Primary CGs of a family member (aged 65+) with a physician diagnosis of AD or meeting the NINCDS-ADRDA criteria for possible AD [28
] will be recruited from psychiatric/psychogeriatric clinics, NGOs, and the community. Primary CGs are those who provides the greatest amount of unpaid care (at no less than 14 hours per week) to the CR and is primarily responsible for making day-to-day care decisions. Care includes providing assistance in any of the following: feeding, dressing, appearance, moving around, bed transfer, bathing, toileting, using the telephone, transportation, shopping, preparing meals, housework, taking medication, and handling money. Other than CR diagnosis and the number of caregiving hours per week, the inclusion criteria are: (a) CR being in the mild to moderate range of dementia severity as determined by Clinical Dementia Rating [29
], (b) CG aged 18+ years, and (c) CG without cognitive impairment (that is, Mini Mental State Examination (MMSE) ≥23) [30
]. Exclusion criteria were CR having parkinsonism or other forms of dementia (vascular dementia, dementia with Lewy bodies, mixed dementia, and so on). Some 120 participants will be recruited for Trial 1 and 100 for Trial 2. Informed consent will be obtained from the caregivers. Ethics approval for the study was obtained from the Joint CUHK-NTEC Clinical Research Ethics Committee.
Randomization and masking
A computer software generating true random numbers will be used for assigning participants into interventions. For Trial 1, participants will be randomized by center or clinic (that is, participants recruited from the same setting will be randomized in one unit). For Trial 2, participants will be randomized individually. In both trials, participants will be told on the consent form that they would receive ‘a course related to the caring of patients with dementia,’ without being told the different interventions. The trainers will not be used as interviewers so as to ensure that the interviewers are blind to the participants’ assignment status as well. Hence, both trials are double-blind randomized controlled trials, but Trial 1 differs from Trial 2 in being a cluster trial.
Trainers should have a background in psychology, social work, occupational therapy or related fields, with experiences in gerontology and dementia. Each treatment condition will last for eight weeks. In the group mode, this will be accomplished by eight weekly group interventions lasting one-and-a-half to two hours each. To facilitate interaction during group sessions, each group will have a maximum of 10 caregivers. In the individual mode, there will be four biweekly sessions of three hours each conducted at the caregiver’s home. The number of face-to-face contact hours between trainer and caregiver is held to be constant across the three intervention groups to control for possible dosage effects.
The psychoeducation group will receive information on aging in general, AD, BPSD, general coping skills (including relaxation, but not positive reappraisal), community resources, ways to handle BPSD and activities of daily living (ADL) needs, and communication skills with the CR. Application of skills will be discussed in relation to CRs’ needs and perspectives, and the importance of developing alternative approaches to match CR conditions.
The benefit-finding group will receive the same curriculum as psychoeducation (but not spending as much time on each topic) plus training in positive reappraisal coping and finding benefits. In addition, during this period, they will record ‘benefit diaries’ up to three times per week; during the evening, caregivers will reflect on the events and experiences in the past few days and mention three events/experiences from which benefits are constructed. Because literacy rates are low in the older population in Hong Kong, in order to standardize the diary method, all participants will be provided with a voice recorder to record their reflections. This also serves as a quality check that the diaries are produced by the participant personally.
Finally, the simplified psychoeducation group will receive an identical curriculum as the psychoeducational group, except that the application and practical components will be delivered in talks only, without hands-on exercises and follow-up discussions. In other words, lecture will be the primary mode of delivery, as is typical of local psychoeducation programs.
On the whole, the simplified psychoeducation group is characterized by information provision only. By comparison, the psychoeducation group is also focused on information provision but it is done in a way that facilitates daily application through dedicated discussions, practical exercises, and feasibility assessments that incorporate the potential needs and perspectives of the CR. Compared with psychoeducation, the benefit-finding group further adds various activities, including diaries, to promote positive reappraisal and finding positive gains in caregiving.
Further to the baseline interventions, boosters aimed to refresh and extend learning will be provided for all three groups at roughly 14
months after the conclusion of the initial treatment. The specific contents of the training programs are shown in Tables , , and .
Baseline intervention - Trial 1 (Group)*
Booster intervention - Trial 1 (Group)*
Baseline intervention - Trial 2 (Individual)*
Booster intervention - Trial 2 (Individual)*
An important consideration when running interventions at centers or clinics is that the CRs will need to be looked after simultaneously when the caregivers attend the training. Thus, volunteer or student assistants will be needed to look after and run activities for the CRs with AD. We recommend a ratio of three assistants to one CR.
It is anticipated that no adverse effects due to the training will occur. This is consistent with trials on caregiver training around the world.
The primary outcomes are CG stress (subjective burden [31
], role overload [33
], and cortisol), perceived benefits (from open-ended questions), subjective health [34
], psychological well-being [35
], and depression [37
]. The secondary outcomes are CG coping [39
], and CR’s BPSD [40
] and functional impairment [42
]. Same for group and individual interventions, all outcome measures, except cortisol, will be obtained at baseline (0
month; T1), post-treatment (2
months; T2), and 6 (T3), 12 (T4), 18 (T5) and 30
months (T6). Salivary cortisol, collected using the Salivette five times during the day (immediately after awakening, 30 minutes post-awakening, 45 minutes post-awakening, and at 11:00 and 21:00) for two consecutive days, will be obtained at T1-T4 only. This design will allow an assessment of the long-term effects of the interventions. The last assessment (T6) is not only necessary to assess the more long-term impacts of the treatment, but is spaced out at a longer interval in order to provide a more stringent test of the effects of the treatment. It is possible that the half-yearly assessment conducted from six months onward may serve as soft reminders for benefit searching, from which the benefit group may gain more. The 12-month interval from T5 to T6 eliminates this possibility and enhances the generalizability of the findings to naturalistic settings. Data analysis will be performed using the principle of intent-to-treat, and missing values will be estimated by multiple imputations using Markov chain Monte Carlo method.
We expect that participants in the benefit-finding group will report more benefits, lower burden, lower overload, better subjective health, better psychological well-being, and lower depression than both the psychoeducation and the simplified psychoeducation group, whereas the latter two would not differ from each other. For cortisol, the benefit-finding group is expected to have a steeper slope of diurnal decline and lower cortisol level at 21:00 than the other two groups. With respect to cortisol in the awakening period, the benefit-finding group is expected to have higher levels at 0, +30 and +45 minutes post-awakening than the other two groups. We also predict an improvement in CG coping, CR BPSD, and a slower rate of decline of CR functional impairment for the benefit-finding group, over the other two groups, over time.
The data structures contain multiple levels. For Trial 1, repeated measurements (level 1) is nested within persons (level 2), which then is nested with the cluster (level 3) from which they were recruited. For Trial 2, the data contain the first two levels only. Because of the multilevel structure, data will be analyzed using mixed-effects regression with full-information maximum-likelihood estimation in Stata version 11.1. This procedure does not require imputation of missing data common in longitudinal studies because the conditional distribution of missing data based on data observed are incorporated into estimation of parameters [44
In mixed-effects regression, the intercepts and effect of time (centered) will be specified to vary randomly at cluster (Trial 1) and person (both Trial 1 and Trial 2) levels. Two dummies will be created for the treatment groups with simplified psychoeducation as the reference category. Thus the effects of benefit-finding and psychoeducation will be compared against simplified psychoeducation in the first set of analyses. Product terms of benefit-finding x time and psychoeducation x time will be created to test whether changes in the outcome measures over time are different from those of the simplified psychoeducation group. Additional analyses will be conducted by dropping participants in the simplified psychoeducation condition while coding psychoeducation as the reference group, so that the effects of benefit-finding can be compared directly with psychoeducation as well. In this case, only two treatment groups will be included in the analyses.
With six repeated measurements (month 0, 2, 6, 12, 18, and 30), a medium treatment by time interaction effect (Cohen’s d
0.80, two-tailed alpha
.017 (Bonferroni-corrected for multiple comparison among three treatment groups), ratio of time-effect variance at person-level to the sum of random-intercept variance and time-level residual variance
0.20 (which is quite large to yield conservative estimates of power), 17 participants per experimental condition are sufficient, no matter whether there are two or three experimental conditions in the regression analysis. For the cluster-randomized controlled trial, with the addition of the ratio of time-effect variance at home-level to the sum of random-intercept variance and time-level residual variance
0.02, four homes per condition and 10 participants per home are adequate, regardless of whether there are two or three experimental conditions in the regression analysis [45