Behavioral interventions are a non-pharmacological treatment that shows improvement in the everyday functioning of people with Mild Cognitive Impairment (MCI). Multiple studies have focused on examining factors that can reduce or enhance adherence to behavioral interventions. However, few studies use adherence as a predictor of functional changes. The goal of this study was to analyze the association between adherence, age, and education in factor score changes of participant impairment, participant adjustment, and partner adjustment in a sample of participants with amnestic MCI (aMCI) and their study partners.

We included fifty-two dyads of a person with aMCI and their study partner with intervention data at baseline and 24-week follow-up from the Physical Exercise and Cognitive Engagement Outcomes for Mild Neurocognitive Disorder (PEACEOFMND) study. At baseline, participants were randomized to one of three behavioral interventions: computerized cognitive training (BrainHQ; n=19), yoga (n=15), or wellness education (n=18). Factors were established from a larger clinical sample that used the same measures as PEACEOFMND. The three-factor latent structure was constructed as the following: 1) participant adjustment combined scores of the Center for Epidemiologic Studies Depression Scale (CES-D), Quality of Life in Alzheimer’s Disease (QoL-AD), and Self-Efficacy for managing MCI scales; 2) partner adjustment included study partner’s scores in CES-D, QoLAD and Caregiving Competence and Mastery Components (CCMC) of the Pearlin scales; 3) participant impairment included participant’s scores in E-Cog memory domain, and study partner’s scores in the Functional Activity Questionnaire (FAQ) and Zarit Burden Interview. We calculated factor changes by obtaining the difference between factor scores at follow-up and baseline. Bayesian correlation analysis was performed to investigate the association between age, education, adherence to the combined behavioral interventions, participant adjustment, participant impairment, and partner adjustment.

The Bayesian correlation results showed moderate evidence (BF10=6.8, Pearson’s r=0.38) supporting a positive correlation between adherence and change in participant adjustment. Additionally, there was moderate evidence (BF10=2.18, Pearson’s r=0.32) supporting a positive correlation between change in participant impairment and participant level of education as well as participant age and change in partner adjustment (BF10=2.8, Pearson’s r=0.33).

Bayesian correlations replicated results from previous analysis using a traditional method, showing that increased adherence to combined behavioral interventions is associated with an increase in participant’s quality of life, self-efficacy, and better mood. Thus, commitment to behavioral intervention completion in aMCI participants is related to overall participant adjustment.

While declarative learning is dependent on the hippocampus, procedural learning and repetition priming can operate independently from the hippocampus, making them potential targets for behavioral interventions that utilize non-declarative memory systems to compensate for the declarative learning deficits associated with hippocampal insult. Few studies have assessed procedural learning and repetition priming in individuals with amnestic mild cognitive impairment (aMCI).

This study offers an overview across declarative, conceptual repetition priming, and procedural learning tasks by providing between-group effect sizes and Bayes Factors (BFs) comparing individuals with aMCI and controls. Seventy-six individuals with aMCI and 83 cognitively unimpaired controls were assessed. We hypothesized to see the largest differences between individuals with aMCI and controls on declarative learning, followed by conceptual repetition priming, with the smallest differences on procedural learning.

Consistent with our hypotheses, we found large differences between groups with supporting BFs on declarative learning. For conceptual repetition priming, we found a small-to-moderate between-group effect size and a non-conclusive BF somewhat in favor of a difference between groups. We found more variable but overall trivial differences on procedural learning tasks, with inconclusive BFs, in line with expectations.

The current results suggest that conceptual repetition priming does not remain intact in individuals with aMCI while procedural learning may remain intact. While additional studies are needed, our results contribute to the evidence-base that suggests that procedural learning may remain spared in aMCI and helps inform behavioral interventions that aim to utilize procedural learning in this population.