Cognitive Stimulation in Activities of Daily Living for individuals living with mild-to-moderate dementia (CS-ADL): a feasibility study

Study design

A non-randomised, feasibility design with a pre-post evaluation of intervention outcomes was conducted. The study ran from November 2023 to June 2024. While planned, randomisation and between-group comparisons were not feasible due to slow and differing rates of recruitment across study sites, which will be discussed further in the results section. Participants were instead non-randomly allocated to either intervention (CS-ADL) or control conditions (TAU). It was not possible to blind participants to treatment allocation due to the nature of intervention. Blinding of outcome assessors was not feasible for this study due to resource and practical constraints.

Participant sample

Participants were eligible for inclusion if they had a formal diagnosis of major neurocognitive disorder (dementia) (American Psychiatric Association, 2013) and mild-to-moderate cognitive impairment classified by the Montreal Cognitive Assessment (MoCA; ≥ 10) (Nasreddine et al., 2005). Participants were required to have some ability to communicate and understand communication, determined by a score of 0 or 1 on questions 12 and 13 of the Clifton Assessment Procedures for the Elderly-Behaviour Rating Scale (Pattie & Gilleard, 1975). Exclusion criteria included the presence of pre-morbid intellectual disability, or significant visual/hearing impairment that would interfere with participation in a group intervention. Individuals with significant physical illness/disability that would interfere with group participation were excluded i.e. medical instability, terminal illnesses, severe cardiac or respiratory conditions, or conditions that severely limit mobility. Individuals with uncontrolled disruptive behaviours that would interfere with group participation, such as aggression, were also excluded. This criterion was assessed informally by gatekeepers using clinical reasoning. Participants taking dementia medication continued to do so during the study. As several outcomes were proxy-rated, caregivers were required to be available during data collection procedures.

A sample size of 34 participants was calculated to achieve 80% power for detecting a two-point mean difference, aligning with recommendations for sample size of pilot and feasibility studies (Julious & Owen, 2006).

Recruitment and consenting procedures

Community-dwelling individuals with dementia were initially recruited from primary care, memory technology resource rooms (MTRR), and psychiatry of later life services across four counties in Ireland. Where challenges with recruitment arose, additional ethical approval was obtained to supplement recruitment through the Alzheimer Society of Ireland (ASI), outreach events, print and radio advertisements.

The consent process for this study aligned with Irish capacity legislation (Assisted Decision-Making (Capacity) Act, 2015), the Health Service Executive (HSE) National Consent Policy (2022) and the Data Protection Act 2018 Health Research Regulations (Government of Ireland, 2018). Written informed consent was sought from the individual with dementia to participate in the study with the automatic presumption of capacity. While it is possible that individuals contacted regarding the study had challenges in their ability to understand, retain, use or weight, and communication information during the decision-making process, it was not within the scope nor ethical remit of this study to request a functional capacity assessment related to the individual’s decision to participate in the research. If an individual refused consent to participate, no further actions were taken to uphold and respect their dignity and autonomy. Nevertheless, all actionable steps were taken to support the individual’s decision-making process. For example, all participant-facing documentation followed dementia-friendly document guidelines. Furthermore, the individual’s primary caregiver was informed of the research where possible.

Outcomes and data collection procedures

Primary outcomes of this study related to various feasibility indicators (recruitment, retention, adherence, intervention integrity, adverse events). Acceptability of CS-ADL has been evaluated previously (Ryan, Chockalingam & Brady, 2024a). The total number of people screened and recruited to the study were recorded, alongside the total number of dropouts and their reasons. The number of intervention sessions each participant attends, as well as the number of adverse events that occurred, were recorded. Any deviations from the intervention manual were recorded to capture intervention integrity.

Secondary outcomes of this study related to intervention outcomes, including ADL functioning, cognition, QOL, communication, neuropsychiatric symptoms, and occupational performance within a group setting. Secondary outcomes were measured in-person at baseline and within one month of study completion. Data were collected at recruitment sites by the principal investigator.

ADL functioning was measured using the caregiver-rated Alzheimer’s Disease Cooperative Study-Activities of Daily Living (ADCS-ADL) scale (Galasko et al., 1997). The ADCS-ADL was chosen as it is disease-specific and comprehensively addresses all relevant ADL outcomes for people living with dementia. It has also demonstrated high internal consistency and high sensitivity to functional changes in mild-to-moderate dementia.

Cognition was measured using the Repeatable Battery for Assessment of Neuropsychological Status (RBANS) (Randolph et al., 1998), assessing immediate memory, visuospatial/constructional ability, language, attention and delayed memory. Alternate forms of the RBANS were used pre- and post-test to minimise practice effects. The RBANS has good reliability and adequate validity indicators, with high sensitivity and specificity in dementia. As blinding was not possible for this outcome domain, standardised manual instructions were followed to minimise bias.

QOL was measured using the Quality of Life-Alzheimer’s Disease (QOL-AD) (Logsdon et al., 1999), completed through both patient self-report and caregiver ratings. The QOL-AD has demonstrated good psychometric properties and is recommended to measure QOL in psychosocial intervention research in dementia care (Moniz-Cook et al., 2008).

Communication was measured using the caregiver-rated Holden Communication Scale (HCS). The HCS is commonly used to evaluate communication outcomes of cognitive stimulation interventions (Woods et al., 2023) and has demonstrated good psychometric properties for people with dementia, with high internal consistency and test-retest reliability.

Severity of NPS and associated caregiver distress were measured using the caregiver-rated Neuropsychiatric Inventory-Questionnaire (NPI-Q). The NPI-Q has demonstrated adequate test-retest reliability, convergent validity and internal consistency, and is suitable for measuring mood and depressive symptoms in dementia patients.

Occupational performance within a group setting was measured within the treatment group using the Occupational Therapy Task Observation Scale (OTTOS) (Margolis et al., 1996). The OTTOS was used as a supplementary measure to aid interpretation of the overall effect of CS-ADL on everyday functioning.

Treatment conditions

The intervention arm received CS-ADL, an occupational therapist-led, group cognitive stimulation program that has been described elsewhere (Ryan, Chockalingam & Brady, 2024a). An overview of the weekly session plans is provided in Table 1. In brief, CS-ADL is a multi-component group programme, with each session consisting of cognitive stimulation, physical activity and ADL practice. Sessions involve stimulation of a range of ADLs including dressing, grooming, meal preparation, laundry and leisure. CS-ADL sessions lasted approximately two hours and were delivered once weekly for seven weeks at two centres. The principal investigator delivered the intervention independently at one site, with assistance from occupational therapy and nursing doctoral students at the second site. Intervention was delivered in the occupational therapy kitchen and ADL room at each site, with an OT kitchen, dining, bed and bathroom area in situ at each centre. A separate room was made available at the second site for caregivers of CS-ADL participants; this was not provided at the first site due to space constraints.

The control arm of the study received TAU. At follow-up, control group participants were offered a home programme with cognitive stimulation resources.

Analysis

Participant characteristics were analysed using descriptive statistics via Statistical Package for Social Sciences (SPSS) version 28.00. A CONSORT flow diagram was used to present information relating to the feasibility indicators of recruitment and retention. Mean differences and 95% confidence intervals were calculated for secondary outcome data.

Ethical approval and informed consent statements

Ethical approval was obtained from the Health Service Executive North East Research Ethics Committee (REC/23/042), the Reference Research Ethics Committee Midlands Area and Corporate (RRECB1223SR), and the Saolta Clinical Research Ethics Committee (C.A. 3138). The consent process for this study adhered to the Irish Assisted Decision-Making (Capacity) Act (2015). Written informed consent was obtained from each participant.

Recruitment. randomisation and retention

Thirty individuals were screened, with 16 participants enrolled in the study during the study period (November 2023–June 2024). Recruitment ceased due to time constraints. Reasons for non-enrolment included no formal diagnosis of dementia (n = 5; 16.67%), presence of severe cognitive impairment (n = 1; 3.33%), and self-withdrawal due to medical instability, onset of uncontrolled neuropsychiatric symptoms, change of interest, or participant self-reported feelings of unsuitability for intervention (n = 8; 26.67%).

Recruitment windows differed across sites due to delays in ethical approval. Figure 1 provides a breakdown of recruitment windows at each site, with further detail of the source of participant referral to the study.

Figure 1. Results of recruitment at each site.

It was initially planned to randomise participants into intervention and control conditions at each site; however, randomisation was not feasible due to the slow and differing rates of recruitment across study sites. Ultimately one site was allocated as the control condition as there were insufficient numbers to randomise within each recruitment site, nor to complete cluster randomisation. Recruitment was ultimately ceased due to time constraints.

Thirteen participants were thus allocated to the treatment group and three to the control group. One participant allocated to the treatment condition withdrew following baseline assessment, and another was lost to follow-up. In the control condition, it was not possible to complete follow-up with one participant due to logistical challenges, and another was lost to follow-up due to hospital admission, resulting in a total dropout rate across both study arms of 6.25%.

A CONSORT flow diagram (Eldridge et al., 2016) depicts the flow of participants through the study ( Figure 2).

Figure 2. CONSORT flow diagram.

Participant characteristics

The mean age of the sample was 77.4 years, with eleven women and five men enrolled in the study. Most study participants reported a diagnosis of Alzheimer’s disease (n = 6; 37.5%). Remaining participants reported a diagnosis of vascular dementia (n = 2; 12.5%), mixed dementia (n = 1; 6.25%), frontotemporal dementia (n = 1; 6.25%), primary progressive aphasia (n = 1; 6.25%) and an unspecified aetiology (n = 5; 31.25%). The mean MoCA score was 20.27 (SD 4.5). All participants were of Irish ethnicity, with only two participants living independently. Comparative analyses were not performed due to sample size disparity. Further demographic details of study participants are provided in Table 2.

Outcomes

Due to the difference in sample size across study arms, differences between groups at follow-up were not analysed. Table 3 details the treatment group’s baseline and follow-up outcomes. Table 4 outlines the sole control group participants results.

For cognitive domains, while confidence intervals were wide and crossed the null suggesting imprecision, magnitude of the mean differences may indicate a potential beneficial effect of intervention. For instance, improvements were observed in RBANS total score (mean difference = 2.80, 95% CI −3.55 to 9.15), immediate memory (mean difference = 5.00, 95% CI −3.76 to 13.76), attention (mean difference = 4.70, 95% CI −6.22, 15.62), language (mean difference = 4.20, 95% CI −1.85, 10.25), delayed memory (mean difference = 2.20, 95% CI −5.61, 10.01).

A reduction in caregiver distress as measured by NPI-Q was observed (mean difference = −2.82, 95% CI −5.52 to −0.12).

For several outcomes, no meaningful change from baseline to follow-up was identified within the treatment group: ADCS-ADL (mean difference = 0.18, 95% CI −3.21 to 3.58); HCS (mean difference = 0.00, 95% CI −1.78 to 1.78); QOL-AD (mean difference = −1.80, 95% CI −6.31 to 2.71); NPI-Q symptom severity (mean difference = −0.45, 95% CI −3.06 to 2.15); OTTOS (mean difference = 2.22, 95% CI -11.16, 15.61).

Intervention adherence

The mean attendance for CS-ADL sessions was 5.75 out of 7. Reasons for non-attendance included illness, mood instability, forgetting, or conflict with medical appointments. No dropouts occurred after intervention commencement.

Intervention integrity

Intervention integrity was maintained across study sites concerning the implementation of essential content of CS-ADL. Some activities were adapted to ensure appropriate engagement of all participants; this is inherent to the principles of cognitive stimulation and CST (Spector et al., 2020). For example, modification was made to the ADL practice component of session six (theme: ‘Baking) of one CS-ADL group as there was an unequal balance of men and women in this group. It was decided to allow participants to choose between baking and a DIY activity to accommodate for meaningful engagement of each gender. Despite this modification, the core function and purpose of CS-ADL remained intact and is within the parameters of intervention fidelity.

Adherence to the intervention protocol was impacted by the availability of equipment and resources at each site. Furthermore, timing of sessions impacted adherence to the intervention protocol; it was ensured that sessions did not exceed the two-hour timeframe, meaning that some activities had to be excluded. Any excluded activities were later reintroduced in following sessions when time allowed. Reality orientation activities were also excluded from one CS-ADL group following participant requests to discontinue these activities as the purpose was unclear, and it was not deemed relevant or meaningful to them.

Adverse events

One adverse event was reported in the intervention group; one participant choked on food prepared in an intervention session. The Heimlich manoeuvre was performed and was successful. An incident report form was filed, and a speech and language therapy referral for review of the participant’s swallow was completed. The participant’s caregiver was informed.

Discussion of trial feasibility

Recruitment to this study proved challenging, failing to meet the proposed sample size within the study timeframe. Recruitment difficulties in non-pharmacological dementia trials are well-documented (Field et al., 2019). Strict eligibility criteria and the high prevalence of underdiagnosis in community services compounded these challenges (Beattie et al., 2018; Begley et al., 2023). However, restrictive inclusion criteria ensure scientific rigour.

Healthcare professionals’ competing priorities further hindered recruitment, a common issue in under-resourced health systems as observed in the Irish health service (Fleming et al., 2022). Additionally, short, inconsistent recruitment windows across sites approval may have exacerbated recruitment challenges. This was due to delays in ethical approval; the present ethical review process in Ireland involves several dozen committee operating at a local or institutional level. Applications were made to three separate ethical committees for approval of this study, with inconsistencies noted across each submission and review process, subsequently impacting the timeline of the study. However, supplementary recruitment strategies used in this study, such as community outreach, patient registries, and media engagement yielded successful results.

Caregiver involvement in data collection may have impeded recruitment, as caregivers may perceive research participation as an additional burden (Whitebird et al., 2011). Provision of incentives can help minimise this burden. For instance, room space and refreshments were provided to caregivers of one intervention group, potentially alleviating burden in relation to travel and offering peer support opportunities.

The study’s non-randomised design was a deviation from initial study protocol due to recruitment challenges. While initially intended, randomising within each site was impractical, reflecting challenges noted previously in clinical research (Colliver & McGaghie et al., 2008). Instead, participants were non-randomly allocated depending on their intervention site, and due to differing rates of recruitment, resulting in low numbers within the control group. While non-randomised designs are pragmatic, they introduce biases and confounds that threaten study validity (Colliver & McGaghie et al., 2008). Future research should consider cluster randomisation to enhance practicality and rigour of the methodology. Previous non-pharmacological dementia intervention trials, including cognitive stimulation, have demonstrated the feasibility of this approach (Jøranson et al., 2015; Mraz, Tong & Liu, 2023).

Intervention feasibility varied by site. Concerns regarding the safety and quality of facilitation arose when only one facilitator was present. Some participants needed extra encouragement or activity adaptation, which was challenging for a sole facilitator to undertake successfully. These issues were mitigated with additional co-facilitators. A minimum of two facilitators is thus recommended for CS-ADL groups larger than three participants, aligning with guidance for the widely implemented CST. While CS-ADL must be led by an occupational therapist, co-facilitators can be from other professions like nursing or psychology, as intervention may also align with the service goals of these professions.

Intervention fidelity concerns emerged when session components were omitted due to time or resource constraints. Participant feedback at one site also led to the exclusion of reality orientation activities, as they felt activities were irrelevant to their daily lives. Reality orientation has previously been suggested to cause frustration in dementia (Savorani et al., 2004). Reconfiguration of the intervention structure and timings is recommended to maintain fidelity in future trials.

Discussion of study outcomes

Comparing outcomes between intervention and control groups was not possible. Within the intervention group, CS-ADL did not demonstrate a meaningful improvement in ADL functioning, a contrast from similar multi-faceted cognitive stimulation interventions (Palomares et al., 2021). This may be attributed to the small sample size. Furthermore, reliance on within-group analyses limits the strength of conclusions; cognitive stimulation has been shown to contribute to a delay in deterioration when compared to a no treatment control (Chapman et al., 2004).

In addition, the ADCS-ADL may not align with the aim of CS-ADL; it is possible that CS-ADL does not contribute to direct improvements in ADL performance, but rather to an enhancement of occupational engagement and participation, as previous qualitative research suggests (Liu et al., 2020). Few measures of everyday occupational engagement and participation in dementia exist, thus qualitative evaluation may better capture the subjective dimensions of these constructs.

While improvements were observed in several cognitive domains within the treatment group, results are imprecise and uncertainty remains. These findings contrast with previous research that shows small, short-term cognitive benefits from cognitive stimulation (Woods et al., 2023). The RBANS may be confounded by the tool’s dependence on language comprehension and expression, abilities often impaired in dementia. Furthermore, neuropsychological measures may not be the most appropriate for evaluating the effect of CS-ADL on cognitive abilities, as they are not reflective of everyday performance. CS-ADL targets cognitive functions such as initiation, organisation, and planning of everyday activities; processes that underpin the construct of ‘functional cognition’. Measures of functional cognition may thus be more appropriate in future research.

While no changes in NPS severity were found, a reduction in caregiver distress due to NPS was observed. This finding is supported by previous research where caregivers report lower burden following their care-recipient’s participation in cognitive stimulation (Farina et al., 2007). Providing room space to caregivers might have alleviated distress through peer-support opportunities. As caregiver wellbeing impacts the quality of care provided to persons living with dementia, future implementation of CS-ADL should consider the provision of opportunities for caregiver peer-support, which may in turn enhance recruitment and retention.

No changes in QOL or communication were identified post-intervention. These findings contrast with literature (Ryan, Chockalingam & Brady, 2024a; Woods et al., 2023; Chao et al., 2020). Outcome measures may have been insensitive to post-intervention changes. In addition, reduced quality of intervention facilitation at one site may have impacted participant engagement, thus affecting gains in communication. Ensuring an appropriate therapist-participant ratio is crucial for future research.

Lastly, no changes in occupational performance were identified. Several OTTOS items including ‘engagement’, ‘quality of work’ and ‘activity level’ often depended on participants interest and prior experience with intervention activities. Participants lacking prior experience or interest may thus have received lower OTTOS ratings, possibly distorting overall group mean. Additionally, factors like fluctuations in mood, poor sleep and physical health impacted performance. Future trials should consider alternative observational tools validated for dementia.

Limitations

The study’s validity and generalisability are limited by several factors. Outcomes were assessed by the intervention facilitator, and thus not blinded to treatment allocation, risking detection bias. This was due to resource constraints common within feasibility studies, and future trials will ensure blinding of outcome assessors to enhance validity of findings. The sample size was not reached, limiting analyses and conclusions regarding CS-ADL’s effect on study outcomes are therefore limited. However, the primary aims of this study were to evaluate the feasibility of the methodology, therefore findings of this study are not anticipated to be confirmatory. In addition, the lack of Public-Patient-Involvement (PPI) questions the study’s relevance to the population. It is acknowledged that variation in co-facilitators available to support intervention at each site may have influenced overall feasibility and intervention outcomes, as discussed above. However, this was due to practical and resource constraints and provides valuable insights for feasibility of intervention delivery for future trials and implementation.

Implications for future practice and research

This research furthers the evidence-base for CS-ADL and similar occupational therapy-led cognitive stimulation interventions. However, further development and evaluation of CS-ADL is required prior to a definitive trial, as several uncertainties remain in relation to outcome assessment, study design and intervention, therefore requiring further evaluation.

The challenges encountered in recruitment and retention highlight the need for careful consideration in the design of future trials evaluating CS-ADL in people living with mild-to-moderate dementia. Future research should ensure a comprehensive recruitment strategy, with consideration of alternative methods of randomisation to enhance feasibility, for example cluster randomisation.

Public-Patient-Involvement (PPI) should be integrated within the design and conduct of future CS-ADL research to aid development of recruitment and retention strategies, advise on study design, and oversee modification of intervention protocol to ensure meaning and relevance to the participant population.

Alternative methods of data collection are also recommended for future research, with consideration of measures of functional cognition and occupational engagement, alongside consideration of a mixed methods approach to evaluation to enhance understanding of the effects of intervention.