The Association Between Cognitive Impairment and Patterns of Activity Engagement Among Older Adults

Abstract

This study aims to assess the association between cognitive impairment and activity engagement patterns. Data from the 2012 Health and Retirement Study were used. A total of 3,943 participants aged 65 or older were included in analyses. Latent class analysis and multinomial logistic regression analysis were used. Four activity engagement profiles were identified: high activity (31.2%), active leisure (18.9%), passive leisure (28.2%), and low activity (21.7%). People in the high activity group engaged in all activities more than people in any other group, whereas people in the low activity group did not actively engage in most activities. Multinomial logistic regression analysis showed that cognitive impairment had an independent effect on the probability of being assigned to the low activity group compared to other groups. Cognitive impairment was associated with inactivity in a variety of activities. Future studies should examine supportive factors, which facilitate active patterns among people with cognitive impairment.

Keywords

cognition activity profile latent class analysis

It is well established that an active, engaged lifestyle is associated with health and well-being. Previous studies have shown that people who engaged in more activities had better self-rated health (Morrow-Howell, Hinterlong, Rozario, & Tang, 2003), better subjective well-being (Baker, Cahalin, Gerst, & Burr, 2005), lower depression (Hertzog, Hultsch, & Dixon, 1999), lower mortality (Paffenbarger, Hyde, Wing, & Hsieh, 1986), and better cognitive function (Bielak, Hughes, Small, & Dixon, 2007; Christensen et al., 1996). Using a person-centered perspective, a limited but growing number of studies have attempted to investigate the heterogeneous patterns and disparate nature of activities as they relate to health (e.g., Burr, Mutchler, & Caro, 2007). Nevertheless, there is limited understanding of the relationship between patterns of activity engagement and cognitive impairment. The current study focuses on the relationship between cognitive impairment and patterns of activity engagement and aims to extend knowledge in this line of inquiry.

At present, the relationship between cognitive impairment and patterns of activity engagement is not well understood. Previous studies have suggested that some physical- and mental health–related factors such as performance of activities of daily living (ADL; Burr et al., 2007), mental health (Croezen, Haveman-Nies, Alvarado, Veer, & De Groot, 2009), and self-rated health (Morrow-Howell et al., 2014) are associated with profiles of activity engagement patterns. However, few studies have investigated the relationship between cognitive impairment, which is a major health-related concern in later life, and profiles of activity engagement patterns. Specifically, little is known regarding whether and/or to what extent older people with cognitive impairment engage in various activities. Some have suggested that people with cognitive impairment or dementia who live in their own homes actually engage in variety of activities and have a strong desire to be active (Parisi, Roberts, Szanton, Hodgson, & Gitlin, 2015; Phinney, Chaudhury, & O’Connor, 2007) regardless of the functional limitations caused by their impairment. Although these studies imply that cognitive impairment does not prevent people from engaging in particular activities, the extent to which cognitive impairment affects patterns of activity engagement is still unclear.

This study thus aims to (1) identify profiles of activity engagement patterns, (2) describe the characteristics of each profile with a focus on cognitive impairment, and (3) examine the extent to which older adults’ cognitive impairment is associated with their activity engagement patterns.

Literature Review

Heterogeneous Patterns of Activity Engagement in Later Life

A robust body of theoretical and empirical literature has established that more active engagement in activities leads to better health and well-being in later life. However, most of these previous studies used one or two activities to measure activity engagement or categorized activities into sets of certain types of activities (e.g., Buchman, Yu, Boyle, Wilson, & Bennett, 2012; Chang, Wray, & Lin, 2014; Pahor et al., 2014; Strawbridge, Wallhagen, & Cohen, 2002). These approaches to capturing activity engagement is limited because some activities may be complementary and others competitive (Burr et al., 2007). Considering that most people engage in multiple activities simultaneously, using one activity or sets of certain types of activities as a measure may not offer a full picture of an individual’s activity engagement patterns.

Departing from conventional approaches to investigating activity engagement, some researchers have suggested that people can be assigned to latent clusters of activities and that cluster members demonstrate similar patterns of engagement. Burr, Mutchler, and Caro (2007) used latent class analysis (LCA) to identify clusters of engagement in productive activities. Analyzing participants’ engagement in productive activities, researchers found four patterns: helpers, home maintainers, worker/volunteers, and super helpers. Croezen, Haveman-Nies, Alvarado, Veer, and De Groot (2009) used cluster analysis with the K-means method and found that participants could be divided into five groups—less social engaged, less social-engaged caregivers, social-engaged caregivers, leisure engaged, and productive engaged—according to their patterns of social engagement. They concluded that people who were grouped as less socially engaged had poorer self-rated health, more mental health problems, more ADL limitations, and more severe loneliness. Using LCA, Morrow-Howell and colleagues (2014) found that the sample could be divided into five profile groups, depending on their activity engagement: low activity, moderate activity, high activity, working, and physically active. These five categories were related to demographic, health, and environmental factors and were associated with well-being outcomes such as self-rated health and depressive symptoms. Matz-Costa, Carr, McNamara, and James (2016) identified four patterns (working, traditional leisure, moderate activity, and high activity) of activity engagement. They indicated that more active engagement patterns were associated with better health outcome (less depressive symptoms and less frailty). Further, the result suggested that positive effects of activity engagement patterns on health outcome were largest when people participated in activities of all four domains (use of body, use of mind, social interaction, and benefit to other). Although these innovative studies have explored activity engagement as a pattern, this approach has rarely been applied to the investigation of the association between cognitive impairment and activity engagement.

Cognitive Impairment and Activity Engagement Among Older Adults

Many previous studies have shown the association between cognitive impairment and less active engagement in physical, social, and cognitive activities (e.g., Bauman et al., 2012; Blondell, Hammersley-Mather, & Veerman, 2014; Choi, Park, Cho, Chun, & Park, 2016; Fratiglioni, Paillard-Borg, & Winblad, 2004; Krueger et al., 2009).

Despite this evidence, two clear gaps exist in this line of research. First, there is a lack of focus on patterns of engagement. As stated above, using only one activity or sets of particular activities may be problematic considering competing nature of activities. Focusing on patterns of engagement may provide a fuller picture of activities of people with cognitive impairment and allow researchers to capture the interconnected nature of them. To date, few studies have examined whether cognitive impairment is associated with an individual’s activity engagement pattern. Fernández-Mayoralas et al. (2015) identified three profiles (active, moderate, and inactive) of leisure activity engagement and found that people in active profile groups had better cognitive function. However, the study sample was limited to people living in nursing homes—a small proportion of the older population. The current study explores whether the association between cognitive impairment and patterns of activity engagement can be seen in a sample of community-dwelling older adults.

Second, researchers and practitioners lack a complete understanding of heterogeneity in activity engagement among people with cognitive impairment. Although several studies have suggested that people who have dementia or cognitive impairment engage in a wide range of activities (Genoe & Dupius, 2014; Phinney et al., 2007), it is unclear whether various engagement patterns among this subpopulation can be identified. Moreover, sociodemographic and health-related characteristics associated with heterogeneous subgroups of cognitively impaired older adults are also not fully understood. Sociodemographic factors and self-rated health are the significant predictors of more active patterns of engagement among older adults without cognitive impairment (Burr et al., 2007; Croezen et al., 2009; Morrow-Howell et al., 2014), but no known study has examined whether these characteristics are associated with more active profiles among people with cognitive impairment.

Purposes of This Study

This study addresses three research questions. First, we ask if patterns of activity engagement can be identified. We expect to find similar patterns of activity subgroups as identified in the literature, such as active in all activities, active in only certain activities, and very limited in all activities. Second, we examine the extent to which cognitive impairment is associated with the probability of belonging to certain activity engagement profiles. Although this study is the first to examine the association between cognitive impairment and activity patterns among community-dwelling older people, we expect that cognitive impairment is associated with lower probability of belonging to a more active profile. Finally, we investigate the possibility of pattern variations among people with cognitive impairment by examining the individual characteristics associated with more active engagement patterns among members of this group. We hypothesize that those cognitively impaired individuals with active engagement patterns have better sociodemographic and health-related characteristics.

Research Design

Sample

Data were drawn from the 2012 wave of the Health and Retirement Study (HRS). Launched in the United States in 1992, HRS is a national longitudinal study that biennially surveys more than 22,000 older adults aged 50 and older and their spouses. The HRS sample is considered statistically representative of households in the United States. Core HRS interviews were conducted in participants’ homes or via telephone. Since 2006, HRS has collected psychosocial and lifestyle data using self-administered questionnaires. The data were obtained from a randomly selected sample comprised of 50% of participants who completed core HRS interviews in 2006 and 2010. The remaining 50% of participants were administered questionnaires in 2008 and 2012. The present study used data from the subsample of respondents completing the psychosocial questionnaires either by mail or telephone in 2012.

We selected respondents based on two criteria: First, to minimize the possibility of including young older adults whose cognitive impairments are not age related, we limited respondents to those aged 65 or older (Comijs, Dik, Aartsen, Deeg, & Jonker, 2005). Second, we excluded respondents who were institutionalized or unable to independently participate in interviews or complete questionnaires. As a result, our final sample included data from 4,073 HRS participants.

For LCA, we used this original sample of 4,073 HRS participants to obtain clusters of activity engagement. For further analyses, however, we built a model to examine the relationship between cognitive impairment and patterns of activity engagement. For these analyses, we used a reduced sample. Of respondents in our sample, 130 (3.2%) had at least one missing value for all dependent and independent variables used to build the model. Of this 130, none had three or more missing values, 104 (80.0%) respondents had two, and 26 (20.0%) respondents had only one. The variable that had most missing cases was cognitive impairment (N = 107). After omitting data from participants whose responses had missing variables, we obtained a reduced sample of 3,943 (96.8% of original sample) participants.

Measures

Activity measurement

Frequency of engagement was measured with a list of 20 activities in the HRS psychosocial questionnaires (see Figure 1 for all activities). Respondents were asked to rate how often they do each activity on a scale of 1 (never) to 7 (daily). Although the list did not specify all everyday activities, it did include items from seven “general domains” of activities which were identified in a previous systematic review (Adams et al., 2011). These seven domains were social, leisure, productive, physical, intellectual/cultural, solitary, and spiritual/serving others. According to Adams and colleagues (2014), social activities can be either formal or informal. The list of items we used did not include any informal social activities (e.g., interacting with others or visiting friends), because these can be classified as social network or social support (Morrow-Howell & Gehlert, 2012). Informal activities may exert a different influence on well-being than other activities. Therefore, we concluded that informal activities were not suitable for measuring activity engagement in this study. Because never engaging in the activity shows clear absence of engagement, response categories for the 20-item list were collapsed into two: 0 = never and 1 = at least sometimes.

Cognitive function

The HRS assessed respondents’ cognitive function using a standardized measure of the Telephone Interview for Cognitive Status (TICS; Brandt, Spencer, & Folstein, 1988). The TICS can be administered in an interview conducted either over the telephone or in person and has been validated in the original study as a cognitive screening instrument. Scores for the TICS range from 0 to 35, and higher scores indicate better cognitive function. Psychometric properties of the TICS have also been validated in previous studies (e.g., Fong et al., 2009; Knopman et al., 2010). According to earlier research, a score of 10 or less indicates the presence of cognitive impairment (Cigolle, Ofstedal, Tian, & Blaum, 2009).

Health-related factors

Some health-related factors are associated with cognitive function. In this study, we included functional status, self-rated health, and mental health in the analysis. Functional status was measured by the presence of ADL difficulties such as bathing, eating, dressing, walking across a room, and entering or leaving bed (no ADL limitations = 0, 1 or more limitations = 1). Self-rated health was measured by respondents’ rating of their health on a 5-point scale (1 = poor condition, 2 = fair, 3 = good, 4 = very good, 5 = excellent). Finally, mental health was measured by Center for Epidemiologic Studies Depression (Radloff, 1977), an 8-item scale of depressive symptoms. Respondents are asked to answer eight questions about their depressive symptoms with yes or no. The score ranges from 0 to 8 and is calculated by summing responses to questions (1 = yes, 0 = no). A higher score indicates the respondent has more depressive symptoms.

Covariates

We included sociodemographic covariates: age cohort groups (aged 65–74 = 0, aged 75–84 = 1, 85 years and above = 2), current marital status (not married = 0, married = 1), sex (male = 0, female = 1), education (less than high school = 0, high school = 1, more than high school = 2), race/ethnicity (non-White = 0, White = 1), quartile of total nonhousing wealth (original data were continuous and given in U.S. dollars), quartile of total income (original data were continuous and given in U.S. dollars), and respondent’s residential region (living urban area = 0, suburban area = 1, exurban area = 2).

Statistical Analyses

Analysis of patterns of activity engagement

We used LCA to investigate patterns of activity engagement and identify profiles among respondents in the sample. LCA assesses the relationship between manifest data and unobserved variables (latent classes) and allows researchers to identify these latent classes from multivariate categorical data. Classes identified by LCA are categorical; therefore, cases within the data can be assigned into exhaustive and exclusive subsets (Eshghi & Haughton, 2011). In this study, we sought to identify latent classes in which members have similar patterns of activity engagement.

We used LCA because it has several advantages over other, traditional clustering methods. First, LCA is not restricted to continuous variables. Second, it provides several model fit statistics, and the optimal number of classes can be empirically determined. Finally, LCA is a robust method against violations of the assumptions such as homogeneity of variance, linearity, and local independence (Vermunt & Magidson, 2002).

We used data from participants’ responses to 20 activities for LCA. Based on these observed data, LCA empirically determined exclusive latent classes into which older adults with similar activity engagement patterns can be assigned. Moreover, LCA also provided the probability of each class engaging in each activity (Muthén & Muthén, 2000). The entire sample was categorized into classes that shared similar engagement patterns.

We used three model fit statistics to determine the number of classes. A significant result on the Lo–Mendell–Rubin (LMR) test indicates a significant improvement in model fit between k-class and (k − 1)-class models (Lo, Mendell, & Rubin, 2001). We also used the Bayesian information criterion (BIC). Lower BIC indicates better model fit. Finally, we referred to entropy, a measure of uncertainty in classification. A higher entropy value indicates high certainty. Values range from 0 to 1. We conducted LCA using Mplus Version 7.4.

Association between cognitive impairment and activity profiles

To address the second research question, we conducted multinomial logistic regression analysis. This analysis allowed us to assess how cognitive impairment was associated with the probability of membership in latent classes of activity engagement after controlling for sociodemographic and health factors. The model included dependent variables of the probability of membership in latent classes and all sociodemographic and health-related covariates. These analyses were conducted using STATA Version 14.

Heterogeneous patterns of engagement among people with cognitive impairment

The third research question was addressed in two ways. First, we divided the sample into two groups: people with, and people without, cognitive impairment. For each group, we calculated proportions assigned to each activity engagement pattern. Second, we divided the group of those who had cognitive impairment into two subgroups: those with less active patterns and those with more active patterns. We tested difference between these two groups using analysis of variance and χ² test for all health-related factors and covariates included in the multinomial logistic regression model.

Results

Profiles of Activity Engagement

As numbers of class increased, model-fit statistics of BIC decreased but entropy did not reach below .7 until the five-class model (Entropy = .73 for four-class model and Entropy = .70 for five-class model). These statistics indicated that having more classes improved model fit. However, results from the LMR test comparing the four-class and five-class models were not significant (LMR = 450.6, p = .13), indicating that the five-class model was not better than the four-class model. Therefore, we selected the four-class model as the final model for further analyses. Then, we labeled four classes based on the visual interpretation of Figure 1, which show the response rates of each cluster for each activity engagement.

Four activity engagement profiles were identified as a result of LCA: high activity (31.2%), active leisure (18.9%), passive leisure (28.2%), and low activity (21.7%). As shown in Figure 1, people in the high activity group engaged in all activities more than people in any other group. Those in the active leisure group engaged in physical leisure activities such as going to sports clubs, playing sports, or exercising as much as people in the high activity group. Those in the passive leisure group participated in activities such as playing crossword games; playing cards or chess; baking or cooking; and making clothes, knitting, embroidering, and so on. Those in the low activity group engaged in all activities less than any other groups. Nevertheless, they engaged in activities such as praying privately, reading, and watching TV.

Description of Activity Profiles

Table 1 shows the result of descriptive analyses. The low activity group had highest proportion of oldest participants (18.8%). Race/ethnicity was not equally distributed across activity profiles, as the low activity group had the lowest proportion of White people (76.5%). A significant association between education and activity profiles was identified: The low activity group was characterized having the lowest level of education (42.0%), the lowest proportion of unmarried (51.7%), the highest proportion of people in the lowest quartile of nonhousing assets (42.4%) and household income (41.1%), and the highest proportion of people who had more than two ADL difficulties (19.2%) and cognitively impairment (9.2%). In addition, the low activity group had the highest average score of depressive symptoms (M = 1.90) and lowest average score of self-rated health (M = 1.65).

Table 1.

Characteristics of Sample and Activity Profiles.

Variables	All (N = 3,943)					Cognitively Impaired (N = 119)
	High	Active Leisure	Passive Leisure	Low		Low	More Active
	(31.2%)	(18.9%)	(28.2%)	(21.7%)	Statistics	(67.2%)	(32.7%)	Statistics
Sociodemographic factors
Age (%)
65–74 Years old	57.2	47.9	50.5	37.0	χ²(6) = 115.5**	25.0	23.1	χ²(2) = 0.6
75–84 Years old	35.4	41.7	39.4	44.2		36.3	43.6
85 or older	7.4	10.4	10.1	18.8		38.8	33.3
Sex (%)
Female	64.1	29.6	75.6	56.1	χ²(3) = 407.3**	51.3	69.2	χ²(2) = 3.5
Race/ethnicity (%)
White	87.6	85.2	86.8	76.5	χ²(3) = 56.6**	62.5	74.4	χ²(2) = 1.7
Education (%)
Less than high school	9.4	16.0	21.9	42.0	χ²(6) = 504.9**	65.0	46.2	χ²(2) = 4.0
High school	28.8	33.5	42.4	35.9		18.8	25.6
Some college or more	61.8	50.5	35.8	22.1		16.3	28.2
Marital status (%)
Married	65.3	70.4	56.0	48.3	χ²(3) = 104.8**	46.3	33.3	χ²(2) = 1.8
Quartile nonhouse wealth (%)
Lowest quartile	14.7	17.0	26.2	42.4	χ²(9) = 323.0**	—	—	χ²(2) = 6.8**
Second lowest quartile	22.2	24.6	27.8	25.9		81.3	59.0
Second highest quartile	28.6	27.0	25.7	18.2		—	—
Highest quartile	34.5	31.5	20.3	13.4		18.8	41.0
Quartile household income (%)
Lowest quartile	16.4	15.5	26.7	41.1	χ²(9) = 303.8**	—	—	χ²(2) = 0.5
Second lowest quartile	23.0	23.6	26.3	27.9		82.5	76.9
Second highest quartile	26.4	28.5	25.8	19.6		—	—
Highest quartile	34.1	32.4	21.1	11.5		17.5	23.1
Residential region (%)
Urban	44.5	47.6	45.1	46.2	χ²(6) = 2.3	52.5	46.2	χ²(2) = 0.9
Suburban	22.4	21.3	21.9	22.2		17.5	15.4
Ex-urban	33.1	31.1	33.0	31.5		30.0	38.5
Health factors
ADL difficulty (%)
0	90.0	87.2	83.6	64.9	χ²(6) = 287.7**	56.3	71.8	χ²(2) = 5.7
1	6.6	9.0	10.5	15.9		13.8	17.9
≥2	3.4	3.8	5.8	19.2		30.0	10.3
Self-rated health (M)	2.45	2.25	2.07	1.65	F(3,939) = 114.4**	1.64	2.21	t(117) = 6.3*
Depressive symptoms (CES-D, M)	0.89	1.06	1.35	1.93	F(3,939) = 63.1**	2.61	1.95	t(117) = 2.4
Cognitive function
Cognitive impairment (%)
No impairment (TICS ≥ 11)	98.9	98.6	98.7	90.8	χ²(3) = 145.1**	—	—
Impaired (TICS ≤ 10)	1.1	1.4	1.3	9.2		—	—

Note. ADL = activities of daily living; CES-D = Center for Epidemiologic Study Depression Scale; TICS = Telephone Interview for Cognitive Status.

*p < .05. **p < .01.

Table 2 presents the results of multinomial logistic regression analysis. The low activity group served as the reference group for this analysis. The probability of belonging to one activity profile group compared to the reference group was calculated for independent variable and covariates. After controlling for covariates, in comparison to the reference group, we found that high activity, active leisure, and passive leisure groups are associated with having no cognitive impairment.

Table 2.

Result of Multinomial Logistic Regression Analysis.

N = 3,943	High Versus Low		Active Versus Low		Passive Versus Low
Variable	RRR	(SE)	RRR	(SE)	RRR	(SE)
Sociodemographic factors
Age (65–74 years old)
75–84 Years old	0.57	(0.06)**	0.76	(.09)*	0.71	(.08)**
85 Years or older	0.29	(0.05)**	0.47	(.09)**	0.46	(.07)**
Sex (male)
Female	1.93	(0.22)**	0.42	(.05)**	3.04	(.34)**
Race/ethnicity (non-White)
White	1.23	(0.18)	1.01	(.15)	1.63	(.22)**
Education (less than high school)
High school	2.19	(0.32)**	1.67	(.25)**	1.61	(.19)**
Some college or more	6.89	(1.05)**	3.26	(.51)**	2.30	(.32)**
Marital status (not married)
Married	1.03	(0.13)	0.95	(.13)	1.08	(.13)
Nonhousing asset (lowest quartile)
Second lowest quartile	1.63	(0.24)**	1.67	(.27)**	1.37	(.18)*
Second highest quartile	2.30	(0.36)**	1.93	(.33)**	1.49	(.22)**
Highest quartile	2.32	(0.41)**	2.08	(.39)**	1.22	(.21)
Household income (lowest quartile)
Second lowest quartile	1.14	(0.17)	1.26	(.21)	1.04	(.14)
Second highest quartile	1.29	(0.21)	1.56	(.28)*	1.28	(.20)
Highest quartile	1.71	(0.32)**	1.95	(.40)**	1.56	(.29)*
Residential region (urban)
Suburban	1.17	(0.15)	0.98	(.14)	0.98	(.12)
Ex-urban	1.38	(0.17)**	1.12	(.15)	1.10	(.13)
Health factors
ADL difficulty (no difficulty)
1	0.58	(0.10)**	0.66	(.12)*	0.70	(.11)*
≥2	0.31	(0.07)**	0.28	(.07)**	0.35	(.06)**
Self-rated health	1.48	(0.09)**	1.32	(.08)**	1.16	(.06)**
Depressive symptoms (CES-D)	0.94	(0.03)*	1.00	(.03)	0.98	(.03)
Cognitive function
Cognitive impairment (no)
Impaired (TICS ≤ 10)	0.34	(0.11)**	0.29	(.11)**	0.26	(.08)**
(intercept)	0.10	(0.02)**	0.28	(.07)**	0.22	(.05)**
Pseudo R ²			0.136
χ²			1,463.41**

Note. ADL = activities of daily living; CES-D = Center for Epidemiologic Study Depression Scale; RRR = relative risk ratio; TICS = Telephone Interview for Cognitive Status.

*p < .05. **p < .01.

Because cognitive impairment was significantly associated with activity profiles, we examined whether variation exists in activity profile distribution between the unimpaired group and impaired group. Figure 2 shows the distribution of activity profiles in cognitive impairment and no impairment groups. Among people without cognitive impairment, 31.6% were assigned to high activity, 19.0% to active leisure, 28.7% to passive leisure, and 20.7% to low activity groups. Although most respondents with cognitive impairment (N = 80, 67.2%) were assigned to the low activity group, fully one third were assigned to groups other than low activity. We used LCA results to assign people with cognitive impairment (N = 119) into two groups: low activity group (N = 80) and higher activity group (N = 39).

Table 1 shows the comparison of characteristics between two groups. Due to the small number of cases, we combined categories of two independent variables. For income, we combined lowest and second lowest and highest and second highest quartiles. Similarly, we combined quartiles for assets. The relationship between income and assets and activity profile was mixed. The group consisting of all those but the low-activity group (relatively active) had significantly fewer people in the lowest and second lowest quartiles of assets, 58.9% and 81.3%, χ²(2) = 6.8, p < .01, than the low activity group. Those in the relatively active group had significantly better self-rated health, 2.21 and 1.64, F(117) = 6.3, p < .05, than people in the low activity group.

Discussion

This study examined the association between cognitive function and patterns of activity engagement. To the best of our knowledge, this is the first empirical study to examine the association between cognitive impairment and profiles of activity engagement identified by LCA among older people living in their own homes.

As expected, we found that the sample could be allocated into profiles and that some people engaged in all activities, some engaged in only some activities, and some engaged in very few activities. These findings aligned with previous studies in which participants were assigned into groups based both on level of their engagement and the type of activities in which they engaged (Burr et al., 2007; Croezen et al., 2009; Morrow-Howell et al., 2014). Although a large proportion of the sample in the present study was assigned to the high activity group, 21.7% was assigned into low activity. Further, descriptive analysis of each group indicated that the low activity group was the most vulnerable when considering sociodemographic and health-related factors. These findings suggest that people with low activity engagement patterns may benefit most from interventions that eliminate barriers to and facilitate participation.

We examined the extent to which cognitive impairment was related to patterns of activity engagement and hypothesized that individuals with cognitive impairment would likely belong to less active groups. As hypothesized, we found that the presence of cognitive impairment significantly distinguished high activity, active leisure, and passive leisure groups from the low activity group, even after controlling for health-related factors and other covariates. This finding was consistent with extant literature, which has suggested that cognitive impairment is associated with less active engagement (e.g., Bauman et al., 2012; Blondell et al., 2014; Krueger et al., 2009). However, this finding has also suggested that cognitive impairment is associated not only with engagement in individual activities or in activity sets but also with patterns of activity engagement. While this finding also supported those of a previous study conducted among people living in nursing homes (Fernández-Mayoralas et al., 2015), it also demonstrated that the finding is generalizable to people living in their own homes.

Informal activities such as visiting friends and talking with family members were not included as part of activity engagement. As such, it is possible that one might find different patterns of engagement if informal social activities were considered. For example, Morrow-Howell et al. (2014) showed that people who were assigned to the “physically active” group were characterized by high levels of engagement in interpersonal exchange. However, as noted above, informal social activities may be classified as social network or social support and therefore be differently associated with well-being than other formal activities are. Because our focus in this study was not on social environmental factors, we intentionally excluded these informal social activities from our analysis. Future studies may need to clarify the association between cognitive impairment and patterns of engagement in relation to engagement in informal social activities.

In addition, this finding implies that cognitive impairment may limit older adults’ opportunity for activity engagement more than their functional ability to participate in them. For example, people in the low activity group, to which most with cognitive impairment were assigned, did not actively engage in even those activities that do not demand high cognitive function (e.g., playing with grandchildren or walking). In a qualitative study, Inness and colleagues (2016) found that people with dementia encounter many barriers that prevent them from participating in leisure activities. They identified three types of barriers: intrapersonal (fear of getting lost), interpersonal (carers perception of whether the person with dementia is able to engage), and structural (accessibility of destinations and attractions/venues and cost of traveling insurance). They also indicated that these barriers were profound and made both people with dementia and their caregivers believe that participation in any activity was extremely difficult. It should be noted that findings from previous studies may not be directly comparable to the results of our studies, as extent and the types of challenge faced by those who have cognitive impairment and those who are diagnosed with dementia when attempting to get involved in activities would be different. Regarding to the barriers to activity engagement for people with cognitive impairment, Parisi, Roberts, Szanton, Hodgson, and Gitlin (2015) showed that limited accessibility (e.g., transportation difficulty) affected activity participation (e.g., going out for enjoyment) for a greater percentage of those with cognitive impairment than those without impairment. These studies suggest that supportive environments—including accessible transportation—may be one promising solution to address barriers to participation; future research is needed to investigate how people with cognitive impairment may cultivate and sustain more active patterns of engagement.

Finally, we examined heterogeneity of engagement patterns among people with cognitive impairment. As expected, most people with cognitive impairment were assigned to the low activity group. However, one third had the relatively active profile (i.e., being assigned in high activity, active leisure, or passive leisure). Comparison of descriptive characteristics between the low activity group and the relatively active group indicated that those in the relatively active group had better sociodemographic and health-related factors. This heterogeneity implies that some underlying mechanisms exist in which demographic and health-related factors help people with cognitive impairment sustain more active engagement patterns. It may be possible that those who reported more active engagement patterns had supportive environmental factors that contributed to such active patterns. This argument is consistent with the result from a previous study which showed that people with dementia prolong their engagement in leisure activities by seeking adequate support and altering their activities to align with their reduced capabilities (Genoe & Dupuis, 2014). Moreover, environmental research on aging suggests that even those who have limited personal resources and capability can adapt optimally if social and physical environmental characteristics support and compensate for their limitations (Lawton & Nahemow, 1973). For example, Trahan, Kuo, Carlson, and Gitlin (2014), in their systematic review, found that modifications to space demands (background music, time of day providing activities, overhead lighting and sound level, and the number of people within an activity space) and to social demands (use of prompting and incorporation of staff or peers to lead activities) consistently yielded positive effects on engagement in social, cognitive, or physical activities among people with dementia living in nursing home or assisted living. In another study, Gitlin and colleagues (2008) suggested the effectiveness of decreasing environmental demands on activity engagement and well-being of people with dementia who were living in their own homes. In this study, they developed an intervention called the Tailored Activity Program (TAP), in which occupational therapists assessed functional capabilities and social and physical environment of people with dementia and introduced tailored activities. The effectiveness was tested through randomized-controlled trial, and the results suggested that TAP was effective in increasing enjoyment in activity engagement and reducing behavioral symptoms. Although these studies tend to have samples of people with dementia and direct comparison may not be possible, they do suggest that environmental support is an underlying mechanism for facilitating active engagement of people with cognitive impairment, as well as those with dementia.

Further investigation should examine two topics. First, researchers should identify environmental factors influencing patterns of activity engagement. As stated above, environmental support may be one key to addressing barriers to participation among people with cognitive impairment. These environmental supports may include informal social activities such as having contact with family and friends. Seeman et al. (2011) indicate that higher frequency of social contact was associated with higher executive functioning among older adults. Although this study excluded informal activities from the analysis, future studies should clarify the influence of supportive environmental factors (including informal social engagement such as contact with family and friends) on overall patterns of activity engagement and cognitive impairment. Second, they should address longitudinal changes in activity patterns to examine the causal relationship between these environmental factors and patterns of activity engagement.

This study has two limitations. First, as this study was cross sectional, the causal relationship between personal and environmental factors and profiles of activities cannot be examined. As a result, we are not capable of answering important questions regarding the directionality of the relationship between cognitive impairment and patterns of activity engagement: That is, does presence of cognitive impairment predict less-active patterns, or do less-active profiles contribute to cognitive impairment? Previous studies have suggested that directionality between health and/or well-being and activity engagement is unclear. For example, Thoits and Hewitt (2001) suggested that volunteering has a reciprocal effect: Engaging in volunteer work may enhance well-being, but those who do volunteer have better well-being than those who do not. Furthermore, Bielak (2010) indicated that the directionality between cognitive impairment and activity engagement has not been confirmed. Although these suggestions may be applicable to the relationship between cognitive impairment and patterns of activity engagement, further study using a longitudinal design is needed to thoroughly investigate the causal relationship between these two constructs.

Second, the sample was limited to a relatively healthy population, because we did not include proxy respondents in the sample. However, the reduced sample nevertheless contained a large proportion of the sample that is nationally representative of older adults in the United States. Therefore, the possibility of generalizing outweighs the limitation of omitting proxy respondents.

Conclusion

This study is one of the first research efforts to empirically demonstrate the association between cognitive impairment and profiles of activity engagement. It demonstrated the possibility and efficacy of grouping participants into similar patterns of activity engagement. People in the least active group (low activity) did not participate in most activities and had many interrelated factors that put them at risk. This finding suggested that cognitive impairment may be associated with reduced engagement in all activities. However, a comparison between those with cognitive impairment in the low activity group and those with cognitive impairment in the relatively active group suggested that additional supportive factors contributed to cultivating and sustaining active engagement despite cognitive impairment. Future research on the supportive environmental factors for activity engagement pattern is important, as is extending research on activity engagement among people with cognitive impairment. Previous theoretical and empirical studies have indicated that people with cognitive impairment may be actively engaged in some activities with supportive environmental factors (Gitlin et al., 2008; Lawton & Nahemow, 1973). Given such indications, heterogeneous patterns of activity engagement among people with cognitive impairment may be accounted for environmental factors.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

Adams

K. B.

Leibbrandt

Moon

(2011). A critical review of the literature on social and leisure activity and wellbeing in later life. Ageing and Society, 31(4), 683–712.

Baker

L. A.

Cahalin

L.P.

Gerst

Burr

J. A.

(2005). Productive activities and subjective well-being among older adults: The influence of number of activities and time commitment. Social Indicators Research, 73, 431–458.

Bauman

A. E.

Reis

R. S.

Sallis

J. F.

Wells

J. C.

Loos

R. J. F.

Martin

B. W.

(2012). Correlates of physical activity: Why are some people physically active and others not? The Lancet, 380, 258–271.

Bielak

A. A. M.

(2010). How can we not ‘lose it’ if we still don’t understand how to ‘use it’?: Unanswered questions about the influence of activity participation on cognitive performance in older age—A mini-review. Gerontology, 56, 507–519.

Bielak

A. A. M.

Hughes

T. F.

Small

B. J.

Dixon

R. A.

(2007). It’s never too late to engage in lifestyle activities: Significant concurrent but not change relationships between lifestyle activities and cognitive speed. The Journals of Gerontology. Series B, Psychological Sciences and Social Sciences, 62, 331–339.

Blondell

S. J.

Hammersley-Mather

Veerman

J. L.

(2014). Does physical activity prevent cognitive decline and dementia? A systematic review and meta-analysis of longitudinal studies. BMC Public Health, 14:510, 1–12.

Brandt

Spencer

Folstein

(1988). The telephone interview for cognitive status. Neuropsychiatry, Neuropsychology, & Behavioral Neurology, 1, 111–118.

Buchman

A. S.

Boyle

P. A.

Wilson

R. S.

Bennett

D. A.

(2012). Total daily physical activity and the risk of AD and cognitive decline in older adults. Neurology, 79, 1071–1071.

Burr

J. A.

Mutchler

J. E.

Caro

F. G.

(2007). Productive activity clusters among middle-aged and older adults: Intersecting forms and time commitments. The Journals of Gerontology. Series B, Psychological Sciences and Social Sciences, 62, S267–S275.

10.

Chang

P.-J.

Wray

Lin

(2014). Social relationships, leisure activity, and health in older adults. Health Psychology, 33, 516–523.

11.

Choi

Park

Cho

K. H.

Chun

S.-Y.

Park

E.-C.

(2016). A change in social activity affect cognitive function in middle-aged and older Koreans: Analysis of a Korean longitudinal study on aging (2006–2012). International Journal of Geriatric Psychiatry, 31, 912–919.

12.

Christensen

Korten

Jorm

A. F.

Henderson

A. S.

Scott

Mackinnon

A. J.

(1996). Activity levels and cognitive functioning in an elderly community sample. Age and Ageing, 25, 72–80.

13.

Cigolle

C. T.

Ofstedal

M. B.

Tian

Blaum

C. S.

(2009). Comparing models of frailty: The health and retirement study. Journal of the American Geriatrics Society, 57, 830–839.

14.

Comijs

H. C.

Dik

M. G.

Aartsen

M. J.

Deeg

D. J. H.

Jonker

(2005). The impact of change in cognitive functioning and cognitive decline on disability, well-being, and the use of healthcare services in older persons: Results of the Longitudinal Aging Study Amsterdam. Dementia and Geriatric Cognitive Disorders, 19, 316–323.

15.

Croezen

Haveman-Nies

Alvarado

V. J.

Veer

P. V.

De Groot

C. P. G. M.

(2009). Characterization of different groups of elderly according to social engagement activity patterns. Journal of Nutrition, Health and Aging, 13, 776–781.

16.

Eshghi

Haughton

(2011). Identifying groups: A comparison of methodologies. Journal of Data Science, 9, 271–291.

17.

Fernández-Mayoralas

Rojo-Pérez

Martínez-Martín

Prieto-Flores

M.-E.

Rodríguez-Blázquez

Martín-García

… Forjaz

M.-J

. (2015). Active ageing and quality of life: Factors associated with participation in leisure activities among institutionalized older adults, with and without dementia. Aging & Mental Health, 19, 1–11.

18.

Fong

T. G.

Fearing

M. A.

Jones

R. N.

Shia

Marcantonioc

E. R.

Rudolpha

J. L.

… Inouye

S. K.

(2009). The telephone interview for cognitive status: Creating a crosswalk with the mini-mental state exam. Alzheimer’s & Dementia, 5, 492–497.

19.

Fratiglioni

Paillard-Borg

& Winblad

(2004). An Active and Socially Integrated Lifestyle in Late Life Might Protect against Dementia. The Lancet Neurology, 3(6), 343–353.

20.

Genoe

M. R.

Dupuis

S. L.

(2014). The role of leisure within the dementia context. Dementia, 13(1), 33–58.

21.

Gitlin

L. N.

Winter

Burke

Chernett

Dennis

M. P.

Hauck

W. W.

(2008). Tailored activities to manage neuropsychiatric behaviors in persons with dementia and reduce caregiver burden: A randomized pilot study. The American Journal of Geriatric Psychiatry, 16, 229–239.

22.

Hertzog

Hultsch

D. F.

Dixon

R. A.

(1999). On the problem of detecting effects of lifestyle on cognitive change in adulthood: Reply to Pushkar et al. (1999). Psychology and Aging, 14, 528–534.

23.

Innes

Page

S. J.

Cutler

(2016). Barriers to leisure participation for people with dementia and their carers: An exploratory analysis of carer and people with dementia’s experiences. Dementia, 15, 1643–1665.

24.

Knopman

D. S.

Roberts

R. O.

Geda

Y. E.

Pankratz

V. S.

Christianson

T. J. H.

Petersen

R. C.

Rocca

W. A.

(2010). Validation of the telephone interview for cognitive status: Modified in subjects with normal cognition, mild cognitive impairment, or dementia. Neuroepidemiology, 34, 34–42.

25.

Krueger

K. R.

Wilson

R. S.

Kamenetsky

J. M.

Barnes

L. L.

Bienias

J. L.

Bennett

D. A.

(2009). Social engagement and cognitive function in old age. Experimental Aging Research, 35, 45–60.

26.

Lawton

M. P.

Nahemow

(1973). Ecology and the aging process. In Eisdorfer

Lawton

M. P.

(Ed.), The psychology of adult development and aging(pp. 619–674). Washington, DC: American Psychological Association.

27.

Mendell

N. R.

Rubin

D. B.

(2001). A note on ‘testing the number of components in a normal mixture.’ Biometrika, 90, 991–994.

28.

Matz-Costa

Carr

D. C.

McNamara

T. K.

James

J. B.

(2016). Physical, cognitive, social, and emotional mediators of activity involvement and health in later life. Research on Aging, 38, 791–815.

29.

Morrow-Howell

Gehlert

(2012). Social engagement and a health aging society. In Prohaska

Anderson

Binstock

(Eds.), Public health for an aging society (pp. 205–227). Baltimore, MD: Johns Hopkins University Press.

30.

Morrow-Howell

Hinterlong

Rozario

P. A.

Tang

(2003). Effects of volunteering on the well-being of older adults. The Journals of Gerontology. Series B, Psychological Sciences and Social Sciences, 58, S137–S145.

31.

Morrow-Howell

Putnam

Lee

Y. S.

Greenfield

J. C.

Inoue

Chen

(2014). An investigation of activity profiles of older adults. The Journal of Gerontology, Series B: Psychological Sciences and Social Sciences, 69, 809–821.

32.

Muthén

L. K.

(2000). Integrating person-centered and variable-centered analyses: Growth mixture modeling with latent trajectory classes. Alcoholism, Clinical and Experimental Research, 24, 882–891.

33.

Paffenbarger

R. S.

Hyde

Wing

A. L.

Hsieh

(1986). Physical activity, all-cause mortality, and longevity of college alumni. New England Journal of Medicine, 314, 605–613.

34.

Pahor

Guralnik

J. M.

Ambrosius

W. T.

Blair

Bonds

D. E.

Church

T. S.

…LIFE study investigators. (2014). Effect of structured physical activity on prevention of major mobility disability in older adults: The LIFE Study randomized clinical trial. Journal of the American Medical Association, 311, 2387–2396.

35.

Parisi

J. M.

Roberts

Szanton

S. L.

Hodgson

N. A.

Gitlin

L.N.

(2015). Valued activities among individuals with and without cognitive impairments: Findings from the National Health and Aging Trends Study. The Gerontologist, 0, 1–11.

36.

Phinney

Chaudhury

O’Connor

D. L.

(2007). Doing as much as I can do: The meaning of activity for people with dementia. Aging & Mental Health, 11, 384–393.

37.

Radloff

L. S.

(1977). The CES-D scale a self-report depression scale for research in the general population. Applied Psychological Measurement, 1, 385–401.

38.

Seeman

T. E.

Miller-Martinez

D. M.

Stein Merkin

Lachman

M. E.

Tun

P. A.

Karlamangla

A. S.

(2011). Histories of social engagement and adult cognition: Midlife in the U.S. study. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 66B, i141–i152.

39.

Strawbridge

W. J.

Wallhagen

M. I.

Cohen

R. D.

(2002). Successful aging and well-being: Self-rated compared with Rowe and Kahn. The Gerontologist, 42, 725–726.

40.

Trahan

M. A.

Kuo

Carlson

M. C.

Gitlin

L. N.

(2014). A systematic review of strategies to foster activity engagement in persons with dementia. Health Education & Behavior, 41, 70S–83S.

41.

Thoits

Hewitt

(2001). Volunteer work and well-being. Journal of Health and Social Behavior, 42, 115–131.

42.

Vermunt

Magidson

(2002). Latent class cluster analysis. In Hagenaars

McCutcheon

(Eds.), Applied latent class analysis (pp. 89–106). Cambridge, MA: Cambridge University Press.