Parkinson’s-adapted cognitive stimulation therapy: feasibility and acceptability in Lewy body spectrum disorders

Background Drug-based therapeutic approaches for Parkinson’s disease dementia (PDD) and dementia with Lewy bodies (DLB) are moderately effective and not always tolerated. Tailoring psychosocial approaches in PDD and DLB may offer additional support and improve outcomes. We adapted home-based, care partner-delivered Cognitive Stimulation Therapy (CST) for individuals with PDD or DLB and their care partners (CST-PD). Objectives To evaluate the feasibility, acceptability, and tolerability of CST-PD. Methods This randomised controlled trial used mixed methods, including a process evaluation. People with PDD, DLB or mild cognitive impairment in PD (PD-MCI) and their care partners were randomised to 12 weeks of treatment as usual (TAU) or CST-PD. Outcomes were feasibility of the study conduct (i.e., recruitment, retention rate) and acceptability and tolerability of the intervention. Measures included rating scales, researcher field notes, therapy diaries, and exploratory clinical and care partner efficacy measures. Results The recruitment target was met with 76 consenting participant-dyads. Retention in both arms was high at over 70%. More than 90% of dyads undertook discrete sessions greater than 20 min duration, but the average number of sessions completed was lower than the recommended dose. Acceptability ratings (i.e., interest, motivation and sense of achievement) of the intervention were high. Participants reported no serious adverse events related to the intervention. Conclusions The field of psychosocial interventions for PDD and DLB is newly emerging, and we demonstrated that this type of intervention is acceptable and well tolerated. Evaluating its clinical effectiveness in a full-scale randomized controlled clinical trial is warranted. Trial registration number The trial is a psychosocial intervention with an allocated ISRCTN number 11455062. Electronic supplementary material The online version of this article (10.1007/s00415-019-09329-6) contains supplementary material, which is available to authorized users.


Introduction
Each year about 11% of people with Parkinson's disease (PD) develop dementia (PDD), which is likely to triple in prevalence by 2060 [1,2]. The prevalence of dementia with Lewy bodies (DLB), which encompasses around 5% of dementias overall, is also increasing [3]. Treatment strategies for PDD and DLB have mostly focussed on medication-based interventions, such as cholinesterase inhibitors or memantine, with effectiveness of these treatments being modest at best. There is a need to find other management strategies specifically adapted for PDD, DLB and related cognitive impairment in PD (termed 'Parkinson's-related dementia'; PRD).
Cognitive Stimulation Therapy (CST) and cognitive rehabilitation therapy can enhance cognitive function, quality of life and care partner outcomes in people with non-Parkinson's-related dementias [4,5] but studies in PDD and DLB are limited [6][7][8]. We, therefore, designed the INVEST (INdiVidualised cognitivE Stimulation Therapy) programme to develop and evaluate a personalised psychosocial intervention specifically for people with PRD. The first phase of INVEST involved adapting the homebased version of CST known as individualised CST (iCST; [4]), for people with PRD and their care partners [9]. Here, in the second phase, we evaluated the newly adapted therapy, CST-PD, to inform a subsequent randomised controlled trial (RCT) of effectiveness. In accordance with the UK's Medical Research Council's guidelines for the development of complex interventions, [10] our objectives were to: (i) conduct an evaluation of the operational aspects of the study including aspects of recruitment, retention, data collection and study design; (ii) assess the acceptability and tolerability of the intervention; (iii) evaluate intervention fidelity, i.e., whether the intervention could be delivered, received and enacted as intended [11]; (iv) explore a possible signal of efficacy.
The first three objectives are reported here, whereas the exploratory pilot trial of efficacy (objective iv) is reported elsewhere [12].

Materials and methods of the trial
The full study protocol has been described previously [13]. Participant-dyads were recruited in the UK from four sites in Greater Manchester (primary site), and three sites in London, Derby, and Warrington. A CONSORT diagram [14] outlines the participant flow through the study, Fig. 1.

Eligibility
Participants with mild cognitive impairment in PD (PD-MCI), PDD or DLB were eligible if they were aged 18 years or older, had a diagnosis based on standard clinical diagnostic criteria [15][16][17] determined by the referring PD specialist, were living at home, and were on stable medication for at least 4 weeks prior to study entry. Care partners were eligible if they were 18 years or older. Capability to participate was based on clinical impression during screening and informed by scores obtained on the Unified Parkinson's Disease Rating Scale-motor examination [ 20]. Participant-dyads were excluded if they could not understand conversational English, were non-literate, had a severe physical or psychiatric illness that precluded participation, or were taking part in another psychosocial intervention study.

Study design, randomisation and ethics
This was a single-blind parallel arm, exploratory pilot trial with an embedded feasibility study and process evaluation. An accredited clinical trials' unit undertook blocked randomisation to CST-PD, or treatment as usual (TAU), on a 1:1 participant-dyad level, without stratification. All participants provided written consent, either voluntarily or via a personal consultee declaration. Demographic data and descriptive clinical measures were collected from participant-dyads during screening. Outcome measures (see Supplementary Table 1: Outcome measure descriptions for people with Parkinson's-related dementia and care partners) were administered by assessors blind to treatment allocation. Assessors were asked to report instances of un-blinding to the study team. The study received favourable ethical opinion from the Yorkshire and The Humber-Bradford Leeds Research Ethics Committee (reference: 15/YH/0531).

Study interventions
Participants in the experimental arm were asked to complete 12 weeks (including a 2-week familiarization lead-in) of CST-PD, delivered at home by the care partner. The therapy involved participating in 30 min of cognitively stimulating activities, two to three times per week. The activities varied in theme and complexity and could be tailored to suit individual needs (see Supplementary Table 2: Themes and topics from the CST-PD manual). The TAU services, offered by the National Health Service in England, were available to those in the experimental arm. At the end of the intervention period, participant-dyads allocated to TAU were offered the opportunity to receive the intervention without data being collected.

Assessment battery
Clinical and care partner measures of potential efficacy, including measures of cognition, behavior, well-being and quality of life (Supplementary Table 1 and described in detail elsewhere [13]), were administered at baseline and immediately after the intervention period.

Feasibility, acceptability, and tolerability evaluation framework
To evaluate INVEST we used a modified version of the ACCEPT [21] checklist based on Thabane et al.'s guidelines for reporting feasibility trials [22]. Our modified checklist provided a systematic basis for making decisions about whether or not to accept components of our study protocol 'as is' or whether to modify it, and in what way. We applied a series of operationalized definitions of feasibility, acceptability and tolerability, with a priori criteria, to arrive at a conclusion of 'accept', 'modify', or 'reject' (Table 2). These criteria were operationalized through a combination of pragmatic clinical judgements regarding acceptability in a population with PRD, as well as previous studies of nonpharmacological interventions for dementia and chronic mental illness [4,30]. In particular, the 'parent' study of INVEST, a full scale RCT of iCST in people with dementia not related to Parkinson's [4], provided valuable insight into appropriate interpretations of acceptability thresholds for the different aspects of the study. The ACCEPT framework was originally designed as a flexible approach that can be applied to pragmatic clinical trials for 'real life' populations regardless of specific diagnosis [2]. Thus, its application to the current study has been appropriate and useful.

Feasibility of the study procedures
'Feasibility' was operationalised through the question, 'can it work?' [23]. Parameters included: eligibility, recruitment  and retention rates, overall trial design (the degree to which the protocol balanced scientific and practical considerations), willingness to be randomised, blinding procedures and data collection (i.e., timing, quality, acceptability).

Acceptability of the intervention
This was defined as the extent to which the participant-dyads considered the intervention 'appropriate' [24]. We assessed care partner's perceptions of their partner's interest, motivation and sense of achievement following each therapy session. These aspects were rated in the therapy diary on a Likert-scale, consisting of the following grades: '1' = strongly disagree; '2' = disagree; '3' = neutral; '4' = agree; '5' = strongly agree (see Supplementary Table 3: Companion's CST-PD diary excerpt).

Tolerability of the intervention
Tolerability was defined as the ability to endure the intervention [25] and was captured by the number of serious adverse events (SAEs) related to either member of the participantdyad feeling burdened or frustrated.

Intervention fidelity (delivery, receipt and enactment)
Intervention delivery was operationalised through two questions, 'can researchers deliver therapy training to care partners?' and 'can care partners deliver the therapy to their partners?' Similarly, receipt was operationalised as, 'did care partners receive the therapy training as intended?' and 'did participants with PRD receive CST-PD as intended?' To evaluate enactment we asked, 'did participant-dyads complete the recommended amount of therapy during the intervention period?' Measures included a training checklist, a training evaluation form, a therapy skills self-assessment, and diary entries (acceptability ratings and free-text entries).

Qualitative investigation
We conducted semi-structured interviews at the end of the intervention with a purposive sample of 11 participant-dyads who undertook the therapy to gather evidence to support or refute our conclusions regarding the categorisations of 'accept', 'modify' or 'reject'. This sample size is considered sufficient to capture all emerging themes, as data saturation (rather than 'theoretical saturation') was likely to have been reached. We followed guidance suggesting that if participants are relatively homogeneous (e.g., as in our case, in which participants all fell within a single diagnostic spectrum and had a close age-range), then a purposive sample size approaching 12 will make it possible to 'render a fairly exhaustive data set' [26].

Sample size
We applied guidance on sample sizes for exploratory work [27,28] and took a conservative approach, estimating the standardised effect size on cognition of the parent form of the intervention to be 0.4, and utilizing a one-sided test and a less stringent significance level of 0.2 to avoid missing a promising effect. Thus, assuming 80% power and a correlation coefficient of 0.5 between baseline and endpoint on cognitive outcomes, the required sample size, including allowance for attrition, was 38 dyads per group at enrolment [13].

Data analysis
Quantitative data were analysed as descriptive statistics, qualitative data were analysed using thematic analysis [29]. Theme identification and initial coding were performed by authors SAM, SV and BK. The coding framework was further developed and themes defined through an iterative process by the INVEST investigators (SAM, SV, SJS and IL). To reach consensus, alternative meanings and interpretations were discussed in the context of all transcripts.

Feasibility of the study procedures
Findings from all data sources regarding the feasibility aspects of the study procedures are outlined in the modified ATTEND checklist, Table 1.

Recruitment/randomisation
The recruitment rate in the first 10 months was 0.8 dyads per site per month, which was lower than expected, and likely influenced by seasonal variation and holidays. Thus, three additional sites were added to enable us to recruit to target at a rate of 0.9 dyads per site per month, approaching the 'modify' threshold of 1-2 dyads per site. Of the 200 participant-dyads referred to the study team, 37 were ineligible, 11 were referred after recruitment had ended and 76 were unwilling to participate. Reasons for ineligibility included: no diagnosis of PD or cognitive impairment in PD; distance from the study centre; being admitted to a care home; and having a medical or psychiatric condition too severe to enable study participation. Reasons for nonparticipation by eligible people included: health issues; worsening dementia and apathy in the participant; care Semi-structured interviews with 11 participant-dyads who undertook the therapy g Most participants found the CST-PD enjoyable, usable, stimulating and versatile, confirming the acceptability and tolerability of the therapy (see Table 3). However, a number of barriers were identified, such as lack of time to deliver the intervention, poor communication skills and diseaserelated aspects preventing the dyad from engaging in therapy sessions

Modify
The timings of the assessment for each component were as follows: a ongoing throughout study, b informed consent, c baseline assessments, d therapy training, e 10-week intervention period, f follow up assessments, g after 10-week intervention, h study end A Based on findings in comparable studies investigating psychosocial interventions in dementia [4,30] partner burden; administrative errors leading to failure to contact the dyad; and unwillingness to be randomised. All participants (n = 76) had the capacity to consent to participation at the start of the trial. During the trial, four participants lost the capacity to consent and a consultee declaration form for continuation was completed by their care partners. Of the 76 participant-dyads consented, 38 were randomised to the CST-PD and 38 to TAU, none withdrew due to group allocation. These data suggest participantdyads were willing to be randomised.
Of the participants with PRD, 19.8% (n = 15) had a diagnosis of PD-MCI, 52.6% (n = 40) had PDD, and 27.6% (n = 21) had DLB. Of this group, 79% (n = 60) were male, 93.4% were white (n = 71) with a mean age of 74.5 years (SD = 6.74). Of the care partners, 84.2% (n = 65) were living with their study partner, 11% (n = 8) were male; 77.6% (n = 59) were spouses or partners, 17.1% (n = 13) were relatives and the remainder 5.3% (n = 4) included a live-in carer, a live-in divorcee, a friend and a grandchild. Baseline demographics, per arm, are shown in Table 2. There was some variability, notably the participant-dyads in CST-PD had a higher level of schooling and a slightly higher proportion of people with PDD. The average number of years members of dyads were known to each other was slightly lower in CST-PD. This case mix imbalance suggests that randomisation was not fully effective.

Eligibility criteria
Of the 200 participant-dyads referred to the study (see Fig. 1), 76 dyads were contacted and screened for eligibility. Of these, all met eligibility criteria and were subsequently randomised. This resulted in an eligibility rate of 36%, meeting the 'accept' threshold.

Retention
The retention rate was 73.7% and meets the 'accept' threshold; 56 of 76 consented participant-dyads completed the trial. Of those who were not followed up, 70% (14 of 20) were in CST-PD and 30% (6 out of 20) were in TAU. Out of the total consented in each inclusion group, 25% (10 of 40) who were not followed up had PDD, 20% (3 of 15) had PD-MCI, and 14% (3 of 21) had DLB. Deteriorating health was the main reason for withdrawal across arms and inclusion groups. Three dyads withdrew from CST-PD as they found it difficult to discuss topics spontaneously or did not like reminiscing.

Study design
We estimated that participant-dyads would remain in the trial for 15 weeks ± 2 weeks. This was achieved in TAU (15 weeks, 5 days), but not in CST-PD (18 weeks, 1 day). In CST-PD, the additional home-visit for therapy training introduced unforeseen delays mainly due to unavailability of participant-dyads.

Blinding
No instances of un-blinding were reported to the study team across sites.

Feasibility of outcome measures
The majority of outcome measures were feasible to administer; however, Goal Attainment Scaling (GAS [30]) was removed early in the trial as assessors reported that it was difficult to determine person-centred and measurable goals in the timeframe allocated. An early audit of the Addenbrooke's Cognitive Examination-III (ACE-III [31]) at the primary site identified inconsistencies in scoring between assessors. To improve accuracy and maintain consistency, all assessors received additional training and a scoring checklist. Missing outcome scale data for trial completers were minimal and met the a priori threshold for acceptability, with all scales having over 83% complete data. Subsequent analyses of the data revealed no differential missingness conditional on the baseline factors [13].
The therapy diary was feasible in terms of recording participant-dyads experiences; 91.7% of diaries were returned by study completers who undertook the therapy, entries were recorded for all completed therapy sessions. Qualitative investigation supported these findings, for example: "It would have been useful to have had more room for feedback on the diary, yes, um, 'cause I'd sort of squashed quite a lot in there." [Care partner, CS17].

Acceptability of the intervention
The intervention was acceptable, based on the care partners' perceptions of the extent to which (in each session) their partner was interested, motivated, and gained a sense of achievement. All a priori targets were exceeded; at least 75% of care partners rated the aspects within the target range (see Table 1). Qualitative findings support the diary ratings, for example: "[name] chose an article he had read in an aeroplane magazine and showed much more interest and enthusiasm in. It is a subject dear to his heart." [Care partner, CS17].

Tolerability of the intervention
Of the 9 SAEs reported, 5 were from CST-PD and 4 from TAU. None of the SAEs reported were due to the trial, suggesting that the intervention was well tolerated.

Intervention for TAU completers
All participant-dyads who completed their TAU arm (n = 32 dyads) were offered the intervention and 50% of dyads received the intervention. Data from these participants, however, was not collected due to ethical constraints.

Delivery and receipt of therapy training
The training protocol checklist revealed that 82.1% of researchers delivered CST-PD training as intended, the remainder completed all but the role-play exercise due to lack of time. The a priori target of CST-PD training being received as intended by the care partner was exceeded; after therapy training 95.6% of care partners (n = 23) rated their CST-PD skills within the 'accept' threshold. Furthermore, the training evaluation form, completed anonymously by care partners (n = 23), revealed that 95.4% found CST-PD training acceptable and 76.2% felt that they had received adequate instruction about how to deliver CST-PD.
Concerning the delivery of CST-PD by the care partner and receipt by the person with PRD, qualitative investigation revealed that some care partners struggled with personalising the activities, and/or coping with their partner's disease symptoms. For example: That said, the acceptability ratings for interest, motivation and sense of achievement (reported above) were high, suggesting that people with PRD received the therapy in the manner intended.

Enactment
During the intervention period, 64% of participants completed 11 or more sessions, with the average number of sessions being 1.76 (SD = 0.72) per week. The majority (92%) of participant-dyads completed sessions of longer than 20 min. Thus, the 'modify' target threshold for enactment was met. The most frequently cited barriers to enactment included disease symptoms and availability. For example: • "A barrier for you, dad, would have been two things. One that you wanted to do the session but I wasn't here to do one, so lack of availability of me, and second your health because your health varies so much day to day." [Care partner, CS9].

Qualitative findings
Eleven participant-dyads, who had completed the 10-week CST-PD protocol, were interviewed. Of these 11 dyads, one dyad completed the CST-PD intervention following their 10-week TAU participation. Of the people with PRD, six males had a diagnosis of PD-MCI, four had a diagnosis of PDD, and one had a diagnosis of DLB.

Discussion
This is the first report of a psychosocial therapy specifically adapted for people with PDD, DLB or PD-MCI and their care partners. The intervention, designed to be pragmatic, flexible and low cost, involved collaboration and cooperation between the person with PRD and their care partner. In sum, we found that undertaking a RCT of the intervention was feasible and that the intervention itself was acceptable and well tolerated by both members of the participant-dyad, although certain modifications will be needed in a future trial. For the components for the trial which we classified as 'modify' on the ACCEPT checklist, possible modifications for a future trial would include: (1) the trial design should include more flexibility in timing of visits since arranging home visits in older, potentially frail participants often incurs delays; (2) the recruitment target, while achieved, was ambitious, and a rate of one dyad per site per 6 weeks may be more realistic; (3) to account for variations in delivery of the intervention, as well as certain challenges faced by care partners in the role of therapists, inclusion of lay therapists or volunteer therapists should be considered; this notion was raised at a recent post-study Patient and Public Voice (PPV) event and met with general approval from the group; (4) strategies to increase the number of sessions per week from two to three could include more engaging, digitised presentations of the therapy sessions with dashboards to record progress and number of sessions achieved; and (5) certain components involving communication skills or a reliance on motor skills could be modified to obviate these aspects since this proved a challenge for certain participants with communication or physical impairments.
Recruitment was challenging because of the overall frailty of the study population; however, we achieved our threshold for feasibility, eligibility and retention. This was a key finding, particularly since our pre-randomisation refusal of the invitation to participate in the study was high (50%). It is likely that the relatively high degree of physical frailty of the study population, as well as extent of care partner burden, may have contributed to this high rate of refusal. Nonetheless, our recruitment target was reached since many potential participants, as well as referring clinicians recognised the importance of such an intervention, and few opportunities for non-medication-based interventions for people with PD-MCI/PDD/DLB exist. Additionally, most dyads who were initially allocated to the TAU arm expressed interest to receive the intervention following their TAU participation but the decline in their health made them rethink their decision. This highlights the challenges of conducting research in this ever-changing and frail population. Randomisation and blinding were both feasible, although slight imbalance in group characteristics may have been due to the relatively modest number of participant-dyads randomised or the lack of stratification by site. The majority of exploratory assessments were feasible. The majority of the study design procedures were also feasible. The process evaluation revealed that participantdyads in the CST-PD arm remained in the trial longer than expected due to difficulty in scheduling the therapy training session. A future study could offer a more flexible therapy training option using online or teleconferencing resources.
Regarding intervention fidelity, our quantitative data indicated that care partners had acquired sufficient knowledge to be able to deliver the therapy; however, our qualitative findings suggested some care partners experienced difficulty personalising and adapting the discussion cues provided in the manual. Our findings regarding enactment indicated that additional strategies are needed to boost adherence. One approach to address both of these aspects would be to include a trained, lay therapist to deliver CST-PD. The lay therapist, perhaps a community-based trained volunteer, could visit the participant-dyad once a week to host an additional session. Observing the lay therapist delivering the therapy effectively may increase therapy-delivery efficacy in the care partner through vicarious experience.
The qualitative findings suggested credibility of CST-PD and possible mechanisms by which this type of intervention may be effective: engendering a sense of mastery through implicit cognitive stimulation that builds on retained abilities and social stimulation with the care partner. For example: "Some of my childhood and things that had gone past, you forgotten all of that you're only assisted, until we sat down in here and we went through the sort of headlines and it was fantastic to thread those paths again and then I can compare with other people's experiences. [Participant with PRD, CS9].
This statement incorporates several potential theoretical mechanisms, both direct and indirect, which are consistent with the multi-faceted nature of the intervention. For example, from the perspective of 'pure CST', change may come about through indirect improvements in cognition resulting from being more relaxed, improved self-esteem due to partner encouragement, greater concentration and memory abilities [32,33]. From the perspective of rehabilitation, change may result from activation of unaffected neural pathways as well as optimisation of remaining abilities, such as in the mechanisms of cognitive rehabilitation therapy applied to dementia [34]. Finally, an important component here is the dyadic aspect and the work undertaken with the spouse, and the person-centred approach resulting from the tailoring of the intervention. The mechanisms underlying dyadic therapies related to improving communication, interpersonal relationships and even intimacy [35,36]. Quality of life, as a multi-dimensional outcome of such interventions, will be affected by all these factors: improved cognition and functional ability, reversal of apathy, and improved communication and relationships with care partners [33,37].
Our study had some limitations that need to be considered. Firstly, since CST-PD was delivered at home by a care partner, we were unable to account in detail for the quality and content of each therapy session, and this is one of the aspects of the intervention that will require modification. Secondly, the imbalance in the number of males compared to female participants may have influenced the reports of acceptability and tolerability of the intervention. This point is important and men with PD are more often studied than women, even out of proportion to the actual prevalence ratio of 2:1; our male: female ratio was still higher at 1:3.75. If studying outcomes such as cognition (e.g., verbal fluency), in which gender differences have been reported [38], such imbalances need to be considered. In addition, there may be some differences in males as compared to females in how PD symptoms affect their lives socially, and this may affect aspects of the intervention. For example, facial masking in women may be contrary to social norms for expressivity in women but not in men, resulting in problems with social relationships for people with PD [39]; however, since our intervention was highly tailored to the individual needs of the participants, such issues and other gender-related aspects are not likely to be significant. Finally, we have found no evidence in the literature of a gender difference in tolerability or acceptability of psychosocial therapies. Stratified randomisation in a future effectiveness trial may preclude such imbalances in baseline characteristics.
In conclusion, our iterative approach to developing and piloting an adapted form of CST is consistent with guidance and represents an important step in the emerging field of psychosocial therapies for this population.