Participants

The trial will be conducted in people’s homes located in metropolitan cities and adjoining rural areas in Australia, Germany, Poland, Norway and the UK.

Inclusion criteria:

• Dyads (cohabiting) who are close in relationship and where one member has a diagnosis of dementia according to the 10th revision of the International Statistical Classification of Diseases and Related Health Problems (ICD-10) criteria (Alzheimer’s disease (AD), frontotemporal dementia, vascular dementia), Lewy body disease or mixed dementia) as determined by a clinician experienced in diagnosing dementia.7 Close in relationship refers to a CG who may be a sibling, spouse, adult child, friend, niece or nephew or any person who has a close relationship to the PwD, that is, anyone who is not a formal paid caregiver.

• Dyads where the PwD has a Neuropsychiatric Inventory-Questionnaire (NPI-Q) score of ≥6 (from a maximum score of 36) and Mini Mental State Examination (MMSE) scores <24 as research indicates that NPI-Q scores >6 occur in PwD who have high MMSE scores.23 The NPI-Q will form part of the screening process, with a trained assessor administering the NPI-Q in the dyad’s home prior to enrolment in the study.

Exclusion criteria:

• Dyads where either or both the CG or PwD have significant hearing impairments that are not resolved through the use of a hearing aid device and limit their capacity to enjoy musical experiences. There will be no further exclusions.

Interventions

Outcomes

At baseline, demographic data (age, gender and dyad history) of both CG and PwD will be collected as well as diagnostic information of PwD (ICD-10).

Where possible, core outcomes (www.comet-initiative.org) for psychosocial intervention research in dementia care were selected.26 For the PwD, the following measures were selected:

• BPSD: the NPI-Q is the most highly regarded and used measure for determining the severity of BPSD in clinical trials. The 12-item scale is used to assess the behaviour of PwD across 12 domains of commonly displayed BPSD. The scale has been translated into >40 languages, has been cross-validated against the NPI as the gold standard (r=0.73) and has demonstrated good validity (sensitivity=74.1%, specificity=79.5%), internal reliability (α=0.783) and excellent test–retest reliability (r=0.99).27 28 Total severity scores range from 0 to 36; higher values are indicative of higher severity. Distress scores range from 0 to 60; higher values represent higher levels of distress. CGs will self-complete the NPI-Q with guidance from the research assessor if required.

• Depression: the Montgomery Asberg Depression Rating Scale (MADRS)29 will be used to assess the severity of depression. Scores for each of the 10 items range from 0 (no symptoms) to 6 (severe symptoms) and are determined through an assessor-led interview with the proxy, in this study, the CG. Total score ranges from 0 to 60 with higher scores indicating more severe depression. The MADRS has been found to have good constructive validity, internal reliability (α=0.84) and test–retest reliability (intraclass correlation coefficient=0.78). The scale has been widely used in clinical trials.29

• QoL: the QoL of PwD will be determined by administering the QoL-AD30 scale. The QoL-AD is recommended by the COMET Initiatve,26 for use in clinical effectiveness trials. It is a simple 13-item self-report measure, which is rated on a 4-point scale, within the structure of a verbally delivered interview. Total scores range from 13 (poor QoL) to 52 (excellent QoL in all areas). Studies indicate that the measure can demonstrate sensitivity to psychosocial intervention, correlates with health-utility measures,31 has excellent inter-rater reliability (κ>0.70) and internal consistency (α=0.82). The QoL-AD is reliable when used with people with MMSE scores of >10. Both a CG proxy and PwD self-report (if possible) will be collected at the three time points. If the PwD is able to complete the MMSE at all time points, then their response will be included in the analysis. If not, then the proxy version at all time points will be used.

• Cognition: the MMSE will be administered preintervention and postintervention (time 1 and time 2) to monitor any change in the PwD’s cognition and to examine the relationship between cognitive decline, BPSD, depression and response to different conditions. The MMSE is a 30-point questionnaire used to estimate the severity and progression of cognitive impairment and to follow the course of cognitive changes in an individual over time.32 The MMSE tests for orientation, attention, memory, language and visual-spatial skills. It is reliable and valid for both diagnosis and longitudinal assessment. Higher scores indicate better cognitive capacity with scores of 24–30 indicating no cognitive impairment; 19–23 indicating mild cognitive impairment; 10–18 indicating moderate cognitive impairment and scores <10 indicating severe cognitive impairment. MMSE scores will be determined through assessor-led interviews with PwD participants.

For the CG, the following measures were selected:

• Depression: CG depression will be measured using the Patient Health Questionnaire-9 (PHQ-9).33 This self-completed 9-item questionnaire asks the participant about how often they experience the descriptors over the last 2 weeks. Each item is scored from 0 (not at all) to 3 (nearly every day). Total scores range from 0 to 27. The PHQ-9 has comparable sensitivity and specificity to other depression measures; high internal reliability (α=0.89) and test–retest reliability (r=0.84).

• Resilience: CG resilience will be measured using the self-completed 14-item Resilience Scale.34 Total scores range from 14 to 98, with higher scores indicative of higher resilience. The measure has been tested and has good concurrent validity, good internal reliability (α=0.8–0.90), good construct validity, test–retest reliability (r=0.67–0.84) and has been translated into 36 languages.35

• Competence: CG competence will be measured using the self-completed Short Sense of Competence Questionnaire.36 The 7-items cover three main domains: self-reported feelings about how the caregiver role impacts the CG’s personal life, satisfaction with their performance as a CG and their satisfaction with how the PwD responds to the CG. Total scores ranged from 7 to 35 with higher scores indicative of a stronger sense of competence. The measure has been cross-validated with the longer 35-item standard Sense of Competence Questionnaire (r=0.88) and has been shown to have high reliability (Cronbach’s α=0.76).

• QoL: CG QoL will be assessed using the self-completed Assessment of Quality of Life-6D instrument.37 Each item asks participants to describe their situation over the past week by ticking the box (from 4 to 6 choices) that best reflects their situation. The psychometric property testing found the instrument to be reliable and valid, and has greater sensitivity to the psychosocial dimensions of QoL than other utility instruments.37 38

• Relationship quality: CG perception of quality of the CG and PwD relationship will be captured by asking the CG to self-complete the Quality of Caregiver-Patient Relationship.39 This 14-item measure aims to capture the strength of the quality of relationship between the PwD and CG, from the CG’s perspective. Total scores range from 14 to 70, with higher scores indicating a higher quality relationship. The measure has demonstrated acceptable internal consistency (α=0.82) and concurrent validity.

• Adherence to the MI/RI intervention will be measured through CG completed diaries. CGs are deemed adherent to the protocol if they have provided >2 sessions of MI or RI per week, for at least 30 min in total. Data on the general use of reading and music by all dyads (including SC) will be collected at post-test. For each diary entry, CGs will be asked to record the date, start and stop time of MI/RI engagement, types of activities used, their experiences during the session (negative, neutral, positive, unsure), effects from the intervention for the remainder of the day until the PwD goes to bed for the evening sleep (negative, neutral, positive, unsure) and any comments. These data will be used in qualitative analyses to gain more nuanced understandings of how the activities are perceived and how these may change over the course of the intervention period.

To evaluate the cost-effectiveness of MI compared with RI and SC on PwD and CG, the following outcomes will be measured:

• Quality-adjusted life-years (QALYs): the EuroQol instrument (EQ-5D-5L),40 is a generic QoL measure that is internationally used to determine the QALYs and used for clinical and economic appraisal.40 The measure is quick to complete and not cognitively demanding. Scores from the five items are not combined arithmetically but using preference weights which arrive at an overall QoL score. These range from lower than 0 (worse than death) to 1 (best possible). PwD will complete the self-report version of the EQ-5D-5L. CGs will complete the measure as a proxy for the PwD and self-report their own health status.26 41–44

• Resource utilisation in dementia (RUD): the RUD is a standardised instrument for resource use data collection in dementia, designed to collect data from formal and informal care across different countries. The RUD assesses resource use of both PwD and CG, including time expended in different daily tasks, and consists of baseline and a follow-up assessment.45 The RUD will be completed through assessor-led interviews with CGs.

CGs will be provided with a set of guidelines as to how to complete the diary and all self-report and proxy measures. The schedule for enrolment, baseline assessments, all outcome measures, and intervention trainings is outlined in figure 3.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Schedule of enrolment, interventions and assessments. AQoL-6D, assessment of quality of life-6D instrument; EQ-5D-5L, the EuroQol instrument; ICD-10, 10th revision of the International Statistical Classification of Diseases and Related Health Problems; MADRS, Montgomery Asberg Depression Rating Scale; MI, music intervention; MMSE, Mini Mental State Examination Score; NPI-Q, Neuropsychiatric Inventory; PHQ-9, Patient Health Questionnaire-9; QALY, quality-adjusted life-years; QCPR, quality of caregiver-patient relationship; QoL-AD, quality of life-Alzheimer’s disease; RI, reading intervention; RS, Resilience Scale; RUD, resource utilisation in dementia; SSCQ, Short Sense of Competence Questionnaire.

Sample size

A total of 165 dyads in each arm of the study, or 495 in total, are needed to detect a difference of 3-points in NPI-Q total severity score (primary outcome) between the MI plus SC and SC arm (primary comparison) at 12 weeks (primary time point). This assumes 90% power, a two-tailed significance level of 5%, equal SD (7.5 points) in the groups, no correlation between baseline and 12 weeks (conservative) and 20% attrition. A 3-point change from baseline in NPI-Q total severity score is considered a clinically meaningful difference.46 A conservative SD of 7.5 points is based on that observed in 1026 community-living participants across eight European countries with mild-to-severe dementia (SD 5.9–6.5 points).47 48 A conservative drop-out proportion (eg, withdrawn by CG, physician or death) of 20% is based on a reported 5.6% (95% CI 1.8% to 12.6%) drop-out at 3 months in 89 inpatients with mild-to-moderate dementia in Finland participating in a three-arm RCT of singing, music or usual care,.25

Recruitment

Randomisation will aim to be distributed equally across five countries (Australia, the UK, Norway, Germany and Poland) to support between-country analyses. Participants will be recruited through established partner organisations. Staff from the partner organisations will introduce the trial to potential participants and invite them to participate in the study. They will be given an information sheet explaining the main aspects of the trial and provided with contact details of the research team who will be available to answer further questions.

Randomisation, allocation concealment and blinding

The randomisation schedule will be computer-generated by an independent statistician and allocation will be carried out through a centralised randomisation service. Block permuted randomisation with stratification by participating site will be used, so that treatment balance within site is achieved. Dyads who meet the inclusion criteria and none of the exclusion criteria will be randomised 1:1:1 into MI, RI or SC. Randomisation will occur after the eligibility checking, informed consent and baseline assessment have been completed. Informed consent/assent will be obtained by a blinded assessor prior to the baseline assessment. The study coordinator in each country will be informed of the allocation and will inform dyads of their group allocation by post, phone or email.

Participant dyads cannot be fully blinded due to the active nature of the interventions; however, plain language statements and consent forms will use neutral wording to maintain equipoise and to avoid expectancy effects. Blinded assessors will collect participants’ data at baseline, postintervention and follow-up. Diaries will be returned in sealed envelopes to minimise risk of assessors becoming unblinded. The success of assessor blinding will be checked by asking the assessor to guess the treatment assignment (or say “I do not know”) after the postintervention and follow-up periods. This treatment guess will then be compared against the actual treatment and the blinding index derived. The independent statistician will not reveal the allocation codes to any of the study team except for the study coordinators of each country in charge of group allocation. All other investigators and the study statistician will remain blinded until the database has been cleaned, a blinded data review has taken place and the database is ready for analysis.

Analysis

Analyses will be performed on an intention-to-treat basis including all randomised dyads in their allocated study arm. The primary outcome (NPI-Q severity total score) will be analysed using a constrained longitudinal data analysis (cLDA) model,49 with response consisting of all scores (baseline, 12 weeks and 6 months) and the model including factors representing intervention, time, intervention by time interaction and site with the restriction of a common baseline mean score across interventions. This refers to the assumption that at baseline there are no differences between the interventions in the mean score, thus assuming the randomisation was effective. This assumption will be enforced statistically in the statistical model. The absolute difference between MI and SC and MI and RI in mean change from baseline will be estimated (including two-sided 95% CI) at 12 weeks (primary time point). A hierarchical fixed sequence testing procedure will allow testing of MI versus RI at 12 weeks at 5% if the comparison of MI versus SC at 12 weeks has a p value <0.05. Secondary analyses will consist of a model adjusted for potential confounders (types of dementia and gender). The cLDA model provides valid inference if the missing data mechanism is at most missing at random. In addition to the intention-to-treat effect, we will obtain the complier average causal effect by making use of the collected adherence data.50 Analyses similar to the primary outcome will be applied to the secondary outcomes for PwD and CGs. Heterogeneity of the intervention effect across subgroups (gender of the PwD/CG, types of dementia, severity of dementia, time of onset dementia, length of time having dementia, relationship between PwD and CG, country) will be assessed by means of interaction tests. The number and percentage of PwDs and CGs with adverse events will be summarised by intervention group.

Cost-effectiveness analysis in a societal perspective will be performed separately for each country using the utilities generated by EQ-5D-5L for both PwD and CG and country-specific weights, to estimate a combined QALY score using a generalised linear model adjusted by the baseline. Health and informal care resources consumed by PwD will be assessed using the RUD and unit cost by country. A generalised linear mixed model will be used to estimate the main predictors of the total costs in the MI, RI and SC groups. Incremental cost-effectiveness ratio will be calculated using the cost and effect estimates comparing MI with RI and SC. The uncertainty around the incremental cost-effectiveness ratio will be estimated using bootstrapping (1000 replications) adjusting to control variables.

Patient and public involvement

All countries have involved user and advocacy organisations in the development and design of the study. The UK, Australia and Germany have been piloting work with formal and informal CGs of PwD for many years. CGs and PwD have been involved designing the diaries which capture adherence data. It was imperative that the diary be user-friendly, not burdensome on the CG, and yet enabled them to document both the positive and negative aspects of the session. Several iterations of the diary were constructed prior to arriving at the final structure. Pilot work in Australia14 involved interviews postintervention to identify strengths, limitations, challenges and experiences in delivering the MI, as well as recommendations for suggested modifications to the intervention training. Representatives of advocacy groups and end users from all countries will be represented on an international Participant and Public Involvement Committee.

Monitoring and oversight

A Trial Operations Committee (TOC) will comprise the principal investigator, chief investigators and clinical trial managers from each of the five countries, the study statistician, health economist and a consumer representative. The TOC will meet at least 6-weekly, and will oversee all aspects of the trial delivery including strategies to support efficient and effective recruitment and retention, reviewing completeness of data sets, monitoring intervention fidelity, management of timelines and milestones, review of country-by-country progress, public and patient involvement or actions and publication and dissemination plans. The role of the members of the TOC is to bring country-specific issues to the international team for discussion to ensure the study is being monitored and delivered according to the agreed protocols. Protocol deviations and any changes or amendments to the operational processes of the trial will be discussed and decisions made by the TOC.

A Trial Steering Committee will comprise members independent of the clinical trial including consumer representatives and representatives from other relevant advocacy organisations. The trial principal investigator (or a proxy in her absence) will also sit on the committee as a non-voting member. The Committee will meet biannually (or more often when needed) to review and monitor all aspects of the study delivery. They will draw on reports provided to them by the TOC and make recommendations to the principal investigator and TOC about whether further actions are required.

A Data Safety Management Committee (DSMC) will review the cumulative study data to evaluate the recruitment, safety, study conduct and scientific validity and integrity of the trial. The committee consists of at least five people with strong methodological, biostatistical and clinical expertise who are independent of the project and an end user representative. The DSMC will be provided with data on recruitment, intervention uptake, any unforeseen and/or adverse events and review serious adverse events. The meetings will consist of an open and a closed part. In the open part, the general progress of the trial will be discussed with the principal investigator. In the closed part, the DSMC will discuss any safety concerns and if considered required, the DSMC will make recommendations to the TOC for appropriate action.

Study data will be collected and managed using REDCap electronic data capture tools hosted at the University of Melbourne.51 The Data Management Coordinating Center will oversee the intrastudy data sharing process between countries, with input from the Data Management Subcommittee. We will develop a data management manual detailing data collection protocols and provide comprehensive training of those members of the research team who collect, check and enter study data. The principal investigator, study statistician and health economist from the University of Melbourne will be given access to the cleaned data sets. Country-specific lead investigators will only have access to their own country’s cleaned data sets. All data sets will be password-protected. To ensure confidentiality, data dispersed to project team members will have any identifying information removed.

Risk management

Processes have been put in place to mitigate risks. One of the most significant risks associated with the project is slow recruitment. To offset the risk of slow recruitment, data are being collected across five countries. If some countries have less difficulty than others in recruiting, then these countries will recruit greater numbers to ensure the required sample size is obtained. Another risk identified is the heterogeneity of intervention delivery. The inclusion of a detailed manual, regular supervision with interventionists, and monitoring the effectiveness of interventionist training will mitigate the risk of poor intervention fidelity.

Ethics and dissemination plan

All research and clinical activities carried out for the HOMESIDE project will be in compliance with fundamental ethical principles including those reflected in the Oviedo convention and the Convention for the Protection of Human Rights and Fundamental Freedoms and legal requirements (Directive 95/46/EC on the protection of individuals with regard to the processing of personal data and on the free movement of such data; and Directives 2001/20/EC, 2005/28/EC relating to the implementation of good clinical practice in the conduct of clinical trials). Ethical conduct will be managed in the following ways:

• The clinical trial coordinator in each country will implement the research in full respect of European/national/ institutional legal and ethical requirements and codes of practice.

• Ethics approvals in each country must be obtained prior to commencement of the trial.

• Informed consent from the PwD’s guardian must be obtained prior to enrolling a participant in the study. Assent from the PwD will always be sought prior to enrolment in the study.

• National and international rules on data protection will be followed. Participating countries in HOMESIDE within the EU and EEA (UK, Germany, Poland and Norway) also relate to the General Data Protection Regulation (GDPR)(Regulation (EU) 2016/679), designed to harmonise data privacy laws across Europe, to protect and empower all EU citizens data privacy and to reshape the way organisations across the region approach data privacy. The HOMESIDE partners have also signed a consortium agreement where they consent to follow national and international rules on collaboration, ethics and data protection.

The report on the main, preplanned analyses of the primary end point and up until the 6-month follow-up will be submitted to a leading medical journal. Further publications may focus on issues such as recruitment and retention strategies for home-based programmes. Publications based on qualitative interviews and video analyses will focus on barriers and facilitators for implementation and promotion of adherence to home-based programmes; experiences of caregivers in delivering the programmes and the development of best practice training guidelines. In addition to publications in academic journals, a number of policy briefing papers for government and aged care/dementia advocacy groups are planned as well as the development of training manuals and guidelines for dissemination.

Data sharing

In accordance with the Australian Code for Responsible Conduct of Research (Universities Australia, 2018), all data will be retained for retrieval and reuse in future research where participant permission is granted. Following project completion, de-identified anonymised data (with participant consent) will be available on the Australian Data Archive https://ada.edu.au and listed on Research Australia’s https://researchaustralia.org website to facilitate access for future research. Data made available will include individual-level deidentified participant data, reports on adverse events and deidentified interview transcripts. According to the GDPR, the consortium have agreed to the reuse of data for 10 years post project completion.

Relevance and benefit to society

As the majority of PwD live in the community and not in residential care settings, quality informal care for PwD is crucial for managing BPSD and enhancing QoL. This protocol details the process for testing the effectiveness and cost-effectiveness of a CG-directed MI and RI designed to manage the BPSD of PwD, the sense of burden and well-being of the CG and provide meaningful possibilities to maintain the relationship between PwD and their CGs. We expect that with support and training, the MI will be easily implemented in the family home by CGs. With the increasing number of people living with dementia and the stress this will place on countries’ economies, our project aims to test an intervention designed to keep people living at home with family CGs for as long as possible, reducing the burden for society and caregivers. Our study will be able to estimate the incremental cost-effectiveness ratio between MI and RI, MI and SC and RI and SC. Data may support aged care policy recommendations and as the interventions will be delivered in five different countries, results will be broadly generalisable.