scores on the classification system range from 0 to 25, with higher scores indicating better quality of life. Scores on the VAS range from 0 to 100, with 100 equating to the best health state. These two scores can be combined into an index value 0.0_–1.0, the higher value indicates better quality of life

Participant (rating self) Carer (rating self) Carer (rating participant)a

Dementia quality of life QoL-AD73 QoL-AD proxy

Takes 10–15 minutes to complete. Includes 13 items. Scores range from 13 to 52 points, with higher scores indicating less impairment

Participant (rating self) Carer (rating participant)a Behavioural symptoms NPI74

Takes 10 minutes to complete. Includes 12 behavioural domains. Scores range from 0 to 144 points, with higher scores indicating greater impairment

Carer (rating participant)

Carer burden ZBI75

Takes 5 minutes to complete. Includes 22 items regarding direct stress to carers. Scores range from 0 to 88 points, with higher scores indicating greater stress

Carer (rating self)

Health- and social-care usage to inform the health economics analysis

CSRI76 _{Administered by trained assessor. Takes} 20 minutes to complete. Includes 29 items covering five domains regarding information about use of health- and social-care services, other economic impacts (such as time off work because of illness) and sociodemographic information. Used to inform health economic study

Carer with participant (rating participant)

ADAS-Cog, Alzheimer’s Disease Assessment Scale–Cognitive Subscale; BADLS, Bristol Activities of Daily Living scale;

CSRI, Client Services Receipt Inventory; EQ-5D-3L, EuroQol-5 Dimensions, three-level version; HRQoL, health-related quality of life; QoL-AD, Quality of Life in Alzheimer’s Disease; NPI, Neuropsychiatric Inventory; VAS, visual analogue scale; ZBI, Zarit Burden Interview.

(CSRI).76_{This detailed questionnaire is designed to be used with a carer and covers all social, health care,}

medication use and out-of-pocket expenses. Initially, mood was to be a secondary outcome, as measured by the Cornell Scale for Depression in Dementia.80_{This was removed prior to the collection of any data}

as mood is covered by the NPI and it would have added unnecessarily to participant burden during data collection.

Carer

We recorded carer age, gender, ethnicity, details about the relationship they had with the person with dementia and how much care they provided. Carers were asked to complete the Zarit Burden Interview (ZBI)75_{and the EQ-5D-3L}72_{to assess their own HRQoL. These outcomes are all among those recommended}

by a consensus recommendation of outcome scales for non-drug interventional studies in dementia.81