Activity Engagement Among Older Adult Spousal Caregivers

Abstract

Objective

Spousal caregiving can have strong implications for health and wellbeing given the strain and burden associated with the role. Maintaining activity engagement is important for late-life health and wellbeing, and may be a possible contributing mechanism to caregiver health and wellbeing. This paper examined longitudinal changes in activity engagement and asks how spousal caregiving status and caregiver age related to longitudinal activity engagement in a sample of older adults.

Method

Data from four waves of the Health and Retirement study were used to model associations between periods of being a spousal caregiver, age, and engagement in physical, social, self-care, passive, and novel information processing activities over a 6-year period.

Results

Caregiving status was associated with declines in physical activity engagement over time. Older age was associated with fewer physical and more self-care and passive activities.

Discussion

Caregivers’ declines in participation in physical activities may be an important indicator for preservation of health and management of caregiving stress.

Many older adults experiencing limitations seek assistance in the form of caregiving from their spouse. Compared to other caregiving relationships, spousal caregiving may be associated with more stress. Spousal caregivers are older than next-generation caregivers (i.e., children) and at higher risk for age-related chronic illness (Schulz & Eden, 2016). They provide more practical and emotional care and are less likely to report roles outside of the home (Pinquart & Sörensen, 2003). Caregiving demands increase risk of depression and health problems in older adults (Burton, Zdaniuk, Schulz, Jackson, & Hirsch, 2003; Martire & Schulz, 2012).

Time restrictions to engage in valued activities may explain the declines some spousal caregivers experience in their own wellbeing (Smith, Williamson, Miller, & Schulz, 2011). Maintaining an actively engaged lifestyle (e.g., physical exercise, socializing, cognitive activities) in midlife and old age promotes well-being (Adams, Leibbrandt, & Moon, 2011), benefits physical health (Everard, Lach, Fisher, & Baum, 2000; Herzog, Franks, Markus, & Holmberg, 1998), protects against dementia and Alzheimer’s disease (Fratiglioni, Paillard-Borg, & Winblad, 2004; Hertzog, Kramer, Wilson, & Lindenberger, 2009) and delays mortality (Menec, 2003).

Caregivers often reallocate their time to meet the needs of the person requiring care. This may involve spending less time in enjoyable leisure or health-promoting activities, such as physical exercise. Increasing caregiving demands have been associated with increases in caregiver activity restrictions, which in turn predicted higher depressive symptoms (Nieboer et al., 1998).

We examined how older spousal caregivers’ activity engagement changed during times of caregiving. We estimated multilevel models using four waves of the Consumption and Activities Mail Survey (CAMS), a biennially administered supplemental survey in the Health and Retirement Study (Sonnega et al., 2014). CAMS asks respondents the amount of time they spent doing selected activities in the past week or month. We defined activity engagement as the relative percent of reported activities in five categories: physical, social, passive information processing, novel information processing, and self/home care. Previous research has identified several of these activity categories being associated with physical and cognitive health and wellbeing (Jopp & Hertzog, 2010).

Our interests were how activity engagement changed during times of caregiving while accounting for known covariates of activity engagement: caregiver age, gender, and education. We hypothesized that older adults would report less engagement in physical activities, more engagement in passive information processing and self-maintenance activities over time (Verbrugge, Gruber-Baldini, & Fozard, 1996), perhaps in response to change in health and physical limitations. We expected that periods of caregiving would be associated with less engagement in physical, social, and novel information processing activities and more engagement in passive information processing and self-care activities (Michelson & Tepperman, 2003). Care demands may impose restrictions (Nieboer et al., 1998), especially on activities done outside the home or with others aside from the spouse.

Method

Sample

Three criteria were applied to select a CAMS subsample for analysis (see Health & Retirement Study, n.d.). First, we identified couples who both participated in at least three waves of CAMS, between 2007 (when CAMS began assessing caregiving roles) and 2013. Second, all respondents who reported being a spousal caregiver during at least one wave of the study were selected for the analytic sample (N = 1083). To limit the sample to those where one member of the dyad was a caregiver, we removed cases where both members of the dyad reported caregiving for each other (N = 360). After removing these cases, the caregiver subsample consisted of 723 individuals who reported being spousal caregivers.

Measures

Caregiving

In each wave, respondents were asked “Did you spend any time last week treating or managing the medical condition of another person?” If yes, respondents indicated to whom they were providing care (check all that apply; spouse, child, parent, other relative, nonrelative). We identified spousal caregivers as those who were treating or managing the medical condition of a spouse at any point during the study.

Activity engagement

CAMS captures several domains of time use, including physical, cognitive, and social (Hurd & Rohwedder, 2007). Respondents estimated the number of hours they spent on 37 activities within the last week (20 activities; watching TV, reading books, listening to music) or the last month (17 activities; volunteering, playing games, playing an instrument). If they did not participate in an activity, respondents checked a box labeled “0 hours”.

Previous literature on activity engagement in older adulthood (Hultsch, Hertzog, Small, & Dixon, 1999) guided our activity categories: physical (e.g., walking, exercising), social (e.g., communicating by phone, letters, or e-mail with friends/relatives/neighbors, visiting others in person), passive information processing (e.g., listening to music, watching television), novel information processing (e.g., reading books, working with a computer), or self/home-maintenance (e.g., doing chores, preparing meals). We identified 20 relevant activities surveyed in CAMS (see Supplementary Table 1) and examined the proportion of activities individuals completed in each category (i.e., out of the total number of activities endorsed, what percentage of activities fell into each of the four categories). If individuals were missing data for an activity, the denominator (number of activities for that category) was adjusted.

Analytic Plan

We used multilevel modeling to examine changes in activity engagement over time. We estimated a model treating the ratio of engagement in each of the five activity categories as separate outcomes.

The focal predictors in the models were time (waves 0–3), a time-varying indicator of spousal caregiving (0 not currently caregiving, 1 currently caregiving), and baseline age (grand mean centered). We included interactions between time and caregiving status and time and baseline age to examine changes in activity engagement over the course of the study. Because activity engagement may vary by age, we thought it was necessary to account for these relationships in order to assess the unique effect of caregiving. Covariates in the models included gender (0 male, 1 female) and years of schooling (grand-mean centered). Given that women are more likely than men to act as spousal caregivers, we additionally tested a gender X spousal caregiving status interaction in each of the models. There were no significant interactions for any of the four activity outcomes. To capture change in activity over survey waves, we included the random effect of survey wave.

$$\begin{array}{c}{Y}_{ij}={\gamma }_{00}+{\gamma }_{01}BaselineAg{e}_i+{\gamma }_{02}Gende{r}_i+{\gamma }_{03}Educatio{n}_i\\ +{\gamma }_{10}SurveyWav{e}_{ij}+{\gamma }_{20}CaregiverStatu{s}_{ij}\\ +{\gamma }_{11}BaselineAg{e}_i*SurveyWav{e}_{ij}\\ +{\gamma }_{21}CaregiverStatu{s}_i*SurveyWav{e}_{ij}\\ +[{\zeta }_{0i}+{\zeta }_{1i}Tim{e}_{ij}+{\epsilon }_{ij}]\end{array}$$

Results

More women than men identified as spousal caregivers (61.69%). Caregivers’ average baseline age was 66.26 (SD = 9.44) years old. They achieved an average of 12.97 years of education (SD = 2.89). The majority of the sample spent 1 survey wave in a caregiving role (62%), 24% spent two survey waves, and 14% spent three to four survey waves caregiving for a spouse.

Activity Engagement

Estimates obtained from the multilevel models are presented in Table 1. Before describing results of the conditional models, we begin each section with the between- versus within-person variance portioned for each activity outcome.

Table 1.

Model Estimates by Activity Category

	Physical activities		Social activities		Self-care activities		Passive info processing		Novel info processing
Outcome	Estimate (SE)	p	Estimate (SE)	p	Estimate (SE)	p	Estimate (SE)	p	Estimate (SE)	p
Fixed effects
Intercept	0.16 (0.003)	<.001	0.20 (0.004)	<.001	0.24 (0.004)	<.001	0.20 (0.003)	<.001	0.20 (0.004)	<.001
Gender	−0.03 (0.003)	<.001	−0.004 (0.004)	.34	0.01(0.004)	.05	0.01 (0.004)	<.01	0.02 (0.004)	<.001
Education	0.0001 (0.001)	.80	0.001 (0.001)	.09	−0.003 (0.00)	<.001	−0.004 (0.00)	<.001	0.001 (0.00)	<.001
Baseline age	−0.0004 (0.0002)	.1	−0.000 (0.0002)	.9	0.001 (0.00)	<.01	−0.0005 (0.0002)	.02	0.0002 (0.0002)	.52
Time	−0.002 (0.001)	.08	0.0004 (0.001)	.78	0.003 0.001)	.01	0.003 (0.001)	.02	−0.005 (0.001)	<.001
Caregiver status	−0.007 (0.003)	.04	0.004 (0.004)	.24	0.005 (0.004)	.16	0.0005 (0.003)	.87	−0.004 (0.004)	.31
Time × Baseline age	−0.0003 (0.00)	.01	0.00 (0.00)	.94	−0.00 (0.00)	.66	0.0002 (0.00)	.03	0.0001 (0.00)	.32
Time × Caregiver status	0.005 (0.001)	.01	−0.003 (0.002)	.09	−0.00 (0.002)	.83	−0.001 (0.001)	.45	0.001 (0.002)	.73
Random effects
Within person	0.002 (0.00)	<.001	0.002 (0.00)	<.001	0.002 (0.00)	<.001	0.002 (0.00)	<.001	0.002 (0.00)	<.001
Initial status	0.001 (0.00)	<.001	0.001 (0.00)	<.001	0.002 (0.00)	<.001	0.001 (0.0001)	<.001	0.002 (0.00)	<.001
Rate of change	0.0001 (0.00)	<.001	-	-	0.00 (0.00)	.05	0.0001 (0.00)	<.01	-	-
Covariance	−0.00 (0.00)	.26	-	-	0.00 (0.00)	.64	0.00 (0.00)	.21	-	-

Note: Gender, 0 = male 1 = female; Education, 0–17 years; Time, 0–3; Caregiver status, 0 = not caregiving 1 = caregiving for spouse.

Physical Activities

Forty-seven percent of the variance in ratio of engagement in physical activities was between person. Changes in physical activity engagement were related to both age and caregiving status. A significant time × age interaction suggested that older adults experienced steeper declines in physical activities over time. A significant time × caregiver status interaction suggested that spousal caregivers displayed a steeper decline in physical activities over the survey period. Additionally, women tended to engage in less physical activities compared to men.

Social Activities

Thirty-seven percent of the variance in ratio of engagement in social activities was between person. We were unable to successfully estimate a model for social activities that included the random effect of time, thus time was dropped from the model. There were no significant effects of time, age, caregiving status, or model covariates in predicting social activity engagement.

Self/Home-Maintenance Activities

Forty-six percent of the variance in ratio of engagement in self/home-maintenance activities was between person. A main effect of survey wave suggested that respondents engaged in more of these types of activities over the course of the survey. A main effect of baseline age indicated that older adults engaged in more self/home-maintenance activities. Higher education was association with doing less self/home-maintenance activities.

Passive Information Processing

Forty-nine percent of the variance in ratio of passive activities was between person. A significant survey wave × age interaction indicated that older adults experienced steeper increases in passive activities over the course of the survey. Women tended to engage in more passive activities than men and higher educational achievement was associated with less engagement in these types of activities.

Novel Information Processing

Fifty percent of the variance in ratio of engagement in social activities was between person. We were unable to successfully estimate a growth model for these activities; thus, a random effect of survey wave was not included in the model. A main effect of survey wave suggested that respondents engaged in less novel information processing activities over the course of the survey period. Women and those with higher education engaged in more of these activities.

Discussion

Our analyses revealed several associations between activity engagement, age, and survey wave; however, engagement in physical activities was the only activity associated with spousal caregiver status. Periods of spousal caregiving were associated with less engagement in physical activities over time; however, caregiving involves a certain level of physical activity and previous research suggests that caregivers are physically active but tend not to report their activity as exercise (Fredman, Bertrand, Martire, Hochberg, & Harris, 2006).

Surprisingly, we did not observe relationships between caregiving and other types of activities associated with healthy aging. It is possible that caregivers viewed their duties as social activities, which is why no differences emerged in that category. Additionally, although spousal caregiving is not uncommon in older adulthood, younger caregivers, such as children, are more likely to be involved in high-need caregiving situations than spouses (Schulz & Eden, 2016). The persons receiving care in this sample may have had lower needs and required less care, or shorter-term care as evidenced by caregivers’ time spent in their role. In creating our analytic sample, we first selected respondents where both members of the dyad participated in the survey, which may have resulted in a healthier sample. Future work may assess the relationship between activity engagement and activity demands (e.g., intensity of the caregiving role).

As expected, engagement in both self-care and passive information processing were higher among older participants, perhaps in response to changes in health and physical limitations.

This study provides a longitudinal view on the relationships between caregiving and activity engagement, however, we were constrained to only four survey waves and limited information on the caregiving relationship. We were also unable to evaluate whether these activities overlapped with caregiving tasks. This may be important as, for some caregivers, wellbeing is negatively related to daily household tasks but positively related to caregiving itself (Freedman, Cornman, & Carr, 2014). We attempted to counter the challenges of open-ended responses in a mail survey, such as improbable lengths of time spent on activities, by using activity ratios as an outcome. However, this strategy limits the amount of variability in activity engagement we were able to capture.

The results point to the importance of examining caregiving as an activity that varies across one’s late adult years that may interfere with physical activity engagement.

Supplementary Material

Supplementary data is available at The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences online.

Funding

This work was supported by the National Institute of Aging at the National Institutes of Health (7R03AG048333-02 to T. L. Queen).

Author Contributions

T. L. Queen planned the study, built datasets, lead analysis, and writing. J. Butner advised on analyses. C. A. Berg and J. Smith helped to plan the study and revised the manuscript.

Conflict of Interest

None reported.