Environment profiles, social participation patterns, depressive symptoms and quality of life of disabled older adults: a longitudinal investigation

Introduction

Disability in later life can be defined in several ways. Having difficulty in executing activities that are necessary to live independently, including activities of daily living (ADL) (e.g. bathing, dressing, or toileting) and instrumental activities of daily living (IADL) (e.g. preparing meal, housekeeping, or shopping), is the most widely used way of measuring disability in aging research (Gobbens, Citation2018). Recently obtained Senior Citizen Condition Survey data from Taiwan suggest that 13.03% of adults aged 65 years or over have difficulty in performing at least one ADL; 28.08% of them reported at least one problem with IADL, and the prevalence of difficulties with ADL or IADL increases with age (Ministry of Health & Welfare, Citation2018). Literature on older adults has consistently established that increased impairment or disability is associated with more disruptions in SP (e.g. Rosso et al., Citation2013). Social activities involve more complex cognitive performance, social skills and physical demands than self-care activities, so they are more difficult for disabled older adults to perform. Disabled older adults value social activities more than other activities, such as ADL or IADL, but claim that more of their needs in involving outdoor social activities are unmet (Turcotte et al., Citation2015). Thus, this study concerns the social participation (SP) of community-dwelling older adults with disabilities. While disability is inevitable for some and mostly irreversible, optimizing SP has the potential to mitigate the impact of disability on subjective well-being (SWB) in later life (Turcotte et al., Citation2015). SWB encompasses various aspects, including cognitive evaluation or satisfaction of one’s life, positive emotions and negative emotions (Skevington & Böhnke, Citation2018). World Health Organization (WHO) defines quality of life (QOL) as one’s perception of life under the influence of external cultural and internal value systems, which resembles the cognition component of SWB (Skevington & Böhnke, Citation2018; WHOQOL Group, Citation1995). Thus, QOL was used to represent the cognitive evaluation aspect, whereas depressive symptoms (DS) were applied to denote the negative emotions, also as opposite to the negative emotions part of SWB.

The healthy ageing initiative from WHO defines heathy aging as a process of developing and maintaining the functional ability to enhance wellbeing in older age. Functional ability consists of the environment a person lives, the person’s intrinsic capacity, and their interaction (WHO, Citation2020a). Environments comprise all factors that can be related to older people’ lives at home, community and society, including products, equipment and technology that can affect daily functioning, the natural or built environment, support and assistance from others, attitudes, and services, system and policies (WHO, Citation2020a, p. 13). The healthy ageing framework proposes that a supportive environment that tailors to the physical and mental state of older people (resembles intrinsic capacity) can enhance functional ability, which can be manifested by facilitating their activity participation, strengthening their ability to build and maintain relationship, and enabling them to maintain preferred lifestyle (WHO, Citation2020b). Guided by the healthy ageing assumption, this study aims to investigate how environments that consist of various domains can affect the SP and the SWB status of disabled community-dwelling older adults.

While participation mostly concerns actions and behaviors that a person exhibits, SP emphasizes more strongly preferred activities and social roles that are valued by the individual and his/her cultural or social environment (Desrosiers et al., Citation2004; Eyssen et al., Citation2011). However, SP has been operationalized and measured differently in prior studies. For example, Chiao et al. (Citation2011) used six group activities that are performed in social organizations to evaluate SP and found that continuously participating or initiating participation in group activities was associated with fewer DSs. Levasseur et al. (Citation2004) assessed SP in terms of 12 categories of 69 life habits, ranging from personal care to recreation activities, and concluded that engagement in every SP category was associated with improved QOL of disabled older persons. However, inconsistencies in typology and measurement in the field of SP has made the results in earlier studies difficult to compare. Furthermore, despite a growing literature on the SP of older adults, very little attention has been paid to those with disabilities. Therefore, the extent to which and the ways in which the SWB of disabled older adults, including their DSs or QOL, can be affected by a more comprehensive and consistent assessment of SP, remain unclear.

As the aging population increased, environmental gerontology was introduced in the 1960s to foster the development of supportive housing or community and delay older people’s relocation to long-term care facilities (Schwarz, Citation2013). Mechanisms of how a supportive environment can promote aging in place can be explained by various theoretical perspectives. The competence-press model that based on ecological perspective from Lawton & Nahemow (1973) and its succeeding theory of person-environment fit are the most dominant approaches in the literature. From these perspectives, an optimal well-being status can be achieved when a balance between environment press and the abilities of an individual is accomplished (Bigonnesse & Chaudhury, Citation2020). Thus, environmental gerontology involves varied research agenda addressing the dynamic transactions among varieties of housing arrangement, different layers and constructs of environment and older people with diverse physical and mental states (Schwarz, Citation2013). However, there is no consensus on how the environment can be comprehensively measured. For example, WHO proposes that a community’s age-friendliness in fostering active ageing should involve eight determinants that across structures, environment, services and policies inside and outside the community (WHO, Citation2018). In the meantime, as mentioned earlier, the healthy ageing initiative describes that environment involves five major domains. When turning from theory to empirical research, scientific studies concerning how physical layout design of neighbor or community, or social environment (e.g. support network or social disorder) influences the physical and mental health of older adults are more common in the past decades (Golant, Citation2013). To make the environment measure consistent with our theoretical framework, we use the environment classification proposed by healthy ageing initiative and characterize environment should include products, equipment and technology that can assist older people’s daily lives, the natural or built environment, support and assistance from others as well as attitudes, and services, system and policies surrounding them.

Disabled older adults are more likely to be affected by the environment than their healthier counterparts because they experience more obstacles in the environment and more restrictions on their social engagement as a result of those barriers (Vaughan et al., Citation2016). However, only a few empirical studies have examined how environment is related to the SP of community-dwelling older individuals with disabilities. For example, the outdoor built environment (including curb ramps and street lights, for example) is related to the physical activity of disabled older persons (Rosenberg et al., Citation2013). Greater proximity to neighborhood resources, such as more services and amenities within a 5-min walking distance, is positively associated with more social activities inside the community (Levasseur et al., Citation2011). Notably, most of the literature considers only a particular aspect of environment, such as the physical or social environment, and its influence on participation in a certain type of activity (including, physical or social activity), neglecting the interrelationships among varied environment aspects. The work done by Chao and Chen (Citation2019) found that 12 environmental indicators for long-term care institutions can be factored into three environment domains, including physical, social, and attitudinal environment. Disabled older adults in a high supportive environment across these three domains were more likely to engage in more activities and also exhibited better mental health. The study also revealed that persons who rated high at one environmental domain tended to value other domains as high, suggesting that varied environmental domains should be considered simultaneously to investigate their influences on activity participation or mental health. Since Chao & Chen’s work was conducted in the context of residential settings, how the consideration of multiple environmental domains can support various SP-related activities across life areas, and affect the SWB of disabled older adults residing in communities requires further investigation.

Very few studies have used latent class analysis (LCA) to consolidate multiple activities into broader categories, classify them into several mutually exclusive subgroups, and examine differences in depression or QOL among subgroups (e.g. Park et al., Citation2018; van Hees et al., Citation2020). However, to the best of our knowledge, no study has simultaneously examined longitudinal associations among environment profiles, SP profiles and subsequent SWB outcomes. Therefore, this study used the healthy ageing framework to explore how are environment and SP profiles associated with DSs or QOL (WHO, Citation2020a). Based on reviewed works, the following hypotheses are proposed: (1) A highly facilitative environment which consists of five environmental domains at T1 is associated with high SP engagement and high SWB at T1; (2) A highly facilitative environment at T1 is also related to a high likelihood of belonging to an active SP profile at T2; (3) A highly facilitative environment at T1 and belonging to an active SP profile at T2 are related to high SWB at T2.

Method

Results

Selection of environment profiles at T1

Table 2 lists the 16 environmental variables and results for EFA, using the weighted least square means and variance adjusted (WLSMV) estimation. EFA was carried out to determine the underlying structure of T1 environmental domains, and WLSMV was used to adapt the discrete responses (Barendse et al., Citation2015). The EFA results showed that a five-factor structure yielded a satisfactory model fit (χ²₍₇₃₎=302.096, p<.001; RMSEA = 0.049; CFI = 0.996; TLI = 0.991).

The upper panel of Table 3 presents the results of the selection of T1 environment profiles, using the five environmental domains identified from EFA as indicators. The 3-class LCA model was used because it yielded the smallest BIC value of 15,780 and the greatest entropy of 0.659. The LMR test also revealed that the 4-class model did not differ significantly from the 3-class model (LMR test = 65.588, p > 0.5). Figure 1 visually represents the three identified environment profiles across the five environmental domains. The first profile contained 34.4% of the sample and was labeled, ‘Highly Facilitative Environment’ (Highly-FE) because participants in this profile were the most likely to report that their environment is helpful in all five environmental domains, especially over 70% of the respondents in the ‘Highly-FE’ class scored the ‘information, transportation & medical care’, ‘attitude and help from family’, and ‘attitude and help from community’ environmental domains as a facilitator. The second profile included 56.0% of the sample, of which a moderate percentage identified their environment as helpful in all five domains, and so was characterized ‘Moderately Facilitative Environment’ (Moderately-FE). Respondents in the ‘Moderately-FE’ class had slightly lower probabilities of evaluating the ‘attitude and help from family’ and ‘attitude and help from community’ environmental domains as helpful (60 to 70%), compared to those in the ‘Highly-FE’ class. However, they had very low probabilities (10 to 20%) of accessing the other three environment domains as facilitative. Profile 3 comprised 9.6% of the participants and was named ‘Weakly Facilitative Environment’ (Weakly-FE) because the associated older adults were the least likely to regard their environment as facilitative in all domains. Among the five environmental domains, less than 2% of the older adults in the ‘Weakly-FE’ class reported the ‘physical environment’ and ‘programs and policies’ environmental domains as helpful; whereas less than 20% of them considered the other three environment domains as beneficial. In general, participants in the three classes consistently exerted lower probabilities of rating the ‘physical environment’ and ‘programs and policies’ environmental domains as helpful than the other three environmental domains. In contrast, older individuals in the three identified classes yielded significantly varied likelihood of reporting the other three environmental aspects as supportive, including the ‘information, transportation and medical care’, ‘attitude and help from family’ and ‘attitude and help from community’ domains.

Figure 1. 3-class environment profiles at T1.

Display full size

Selection of social participation profiles at T1 and T2

The middle panel of Table 3 displays LCA results concerning the selection of T1 SP profiles. Although the 3-class model did not yield a smaller BIC value than the 4-class model, the BIC value steeply declined from the 2-class model (BIC = 12,776) to 3-class (BIC = 12,302), the change from the 3-class model to the 4-class model (BIC = 12,262) was small. Moreover, results of LMR test revealed that the 4-class model was not significantly superior to the 3-class model (LMR test 118.111, p > 0.5). Therefore, the 3-class LCA model was used in subsequent analyses. Figure 2 shows the three identified profiles in SP. Profile 1 contained 43.3% of the sample at T1 and was named, ‘High-SP’. Disabled older adults in this profile were the most likely to engage in the ten SP-related activities. Profile 2, labeled as ‘Moderate-SP’, was associated with 33.9% of the respondents who had moderate probabilities of performing all SP-related activities. Profile 3 included 22.8% of the sample and was labeled as ‘Low-SP’. The three parallel lines shown in Figure 2 illustrated that the probabilities of engagement in the ten SP activities consistently differ across classes. Except for volunteering and active recreation, over 90% of the older adults in the ‘High-SP’ class had participated the rest eight SP-related activities. Similar to the ‘High-SP’ class, participants in the ‘Moderate-SP’ class also showed a low probabilities of engagement in volunteering and active recreation. They also had a lower likelihood of involvement in the other eight SP activities than did those in the ‘High-SP’ class. Lastly, older adults in this profile were not involved in most SP-related activities, although nearly 10% of them visited others and did physical exercise and close to 30% of them invited others to their home.

Figure 2. 3-class social participation profiles at T1.

Display full size

The lower panel of Table 3 and Figure 3 reveal the LCA results for the selection of T2 SP profiles. Consistent with the classification at T1, the 3-class model was used and also labeled, ‘High-SP’ (26.0%), ‘Moderate-SP’ (39.3%), and ‘Low-SP’ (34.7%). Although the 3-class model did not yield smaller AIC and BIC values than the 4-class model, both values decreased rapidly from the 2-class model (AIC = 7,893; BIC = 7,992) to the 3-class (AIC = 7,489; BIC = 7,640), but the differences between the 3-class model and the 4-class model (AIC = 7,360; BIC = 7,563) were much smaller. Moreover, with an entropy value of 0.846, the 3-class model was considered to fit the data well (Weller et al., Citation2020). The visualization of the three identified groups presented in Figure 3 was similar to that shown in Figure 2, suggesting that the probabilities of participating the ten SP-related activities across three classes at T1 and T2 were alike. However, the distribution of participants in the three SP classes was different between waves, while 43.3% of the respondents belonged to the ‘High-SP’ class at T1 but only 26.0% of the reviewees were classified as ‘High-SP’ at T2. In contrast, the percentage of participants belonging to the ‘Low-SP’ class was 22.8% at T1 but increased to 34.7% at T2.

Figure 3. 3-class social participation profiles at T2.

Display full size

Discussion

This work advances current knowledge in its field by using a clustering method to examine comprehensively the influence of a wide range of environment and SP measures on DSs or QOL and to consider holistically the interrelationships within numerous environment or SP indicators. This study also contributes to our understanding by using panel data to explore the longitudinal effects of prior environment profiles on subsequent SP profile memberships, and DSs or QOL.

Three unique environment profiles, labeled Highly-, Moderately- and Weakly-FE, were identified from T1 data based on five environmental domains that were factored from 16 indicators. The classification of environment profiles is similar to that by Chao and Chen (Citation2019) of the environment profiles of long-term care facilities. These classes contained disabled older persons who consistently evaluated all aspects of their environment as highly or weakly facilitative, from tangible to intangible and from physical to policies (Eronen et al., Citation2014; Morrow-Howell et al., Citation2014). Thus, the findings in this work demonstrate that a facilitative environment may comprise multiple factors, which are likely to be related, so all of the surroundings should be considered simultaneously to understand fully the effect of environment on a disabled older person.

The three SP classes that were obtained from T1 and T2 data exhibited similar patterns, both named High-, Moderate- and Low-SP. The classification in this study generally resembles the typologies that were developed by Morrow-Howell et al. (Citation2014) for older adults in general. The SP profiles herein suggest that disabled older adults in different classes had consistently higher or lower probabilities of engaging in all ten SP-related activities—and not just one. Thus, the SP-related activities can be interconnected with respect to involvement and disabled older persons who are highly involved in one activity tended to be more engaged in all other activities. Moreover, the findings also reveal the activity preferences and lifestyle heterogeneity of disabled older individuals. For instance, 43% and 26% of the participants were in the high-SP class and were actively involved in almost every SP-related activity. Accordingly, being disabled does not entail withdrawing from society, and this study found that a considerable fraction of older persons actually continues to participate in various social activities and their SP is relatively robust against the constraints that are imposed by disability.

Consistent with the first research hypothesis, older individuals in the ‘Highly-FE’ and ‘Moderately-FE’ classes were more likely to belong to the ‘High-SP’ or ‘Moderate-SP’ subgroups than those in the ‘Weakly-FE’ class. Moreover, participants in the ‘Highly-FE’ or ‘Moderately-FE’ class or in the ‘High-SP’ or ‘Moderate-SP’ group consistently exhibited significantly fewer DSs and better QOL than those in the ‘Weakly-FE’ or ‘Low-SP’ class in the same wave. The cross-sectional relationships that are identified in this work are consistent with earlier researches on SP profile membership and positive SWB outcomes (Chao & Chen, Citation2019; Park et al., Citation2018; van Hees et al., Citation2020). The results also confirm that a supportive environment across various domains can exhibit joint effects and directly improves disabled older adults’ mental health and QOL.

With regard to second hypothesis, members of the ‘Highly-FE’ subgroup at T1 were more likely to remain in the ‘High-SP’ or ‘Moderate-SP’ subgroup at T2, and participants with these two T2 SP classes yielded significantly fewer DSs or QOL at T2. The findings suggest that a prior highly facilitative environment was directly associated with belonging to more active SP classes at T2, and indirectly reduced the manifestation of DSs and increased QOL two years later. The results provide empirical evidence for longitudinal connections between the environment, as an antecedent, to subsequent SP patterns, which lead to DSs and QOL afterward.

With respect to changes in SP class membership between waves, the transition patterns revealed a predominantly stable tendency because over 50% of the sample remained in their initial status over time. However, a sizable proportion of disabled older individuals shifted upwards from ‘Low- to Moderate-SP’ (22.9%) and from ‘Moderate- to High-SP’ classes (32.6%), representing a shift to a more socially active lifestyle. Finally, only a relatively small fraction of the sample exhibited to a deterioration of SP level between waves, such as from ‘Moderate- to Low-SP’ (13.5%) and from ‘High- to Moderate-SP’ (18.8%) classes. The LTA results revealed that the SP patterns of disabled older individuals were mostly stable over two years. Additionally, a considerable (20–30%) proportion of the older individuals studied herein increased their probability of engagement in various SP-related activities and exhibited a positive shift toward a more active SP profile afterward. This finding not only destroys the myth that being disabled entails a lack of SP, but also demonstrates that some disabled older individuals have the potential to exhibit more SP over time.

In summary, the results in this study have several implications. First, the heterogeneity of disabled older people should be recognized as manifested in their membership in diverse SP class and varied activity transition patterns. In addition to promoting ADL and IADL functioning, sustaining and supporting SP should be included as a part of basic needs and an essential element of functional ability. Thus, both universal and targeted programs should be designed to meeting the SP needs of disabled older persons according to their existing SP class status. Second, the typologies of the environment herein provide valuable insights into how a wide variety of features of the environment can be integrated and jointly influence the SP profiles, mental health and QOL of community-dwelling disabled older adults. Positive environmental attributes in tangible/intangible and physical/social/attitudinal environment domains should be considered simultaneously to determine their cumulative effects. Third, SP-related activities are interconnected and should be evaluated concurrently to capture fully older persons’ heterogeneous lifestyles: determination of the frequency or number of instances of involvement of a certain activity is inadequate. Thus, preventive interventions should be developed to promote or maintain preferred meaningful SP in various areas of life to optimize current and future SWB. Fourth, this study supports the healthy ageing assumption, finding that a highly facilitative environment is linked to high SP, fewer DSs and better QOL, both cross-sectionally and longitudinally. Thus, perceived environmental barriers and SP status can be used as a screening tool for current and future mental well-being and QOL. Specifically, disabled older persons who reside in an environment with more barriers, such as poorly designed and laid out buildings, poor accessibility of public transportation or health care services, or strong discrimination against older and disabled populations, and who exhibit low engagement in SP-related activities are particularly vulnerable to DSs and deteriorating QOL both now and two years in the future.

Several limitations of this work should be addressed. First, the significant disparities in personal backgrounds between the personal backgrounds of respondents with follow-up and those who dropped out could biased the study’s results. Therefore, the results should be generalized to the disabled older population who are younger and have better ADL or cognitive functioning. Additionally, the participants were mainly local long-term care or community care services users and only 10% of the respondents were not a recipient of any service, thus also limiting the generalizability of the results. Second, this study did not exclude participants based on their SPMSQ scores because less than 10% of the respondents had more than five errors out of ten items and all eligible respondents were required to be able to comprehend survey questions. Third, although the use LCA is superior in assigning individuals to class on a basis of their probability of being in classes that are identified from the pattern they have on various indicators. Given the complexity of classes, researcher usually need to assign names to represent the identified classes, causing possible constraint of ‘naming fallacy’. That is, the name of the class fails to accurately symbolize the class membership (Weller et al., Citation2020). Moreover, because T1 and T2 LCA was performed separately, results of sensitivity analysis indicated that respondents in the same SP class between waves may not be entirely matching. Thus, the changes in LCA SP class between waves that were observed might be partially attributed to both statistical procedure and intrapersonal change. Fourth, the longitudinal nature of the data used that were used herein favor the exploration of the temporal ordering of environment, SP profiles and DSs or QOL. However, further longitudinal studies over a longer period and more observations would permit the use of more rigid analysis methods and strengthen the causal inferences drawn in this study. Fifth, the environment was assessed based on respondents’ subjective appraisals of various features of the environment. Thus, the same objective environment might be evaluated differently by different respondents. Thus, the inclusion of objective environmental measures or data that are collected by means other than self-reports would help fully reflect the concept of ‘the environment’. Sixth, although the analysis involved 16 environmental indicators and ten SP-related activities, not all environmental measures and activities were considered and the selection or incorporation of indicators might have affected the determination of environment or SP profiles. Therefore, a single profile may incorporate various indicators in different studies and should be applied with caution. Seventh, this study only considered ADL or IADL limitations as a manifestation of disability, however, different types of disabilities might be affected by varied environmental factors and also influence SP activities differently. Thus, future study would benefit from further examining the associations between varied environment factors and specific types of SP in persons with different types of disabilities.