Effects of care partner type on outcomes from a multicomponent behavioral intervention for mild cognitive impairment

Abstract

INTRODUCTION

Multicomponent non‐pharmacological interventions, such as the HABIT Healthy Action to Benefit Independence & Thinking® program, have shown promise for improving emotional well‐being, functioning, and self‐efficacy in both people with mild cognitive impairment (pwMCI) and their care partners. However, limited research has examined whether outcomes differ based on type of care partner who co‐enrolls with the pwMCI.

METHODS

We analyzed data from 875 pwMCI−care partner dyads who completed the HABIT program. Multivariate linear mixed‐effects regression models evaluated changes from baseline in outcomes by care partner type (spouse vs. non‐spouse) at post‐intervention, 6‐months, and 12‐ months.

RESULTS

At post‐intervention, pwMCI with spousal care partners demonstrated significant improvements across all outcomes from baseline, whereas those with non‐spousal partners improved in five out of seven outcomes. Across subsequent follow‐ups, pwMCI with spousal care partners showed sustained mean reductions in anxiety (post‐intervention: −2.0; 12‐month: −0.9) and increased compliance (post‐intervention: 5.7; 12‐month: 0.9). Everyday functioning in memory (post‐intervention: −0.9; 12‐months: 1.8) and executive functioning (post‐intervention: −0.7; 12‐months: 4.8) improved at post‐intervention but declined thereafter. Spousal care partners reported reduced anxiety (−1.2) and depression (−1.3) at post‐intervention, but by 12‐months they experienced increased burden, anxiety, and depression relative to baseline. Non‐spousal pwMCI–care partner dyads did not show significant change at 6 or 12‐months.

DISCUSSION

The HABIT intervention may have more favorable outcomes for pwMCI and care partners in spousal compared to non‐spousal dyads. These findings highlight the potential impact of care partner characteristics on both immediate and long‐term response to behavioral interventions for MCI.

Highlights

• At post‐intervention, people with mild cognitive impairment (pwMCI) with spousal care partners showed significant improvements across all outcomes, while those with non‐spousal partners improved in five of seven outcomes.

• Gains attenuated over time for both spousal and non‐spousal dyads

• Spousal care partners reported reduced anxiety and depression at post‐intervention, but by 12 months they experienced increased burden, anxiety, and depression relative to baseline.

• Non‐spousal pwMCI–care partner dyads did not show significant change at 6 or 12 months.

1. BACKGROUND

Mild cognitive impairment (MCI) represents an early stage of cognitive decline between normal aging and dementia that is characterized by noticeable memory loss or cognitive domain impairment—such as language—while individuals remain largely independent in activities of daily living. ¹ The incidence of MCI increases with age, ranging from 22.5 per 1,000 person‐years among individuals aged 75 to 79 to 60.1 per 1,000 person‐years among those aged 85 and older. ² Alzheimer's disease (AD) is the most common cause of MCI, ² and approximately 15% of people with mild cognitive impairment (pwMCI) due to AD will progress to dementia within 2 years ³ and about one‐third within 5 years. ⁴ MCI is increasingly recognized as a critical window for early intervention strategies aimed at preventing or delaying the onset of dementia.

The United States Food and Drug Administration has recently approved lecanemab and donanemab‐azbt for AD in the earlier disease stages (i.e., MCI or mild dementia), with the goal of slowing disease progression through beta‐amyloid reduction. ⁵ These pharmacological treatments to date have demonstrated statistically significant slowing of decline in functional outcomes, but the clinical meaningfulness of this slowing remains debated, and these medications are accompanied by risk of adverse effects. ⁶ , ⁷ , ⁸ As a result, patients and their families are increasingly encouraged to pursue non‐pharmacological interventions to slow or prevent progression. Emerging evidence suggests that behavioral interventions may yield beneficial effects in pwMCI, including improvements in cognitive function and performance of daily activities. ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ One such intervention is the Healthy Action to Benefit Independence & Thinking^® program (hereafter HABIT program), designed specifically for persons with amnestic MCI and their care partners. The HABIT program delivers a 10‐day intervention encompassing five core components: physical exercise (yoga), computerized cognitive training, wellness education, support groups, and memory support system training. ¹⁴ , ¹⁵ Quality of life (QoL), mental health, and cognitive functioning for pwMCI are considered key outcomes of interest as they are of high priority to both pwMCI and their care partners. ¹⁶ Previous evaluations of the HABIT program have demonstrated overall positive effects on QoL, mood, cognitive functioning, and engagement in memory‐related activities of daily living for pwMCI and mood, anxiety, and burden for their care partners. ¹⁴ , ¹⁷ , ¹⁸

A key strength of the HABIT program is its intentional inclusion of informal care partners (i.e., unpaid individuals such as spouses, adult children, friends), a group often overlooked in the majority of clinical trials. ¹⁹ In fact, informal care partners play a critical role in the care of individuals with MCI and dementia. In the United States alone, over 15 million informal care partners provide more than 18 billion hours of unpaid care annually to people living with dementia. ²⁰ Despite their essential contributions, care partners frequently face considerable challenges, including physical and mental health issues, care partner burnout, and financial strain. ²¹ Moreover, partner‐related factors—such as mood, psychological well‐being, physical health, perceptions of caregiving, and the quality of the care partner‐care recipient relationship—have been shown to significantly influence the health and well‐being of the person receiving care. ²⁰ , ²² Findings from the HABIT program have shown overall positive effects on care partners’ mood, anxiety, and burden, demonstrating the program's utility in improving these outcomes for care partners. ¹⁷

Previous evaluations of the HABIT Program have focused primarily on assessing the overall effectiveness of the intervention and the five individual components on select clinical outcomes for both pwMCI and their care partners. ¹⁴ , ¹⁸ However, little is known about the extent to which the type of care partner involved (i.e., spouse vs. non‐spouse) influences the effectiveness of the intervention. Therefore, the present study aimed to examine the associations between care partner type and intervention outcomes for both pwMCI and their care partners across multiple follow‐up time points (i.e., end of treatment, 6 months, and 12 months). Given the distinct nature of spousal and non‐spousal caregiving relationships, we hypothesized that: (1) Dyads with spousal care partners would demonstrate more significant and sustained improvements in patient and caregiver outcomes due to higher levels of daily engagement and shared living environments; (2) non‐spousal care partners, who may be newer to the intensive support role, would report a greater initial increase in burden compared to spousal partners.

RESEARCH IN CONTEXT

1. Systematic review: A targeted literature search was conducted using PubMed, CINAHL, PsycINFO, and Google Scholar. Emerging evidence indicates that behavioral interventions may confer benefits for people with mild cognitive impairment (pwMCI) and similar at‐risk groups, including improvements in cognitive function and daily functioning.

2. Interpretation: This study demonstrated significant post‐intervention improvements among pwMCI enrolled in the HABIT Healthy Action to Benefit Independence & Thinking® program. Although gains attenuated over time for both spousal and non‐spousal dyads, pwMCI and care partners in spousal dyads maintained more favorable outcomes overall.

3. Future directions: Future research should explore how relationship characteristics influence the effectiveness of non‐pharmacological interventions. Studies with larger samples of non‐spousal care partners are also needed to enable subgroup analyses (e.g., adult children vs. siblings vs. friends) and disentangle potential differences across these groups.

2. Methods

2.1. Data source

Data for this study were derived from the HABIT Program that employed longitudinal time points to collect data from participants (pwMCI and their care partners). This analysis focused on clinical outcomes for both pwMCI and their care partners assessed at four key time points: baseline, end of treatment, and at 6‐ and 12‐months post‐intervention. Comprehensive details of the HABIT study protocol, including recruitment and implementation, have been described elsewhere. ¹⁵ The study was reviewed and approved by the Institutional Review Boards at both Mayo Clinic and the University of Washington. All participants provided written informed consent prior to enrollment, and all data were deidentified.

2.2. Study sample

Because the HABIT Program evolved over time from a single‐component intervention to the current five‐component format, dyads received differing numbers of intervention components depending on the year of enrollment. To increase analytic homogeneity and maintain sufficient sample size, we restricted the analytic sample to 985 pwMCI‐care partner dyads who received four or five intervention components enrolled in the HABIT program between 2014 and 2024. We further restricted the sample to those with data available for the primary independent variable (care partner type) and individual characteristics (age, sex, race, and educational attainment), resulting in a final analytic sample of 875 dyads.

2.3. Independent variable

The primary independent variable of interest was care partner type. For the purposes of this analysis, care partner type was dichotomized into two groups: spousal care partner (including spouses and significant others/romantic partners) and non‐spousal care partner (including children, siblings, other relatives, friends, and others [e.g., a neighbor, spiritual leader]), given spousal care partners occupy unique roles and often face greater vulnerability compared to other types of care partners. ²³

2.4. Outcome measures

A total of seven clinically relevant outcome measures were assessed for pwMCI, and three for care partners. These outcomes were measured at baseline, end of treatment, 6‐month follow‐up, and 12‐month follow‐up. ¹⁵

QoL (pwMCI only) : Assessed using the 13‐item Quality of Life–Alzheimer's Disease (QoL–AD) scale, ²⁴ which evaluates multiple domains including relationships, finances, physical condition, mood, energy, memory, and daily functioning. Each item is rated on a four‐point scale (total scores range from 13 to 52), with higher scores indicating better QoL.

Memory self‐efficacy (pwMCI only) : Evaluated using a modified version of the Chronic Disease Self‐Efficacy Scale, ²⁵ adapted for MCI. ²⁶ The resulting nine‐item Self‐Efficacy in MCI scale includes the three‐item Do Chores Scale, two‐item Social/Recreational Activities Scale, and four items from the five‐item Manage Disease in General Scale, with higher scores indicating higher self‐efficacy.

Everyday cognition–memory and executive functioning (pwMCI only) : Care partners completed the memory and executive functioning domains of the Everyday Cognition (ECog) questionnaire as informant ratings of the pwMCI's daily functioning. ²⁷ The memory domain scores range from eight to 32, and executive functioning from 15 to 60, with higher scores indicating greater impairment.

Compliance score (pwMCI only) : Compliance refers to use of the Memory Support System (MSS), one component of the HABIT intervention. The MSS is a portable, compensatory memory aid calendar, to‐do, and note‐taking system. The overall compliance score ranged from 0‐10 and measured use of each section of the MSS on two randomly selected days from the preceding week. Compliance was measured by the trainer and defined as achieving a score ≥ 7. Assessments were conducted at five time points: the first day of the intervention, the final day of the intervention, and at 6‐, 12‐, and 18‐months post‐intervention.

Anxiety (pwMCI and care partner) : Measured using the 10‐item Anxiety Inventory Form, ²⁸ adapted from the State‐Trait Anxiety Inventory by the REACH project. Scores range from 10 to 40, with higher values reflecting more anxiety symptoms.

Depression (pwMCI and care partner) : Assessed via the Center for Epidemiologic Studies Depression Scale (CES‐D), ²⁹ a 20‐item self‐report instrument. Scores range from 0 to 60, with higher scores indicating more severe depressive symptoms.

Care partner burden (care partner only) : Measured using the 12‐item short form of the Care Partner Burden Inventory, ³⁰ which evaluates multiple dimensions of care partner stress and the impact of caregiving on daily life. Higher composite scores indicate greater perceived burden.

2.5. Statistical analysis

First, comparisons of demographic and baseline characteristics of pwMCI and care partners by care partner type (spouse vs. non‐spouse) were conducted using independent samples t‐tests for continuous variables and chi‐square tests for categorical variables. Next, we used independent samples t‐tests to compare the selected outcome measures for pwMCI and care partners by care partner type at each of the four time points: baseline, treatment end, 6‐month follow‐up, and 12‐month follow‐up. Finally, we conducted separate linear mixed effects regression models for each outcome measure to evaluate change over time and whether this varied by care partner type. Specifically, these models tested the main effects of time (categorical: baseline, treatment end, 6‐month, 12‐month), care partner type (categorical: spouse, non‐spouse), and the interaction of time X care partner type. All mixed models adjusted for the following covariates: pwMCI age, sex, years of education, number of intervention components received (four or five), and baseline outcome measure scores. A two‐sided adjusted p‐value of < 0.05 was considered statistically significant. All analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC).

3. RESULTS

Of the 875 pwMCI–care partner dyads included in the analytic sample (Table 1), 492 pwMCI−care partner dyads (56.2%) received all five components of HABIT, while the remaining dyads completed at least 4 of the 5 interventions. Overall, the mean age was 75.0 years (SD = 6.9) for pwMCI and 70.9 years (SD = 9.2) for care partners; 58.7% of pwMCI and 34.1% of care partners were male. The two groups had a similar racial composition (7.5% vs. 7.7% non‐Hispanic White) and a median of 16 years of education. When stratified by care partner type (spouse vs. non‐spouse), 777 pwMCI (91.1%) had a spousal care partner, while 98 (8.9%) had a non‐spousal partner (e.g., an adult child, sibling, other relative, or friend). Compared to pwMCI with a spousal care partner, pwMCI with a non‐spousal care partner were significantly older (mean age [SD]: 76.6 [7.8] years vs. 74.8 [6.8] years; p = 0.040), more likely to be female (79.6% vs. 36.4%; p < 0.001), and had fewer years of education (median [range]: 15 [6–20] vs. 16 [6–20] years; p < 0.001). In contrast, non‐spousal care partners were significantly younger than spousal care partners (mean age [SD]: 57.9 [12.0] vs. 72.6 [7.3] years; p < 0.001), and proportions of female care partners were higher than male ones in both spouse and non‐spouse groups (83.7% vs. 63.7%; p < 0.001).

TABLE 1.

Patient and care partner characteristics by care partner type

Characteristics	Overall (875)	Spouse a (n = 777)	Non‐spouse b (n = 98)	p‐Value
Patient demographics
Age, mean (SD), years	75.0 (6.9)	74.8 (6.8)	76.6 (7.8)	0.04
Female, no. (%)	361 (41.3)	283 (36.4)	78 (79.6)	< 0.001
Race other than white, no. (%)	66 (7.5)	57 (7.3)	9 (9.2)	0.51
Education duration, median (range), years	16 (6–20)	16 (6–20)	15 (6–20)	< 0.001
Care partner demographics
Age, mean (SD), years	70.9 (9.2)	72.6 (7.3)	57.9 (12.0)	< 0.001
Female, no. (%)	577 (65.9)	495 (63.7)	82 (83.7)	< 0.001
Race other than white, no. (%)	67 (7.7)	58 (7.5)	9 (9.2)	0.55
Education duration, median (range), years	16 (2–20)	16 (2–20)	16 (12–20)	0.53
Patient and care partner
# Intervention received (%)				0.19
4	383 (43.8)	334 (43.0)	49 (50.0)
5	492 (56.2)	443 (57.0)	49 (50.0)

^a Spouse and partner.

^b Adult child, sibling, other family member, friend, and other.

Table 2 presents unadjusted comparisons of clinical outcome measures between pwMCI and care partner dyads by care partner type (spouse vs. non‐spouse) across four time points: baseline, end of treatment, and 6‐month and 12‐month follow‐up. Overall, there were few statistically significant differences between the two groups. Notable exceptions included pwMCI informant rating of everyday memory functioning at treatment end, which were significantly lower (lower scores indicate less impairment) in the spousal care partner group compared to the non‐spousal group, suggesting less impact of memory on daily functioning (mean [SD]: 18.1 [4.9] vs. 19.4 [4.6]; p = 0.01). Additionally, care partner burden scores were significantly lower at the end of treatment in the spousal care partner dyads than non‐spousal dyads (mean [SD]: 11.9 [7.0] vs. 14.9 [7.9]; p < 0.001). At the 6‐month follow‐up, the spousal care partners showed higher depression (mean [SD]: 9.2 [7.1] vs. 6.2 [4.3]; p < 0.001) and anxiety scores (mean [SD]: 17.5 [5.2] vs. 15.7 [3.9]; p = 0.01) than non‐spousal partners. No other clinical outcome measures demonstrated statistically significant differences between care partner types at any of the four time points.

TABLE 2.

Patient and care partner clinical outcome measures by follow‐up time

	Baseline			At treatment end			At 6‐month			At 12‐month
Characteristics	Spouse a	Non‐spouse b	p‐Value	Spouse a	Non‐spouse b	p‐Value	Spouse a	Non‐spouse b	p‐Value	Spouse a	Non‐spouse b	p‐Value
Patient
Quality of life (QOL‐AD)	40.2 (5.3)	39.7 (5.9)	0.39	41.5 (5.1)	40.7 (5.7)	0.20	40.6 (5.6)	39.0 (6.8)	0.11	40.8 (5.5)	40.3 (6.2)	0.53
Self‐efficacy (Mem‐SE)	71.9 (14.6)	72.7 (15.0)	0.58	75.3 (13.2)	76.6 (12.5)	0.38	71.9 (14.9)	74.4 (14.2)	0.35	71.5 (16.0)	71.0 (16.1)	0.83
Anxiety (AIF)	18.3 (5.2)	17.3 (5.3)	0.08	16.2 (4.6)	16.0 (4.8)	0.76	17.3 (5.2)	17.1 (5.9)	0.85	17.1 (5.1)	17.2 (5.5)	0.95
Depression (CES‐D)	12.3 (8.2)	12.9 (8.7)	0.47	9.5 (6.7)	10.0 (7.9)	0.56	11.7 (8.7)	14.7 (9.6)	0.06	11.2 (8.2)	13.2 (9.2)	0.12
ECog‐mem	18.7 (5.1)	19.8 (5.5)	0.06	18.1 (4.9)	19.4 (4.6)	0.01	19.0 (4.8)	20.3 (4.8)	0.12	20.3 (5.7)	20.9 (6.0)	0.47
ECog‐exe	30.0 (10.0)	31.0 (8.8)	0.40	29.4 (9.7)	30.5 (8.4)	0.34	31.7 (10.2)	34.2 (11.3)	0.18	34.6 (11.4)	35.1 (11.4)	0.80
Compliance	2.8 (2.0)	2.6 (2.1)	0.44	8.6 (2.0)	8.7 (1.8)	0.74	4.4 (3.4)	3.6 (3.8)	0.18	3.8 (3.5)	3.3 (3.1)	0.49
Care partner
Burden (CBI)	11.8 (7.4)	13.2 (7.9)	0.10	11.9 (7.0)	14.9 (7.9)	< 0.001	12.8 (7.7)	13.0 (8.4)	0.87	12.8 (7.6)	13.2 (8.0)	0.73
Anxiety (AIF)	17.3 (5.2)	16.6 (4.7)	0.19	16.1 (5.1)	16.3 (5.1)	0.71	17.5 (5.2)	15.7 (3.9)	0.01	17.7 (5.2)	16.5 (4.4)	0.11
Depression (CES‐D)	9.6 (7.0)	9.1 (8.0)	0.50	8.3 (6.6)	8.6 (7.7)	0.71	9.2 (7.1)	6.2 (4.3)	< 0.001	10.0 (7.3)	7.9 (7.1)	0.05

Data are presented as mean (standard deviation).

Abbreviations: AIF, Anxiety Inventory Form; CBI, caregiver burden inventory; CES‐D, Center for Epidemiologic Studies Depression Scale; ECog exe, everyday cognition executive functioning; ECog‐mem, everyday cognition memory; Mem‐SE, memory self‐efficacy; QOL‐AD, quality of life‐AD.

^a Spouse and partner.

^b Adult child, sibling, other family member, friend, and other.

Table 3 displays the estimated clinical outcome measures at 12 months and the differences between spousal and non‐spousal care partner groups, as derived from multivariate linear mixed‐effects regression models. All clinical outcomes in pwMCI were comparable regardless of care partner type, and similar findings were observed for care partners’ outcomes.

TABLE 3.

Estimated clinical outcome measures at 12‐month follow‐up by care partner type using linear mixed effects regression models

Clinical outcome measures	Care partner type, mean score (95% CI)
Patient	Spouse a	Non‐spouse b	Difference c
Quality of life (QOL‐AD)	40.6 (40.2 to 40.9)	40.9 (39.9 to 41.9)	−0.4 (−1.4 to 0.7)
Self‐efficacy (Mem‐SE)	70.8 (69.9 to 71.7)	71.1 (68.3 to 73.9)	−0.3 (−3.3 to 2.7)
Anxiety (AIF)	17.3 (16.9 to 17.6)	17.4 (16.4 to 18.5)	−0.2 (−1.3 to 1.0)
Depression (CES‐D)	11.5 (11.0 to 12.0)	12.2 (10.6 to 13.9)	−0.8 (−2.5 to 0.9)
ECog‐mem	20.4 (20.2 to 20.8)	20.5 (19.6 to 21.4)	0.0 (‐0.9 to 0.9)
ECog‐exe	34.7 (34.1 to 35.3)	34.8 (33.0 to 36.6)	−0.1 (−2.0 to 1.8)
Compliance	3.7 (3.4 to 4.0)	2.9 (2.0 to 3.8)	0.8 (‐0.1 to 1.8)
Care partner
Burden (CBI)	13.1 (12.7 to 13.5)	13.5 (12.3 to 14.6)	−0.4 (−1.6 to 0.9)
Anxiety (AIF)	17.9 (17.5 to 18.2)	17.7 (16.7 to 18.7)	0.1 (−0.9 to 1.2)
Depression (CES‐D)	10.1 (9.5 to 10.6)	9.6 (8.1 to 11.1)	0.5 (−1.2 to 2.1)

Adjusted for patient age, sex, education, number of interventions received, and baseline of the outcome.

Abbreviation: AIF, Anxiety Inventory Form; CBI, caregiver burden inventory; CES‐D, Center for Epidemiologic Studies Depression Scale; CI, confidence interval; ECog exe, everyday cognition executive functioning; ECog‐mem, everyday cognition memory; Mem‐SE, memory self‐efficacy; QOL‐AD, quality of life‐AD.

^a Spouse and partner.

^b Adult child, sibling, other family member, friend, and other.

^c The difference represents the estimated mean score between spouse and non‐spouse care partners. The difference is considered statistically significant if the 95%CI does not include zero.

Table 4 summarizes the estimated changes in clinical outcomes for pwMCI from baseline to each follow‐up, stratified by care partner type. Among those with spousal care partners, significant improvements were observed in all seven measures at the end of treatment. In this group, reduced anxiety and depression and increased compliance were maintained at 6 and 12 months, despite caregiver report of declining functional ability related to memory and executive function. For pwMCI with non‐spousal care partners, five of seven outcomes (excluding memory and executive functioning related daily functioning) also improved significantly at treatment end. However, both functioning measures worsened significantly by 12 months compared with baseline, and other outcome improvements were no longer significant at 6 and 12 months.

TABLE 4.

Estimated changes in patient clinical outcome measures by care partner type using linear mixed effects regression models

Clinical outcome measures	Mean change score (95% CI)
At treatment end vs. baseline	Spouse a	Non‐spouse b
Quality of life (QOL‐AD)	1.3 (1.0 to 1.6)	1.0 (0.1 to 1.9)
Self‐efficacy (Mem‐SE)	3.3 (2.5 to 4.2)	3.7 (1.1 to 6.2)
Anxiety (AIF)	−2.0 (−2.4 to −1.7)	−1.2 (−2.2 to −0.3)
Depression (CES‐D)	−3.0 (−4.2 to −1.8)	−2.9 (−4.4 to −1.4)
ECog‐mem	−0.9 (−1.6 to −0.2)	−0.2 (−1.1 to 0.6)
ECog‐exe	−0.7 (−1.2 to −0.1)	−0.1 (−1.8 to 1.4)
Compliance	5.7 (5.5 to 6.0)	6.0 (5.3 to 6.6)
At 6‐month vs. baseline
Quality of life (QOL‐AD)	0.3 (−0.1 to 0.7)	−0.4 (−1.6 to 0.8)
Self‐efficacy (Mem‐SE)	−0.4 (−1.5 to 0.8)	0.6 (−3.0 to 4.1)
Anxiety (AIF)	−0.9 (−1.4 to −0.5)	−0.2 (−1.5 to 1.2)
Depression (CES‐D)	−0.5 (−1.1 to 0.2)	1.2 (−0.8 to 3.2)
ECog‐mem	0.6 (0.3 to 1.0)	0.6 (−0.5 to 1.8)
ECog‐exe	2.3 (1.6 to 3.0)	2.1 (−0.1 to 4.4)
Compliance	1.7 (1.4 to 2.0)	0.7 (−0.3 to 1.6)
At 12‐month vs. baseline
Quality of life (QOL‐AD)	0.3 (−0.1 to 0.7)	0.9 (−0.2 to 2.0)
Self‐efficacy (Mem‐SE)	−1.4 (−2.5 to −0.4)	−1.1 (−4.4 to 2.1)
Anxiety (AIF)	−0.9 (−1.3 to −0.5)	−0.4 (−1.6 to 0.9)
Depression (CES‐D)	−0.8 (−1.4 to −0.2)	−0.3 (−2.1 to 1.6)
ECog‐mem	1.8 (1.5 to 2.2)	1.6 (0.6 to 2.6)
ECog‐exe	4.8 (4.2 to 5.5)	4.7 (2.7 to 6.7)
Compliance	0.9 (0.5 to 1.2)	0.1 (−0.9 to 1.2)

Note: Data are presented as the mean change in each clinical outcome measure from baseline within each group. A change is considered statistically significant if the 95% confidence interval does not include zero. The difference in changes between the two groups is considered statistically significant if their 95% confidence intervals do not overlap. Adjusted for patient age, sex, education, number of interventions received, and baseline of the outcome.

Abbreviation: AIF, Anxiety Inventory Form; CES‐D, Center for Epidemiologic Studies Depression Scale; CI, confidence interval; ECog exe, everyday cognition executive functioning; ECog‐mem, everyday cognition memory; Mem‐SE, memory self‐efficacy; QOL‐AD, quality of life‐AD;

^a spouse and partner.

^b Adult child, sibling, other family member, friend, and other.

Table 5 presents the estimated changes in care partner clinical outcome measures from baseline to follow‐up time points, stratified by care partner type. Spousal care partners exhibited significant improvements at the end of treatment, including reductions in anxiety (mean change: −1.2; 95% CI: −1.6 to −0.9) and depression (mean change: −1.3; 95% CI: −1.7 to −0.8). However, by the 12‐month follow‐up, spousal care partners showed a significant increase in burden scores (mean change: 1.3; 95% CI: 0.8 to 1.8), anxiety (mean change: 0.7; 95% CI: 0.3 to 1.1), and depression (mean change: 0.8; 95% CI: 0.3 to 1.4). For non‐spouse partners, only the burden score at the end of treatment (mean change: 1.8; 95% CI: 0.6 to 2.9) increased significantly. Additionally, none of the observed changes in care partner outcomes differed significantly between spouse and non‐spouse groups at any time point.

TABLE 5.

Estimated changes in care partner clinical outcome measures by care partner type using linear mixed effects regression models

Clinical outcome measures	Mean change score (95% CI)
At Treatment end vs. baseline	Spouse a	Non‐spouse b
Burden (CBI)	0.1 (−0.3 to 0.5)	1.8 (0.6 to 2.9)
Anxiety (AIF)	−1.2 (−1.6 to −0.9)	−0.3 (−1.3 to 0.7)
Depression (CES‐D)	−1.3 (−1.7 to −0.8)	−0.5 (−1.8 to 0.8)
At 6‐month vs. Baseline
Burden (CBI)	1.2 (0.7 to 1.7)	1.4 (−0.2 to 3.0)
Anxiety (AIF)	0.3 (−0.1 to 0.8)	−0.1 (−1.4 to 1.3)
Depression (CES‐D)	0.0 (−0.6 to0.6)	−0.7 (−2.6 to 1.1)
At 12‐month vs. Baseline
Burden (CBI)	1.3 (0.8 to 1.8)	1.4 (0.0 to 2.8)
Anxiety (AIF)	0.7 (0.3 to 1.1)	0.7 (−0.5 to 1.8)
Depression (CES‐D)	0.8 (0.3 to 1.4)	0.4 (−1.2 to 2.0)

Note: Data are presented as the mean change in each clinical outcome measure from baseline within each group. A change is considered statistically significant if the 95% confidence interval does not include zero. The difference in changes between the two groups is considered statistically significant if their 95% confidence intervals do not overlap. Adjusted for patient age, sex, education, number of interventions received, and baseline of the outcome.

Abbreviation: AIF, Anxiety Inventory Form; CBI, caregiver burden inventory; CES‐D, Center for Epidemiologic Studies Depression Scale; CI, confidence interval.

^a Spouse and partner.

^b Adult child, sibling, other family member, friend, and other.

4. DISCUSSION

This study examined whether patterns of clinical outcomes among pwMCI and their informal care partners differed by care partner type (spouse vs. non‐spouse) after participation in a multicomponent non‐pharmacological intervention program for MCI, the HABIT Program. Overall, the effectiveness of the intervention was comparable between spousal and non‐spousal care partner groups for both pwMCI (QoL, self‐efficacy, anxiety, depression, everyday memory/ functioning, intervention compliance) and care partners (burden, anxiety, depression). However, certain clinical outcomes—namely, care partner burden, everyday cognition of pwMCI, treatment compliance, and anxiety and depression in both members of the dyad—exhibited distinct trajectories over time by care partner type.

By the training end of HABIT, non‐spousal care partners reported an increased sense of burden, while spousal care partners did not. One potential explanation for this is that many spousal care partners may already have taken on caring for their spouses prior to beginning the HABIT program, while non‐spousal care partners may be new to this support role, leading to an increased sense of burden during their heightened involvement in the HABIT intervention period. A more pronounced awareness of the cognitive deficits among non‐spousal care partners may have also contributed to this difference. Moreover, adult children who served as non‐spousal care partners may be more likely to have other responsibilities, such as caring for their own children and work duties. As members of the “sandwich generation,” they are particularly vulnerable to the added burden of additional caregiving responsibilities. ³¹ Declines in functional ability over time are anticipated in cohorts of pwMCI. Consistent with this, increased caregiver burden was found in the spousal dyads, but not the non‐spousal dyads by 6 and 12‐month follow‐ups. Likely due to the intimate nature of the relationship, spousal care partners tend to provide the highest level of care, but are vulnerable to neglecting their own self‐care, which can lead to care partner burnout. ³² In contrast, following program completion, the level of engagement that non‐spousal care partners have with the pwMCI may revert to a relatively lower level compared with that of spousal care partners.

Another notable finding was the greater concordance, observed among spousal dyads, between pwMCI‐reported memory self‐efficacy and spousal care partner‐reported memory‐ and executive functioning‐ related daily functioning across follow‐up time point. For example, in the spousal care partner group (Table 4), PwMCI self‐reported memory self‐efficacy increased from baseline to treatment end, while care partner reported everyday memory and executive functioning significantly improved in the same time period. In contrast, this concordance pattern was not seen in the non‐spousal‐dyads. This may reflect less consistent day‐to‐day involvement commonly seen among non‐spousal care partners. For instance, adult children, siblings, extended family, or friends may each enter the program with different baseline expectations and levels of involvement. In contrast, spousal care partners are often deeply involved in daily routines and may be more sensitive and aware of subtle treatment effects, ³³ leading to earlier detection and reporting of gains. In fact, compared to non‐spousal care partners, spousal care partners have been found to report greater deficits in pwMCI, ³⁴ and their ratings tend to be more accurate as they correspond more closely to concurrently measured objective cognitive abilities and better predict subsequent cognitive decline ³⁵ , ³⁶ , ³⁷ . Given that independent functioning in activities of daily living is a core distinguishing feature between MCI and dementia, and that functional assessments typically rely on informant reports, it is critical to account for the source of observation and recognize potential discrepancies between care partner types in evaluating functional status.

Interestingly, with spousal and non‐spousal care partners the pwMCI achieved improved compliance with the MSS as rated by a HABIT trainer by the end of treatment, but compliance persisted longer in pwMCI with a spousal partner than in those with non‐spousal care partners. Compliance remained positive across all follow‐ups (end of treatment, 6 months, and 12 months) for spousal care partner dyads, but only at the end of treatment for the non‐spousal care partner dyads. While the change did not differ significantly between groups, the lack of sustained compliance among patients with non‐spousal care partners could affect long‐term treatment effectiveness, as treatment compliance has been linked to functional outcomes at both 6 and 12 months in prior HABIT analyses. ³⁸ , ³⁹ Compliance to complex multimodal behavioral interventions such as HABIT requires considerable effort and ongoing support from care partners. Greater engagement and emotional investment of spousal care partners may contribute to a higher motivation and improved long‐term intervention uptake.

Lastly, anxiety and depression are common among both pwMCI and their care partners, with reported prevalence rates in pwMCI of approximately 21% for anxiety and 14.3%–31.2% for depression. ⁴⁰ Mood symptom severity has also been associated with cognitive deficits and progression to dementia. ⁴¹ Moreover, anxiety and depression emerged as two key components of the pwMCI adjustment factor identified in an exploratory factor analysis of HABIT outcomes. ⁴² In our study, pwMCI accompanied by spousal care partners demonstrated significant reductions in anxiety at the end of treatment. Although this improvement attenuated over time, it remained significantly lower at 12 months compared with baseline. One possible explanation for this pattern is that spousal care partners may provide more consistent emotional support and reinforcement of intervention strategies, which could enhance pwMCI coping and perceived self‐efficacy, leading to sustained reductions in anxiety even after the formal intervention period. Spousal care partners experienced similar reductions in mood symptoms immediately post‐intervention; however, these symptoms increased by the 12‐month follow‐up. This increased mood disturbance in spousal care partners aligns with observed increases in care partner burden over time and may underscore the impact of the intensive involvement and emotional closeness of a spousal dyad. In contrast, among pwMCI accompanied by non‐spousal care partners, improvements were limited to reduced anxiety and depression at the end of treatment, with no other significant changes from baseline for either pwMCI or care partners at later follow‐up points.

4.1. Limitations

There are several limitations to this study and important future directions that warrant consideration. The lack of an untreated control group limits our ability to attribute significant effects fully to the HABIT intervention. That is, there may be other confounding factors that contributed to the observed differences in outcomes between spousal vs. non‐spousal dyads that we did not measure or account for by including matched controls. For example, these groups may differ systematically in their levels of motivation, interest, or engagement in HABIT. Future work should also consider additional care partner characteristics that may influence response to intervention and reporting of function such as demographics, cohabitation status, frequency of contact, and mental health and burden of the care partners at baseline, given that these factors have the potential to bias results. ⁴³ , ⁴⁴ , ⁴⁵ , ⁴⁶ Another limitation is the disproportionately large proportion of dyads with spousal care partners (88.8%), which is consistent with other large‐scale AD‐focused clinical trials but limits the generalizability of our findings. ⁴⁷ Although we had a large sample size (777 dyads with spouses, 98 dyads with non‐spouses), we may have still been underpowered to detect some small effects in the non‐spouse group. Drawing on effect sizes between 0.16 and 0.30 reported in prior analyses, ¹⁰ our analytic sample was estimated to provide roughly 32% to 80% power to detect effects at the two‐sided α = 0.05 level. Furthermore, because we did not collect data on the amount of time care partners spent with the pwMCI. As a result, we are unable to disentangle the extent to which observed differences reflected the type of care partner relationship versus differences in caregiving time or intensity. Also, it is important to acknowledge that different sub‐groups within non‐spouses (e.g., adult children, siblings, friends) may differ substantially from one another in their response to interventions like HABIT, resulting in more heterogeneity within that group. Future research with larger samples of non‐spousal partners is needed to conduct subgroup analyses (e.g., adult children vs. siblings vs. friends) to disentangle these potentially different effects. Heterogeneity also exists in the spousal group along dimensions such as length and quality of the marriage, dyadic coping style, major age differences between spouses and the health status of the spousal care partner. Future studies should further investigate how various relationship characteristics impact the effectiveness of these non‐pharmacological interventions. The generalizability of our results is also limited by the lack of representativeness in this majority white, college‐educated sample. Thus, future studies are needed that include more diverse samples and sufficient power to replicate and extend our findings.

5. Conclusions

This study provides preliminary evidence that the HABIT Program may exert differential effects on individuals with MCI and their care partners depending on partner type. Dyads with spousal care partners demonstrated modestly greater benefits and better compliance up to 12 months post‐intervention. Although effects were small and require replication with control groups, these findings suggest that care partner type may moderate responses to behavioral interventions and should be considered in the design and refinement of future non‐pharmacological trials (e.g., US Pointer, FINGER trial).

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest. Author disclosures are available in the Supporting Information.

CONSENT STATEMENT

All human subjects provided informed consent.

Supporting information

Supporting Information