LIVE@Home.Path—innovating the clinical pathway for home-dwelling people with dementia and their caregivers: study protocol for a mixed-method, stepped-wedge, randomized controlled trial

Among the top ten causes of death globally, dementia is the only one that cannot be prevented, cured or substantially slowed [7], leaving disease management, caregiver support and service innovation as the top priority for health policy-makers in the reduction of disease burden. Due to expected positive interactions within the family, interventions supporting them as caregivers not only potentially lessen the caregivers’ burden [8], but could also be beneficial for the PWD (e.g. reducing neuropsychiatric symptoms and delaying nursing home admission) [9, 10]. As such, interventions supporting caregivers hold the potential for better overall resource allocation and utilization [11].

Caring for a PWD comes at a high cost, both individually and at societal level. Caregivers to PWDs have lower perceived health and higher rates of mortality relative to their noncaregiver counterparts [12]. The effect of practical assistance and psychoeducational programs have been evaluated, but most single initiatives have fallen short in reducing the caregivers’ burden [13]. The Maximizing Independence (MIND) at HOME study undertaken in Baltimore, USA, during 2008–2010 included approximately 300 home-dwelling persons with cognitive impairment or dementia in a parallel randomized multicomponent trial [14, 15]. This study showed that 18 months of care coordination through individualized care planning, implementation of a care plan, monitoring and reassessment had beneficial effects on the time to transition from home, number of dementia-related unmet needs, quality of life (QoL) and, importantly, a potentially clinically relevant reduction in self-reported number of hours spent on caregiving tasks, as a measure of caregiver burden [14, 15]. Developing this model further, the MIND at Home-Plus study included an additional 340 persons to evaluate the effect on long-term care placement, hospitalization and health-care expenditures of a 24-month homecare coordination program for PWD [16]. The MIND at Home-Streamlined trial is now refining the intervention to investigate its impact on time to long-term care placement, needs, burdens and QoL in PWDs and their caregivers, as well as cost utilization [17]. Results of the latter study are highly anticipated due to the potential for effective system-level approaches to dementia care [17]. Yet, due to fairly large regional and cultural differences in care organization, there is a need for implementation studies in other countries to explore the generalizability of the program.

A multicomponent intervention is not merely a discrete package of separate components, but a process of changing what complex systems do [18]. Intervening within a complex system involves disrupting prior ways of working while introducing new ones [19]. The degree of complexity can be understood as a relative construct, defined by the number of components, diversity of the intended outcome, number of targeted organizational levels and level of skill required to deliver the intervention [20], while additionally considering the interplay between context, setting and the implementation process [21]. In the COSMOS trial, a randomized implementation hybrid trial carried out in Norwegian nursing homes during 2014–2015, our group successfully developed, implemented and effect evaluated a multicomponent intervention addressing COmmunication, Systematic assessment and treatment of pain, Medication review, Organization of activities and Safety [22]. Overall, the intervention resulted in improved QoL and activities of daily living (ADL), in addition to a decrease in neuropsychiatric symptoms such as agitation and depression as well as a reduction in the number of medications used among nursing home residents [23‐27].

To provide cost-effective care while securing the needs of PWDs and caregivers represents a complex health challenge warranting multicomponent interventions implemented in daily clinical practice. Aiming at system-level change, such interventions require stakeholder involvement as well as collaboration within and between different levels of primary and specialist health-care services, nongovernmental institutions, users and researchers, addressing the need for appropriate and coordinated cross-sector action.

The LIVE@Home.Path trial aims to develop, adapt, implement and effect-evaluate a multicomponent intervention for home-dwelling dyads of PWDs and their caregivers, aiding them to stay safer, longer and more independently at home with cost-effectiveness. In this study, caregivers are defined as family or close friends, equaling informal caregivers. LIVE@Home.Path is an acronym referring to each component of the complex intervention: Learning, Innovation, Volunteer support and Empowerment—At Home Pathway. The primary outcome is resource utilization. This is measured by the Resource Utilization in Dementia (RUD) instrument and the Relative Stress Scale (RSS), reflecting that resource utilization is more than the actual time required for caregiving tasks, but also how burdensome the task is experienced by the caregiver. Importantly, the caregiver burden is individual, and may be related to economic hardship, anxiety, depression, hopelessness, impaired QoL or lack of sleep and time for recreation. This individual perspective underlines the significance of user involvement, reflected in the trial’s slogan: what matters to you? Secondary outcomes include neuropsychiatric symptoms, number of adverse events, use of assistive technology, involvement of volunteers, QoL and clinical global impression of change for the PWD as well as caregivers’ depression, QoL and work performance.

The LIVE intervention will reduce time and resources that caregivers spend in organizing and supporting PWDs’ daily activities, thereby reducing the caregiver burden.

The required sample size was calculated to detect a difference of 7 h/week for the primary outcome RUD. Based on the literature, we assumed that the mean number of hours of informal care is 46 h/week with a standard deviation (SD) of 20 h/week [58]. With 80% power and a significance level of 5%, the required sample size was estimated to be 260 dyads. To allow for 20% loss to follow-up, a total of 315 dyads, equaling 105 per municipality, must be included.

Participants will be recruited from memory clinics at local hospitals, from municipal memory teams and after advertisements in general media such as newspapers, radio and TV in Bergen, Bærum and Kristiansand. Bergen is the second largest municipality of Norway with approximately 280,000 inhabitants in 2018, Bærum is ranked the fifth largest with 127,000 inhabitants, while the 92,000 inhabitants of Kristiansand constitute Norway’s sixth largest municipality [59].

PWDs are eligible for inclusion if they: are aged ≥ 65 years; are home-dwelling; have a minimum 1 h/week regular face-to-face contact with the caregiver; are diagnosed with dementia according to standardized protocol [60]; have Mini-Mental State Examination (MMSE) score of 15–25; have a Functional Assessment Staging Test (FAST) score of 4–7; and provide written informed consent. Exclusion criteria are: participation in another ongoing intervention trial; or expected survival < 4 weeks. PWDs are eligible for inclusion regardless of etiology of the dementia and presence of other disorders. Caregivers are eligible for inclusion if they have a minimum of 1 h/week regular face-to-face contact with the PMD and provide written informed consent. As such, both the PWD and the caregiver will be included in the trial, representing a dyad.

Data from all 315 dyads will be assessed every 6 months from baseline to the end of study period after 24 months, death or permanent residency in a nursing home—in total, five waves of data collection. The stepped-wedge randomized control design [61] implies that all participants will receive the 6-month intervention program during the study period, for which the timing of the intervention is determined by the randomization (Fig. 2). The control group constitutes the dyads waiting for the intervention at a given time during the study; this group will have access to health care and receive treatment as usual. Criteria for discontinuing the intervention or participation are requested from participants to withdraw from the trial. The trial’s user-oriented approach, aiming at minimizing the participant burden associated with follow-up visits, in addition to flexibility in scheduling of the visits are sought to promote retention and prevent loss to follow-up over the trial. No distinct adverse events are expected before the start of the trial or during the trial, while possible adverse events related to the change in prescribed medication during the general practitioner’s medication review might occur. If so, they will be reported by the coordinators to the researchers, either immediately or at their regular follow-up every 2 weeks (physical meeting, by phone or by e-mail), in addition to feedback from the coordinator to the general practitioner. A statistician will randomly allocate the order of the intervention using block randomization; the dyads are randomized in clusters within each geographical location. The random sequence will be generated using a computerized random number generator undertaken for all three municipalities after the inclusion and baseline assessments are completed for all participants. Research assistants, researchers conducting the analyses and other study personal conducting data collection will be blind to the randomization order and to the implementation process of the intervention. Participants will not be informed of the intervention and implementation strategy to secure blinding until they are allocated to their coordinator during the intervention period. From this point of time, they become unblinded. Given the practice change of the intervention, the municipality homecare services will be aware when their cluster enters the intervention period.

When developing a pathway for dementia care, incorporating experiences and perspectives from the PWDs and their caregivers is fundamental. In line with the INVOLVE framework [62], this trial is developed through user involvement from the conception of the idea, via design through the implementation phase. At the structural level, user involvement is secured via collaboration with the head of research at the Norwegian Health Associations [51], participating in the Steering committee, and locally grounded by dementia coordinators in the municipalities. At the individual level, the Centre for Elderly and Nursing Home Studies (SEFAS), responsible for conducting the trial, employs a user-representative as a co-researcher in a 10% position, who participates in the study’s advisory board and working group. The mixed-method design [63] encompasses the integration of data from quantitative assessment of validated outcomes with material from qualitative interviews and participant observation. Utilizing an exploratory hermeneutic design [64], in-depth and focus group interviews with PWDs (n = 15), caregivers (n = 15), municipality health-care staff (n = 20), general practitioners (n = 10), volunteers (n = 18) and volunteer coordinators (n = 6) will be conducted. To evaluate the acceptability and feasibility of the communication platform, interviews with caregivers and care staff will be made, as well as real-life observations form use among PWDs and caregivers.

Table 2 presents the primary and secondary outcomes according to domain, specific measurement, metric, method of aggregation and time points. The primary outcome of the LIVE@Home.Path trial is formal and informal resource utilization, measured by the RUD instrument [65, 66] and the RSS [67] (Table 2). As such, we consider overall resource utilization as more than the time required to care for the PWD; it also encompasses how burdensome the task is experienced by the caregiver. The informal care time use is measured in hours/month [65, 68], in addition to numbers of contacts with the health-care system and use of medications. The RUD is a standardized and widely used instrument assessing dementia care, proven useful across different care systems and countries and in both clinical trials and observational studies [65, 66]. Caregivers stress will be assessed by the RSS, a self-report instrument covering three dimensions of “emotional distress”, “social distress” and “negative feelings”. It is regarded as a useful instrument to stratify careers according to the risk of psychiatric morbidity [69, 70].

The secondary outcomes presented in Table 2 include measures of QoL, psychiatric symptom load, ADL, comorbidity and pain as well as measure of goal achievements. The QoL for both the PWD and the caregiver will be measured by self-report using the Quality of Life in Alzheimer’s disease scale (QoL-AD) [71] and the generic quality of life measure EQ-5D-5L [72], including the EQ-5D-VAS scale [73]. Additionally, QoL for the PWD will be assessed by proxy by the caregiver with the QoL-AD [71]. Psychiatric symptoms for the PWD will be proxy rated by the caregiver using the Neuropsychiatric Inventory Questionnaire (NPI-12) [74], the Cohen-Mansfield Agitation Inventory (CMAI) [75, 76] and the Cornell Scale for Depression in Dementia (CSDD) [77], while caregiver psychiatric symptoms will be self-reported using the Geriatric Depression Scale (GDS) [78] in addition to the RSS [67]. Data on ADL for the PWD will be proxy rated by the caregiver utilizing instrumental (I-ADL) and personal (P-ADL) measures [79]. Data on pain will be obtained by self-report from the PWD using the MOBID-2 Pain Scale [80‐84] and the level of comorbidity will be evaluated by the interviewer according to the General Medical Health Rating Scale (GMRH) [85]. The Clinical Global Impression of Change Scale (CGIC) will be assessed after the intervention to quantify and track patient progress and treatment response [86]. In addition to the instruments presented in Table 2, other outcome measures include the number of adverse events (falls, disappearances outdoors, fire hazard), use of assistive technology (number of technical aids, cognitive intervention devices and assisted living systems), involvement of volunteers (number of participants with contact with a volunteer, number of hours spent with a volunteer), number of medications used (both regular and on demand) and participation in educational programs for the PWD and the caregiver. These outcome measures will be described as the mean change in sum of events (number devices, hours, medications, educational programs) over the intervention period compared to controls (as outlined in Table 2).

Prior to inclusion and baseline data collection, a 1-day seminar will be arranged for the study personal to secure training in the use of tablets and scoring of relevant psychometric scales. A study manual has been developed to guide data collectors during their visits to secure standardized reporting. Close to 24 h/day, telephone and mail support will be offered by the research team during times of data collection. Researchers and municipal study personal will collect data at baseline as well as 6, 12, 18 and 24-month follow-up. The municipalities will receive 5000 NOK per enrolled dyad to compensate for extra administrative work. At baseline, demographic data such as year of birth, gender, marital status, housing characteristics, education and employment will be collected, as well as data on the dementia syndrome, including the current score on the Mini-Mental State Examination, Norwegian Version (MMSE-NR3) [87, 88], Functional Assessment Staging Test (FAST) [89] and The Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE) [90, 91]. The MMSE-NR3 [88] will be assessed every 12 months during the trial.

Intention-to-treat analyses will be performed accounting for municipality as a random effect in mixed-effect models and the generalized estimating equation (GEE) with nonlinear effect comparing the intervention groups to controls. Repeated observations within persons will be accounted for with a correlation matrix. All secondary outcomes will be adjusted for multiple comparisons using the Hochberg method [92]. Given the potentially informative censoring due to dropout, institutionalization and death, we will jointly model the primary outcome and attrition through a shared person-specific random intercept. Missing data will be handled using multiple imputations by chained equations (MICE).

The study was approved in May 2019 by the Regional Committee for Medical and Health Research Ethics, North Norway (2019/385) and West Norway (2017/1519) (the pilot), and registered at ClinicalTrials.gov (NCT04043364). Assessment and utilization of personal data from the dyads as well as from volunteers and volunteer coordinators from nonprofit organizations are approved by the Norwegian Centre for Research Data (NSD) (ref. 514093). After verbal and written information, spoken and written informed consent was obtained in direct conversation with the caregiver and the PWD, if capable of providing consent for participation. If not, the next of kin or a legal advocate provided consent based on their determination on whether the PWD, when they were able, would have agreed to participate in the trial.

Participation in research is based on affirmative, unambiguous, informed and specific consent [93]. Persons with cognitive impairment will often not be able to provide such a comprehensive consent or understand the scope and consequences of data assessment. Local legislation for obtaining ethical permission in studies varies substantially between European countries [94]. In Norway, the next of kin or a legal advocate can provide consent based on their determination of whether the person, when they were able, would have agreed to participate in the trial [23]. These principles for obtaining informed consent were applied in the LIVE@Home.Path trial. From 2018, the European Union-wide law on data protection, the General Data Protection Regulation (GDPR), represents a significant step toward protection of participants in research [95]. In particular, Article 6 protects PWDs and their relatives from being coerced to consent without awareness of how their data will be used [96, 97]. When assessing sensitive data such as mental health, Article 35 requires a Data Protection Impact Assessment (DPIA), a formal process systematically analyzing, identifying and minimizing the data protection risks of a project. We developed a DPIA (ePhorte UiB: 2019/5569) for the LIVE@Home.Path trial in collaboration with the Data Protection Official at the University of Bergen, encouraging us to again evaluate which data to assess, as well as focus on safe data management. Nonetheless, we anticipate the participation in the LIVE@Home.Path trial to be less burdensome relative to, for example, RCTs on effect of medications, due to the user-oriented approach emphasizing the investigation of the perspective “What matters to you?”

Stakeholders and research funders increasingly require patient and public involvement (PPI) at all stages of research from design, implementation and dissemination of results, shifting focus from research “about” or “for” to research “with” or “by” someone [98, 99]. Our user-representative has provided feedback on a close to weekly basis through participation in the working group and advisory board of the trial. A related principle, Responsible Research and Innovation (RRI), is defined as a transparent, interactive process making societal actors and innovators mutually responsive to each other, and encouraging them to set up a critical perspective when evaluating the innovation and marketability of products [100, 101]. Taken together, these components constitute a framework for sustainable ethic innovation in dementia research (Fig. 3), a model that easily can be applied when designing and conducting research on other vulnerable patient groups.

A stepped-wedge randomized controlled trial design is recommended for evaluation of a multicomponent intervention in health-care services as it provides a number of practical and scientific benefits compared to an ordinary RCT [61]. It is increasingly used in effectiveness studies in the geriatric field [102, 103]. Most importantly, the design allows for providing the intervention to all participants, overcoming ethical and logistical challenges arising from withholding the intervention. This design is, however, more vulnerable to temporal external changes, as more participants are exposed to the intervention toward the end of the study than in earlier stages. If the LIVE intervention fails to prove an effect on resource utilization, we will examine whether this is due to a lack of proper implementation. Thus, if the implementation process is satisfactory, it may suggest that the LIVE components were not tailored to be sufficiently cost-effective if no effects on primary outcome measures are found. An alternative interpretation is that the intervention may not be cost-effective even if primary outcomes change significantly, as resource use by the intervention is more time consuming and/or expensive than the alternative.

Some challenges have emerged during the start of the trial. First, it is demanding to include the estimated number of participants, and, additionally, to keep the number of dropouts low due to the progression of the disease. We should have established closer collaborations with the geriatric specialist health-care services, as we experienced that patients recruited from geriatric outpatient clinics were in the most optimal disease stage for this trial. To increase recruitment, we prolonged the inclusion period to 31 December 2019 and expanded the inclusion criteria to age ≥ 64 years and MMSE range 15–27, while the SEFAS researchers, journalist and co-researcher with user experience continuously work on positive media coverage. Second, data collection from home-dwelling persons in three distinct municipalities is resource and logistically demanding. Third, being selected as a pilot for the data collection software has been challenging, as the file format initially generated handled missing data in a way that was not compatible with our statistical programs. Finally, the participants have so far been recruited in various ways, from home-care services in the municipality and memory clinics at hospitals, to self-referrals after advertisements in the general media. This implies that the dyads included in our trial represent a heterogeneic group of home-dwelling people with dementia.

In conclusion, we expect the implementation of LIVE to lead to a pathway for dementia treatment and care that is cost-effective, feasible and supports independent living, at home.