Evaluation of an Intervention for Living With Mild Cognitive Impairment

The following outcomes from our multicomponent behavioural intervention are expected.

MCI participants will show:

1. Improved functional memory 2) Increased engagement in leisure activities 2) Improved instrumental activities of daily living 4) A reduction in neuropsychiatric symptoms (e.g., depressed mood, irritability)

Family members of MCI relatives will show:

1. Improvement in ability to adapt to and effectively manage challenges posed from living with a relative experiencing cognitive decline
2. Improvement in mood
3. Improvement in one or more health related behaviours

These expected outcomes are based on previously presented pilot showing improved abilities dealing with problems faced from cognitive decline due to MCI, previous research by one of us demonstrating that self-perceptions of better problem solving skills are associated with positive intervention outcomes, previous research showing improvement in mood and well-being in dementia caregivers following psychosocial intervention, and our previous findings of improved memory knowledge and strategy use in people with MCI after experiencing a similar intervention.

In researching these expected outcomes we will additionally investigate the following hypotheses:

1. The degree of cognitive, functional, and neuropsychiatric impairments exhibited by the MCI clients at baseline will influence the degree of benefit experienced by both the MCI clients and their family members on the expected outcomes outlined above.
2. In the MCI client degree of impairment may limit the gains achieved because greater impairment may reduce the ability to self-initiate and sustain behavior change. ii. Conversely, in the family members, degree of impairment in their MCI relative may enhance the gains achieved because past research suggests family members whose relatives with MCI exhibit greater neuropsychiatric symptoms may have the most to gain from early intervention, as these family members report the most significant depressive symptoms and more time spent providing support.
3. Based on our past research we expect that level of program participation (e.g., attendance, engagement with program materials and exercises) will influence degree of benefit experienced by both the MCI clients and their family members on the outcome measures.

METHODS

Participant Identification and Recruitment. Participants for the proposed research will be consenting individuals with MCI and their close family member, recruited from referrals to the clinical program. After referrals are received, and before the program begins, the research assistant will telephone the individuals with MCI and their identified family members in order to explain the study, to obtain verbal consent, and to gather preliminary demographic information. To confirm that the potential MCI participants meet accepted clinical research criteria for amnestic MCI (either single or multiple domain) they will complete a clinical assessment, including a clinical interview and brief cognitive testing, 3 months prior to the start of their participation in the program. Clinical assessment results will be reviewed by two clinical neuropsychologists (KM, AT) involved in the initial development and/or continued provision of the clinical program, who have extensive experience diagnosing MCI for research purposes. In addition to ensuring that the MCI participants meet recognized criteria for amnestic MCI, these descriptive data will be used to investigate our previously mentioned hypotheses examining predictors of effectiveness. To clarify, family members are only permitted to participate if they have a relative who is diagnosed with MCI through our clinical triage process (interview plus brief cognitive assessment). A family member would still be welcome to participate in our program if their relative with MCI declined to attend or dropped out, although to date this has not yet happened. Notably, our program is a clinical service, and volunteering for the research study is not required for participation in the program.

Design. This study will use a randomized waitlist control design. Because of the heavy demand for this clinical program, there is a naturally-occurring waitlist of approximately 3 months. Thus, three months prior to a new program session, research volunteers from the program's clinical waitlist will be randomly assigned to a waitlist-control or treatment group using a random number generator. This random assignment will be conducted after all volunteers have completed baseline testing. Research shows baseline testing can be influenced by participant knowledge and expectations and indeed our own pilot testing showed this influence whereby the treatment group was reported to have poorer functional skills as compared to the waitlist group in the absence of any other discernible group differences. Thus, in our design we will attempt to control for the influence of anticipated treatment on the outcome measures by having both treatment and waitlist-control groups undergo the same three evaluations at the same time periods: baseline pre-test before random assignment; repeat testing at 10 weeks (which will serve as the post-test measure for the waitlist control group and the repeat-pre-test measure for the treatment group) and repeat testing at 20 weeks (which will serve as the post-test measures of the treatment group at 1 month follow-up and the repeat post-test measure for the waitlist group). The group comparisons that will allow us to determine the efficacy of the program in comparison to no program will be through examining the waitlist-control group test schedule 1 versus test schedule 2, and the treatment group test schedule 2 versus test schedule 3. All individuals, regardless of their initial group assignment, will be tested before and after their involvement in the program permitting us to determine individual differences in the degree of benefit obtained from the program. Although there is a 3-month follow-up session in the current program format, testing at this time period is not included in the experimental design because there is no control group for comparison due to program frequency (4 per year) and length of naturally occurring waitlist (3 months). To determine the effectiveness of the intervention, we will use a battery that includes both paper-and-pencil questionnaires that are filled out at home, telephone interviews, and on-site group testing.

Sample size. Sample size estimates for MCI interventions were determined based on our previous research; for caregiver interventions, these were calculated using effect sizes (g) from a meta-analysis and Cohen's power tables. For behavioural interventions aimed at MCI, our previous research showed large and significant behavioural changes (ηp2 = .28 - .33) with sample sizes of 24 per group. For caregiver interventions most similar in nature to ours, effect sizes for improvements in a variety of outcome measures ranged from g = .22 to .52. For comparisons using alpha of .05 and power of .80, we estimate statistically significant effects with sample sizes of 30 to 46 per population group for the smallest effects (i.e., on measures of mood) to less than 8 per group for the largest effects (i.e., on measures of coping). Given the multiple comparisons, we plan to recruit 60 (30 MCI, 30 family members) for the treatment groups and 60 (30 MCI, 30 family members) for the waitlist control group. Based on the most recent annual attendance of 111 people for this clinical program (66 MCI and 45 family), we anticipate no difficulty recruiting this number of participants over the two year granting period. Based on drop-out rates of 11% in our previous studies, we will recruit 144 (72 MCI and 72 family members) over the two years in order to retain 120 (60 MCI and 60 family members) participants for our final sample.

Analyses. Prior to conducting our analyses we will impute missing data in SAS v 9.2 using multiple imputation methods that assume the missing data will be missing at random (MAR) or missing completely at random (MCAR). This model-based strategy for missing data provides significantly better parameter estimates than strategies such as mean substitution or listwise deletion of missing cases. To test our hypotheses that the intervention will result in improved leisure engagement, mood, and functional skills in MCI and in improved health-related behaviours, mood, and adaptive problem solving in family members we will separately consider the MCI participants and their family members and examine program effectiveness by comparing treatment and waitlist-control groups at the respective post-test assessment on each measure using repeated measures analyses of covariance; baseline scores will be used as covariates in order to reduce between-subject variance and to enhance detection of between-group differences. We will also perform a qualitative thematic analysis of responses on our social validity measure to further evaluate the clinical significance of the intervention and to provide information for further improvements to the intervention. To test our hypotheses about the influences of the cognitive and behavioural characteristics of the MCI participants on intervention outcomes [namely that MCI clients with greater impairments will show the least benefit; however, their family members will show the most benefit] we will use multiple regression to investigate which factors moderate change in both the family members' health behaviours, mood, and adaptive skills and in the MCI participants leisure activity engagement, mood, and functional skills. Specifically, we will examine the predictive contribution of (i) the MCI client's abilities (degree of neuropsychiatric symptomatology and level of everyday functioning, obtained from self and family report, and level of cognitive impairment, obtained from direct assessment of the MCI client), and (ii) the MCI clients' intervention-related change in memory strategy knowledge and use (objective and self-report measures obtained directly from the MCI clients) to the benefits obtained as measured by outcome measure change scores from pre-test to post-test.

Feasibility. The program is offered as a regular clinical service, free of cost to participants, with four programs offered each year. In the past year, 86 clients with MCI and 67 of their family members participated. We currently have 13 MCI clients and 12 family members on the waitlist for our next series beginning in April, with additions to the list occurring almost daily. Although participation in research is not a requirement for attendance in this clinical program, we have a high rate of past participant recruitment from people accessing this intervention program, as demonstrated by previous program evaluations. The assembled team of clinician researchers has significant clinical expertise in normal and abnormal cognitive aging, designing and implementing effective cognitive intervention in normal and neurological populations, and in health outcomes associated with caregiving and caregiver interventions. Drs. Murphy (lead investigator) and Rowe (neuropsychologists) and Ms. Climans (social worker) are directly involved in the crafting and current provision of the program and will execute the research study. Drs. Troyer (neuropsychologist also involved in crafting of the intervention protocol), MacKenzie (psychologist), and Dawson (scientist with background in occupational therapy) will provide ongoing consultation regarding project execution, measurement of outcomes, and analysis and interpretation of findings based on their relevant areas of expertise.