Exercise interventions are being explored as a means to minimize cognitive decline in MCI. One study observed significant improvement in memory in participants who engaged in a 24-week physical exercise regimen [15]. Participants were encouraged to perform at least 150 min of moderate intensity exercise per week. Walking was most frequently recommended, but participants were free to choose other forms of exercise. However, these participants did not necessarily meet contemporary criteria for MCI.

In appropriately designed studies, various investigators, including the principal investigator (PI) of this study, have shown that computerized brain fitness training can improve cognitive function in normal older adults [16,22,23]. Barnes et al [24] reported no significant benefit from brain fitness training in MCI. However, the memory effect size they observed was nearly the same as that reported as significant in the larger study in cognitively normal older adults. This suggests that the Barnes and colleagues’ study (and other studies) may be underpowered to see the modest beneficial effect of computer training in MCI. Other recent trials of a computer training program have suggested modest benefit on cognition and mood in patients with MCI [25]. Belleville et al [26] also reported positive outcomes for a program combining education, computer-based attention training, and internal memory compensatory training. They reported improved list-recall memory and face-name association performance, as well as improvements in subjective memory report and sense of subjective well-being. Another randomized case-control trial of a computer training program demonstrated improvements in story recall, abstract reasoning, and behavioral problems in participants with MCI [27].

There is evidence that psychoeducation can bring about positive change in care partners of those with MCI [28,29]. Although most research focuses on dementia education for caregivers, a recent study in patients with MCI found that care partners were more depressed when they had less knowledge of MCI at the time of diagnosis [30]. Studies further show that greater knowledge increases feelings of competence, confidence, and less overall distress in care partners [29].

It is becoming apparent that social support is an important aspect of behavioral interventions. For example, research shows that patients with MCI who participate in group therapy learn acceptance of their diagnoses and what lies ahead of them, and care partners become more aware and accepting of cognitive and behavioral problems [31]. A high level of social engagement can reduce the mortality risk in individuals with MCI [32]. More recent studies are reporting similar findings—there is greater acceptance in participants and care partners undergoing group therapy versus wait-list controls [17,33]. Furthermore, a meta-analysis [33] revealed that a higher risk of AD was associated with loneliness and lower levels of social networking and physical activity.

The Memory Support System (MSS) involves use of a portable calendar and note-taking system that includes three sections: (1) appointments, (2) “to-do” items, and (3) a notes section. MSS trainers provide persons with MCI and a care partner training sessions following a structured curriculum of orientation, modeling, practice, and homework assignments. Standardized forms are used daily to document adherence to the process by the trainers as well as to note individuals’ progress. The MCI state is the ideal target for this intervention as people demonstrating MCI typically have “cognitive reserve” in the form of preserved procedural memory even in the face of clinically significant declarative memory impairments [34]. We have specifically studied this strategy and found significant effect sizes on memory-based everyday functions at training end (d=1.0), 2-month follow-up (d=.88), and 6-month follow-up (d=.56) compared with no-treatment controls [18].

As detailed above, multiple kinds of behavioral interventions have shown promise for different outcomes (eg, quality of life [QoL], functioning, cognition, and self-efficacy). Thus, in the HABIT program and in this trial, we opted to measure all of these outcomes, given that individual studies suggest different interventions may impact different outcomes.

Older adults and especially those with cognitive impairment have unique challenges in adhering to behavioral interventions [35,36]. Yet, in our preliminary studies of various combinations of these behavioral interventions, we have found reasonable enrollment, adherence, and excellent retention of patients with MCI and their partners in this type of research [37]. Preliminary comparative effectiveness data comparing MSS with computerized training showed that the MSS is superior to computerized training on a memory-related functional outcome [38], and both MSS and computer training appeared to contribute to improved partner mood in comparison with providing no treatment [39]. The study described herein used the recruitment, intervention, and evaluation infrastructure of the clinical HABIT program to expeditiously determine which behavioral intervention outcomes related to the prevention and delay of dementia are most important to persons with MCI and their partners.

Participants engaged in daily 45 to 60 min of yoga. We used yoga as it is suited to the constraints of space and the different levels of baseline physical activity of our participants and partners. We used the HABIT framework, which uses an adapted Hatha Yoga practice where participants sit on chairs for some asanas (poses) and use the chair for support for balance during other standing poses and for others parts of the sequence. This adapted Hatha Yoga style is appropriate for older adults and is both beneficial and accessible for those who have limited mobility, including those with walkers or those who are in wheelchairs. HABIT yoga also incorporates breathing and meditation and cultivates an overall sense of connection and support.

Our instructors had at least 200 hours of training and were certified. The appropriately sequenced HABIT yoga practice met the American College of Sports Medicine recommendation for older adults for muscle strengthening and flexibility. The sessions used an armless sturdy chair placed on top of a sticky mat. Instructions were mirrored for the participants (the instructor faced the students and performed a posture on the left side while instructing the right side to reduce confusion). Breathing practice focused on increasing lung capacity and oxygenation. The sessions included meditation practice to support internal focus.

The HABIT yoga intervention was intended to initiate and sustain a schedule rather than a type of physical activity. Following the 10-day program, participants and partners were encouraged to maintain a schedule of 150 min of their preferred exercise per week. Post program, we consider yoga, swimming, walking, running, or formal exercise programming (water aerobics, resistance training, etc) to count equivalently toward this total. Because most clinical trials of yoga include group classes supported by home practice, we provided a customized DVD (digital video disc) as a supplement for continued use and practice after the program to those who opt to continue yoga. The DVD included sections on the following: poses, modifications, benefits, breathing, and meditation practices.

We used the commercially available Posit Science product BrainHQ [41] on tablets (eg, iPads). At the time of the study, this product was the latest generation of the BrainFitness auditory processing speed program studied by Smith et al [16] and Zelinski et al [23] (and included components of the Insight visual processing speed program). Participants completed 45 to 60 min of training daily in the program and were encouraged to maintain 150 min of computerized brain training per week for 18 months post program. Participants were provided a 1-year subscription to the program. Each participant’s adherence and progress were tracked through the clinician portal provided by Posit Science, both during HABIT and for 12 months post program.

This education program involved daily 45- to 60-min group lecture sessions with topics including the following: Introduction to the Program, Living with MCI, Changes in Roles and Relationships, Sleep Hygiene, Steps to Healthy Brain Aging, Preventing Dementia, MCI and Depression, Nutrition and Exercise, Assistive Technologies, Participating in Research, and Community Resources.

We conducted separate support groups with patients and partners. Group size was limited to 10 members at a maximum. The patient group was focused on reminiscence and adaptation, while the partner groups focused on building resources for coping.

We provided each couple with MSS training 45 to 60 min daily with initial and ending adherence sessions. The curriculum is described briefly here.

We assessed the patients’ cognitive function using Cogstate [46], a computerized measure of cognition. We used the Cogstate battery specifically designed for preclinical Alzheimer and MCI populations. This includes measures of simple and choice reaction time (the detection and identification tests, respectively), a test of visual memory (the One-Card Test) and a measure of working memory (One Back Test). This was completed at baseline and at 12-month follow-up.

We administered a timed 400-m walk and the Short Physical Performance Examination [47], which includes a timed 4-m walk, standing side by side, semitandem and full-tandem stance, and a timed arms-folded rise from seated to standing 4 times. This was completed at baseline, intervention completion, and 12-month follow-up.

Activities of daily living (ADL) functional status ratings based on informant assessment were obtained at baseline, intervention completion, and 6 months, 12 months, and 18 months post intervention. The Everyday Cognition (ECog) [48] was used to assess impairments in instrumental ADL. The ECog is an informant-based measure that assesses a participant’s ability to perform everyday tasks in the following areas: memory, language, visuospatial abilities, planning, organization, and divided attention. It was constructed specifically to be sensitive to changes in MCI. Factor analysis supports a 7-factor structure, including one global factor and 6 domain-specific factors. The global factor is strongly correlated with CDR score, MMSE, and clinical diagnosis. In addition to the global factor, the everyday memory factor differentiates MCI from normal cognition, and the everyday language factor differentiates MCI from dementia. Test-retest reliability over an average of 29 days is good (r=.82) [48]. The ECog was modified with its author’s support to assess the participant’s existing functional ability at each time point rather than the original wording comparing functioning with 10 years before to better gauge change from baseline to follow-up. Additionally, in anticipation of future long-term follow-up, ADL and instrumental ADL were further assessed with the Functional Assessment Questionnaire (FAQ) at baseline, intervention completion, and 6, 12, and 18 months post intervention. The FAQ is the standard functional measure required for use throughout the Alzheimer’s Disease Research Center’s network and was developed for use with dementia patients but has also shown to discriminate between normal controls and those with MCI [49]. In this regard, we surmised the FAQ would prove to be more useful in measuring advanced ADL impairments than the ECog in longitudinal follow-up.

At all assessment points, both the patient with MCI and the care partner completed the Center for Epidemiological Studies Depression Scale (CES-D). In addition, both the participant with MCI and the care partner completed the Anxiety Inventory Form, a 10-item rating scale modified from the State-Trait Anxiety Inventory [50] by the Resources for Enhancing Alzheimer’s Caregiver Health project [51].

Both the participant and care partner completed the QoL-ADL [52]. The QoL-ADL is a 13-item measure developed for individuals with dementia that has been utilized in MCI and with care partners. Participants and care partners were asked to rate their relationships, concerns about finances, physical condition, mood, energy level, memory, aspects of daily functioning, and overall life quality on a 4-point scale.

The MCI participants completed a measure of self-efficacy at all assessment points using modified, selected items from the chronic disease Self-Efficacy Scales [53]. The entire 3-item Do Chores Scale, 2-item Social/Recreational Activities Scale, and 4 items of the 5-item Manage Disease in General Scale were utilized based on their relevance to MCI. Original scales have reported internal consistency reliability of r≥.82 and test-retest reliability of r≥.84 [53]. The language from the original scales was modified to be specific to those with MCI (ie, “your memory/cognitive difficulty” rather than more general references to “your health condition”). The result is the 9-item Self-Efficacy in Mild Cognitive Impairment Scale.

Care partners completed the caregiving competence and mastery components of the Pearlin [54] at all assessment points. The measures reflect their titles and range from 4 to 6 items.

Care partners completed the short form of the Caregiver Burden Inventory [55] at all assessment points. This measure is an assessment of degree of stress experienced by family caregivers. It includes 12 questions concerning the effect of the participant’s disability on care partners’ lives. It is scored as a composite measure, combining several aspects of caregivers’ reactions.

Patients were given a CDR [56], which is a structured interview designed to stage dementia. In the CDR, the patient is rated on 6 dimensions: Memory, Orientation, Judgment and Problem Solving, Community Affairs, Home and Hobbies, and Personal Care; they are then assigned a global score, which is generated via an established algorithm. A global score of ≤0.5 was required for enrollment into the study. This measurement was updated at the 18-month visit to stage the patient’s cognitive impairment at study completion.

The Dementia Rating Scale, 2nd edition [57], was administered to help determine the overall level of cognitive functioning at baseline. This was also given at the 12-month follow-up point to characterize global cognitive functioning.

The MMSE [58] is a widely used screening measure of cognitive impairment. The MMSE was given to care partners at the eligibility session to determine whether global cognitive functioning is intact. A score of ˃24 was required for enrollment in the study.

The participant with MCI and the care partner completed a measure of self-compassion at all assessment points using the 12-item short form of the Self-Compassion Scale [59]. This form has high internal consistency reliability of r=.87 and a very high correlation r=.97 with the full 26-item Self-Compassion Scale [60].

The participant with MCI and the care partner completed the 6-item Gratitude Questionnaire Scale [61] at all assessment points. This Likert scale also has high internal consistency reliability of r=.82 and a high fit index (.95) for a single-factor model and good concurrent validity [61]. Table 2 lists the timing of the efficacy and ancillary measures used in the study.

Data analysis for aim 1 is complete and reported elsewhere.

To determine the degree to which the different components of the HABIT program contribute to improvements in each of the targeted outcomes, we will utilize linear mixed effect models approach that accounts for the randomized complete block design that was employed in the assignment of treatment combinations to study sessions within each site . In these models, we will include indicators for the different treatment effects, as well as indicators for potential confounding features such as age, sex, study site, and study session. Using these analyses, we will test for improvements from baseline over follow-up by testing for the significance of interactions between the variables representing study components and those reflecting follow-up period. These tests are similar to paired t tests but are more flexible. For example, they allow simultaneous analysis of data from more than 2 time points, and they enable the inclusion of covariates that may differ from baseline to follow-up.

We will examine the significance of the different treatment effects both with and without adjustment for potential confounders such as gender, age, and baseline scores on the Dementia Rating Scale. We will also explore potential heterogeneity of treatment effects by assessing interactions between treatments and gender, age, and baseline Dementia Rating Scale score of the patient and relationship of the partner (spouse vs other).

We aim to illustrate the efficiency of our novel design and statistical analysis methods for the evaluation of multicomponent interventions. The need for multicomponent studies is so critical that the Food and Drug Administration has gone so far to issue a plan for how to ensure combination (multicomponent) trials meet its standards [63]. Typical clinical trials often contrast one treatment to another (often a placebo). In the case where multiple treatments are studied, it is possible to employ fractional factorial designs [64]. In these designs, subsets of experimental combinations are carefully selected to enable the estimation of treatment effects of interest. As these designs compare an experimental condition with no active treatments to experimental condition with only one or two treatments, we opted to pursue a different experimental approach. In this approach, we formed 5 distinct treatment group combinations by removing a single component of the 5 possible components comprising the HABIT program. As part of this proposed effort, we will assess the efficiency of this study design in comparison with 2-arm studies (ie, treatment vs control) and fractional factorial designs. In our application of this study design, we randomly assigned one of these five treatment combinations to study session, stratified by study location in such a way that imbalance is minimized. Our existing study design meets the following constraints: (1) each study site (Arizona, Florida, Minnesota, and the University of Washington) administered each candidate treatment combination once and (2) each treatment combination was given 4+ times for a total of at least 20 sessions of treatment.