2.1. Design, setting and participants
The two-arm parallel intervention study was designed as a cluster randomized controlled trial. It recruited 2,788 participants: 1509 participants in 6 intervention clusters and 1279 participants in 6 control clusters. The research protocol was approved by the World Health Organization Ethics Review Committee WHO/ERC ID; ERC.0003064, dated 7 March 2019 and Boromrajonani College of Nursing, Lampang Thailand Ethics Review Committee E2562/005, dated 4 March 2019. It was registered at the Thailand Clinical Trial Registry, Trial registration number TCTR20190412004.(21) The trial was conducted over a period of 2 years. The net intervention period was six months.
Study setting
The trial location was in Chiang Mai, Northern Thailand. A total of 284,457 older adults resided in Chiang Mai city according to a provincial report in 2019. This number accounts for 18% of its population and was higher than the national proportion of people aged 60 years and older, which was 16%. The culture in Chiang Mai is similar to that of many neighbouring ASEAN countries in terms of caring for older parents as a family tradition and social value.
Two subdistricts were involved in the study: subdistrict XXX (10 villages) for the intervention arm, and subdistrict YYY (15 villages) for the control arm. They are in the same Mueang Chiang Mai district of Chiang Mai province. The distance between the two subdistricts were far enough to prevent relayed messages although they were of a similar demographic and socio-cultural characteristics. The two subdistricts have a similar population (more than 18,000), and similar ageing rates (more than 18% of people over 60). Residents in both subdistricts are mostly Thai people who have access to health care services under the national health insurance system and social welfare services for older people covered by municipality funds.
Sample size and power
The estimated sample was 1500 participants in each arm to determine the size difference of 0.5 unit with standard deviation (SD) of 4 between the two arms. STATA version 11SE (Stata Corporation, College Station, TX) was utilized for sample and power estimations. The precision levels applied are a P-value of 0.05 with a 95% confidence interval. The sample size was inflated for design effect 1.2 of cluster randomized design application, and for compensation of potential non-responses and drop-outs during the recruitment and study of up to 20%. Moreover, the sample size was estimated to be sufficient for detecting minimal difference in key parameters such as mean CBI, ADL, and health-related QOL indicators within the study population. We calculated backward sample-power calculation at the end.
Sequence generation, random allocation, and blinding
Cluster randomization was used to prevent contamination between the intervention subdistrict and control subdistrict. It was not possible to randomize the villages from different administrative areas, municipalities, and directly allocate them to the intervention and control arms. Therefore, we practiced internal randomization to recruit six eligible clusters randomly within the intervention arm municipality, which has 10-15 villages, and likewise, in the control arm. (22) (Figure 1)
A statistician blinded to the study generated a random number for each arm of the study and recruited villages randomly within each arm. The control arm and intervention arm villages were geographically distant and administratively exclusive, yet they were in the same province and similar demographically, socially and culturally in order for social intervention evaluation.
It was not possible to do double blinding. However, participants were blind to allocation. Participants and research assistants carrying out assessment were not able to know the randomly-allocated clusters before the study begun. Thus, bias and contamination were controlled.
The unit of randomization, cluster, was a village. Participants in the intervention clusters received the community-integrated intermediary care (CIIC) facility and service intervention, after screening for eligibility. Eligibility criteria for a cluster was a village which had more than 300 older persons over 60 years of age, at the time of randomization.
Inclusion criteria for study participants
1. Persons over 60 years of age and their family caregivers
2. Either male or female
3. Resident in the study site districts
3.3. Exclusion criteria
1. Persons over 60 or their family caregivers without informed consent
2. Persons who could not understand the explanation for informed consent although provided with language support
3. Households without an older person over 60 years
4. People with cognitive impairment or severe impairment decision-making abilities
2.2. Outcome measurement
Validated instruments commonly used in ageing and long-term care research were carefully chosen in order to assess the impact of the intervention in objective outcomes. Most of the instruments used were already translated into Thai versions and validated in previous studies and programs in Thailand. We conducted a pilot test for the target population and ensured reliability of all the instruments in the study setting and context.
Primary outcome
Primary outcome was the family caregiver’s burden at the six months follow-up. The caregiver burden inventory (CBI) was used to measure burden of the caregiver. (23, 24) The CBI was applied to measure the family caregiver’s burden at baseline month 0 and month 6. The CBI is an internationally validated, 24-item, 5-point Likert scale which measures the burden of caregiver in 5 dimensions as follows:
a. Time dependence burden (5 items)
b. Developmental burden (5 items)
c. Physical burden (4 items)
d. Social burden (5 items)
e. Emotional burden (4 items).
After summing up the total score, a score greater than 36 indicates a risk of burnout and a score near 24 indicates a need to seek respite care.
Secondary outcomes
Secondary outcomes consisted of biopsychosocial indictors such as functional ability, depression and quality of life of the elderly, measured comparing the intervention and control arms after the first six months of intervention. Functional ability was assessed by applying Barthel’s Activity of Daily Living (ADL) assessment. (25) ADL measures the level of ability for 10 basic items (bowels, bladder, grooming, toileting, feeding, transfer, mobility, dressing, stairs, bathing). Total possible scores range from 0 – 20, with lower scores indicating increased disability. Depression was screened by applying the Geriatric Depression Scale (GDS), which is commonly-used internationally, validated and regularly used in Thailand. (26) Health-related quality of life was measured through EQ 5D questionnaires. (27) A validated Thai version instrument was readily available and used after piloting. Health-related quality of life (EQ-5D-5L) with EQ visual analogue scale (EQ VAS) (27, 28) was applied to measure the health-related quality of life of the elderly persons. The EQ-5D-5L comprises five dimensions: mobility, self-care, usual activities, pain/discomfort and anxiety/ depression. Each dimension has 5 levels. The EQ VAS is a measure of overall self-rated health status. It records the respondent’s self-rated health on a vertical, visual analogue scale where the endpoints are labelled ‘Best imaginable health state’ and ‘Worst imaginable health state’.
2.3 Intervention
The community-integrated intermediary care (CIIC) model is a novel prevention-based, long-term care model created to enhance the family-based long-term care system (Figure 2) which has traditionally been practiced in Thailand and its neighboring Asian countries as well as many low-and middle-income countries globally.(14, 29) Conceptually it is a combination of formal care and informal care in a district, or subdistrict or a city. To implement the CIIC service model a community-based facility had to be established within the community, mobilizing community resources and the municipality funds. It would serve as a five to ten beds respite home and office for CIIC staffs. Two auxiliary nurses would work alternatively for care, coordination, and capacity-building with two assistants and volunteers. The CIIC facility would be geographically community-centered, functionally linked to the primary health care center and administratively linked to the municipality office. However, its human resources and professionals would be separate, not from the primary health care center. The CIIC service model consisted of three components: (1) care prevention exercise, (2) care capacity building of the family caregiver; and (3) community respite service.
Care prevention exercise is to prevent the long-term care need of the older persons, to preserve their functional ability leading to active ageing. (Figure 2) It is a set of functional training exercises particularly designed for older adults either in a sitting or standing position. We provided two options: (a) 45 minutes long community-based group exercise; and (b) 10 minutes long home exercise set for every day of the week to be practiced twice a day individually or with friends and family at home. We provided an exercise DVD, wall posters to guide the exercise techniques, and a calendar for recording exercise. Home sets of exercises consisted of stretch, upper limb, lower limb, trunk and squat components (30), whereas group exercise consisted of dynamic and static stretch, tube training for upper and lower limbs, squat, balance training, walking and brain training. Community-based group exercise activity required a venue, schedule, and a television or projector for a group of 30 persons. We created six such venues in the intervention clusters. Before launching the community-based group exercise, community volunteers were trained to be the leaders of group exercise. The training took four days. Stakeholders made the schedule of group exercises depending on the participant’s available time.
Secondly, CIIC provided training and assistance to enhance the family caregiver’s care capacity. (Figure 2) The training was not a general, regular course. It was an issue-based technical training, relating to the need of the care recipient. CIIC teams arranged home visits when caring families made a request. A care expert nurse and the team reviewed the care techniques and taught the family caregiver on site. This activity was linked to the primary health care service, and community health volunteers coordinated with the municipality authority. A visit to home screen the care need of the older persons, the burden of the family caregiver and technical gaps, and the home modification need indicated what could be provided by the municipality and what could be referred to the primary health care center.
The third service of CIIC was to provide formal long-term care service in the form of a community respite home for eligible older persons. (Figure 2) Eligibility criteria were set by the city stakeholders. In this study the eligibility criteria for the CIIC temporary respite care service was for applicants with full-time, unpaid family caregivers caring for dependent individuals over 60 years of age and meeting the eligibility as assessed using the ADL and CBI.(14)
ADL Criteria
In order to be eligible to be admitted to the CIIC facility, the family caregiver had to be personally providing the care recipient with assistance for at least 2 of the following 6 activities of daily living (ADL): 1) Bathing-the family caregiver is assisting the older adult with bathing, including help with washing, shampooing, getting in or out of the tub or shower, brushing teeth, and other aspects of personal grooming (Bathing ADL score 0); 2) Dressing- the family caregiver is assisting the older adult with dressing, including helping the individual put on or take off clothing and footwear (Dressing ADL score 0 or 1); 3) Toileting-the family caregiver is helping the older adult get on or off the toilet, commode, or bedpan as well as clean themselves, or the individual is incontinent (Toileting ADL score 0 or 1); 4) Transferring- the family caregiver is helping the older adult get to and from a bed or chair (Transfer ADL score 0 or 1); 5) Walking or mobility- the family caregiver is helping the older adult move from one stationary point to another by removing obstacles, opening doors, and assisting with canes, wheelchairs, or other assistive devices (Mobility ADL score 0 or 1); 6) Eating or feeding- the family caregiver is helping the older adult who has difficulty chewing or swallowing without assistance or needs partial or total help with eating (Feeding ADL 0 or 1).
CBI Criteria
On the assessment of caregiver burden, the following 2 eligibility criteria were required to utilize the respite care service: 1) time dependency items score >17; and 2) physical health items score >14. If there was only one person in the family acting as a caregiver for the older adult with the qualifying ADL criteria and the caregiver had to leave the house to travel, the older adult would qualify for admission to the CIIC based on the available capacity of the facility on the appointment days. These criteria could be revised by stakeholders depending on the estimates of the population to be served, budget of the municipality, and capacity of the facility.
Control group
The CIIC intervention was evaluated in comparison with an active comparator which was the existing service in Thailand. Control arm participants received the usual care (ie, the current system of LTC common to all provinces in Thailand), consisting principally of a health volunteers’ assistance to those living alone or severely dependent older persons in addition to traditional family-based care. The difference between the services of the 2 arms refers to the newly-launched CIIC services, as described above.
Screening and Assessment of Family Burden and LTC Capacity
All older adults and caregivers in each cluster who provided informed consent were screened for ADL status and health-related QOL utilizing the EuroQol 5-dimensions 5-levels (EQ-5D-5L) for older adults, and caregiver burden measured using the Caregiver Burden Inventory (CBI). In both arms, a basic health check, blood pressure check, and BMI assessment were provided by the study participants. If any disease was suspected through the health check, appropriate referrals to existing health care services were provided. This was a benefit for all study participants. The control group also received the same assessment with an explanation that the assessment was being conducted as part of a survey and the result would be utilized for research purposes. They also received the benefit of the additional health check followed by appropriate referrals to health care professionals.
2.4. Data analysis
All analysis applied cluster level analysis for all outcome measures. Cluster adjusted t-test and cluster adjusted chi-squared test were applied to compare the outcome between the intervention and control groups. At first the result of the baseline survey was compared between the intervention and control arm clusters to check the balance between the two arms and to detect possible confounders. (Table 2)We compared the level of the indicators measured at the baseline, applying cluster level analysis. Similarity between the two arms after randomization indicated that cluster adjusted analysis was sufficient.
Intention-to treat analysis was applied. Bio, psycho, social indicators were compared between the two arms. The primary outcome was the caregiver burden inventory (CBI). (Table 3-5)
First, we compared outcomes in the categorical data. The CBI scores were categorised into three groups: a low burden (<24), a medium burden (24-36) and a high burden (>36).
Second, we compared the change in CBI level between the baseline level and evaluation level. The decline in CBI within six months follow-up was defined as a desirable outcome event.(Table 4) In this analysis, those lost to follow-up or died were treated as a negative outcome in the intention to treat approach. Finally, we compared the level of the indicator, CBI measured at the evaluation, applying cluster level analysis. The power of the sample was 0.79 for the primary outcome results.
Effectiveness in preventing care need or functional decline was analysed using the secondary outcome ADL. (Table 6-8) First it was analysed in three categories. Then the level of ADL was compared between intervention and control clusters. The change in ADL level was used to detect the outcome functional decline. Sustaining the same level of ADL or an increase in ADL level within six months of follow up was defined as desirable outcome. (Table 7) This analysis also applied intention to treat approach and cluster level analysis. Another secondary outcome GDS was analysed in categories and also comparing the level of GDS. (Table 9 and 10) Health related QOL was analysed comparing EQ5D5L level between the intervention and control arms in cluster level analysis. (Table 11)