ACT for Dementia Caregivers: Evaluating Psychological Flexibility as the Mechanism to Improve Psychological Distress

Design of the current study:

The proposed study has several design aspects that should be identified and justified. It is a waitlist randomized controlled trial, allowing us a rigorous experimental design for between-person comparisons and robust assessments of mediation between treatment (Tx) and waitlist control (W), while also allowing all participants to get the treatment, and facilitating a larger sample size for within person analyses. See Figure 3 for the timing and delivery of the study assessments.

Participants Power analysis: Using RStudio, the investigators stimulated models using the lme4 package and mediation package. Based on pilot data, the investigators assumed a medium effect across all variables (i.e., X --> M, X--> Y, and M --> Y) and a strong intraclass correlation (ICC = .6). The investigators ran 1000 simulations and found that using a sample size of 120 participants (60 Tx, 60 in waitlist), the investigators were able to identify significant average causal mediation effect, significant average direct effect, a significant total effect, and that the proportion mediated was significant (all ps \< .05). The estimated proportion mediated was approximately 25% (95% CI: [10%,57%]), suggesting the ability to identify meaningful indirect pathways from X to Y via M. These results indicate that a sample size of 120 participants provides adequate power for the mediation analyses. Given that the mediation model represents the most complex analysis in the study and accounts for individual pathways, this sample size will be sufficient to power the direct between- and within-person analyses outlined in both Aim 1 and Aim 2. Based on the pilot study where a 27.7% attrition rate was observed between the first and second timepoints, the investigators will aim to recruit 153 individuals into the study to arrive at the desired sample size of 120.

Recruitment: Family caregivers will be recruited through IRB-approved materials. The investigators will utilize the recruitment methods deemed most effective in the prior Stage I and II studies. Of note, the prior Stage II study recruited 90+ participants across nine months in 2023. The investigators will advertise via 1) the Alzheimer's Association's TrialMatch program, 2) social media and print/online stories distributed by Utah State University, 3) health fairs and events such as the Walk to End Alzheimer's, and 4) existing listservs of community partners. These partners include Utah's Alzheimer's Disease and Related Dementia Coordinating Council (Utah Department of Health and Human Services), and USU's Institute for Disability Research, Policy, and Practice. New approaches for the current study will include accessing the research registry of USU's Alzheimer's Disease and Dementia Research Center, and using the newly-established USU Dementia Caregiver Clinic, which brings in an estimated 80 caregivers per year. The investigators also will utilize the Participant Manager's time to actively recruit out of state to increase geographic and ethnic/racial heterogeneity. The investigators will request that the study opportunity be distributed via the 15 newsletters representing NIA supported ADRC's. With IRB approval, the investigators can request to access the Participant Resource Pool Michigan Center for Urban African American Aging Research (MCUAAAR).

Retention: Participants will receive weekly reminder emails to complete a session, on a day of their choosing. If a participant has lapsed in engagement for two weeks, they will receive a reminder phone call. If attrition occurs, the data from any of the assessments or responses in treatment sessions attended are still useful; as described in the analysis section, an intent to treat analytic approach will be used.

ACT for Caregivers Intervention: The ACT for Caregivers online, self-guided intervention is intended to improve psychological flexibility (via learning to be more open, aware, and engaged), and to improve psychological distress. Participants access 6 sessions (each corresponding to one or more tenets of psychological flexibility), which they work through in order, asynchronously at their own pace, at home. Participants are encouraged (but not mandated) to complete one session per a week over six weeks, with sessions taking 20-40 minutes to complete.

Analyses:

An 'intent to treat' approach will be employed for all participants in the study. Analyses will be completed using Stata 18 (data cleaning) and R 4.0.0 or higher (lme4, nlme). A linear mixed-model will be used, where measurement occasions (Level 1) are nested within individuals (Level 2); treatment condition will be a Level 2 predictor that tests differences between the treatment and waitlist groups from T1 to T3. As observations (Level 1) are nested within individuals' (Level 2) over time, the investigators will employ mediation analyses within a mixed modeling framework. As an RCT examining psychological flexibility (putative mediator = M) as a mechanism underlying the effects of the ACT intervention (intervention = X) on reductions in caregiver psychological distress (outcome = Y), the investigators will conduct the following analyses, consistent with the Checklist for Investigating Mechanisms in Behavior-change Research (CLIMBR).

To assess aim 1, Model 1 will be used to investigate whether participants assigned to treatment with the ACT intervention exhibit decreased psychological distress relative to wait-list control participants, establishing the CLIMBR X --> Y path. To assess aim 2, Model 2 will extend Model 1 to include psychological flexibility as a mediator (M), to evaluate the CLIMBR X --> M and X --> M --> Y linkages articulating the extent to which ACT intervention-related changes in distress are mediated by (increased) psychological flexibility.

Mediation with be formally evaluated by percent reduction in the X --> Y association, tests of indirect effects, and computation of mediator effect sizes in RCTs. Supplemental analyses will consider the subcomponents of psychological flexibility, open, aware, and engaged, to explore the specific dimension(s) driving ACT-related change in psychological distress. Further, chi-square tests of nested model comparisons, evaluating model fit when constraining the magnitude of psychological flexibility effects to be equal, will be employed to determine differences in the magnitude of psychological flexibility components as mediating mechanisms. The investigators will also consider false discovery rates to evaluate potential spurious associations.

To assess aim 3, the investigators will investigate the role of program adherence in associations between psychological flexibility and caregiver psychological distress. The investigators will regress changes in T1 --> T2 and T1 --> T3 psychological flexibility on the number of sessions completed at T2, the operant index of adherence (# sessions completed 75 percent or more), to determine any modulation of effect by adherence. All item-level missingness will be treated using full information maximum likelihood (FIML).