The Effect of Home- and Community-Based Services on Social Engagement

Abstract

Objectives: Home- and community-based services (HCBS) help older adults to remain active in community settings. However, it is not known if there is a causal relationship between HCBS and social engagement. Methods: We used data from the 2010 and 2012 Health and Retirement Study and measured the effect of HCBS on social engagement via nearest-neighbor Mahalanobis matching, optimal pair matching, genetic matching, and optimal full matching. Results: Genetic matching showed that the odds of social engagement for participants who received at least one HCBS (congregate meal, home-delivered meal, transportation service, case management, homemaker or housekeeping services, or caregiver services) in the prior two years was 1.07 times more likely than participants who have not received any HCBS (robust SE = .030, p = .040). Discussion: HCBS may remove barriers to social engagement through increasing older adults’ personal resources and personal networks.

Keywords

volunteerism and civic engagement home- and community-based care and services long-term services and supports

What this paper adds

• Defining social engagement as volitional engagement in informal or formal activities, we found that recipients of home- and community-based services were more socially engaged than non-recipients.

Applications of study findings

• Policy interventions that increase older adults’ personal resources and personal networks may be effective for increasing older adults’ social engagement.

Introduction

Older adults who are socially engaged consistently show better health including physical health (B. M. Smith et al., 2018), mental health (Mackenzie & Abdulrazaq, 2021), and cognitive function (Amano et al., 2022). In addition to better health outcomes, social engagement is also associated with improved life satisfaction (Jang et al., 2004) and greater resilience (Morrow-Howell et al., 2018). Although such positive images have been commonly portrayed, researchers are aware of the complexity of these relationships, and increasing attention at the same time has also been placed on piecing out the pathways between social engagement and health (e.g., Brown et al., 2016). The purpose of this study is to explore the role of home- and community-based services (HCBS) as an antecedent to social engagement in this pathway.

While the term “social engagement” has been widely used, a consensus on its definition has not emerged. Activities that are considered socially engaging thus vary, ranging from common types of activities that involve social interaction (e.g., doing unpaid community/volunteer work, going to restaurants, or attending sporting events) (Mendes de Leon et al., 2003) to participating in activities that involve interactions between or among people (Reynolds & Carr, 2022). These interactions take place in social contexts that may be informal (e.g., interacting with friends) or formal (e.g., attending a club) (Ellaway & Macintyre, 2007). Defining social engagement as engaging in social activities that always include social exchange, national data showed that 48% of older adults were involved in informal social engagement only, 18.6% were socially engaged both formally and informally, and 32.8% were socially engaged minimally (Amano et al., 2022). Researchers have also tended to define social engagement as involving activities in which individuals participate voluntarily (Bennett, 2002). Self-determination theory posits that engagement activities and well-being outcomes are more highly related when the activities are volitional in nature (Kahana et al., 2013). Scholars adopting this definition often exclude working and caregiving activities from social engagement, because they are not as discretionary as other forms of participation (Morrow-Howell & Gehlert, 2012).

HCBS services are part of the long-term services and supports (LTSS) that the government provides for individuals to receive services in a home or community setting. In 2020, it was estimated that there were about two million Americans who receive HCBS services (KFF, 2020). In addition to meeting medical needs, HCBS also provides for non-medical needs, the most common of which were found in a survey to be personal care (84%), medication adherence (40%), caregiver supports/training (38%), case management (36%), and transportation (32%) (Norman et al., 2018, p. 4). Both health limitations and transportation have been cited by LTSS recipients in 15 states as the top two main reasons why they could not always do things with others when they wanted to (Human Services Research Institute and ADvancing States, 2022, p. 46). Thus, HCBS services could be an efficient first step for removing barriers to social engagement.

HCBS waivers explicitly aim to provide the supports and services for individuals to successfully participate in the community, as well as provide sufficient accommodations to access the range of opportunities that are available to the general population (Friedman & Spassiani, 2017). For example, many waivers provide services for setting up telephones, which allow individuals to sustain and develop relationships or seek out opportunities for community engagement (Friedman & Spassiani, 2017; Myers et al., 1998). Homemaking services may include reading/writing essential correspondence for visually or physically impaired participants (Missouri Department of Health & Senior Services, 2023). Congregate meals are often held in social settings, such as a church or a community center. These “third places” (i.e., not homes or workplaces) are potentially effective places for developing “weak ties” with other individuals (Granovetter, 1973; Sugiyama et al., 2023). Even home-delivered meals may extend beyond a mere meal and enable meaningful one-on-one interactions with the meal deliverers (Thomas et al., 2020).

According to the Resource and Strategic Mobilization model (RSM), older adults’ engagement in productive activities is strongly related to their personal resources (e.g., household income, home ownership, physical health, and cognition), personal networks (e.g., family networks, social networks, religious attendance, and friends/relatives living nearby), and sociodemographic characteristics (e.g., age, gender, race, and education) (Shen, 2017). For example, the model has been empirically tested to show that older adults who are less likely to volunteer have fewer personal health resources (Shen et al., 2020). For the purposes of this study, we expand the model’s definition of “productive activities” to refer to a broader range of social engagement activities. HCBS services are designed to provide support and assistance to people who have limitations in their ability to perform daily activities due to a disability, aging, or illness. These aims are consistent with the RSM model’s focus on personal resources and personal networks; therefore, we hypothesize that HCBS may lead to increased social engagement. While cross-sectional surveys have already shown that the majority of HCBS recipients do take part in various social activities (Human Services Research Institute and ADvancing States, 2022), no study to date has sought to establish a causal relationship between HCBS and social engagement using a longitudinal dataset.

Methods

Data

We used the 2010 and 2012 waves of the nationally representative Health and Retirement Study (HRS), because the 2012 wave is the only wave in the HRS that has collected information about HCBS utilization within a two-year period (HRS, 2020, HRS, 2021a, HRS, 2021b, HRS, 2022a, HRS, 2022b). While the HRS also collected data on HCBS use in the 2011 Health Care Mail Survey (HCMS), the HCMS survey asked if the respondents “ever used any services” (HRS, 2015, p. 23), compared to the 2012 HCBS module, which asked if the respondents used services “in the last two years” (HRS, 2012, p. 1). The unrestricted time range of the HCMS does not allow us to appropriately control for baseline covariates at or prior to receiving HCBS services, therefore it was not used for this study. The HRS is sponsored by the National Institute on Aging (grant number NIA U01AG009740) and is conducted by the University of Michigan. This analysis also uses Early Release data from the Health and Retirement Study (Cross-Wave Tracker File and Cross-Wave Census Region/Division and Mobility File). These data have not been cleaned and may contain errors that will be corrected in the Final Public Release version of the dataset. The questions about HCBS utilization in the 2012 wave were part of an experimental module administered to approximately a random 10% sample of the core (HRS, 2023). The 2010 baseline covariates were taken from the RAND HRS Longitudinal File and the RAND 2010 HRS Fat File. The respondents were restricted to those who met the following inclusion criteria: (1) community-dwelling older adults age 51 or over in 2010 (n = 18,293), (2) completed the 2012 psychosocial leave-behind questionnaire (LBQ) questions on social engagement (n = 6698), (3) completed the 2012 module on HCBS utilization (n = 625), and (4) had no missing values in the social engagement variable (n = 623).

To account for missing data on the study variables that would result in a loss of 5.5% or 34 cases if listwise deletion were used, we first verified that the data were not missing completely at random (MCAR) according to Little’s (1988) test, χ2 (216) = 330.67, p < .01 and assumed that the data are missing at random (MAR). Based on the rule of thumb that the number of imputations should be equal or greater than the percentage of incomplete cases (White et al., 2011), we then created 10 multiply imputed datasets via chained equations using the “mice” package in R (Buuren & Groothuis-Oudshoorn, 2011). Missing values were imputed using predictive mean modeling for continuous variables, proportional odds model for ordinal variables, and logistic regression for dichotomous variables. Finally, we inspected trace plots for each imputed variable to look for adequate mixing and convergence around a stable mean (Enders, 2022). This study was deemed exempt from human subjects’ review by the Washington University in St. Louis Institutional Review Board.

Measures

Dependent Variable

We defined social engagement as volitional engagement in informal or formal social activities (Amano et al., 2020) and included five items from the 2012 HRS leave-behind questionnaire: (1) do volunteer work with children or young people, (2) do any other volunteer or charity work, (3) attend an educational or training course, (4) go to a sport, social, or other club, and (5) attend meetings of non-religious organizations, such as political, community, or other interest groups (J. Smith et al., 2017). To measure only clear levels of social engagement and to address the skewness of the variable, we created a dichotomous social engagement variable, where 1 = participants who engaged in at least one of the four activities several times a week or more and 0 = participants who did not engage in any of the four activities several times a week or more.

Independent/Treatment Variable

We defined HCBS recipients as 2012 HRS respondents whom responded that they had used any one of the following services within the prior two years (i.e., 2010–2012): (1) congregate meals, (2) home-delivered meals, (3) transportation services, (4) case management, (5) homemaker or housekeeping services, or (6) caregiver services. In this study’s sample, 57.0% attended congregate meals, 17.7% received transportation, 15.2% received home-delivered meals, 11.4% received homemaker or housekeeping services, 7.59% received case management, and 7.59% received caregiver services (HRS, 2021b). We applied propensity score matching techniques (see Analytical Strategy for details), and HCBS service utilization was viewed as the treatment condition, where 1 = treated and 0 = not treated.

Covariates

The 2010 HRS covariates used for both the propensity score models and outcome models consisted of variables related to social engagement in later life and selection into HCBS services (Chen et al., 2019; Muramatsu et al., 2010; Pepin et al., 2017; Robinson et al., 2020). These variables included sociodemographic factors (age, gender, race, marital status, education, annual household income, poverty status, Medicaid coverage, and retirement status); social and physical environment (live alone, religious participation, and rural status); and health-related factors (self-rated health, activities of daily living [ADLs], instrumental activities of daily living [IADLs], vision score, chronic conditions, and depression).

Analytical Strategy

We leveraged the Neyman-Rubin causal framework to assess the effect of HCBS utilization on social engagement among older adults (Neyman et al., 1990; Rubin, 2006). This framework can be thought of as a missing data problem, because only one of the two potential outcomes of an older adult in our sample can exist: We either observe the outcome if they received HCBS services (Y_1i), or we observe the outcome if they did not receive HCBS services (Y_0i). However, both potential outcomes are needed to estimate the causal effect of treatment: π_i = Y_1i – Y_0i. To circumvent this problem, the framework holds that the treatment effect can be estimated as a mean difference between treated versus nontreated individuals (Guo & Fraser, 2015). If we let W = 1 denote HCBS utilization and W = 0 denote non-HCBS utilization, then the treatment effect of HCBS utilization on the social engagement outcome Y can be given by: π = E (Y₁|W = 1) – E (Y₀|W = 0). Furthermore, given the observational design of the HRS, we have to make two assumptions: (1) the strongly ignorable treatment assignment assumption, where conditional on X observable covariates, assignment of older adults to HCBS versus non-HCBS use is independent of the social engagement potential outcomes (Y_1i and Y_0i), and (2) the Stable Unit Treatment Value Assumption (SUTVA), which states that the outcomes of one older adult should not affect the treatment assignment of any other older adults (Stuart, 2010).

Propensity score analysis was applied in the present study to construct matched groups that are essentially identical on a set of observed covariates. The MatchThem package in R was used to compare nearest-neighbor Mahalanobis matching, optimal pair matching, genetic matching, and optimal full matching, to test the robustness of the results with respect to the matching algorithm (Pishgar et al., 2021). Covariate imbalances were assessed using the mean of the standardized mean differences of each imputed data; covariates exceeding a threshold of .1 were considered imbalanced (Stuart et al., 2013). For our outcome analysis, we then used logistic regression analysis to measure the effect of HCBS utilization on social engagement, controlling for the same covariates as the matching models to be “doubly robust,” and we fitted models on both the matched and unmatched samples to assess bias reduction as a result of matching (Griefer, 2023). The results were pooled using Rubin’s (1976) rule. To account for the complex survey design of the HRS, and because our sample is limited to those who completed the LBQ, the LBQ sampling weights were incorporated as a covariate in the matching models (DuGoff et al., 2014; J. Smith et al., 2017).

Results

Table 1 provides descriptive statistics and bivariate analyses of the study variables between HCBS and non-HCBS recipients in the sample. As expected, HCBS recipients had significantly more ADLs and IADLs than non-HCBS recipients; nearly one-third (31.7%) of HCBS participants had one or more ADLs, compared to more than a 10th (11.6%) of non-HCBS participants, and 24.1% of HCBS participants had one or more IADLs, compared to 9.7% of non-HCBS participants. HCBS recipients, compared to non-HCBS participants, also had significantly fewer individuals with good or better vision (70.9% vs. 81.7%, respectively), more chronic conditions (2.59 vs. 2.00), and more depressive symptoms (2.01 vs. 1.26). HCBS recipients also tended to be older (mean age of 69 vs. mean age of 66), had more Medicaid coverage (22.1% vs. 5.8%), and had a larger share of females (65.8% vs. 59.6%). There were no significant differences on race, marital status, education, religious affiliation, self-rated health, rural, household income below poverty threshold, live alone status, annual household income, or social engagement.

Table 1.

Characteristics of HCBS and Non-HCBS Recipients in the Sample.

	n = 623	Non-HCBS	HCBS	p
	n = 623	n = 544	n = 79	p
Independent variables: categorical (n (%))	Female (ref: male)	324 (59.6)	52 (65.8)
	Race
	White/Caucasian	423 (78.6)	55 (69.6)
	Black/African American	83 (15.4)	20 (25.3)
	Other	32 (5.9)	4 (5.1)
	Marital status (ref: not married)	326 (59.9)	41 (51.9)
	Education
	Less than high school	107 (19.7)	19 (24.1)
	High school graduate	162 (29.8)	21 (26.6)
	College and above	275 (50.6)	39 (49.4)
	Religious (ref: not religious)	419 (77.0)	59 (74.7)
	Retirement status			*
	Not retired	224 (42.1)	15 (19.0)
	Completely retired	223 (41.9)	51 (64.6)
	Partly retired	63 (11.8)	11 (13.9)
	Not working for pay or homemaker	22 (4.1)	2 (2.5)
	Self-rated health
	Excellent	57 (10.5)	7 (8.9)
	Very good	177 (32.5)	20 (25.3)
	Good	179 (32.9)	21 (26.6)
	Fair	102 (18.8)	25 (31.6)
	Poor	29 (5.3)	6 (7.6)
	Number of ADLs			*
	0	481 (88.4)	54 (68.4)
	1–2	50 (9.2)	15 (19.0)
	≥3	13 (2.4)	10 (12.7)
	Number of IADLs			*
	0	491 (90.3)	60 (75.9)
	1–2	42 (7.7)	13 (16.5)
	≥3	11 (2.0)	6 (7.6)
	Rural (ref: urban/suburban)	127 (23.5)	19 (24.4)
	Good or better vision (ref: fair, poor, or legally blind)	443 (81.7)	56 (70.9)	*
	Household income below poverty threshold (ref: above poverty threshold)	50 (9.2)	10 (12.7)
	Medicaid (ref: no coverage)	31 (5.8)	17 (22.1)	*
	Live alone (ref: does not live alone)	360 (66.2)	59 (74.7)
Independent variables: continuous (M (SD))	Age	65.87 (10.12)	69.05 (10.18)	*
	Annual household income (logged)	.30 (.35)	.30 (.35)
	Chronic conditions (0–9)	2.00 (1.46)	2.59 (1.57)	*
	Depression (0–8)	1.26 (1.87)	2.01 (2.30)	*
Dependent variable: categorical (n (%))	Social engagement (ref: not socially engaged)	205 (37.7)	34 (43.0)

Note. p-values from chi-square tests of independence, Fisher’s exact tests when expected counts lower than five, and Wilcoxon rank sum tests with continuity correction. ADL, activities of daily living; IADL, instrumental activities of daily living.

*p < .05.

Table 2 presents the pooled effects of receiving HCBS services on social engagement, with and without matching, and the number of covariates that are within the .1 standardized mean difference threshold. Both the unadjusted model without matching and nearest-neighbor Mahalanobis matching did not achieve balance on most of the covariates. Only eight covariates in the nearest-neighbor model were within the .1 standardized mean difference threshold. Optimal pair matching, genetic matching, and optimal full matching all achieved balance on 17 out of 18 of the covariates. Genetic matching had the smallest standardized mean differences in the multiply imputed data, compared to both types of optimal matching. Genetic matching also had the smallest robust standard errors. For the genetic matching model, we found that the odds of social engagement for participants who received at least one HCBS services in the prior two years was 1.07 times more likely than participants who have not received any HCBS services in the prior two years (robust SE = .030, p = .040). We found similar magnitudes and directions of the odds of social engagement using nearest-neighbor Mahalanobis matching, optimal pair matching, and optimal full matching. Except optimal pair matching, all treatment effects were statistically significant using an alpha level of .05.

Table 2.

Effect of HCBS Utilization on Social Engagement Using Different Matching Methods.

Matching method	Odds ratio^a	Robust SE	Z	p	ESS^b	Covariates balanced^c
Unadjusted model	1.13	.061	1.97	.049	544/79	10/18
Nearest-neighbor Mahalanobis matching	1.07	.024	2.75	.0082	79/79	8/18
Optimal pair matching	1.14	.057	2.28	.055	79/79	17/18
Genetic matching	1.07	.030	2.20	.040	79/79	17/18
Optimal full matching	1.16	.072	2.04	.042	138/79	17/18

^aThe odds ratios represent the population average treatment effect on the treated (PATT) using logistic regression. All models controlled for age, gender, race, marital status, education, annual household income, poverty status, Medicaid coverage, and retirement status, live alone, religious participation, and rural status, self-rated health, activities of daily living [ADLs], instrumental activities of daily living [IADLs], vision score, chronic conditions, and depression.

^bEffective sample size (control/treated).

^cCovariates were considered balanced if the mean of the standardized mean differences in the imputed data were within a .1 threshold (Stuart et al., 2013); for the unadjusted model, covariates were considered balanced if their bivariate test was not statistically significant.

Figure 1 shows the standardized mean differences of the covariates before and after using genetic matching on the multiply imputed data. Across all of the imputations, chronic conditions had the largest standardized mean differences, with a mean of .10. All of the other covariates were balanced between HCBS and non-HCBS participants, with standardized mean differences less than the .1 threshold.

Figure 1.

Covariate balance before and after genetic matching on multiply imputed data.

Discussion

This study’s findings add to the literature on the protective benefits of HCBS for older populations. HCBS have previously been linked to outcomes such as reduced risk of nursing home admission and greater likelihood of living in the community (Muramatsu et al., 2007, 2008), but this is the first study to link HCBS use with increased social engagement. According to the RSM model, HCBS services may be related to increased social engagement because of increases in personal resources and personal networks. However, our findings suggest that HCBS services may have a direct, positive effect on social engagement, irrespective of key individual personal resources that were controlled for in our models. This suggests that a helpful extension to the RSM model may be adding a macro-level pathway between social policies, such as HCBS, and social engagement, as well as a mezzo-level pathway between the organizational and service-delivery components of HCBS and social engagement. These pathways may be located farther upstream in the RSM model, such as serving as antecedents to an individual’s personal network. For example, HCBS services tend to “place a greater responsibility on families to support the formal care being received” (Gorges et al., 2019, p. 1117), increasing a recipient’s network of informal caregivers.

There are several possible explanations for the link between HCBS and social engagement. First, HCBS directly removes barriers to being engaged in the community; for example, meeting the physical health needs of older adults increases the likelihood for volunteering or other forms of social engagement (Dury et al., 2015). Second, while a variety of baseline resources were controlled for in this study, HCBS services may increase other personal resources over time, which, in turn, are associated with increased social engagement (McNamara & Gonzales, 2011). For instance, as mentioned previously, HCBS programs tend to encourage family members to support care recipients in lieu of institutional care, such as through consultation for family caregivers (Feinberg & Newman, 2004); this has the effect of increasing or strengthening an individual’s family network. Increased interactions with family members can be considered a form of social engagement, or it may serve as an additional pathway for other types of social engagement (Bath & Deeg, 2005). Finally, a number of studies have explored the psychological ramifications of HCBS support, such as increased life satisfaction (Chen et al., 2019), which may also be related to a person’s likelihood of being actively involved in the community.

This study shows with longitudinal data and causal inference methodology that HCBS is a robust predictor for social engagement. The strength of this study was the use of multiple matching methods, which served as sensitivity analyses for uncovering a robust treatment effect. Another strength was that this study did not the adopt decision of other researchers in combining the HRS, 2012 survey, which asks if respondents used HCBS in the last two years, with an off-year HRS survey in 2011, which asks if respondents ever used HCBS (Pepin et al., 2017; Robinson et al., 2020). This difference in wording is significant, as combining the two samples would not allow one to appropriately control for baseline covariates at or before HCBS services were received.

One limitation of this research is that it does not fully meet the SUTVA assumption due to heterogeneous treatment effects—Individuals whom we defined as having received the “HCBS treatment” may have received different combinations or dosages of HCBS services. Unfortunately, we could not compare the potential outcomes of individuals who received identical HCBS services due to small sample sizes. A second limitation is that our adoption of a two-year observation window, while reducing the chance for heterogeneous treatment effects, may artificially limit our ability to observe longer-term effects of HCBS on future social engagement. Finally, a third limitation is that the LBQ does not include other forms of social engagement, such as gatherings with friends or family. While the LBQ item “do activities with grandchildren, nieces/nephews, or neighborhood children” may reflect social engagement with family and friends, it was excluded on the basis that activities with relatives could include nondiscretionary caregiving. Custodial grandparenting, for example, may be associated with the loss of social relationships (Hayslip et al., 2019). Future research on social engagement should carefully distinguish whether interactions with family member are obligatory or use a continuous measure to probe the degree of perceived choice (e.g., Pertl et al., 2019).

Recent legislative efforts to increase support for HCBS services, such as the Inflation Reduction Act of 2022, have not been funded (Kreider & Werner, 2023), even as evidence on the benefits of HCBS continues to mount. For policymakers, this study shows that investing in HCBS may reap the benefits of social engagement, such as increasing the capacity for older adults to be productively engaged in volunteering and promoting well-being for both older adults and their communities. We also suspect that this linkage may be vitally important both during and after the COVID-19 pandemic, given the experience of social isolation and loneliness among many older adults. Increased infrastructure for community services may alleviate these stressors, but consideration of the modalities (e.g., telehealth) of the services is also needed to address social distancing or reach remote individuals (Isasi et al., 2021). For researchers, the known benefits of social engagement should motivate additional inquiry into how HCBS services are related to social engagement, ideally controlling for the type and intensity of the services, and how this relationship may further promote the health and well-being for older adults.

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) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the HRS (Health and Retirement Study), is sponsored by the National Institute on Aging (grant number NIA U01AG009740), and is conducted by the University of Michigan.

IRB Protocol Approval Number

This study was deemed exempt from human subjects’ review by the Washington University in St. Louis Institutional Review Board (#202305118).

ORCID iD

Peter C. Sun

References

Amano

Morrow-Howell

Park

(2020). Patterns of social engagement among older adults with mild cognitive impairment. The Journals of Gerontology: Series B, 75(7), 1361–1371. https://doi.org/10.1093/geronb/gbz051

Amano

Park

Morrow-Howell

Carpenter

(2022). The association between patterns of social engagement and conversion from mild cognitive impairment to dementia: Evidence from the health and retirement study. Alzheimer Disease and Associated Disorders, 36(1), 7–14. https://doi.org/10.1097/WAD.0000000000000486

Bath

P. A.

Deeg

(2005). Social engagement and health outcomes among older people: Introduction to a special section. European Journal of Ageing, 2(1), 24–30. https://doi.org/10.1007/s10433-005-0019-4

Bennett

K. M.

(2002). Low level social engagement as a precursor of mortality among people in later life. Age and Ageing, 31(3), 165–168. https://doi.org/10.1093/ageing/31.3.165

Brown

C. L.

Robitaille

Zelinski

E. M.

Dixon

R. A.

Hofer

S. M.

Piccinin

A. M.

(2016). Cognitive activity mediates the association between social activity and cognitive performance: A longitudinal study. Psychology and Aging, 31(8), 831–846. https://doi.org/10.1037/pag0000134

Buuren

S. V

Groothuis-Oudshoorn

(2011). Mice: Multivariate imputation by chained equations in R. Journal of Statistical Software, 45(3). 1.https://doi.org/10.18637/jss.v045.i03

Chen

Amano

Park

Kim

(2019). Home and community-based services and life satisfaction among homebound and poor older adults. Journal of Gerontological Social Work, 62(7), 708–727. https://doi.org/10.1080/01634372.2019.1639094

DuGoff

E. H.

Schuler

Stuart

E. A.

(2014). Generalizing observational study results: Applying propensity score methods to complex surveys. Health Services Research, 49(1), 284–303. https://doi.org/10.1111/1475-6773.12090

Dury

De Donder

De Witte

Buffel

Jacquet

Verté

(2015). To volunteer or not: The influence of individual characteristics, resources, and social factors on the likelihood of volunteering by older adults. Nonprofit and Voluntary Sector Quarterly, 44(6), 1107–1128. https://doi.org/10.1177/0899764014556773

10.

Ellaway

Macintyre

(2007). Is social participation associated with cardiovascular disease risk factors? Social Science and Medicine, 64(7), 1384–1391. https://doi.org/10.1016/j.socscimed.2006.11.022

11.

Enders

C. K.

(2022). Applied missing data analysis (2nd ed.). Guilford Publications.

12.

Feinberg

L. F.

Newman

S. L.

(2004). A study of 10 states since passage of the national family caregiver support program: Policies, perceptions, and program development. The Gerontologist, 44(6), 760–769. https://doi.org/10.1093/geront/44.6.760

13.

Friedman

Spassiani

N. A.

(2017). Getting out there: Community support services for people with intellectual and developmental disabilities in Medicaid HCBS waivers. Inclusion, 5(1), 33–44. https://doi.org/10.1352/2326-6988-5.1.33

14.

Gorges

R. J.

Sanghavi

Konetzka

R. T.

(2019). A national examination of long-term care setting, outcomes, and disparities among elderly dual eligibles. Health Affairs, 38(7), 1110–1118. https://doi.org/10.1377/hlthaff.2018.05409

15.

Granovetter

M. S.

(1973). The strength of weak ties. American Journal of Sociology, 78(6), 1360–1380. https://doi.org/10.1086/225469

16.

Griefer

(2023, April 13). Estimating effects after matching. MatchIt. https://kosukeimai.github.io/MatchIt/articles/estimating-effects.html

17.

Guo

Fraser

M. W.

(2015). Propensity Score Analysis. Sage.

18.

Hayslip

Jr. Fruhauf

C. A.

Dolbin-MacNab

M. L.

(2019). Grandparents raising grandchildren: What have we learned over the past decade? The Gerontologist, 59(3), Article e152–e163. https://doi.org/10.1093/geront/gnx106

19.

HRS . (2012). Health and Retirement Study, (HRS 2012 Module 2—Final Version) questionnaire. Produced and distributed by the University of Michigan with funding from the National Institute on Aging (grant number NIA U01AG009740).

20.

HRS . (2015). Health and Retirement Study, (2011 Health Care Mail Survey (English)) Questionnaire: English. Produced and distributed by the University of Michigan with funding from the National Institute on Aging (grant number NIA U01AG009740).

21.

HRS . (2020). Health and Retirement Study, (Cross-Wave Tracker File (Early 2018 V2.0)) public use dataset. Produced and distributed by the University of Michigan with funding from the National Institute on Aging (grant number NIA U01AG009740).

22.

HRS . (2021b). Health and Retirement Study, (RAND HRS 2012 Fat File (V3A)) public use dataset. Produced and distributed by the University of Michigan with funding from the National Institute on Aging (grant number NIA U01AG009740).

23.

HRS . (2021a). Health and Retirement Study, (Cross-Wave Census Region/Division and Mobility File (Early V8.2)) public use dataset. Produced and distributed by the University of Michigan with funding from the National Institute on Aging (grant number NIA U01AG009740).

24.

HRS . (2022a). Health and Retirement Study, (RAND HRS 2010 Fat File (V6A)) public use dataset. Produced and distributed by the University of Michigan with funding from the National Institute on Aging (grant number NIA U01AG009740).

25.

HRS . (2022b). Health and Retirement Study, (RAND HRS Longitudinal File 2018 (V2)) public use dataset. Produced and distributed by the University of Michigan with funding from the National Institute on Aging (grant number NIA U01AG009740).

26.

HRS . (2023). Experimental modules. Health and Retirement Study. https://hrs.isr.umich.edu/documentation/modules

27.

Human Services Research Institute and ADvancing States . (2022). 2021-2022 National Core Indicators—Aging and Disabilities—2021-22 NCI-AD Adult Consumer Survey National Report. Human Services Research Institute and ADvancing States. https://nci-ad.org/upload/reports/2021-22_NCI-AD_Adult_Consumer_Survey_Part_1.pdf.

28.

Isasi

Naylor

M. D.

Skorton

Grabowski

D. C.

Hernández

Rice

V. M.

(2021). Patients, families, and communities COVID-19 impact assessment: Lessons learned and compelling needs. In NAM Perspectives, 2021. Washington, DC: National Academy of Medicine. https://doi.org/10.31478/202111c.

29.

Jang

Mortimer

J. A.

Haley

W. E.

Graves

A. R. B.

(2004). The role of social engagement in life satisfaction: Its significance among older individuals with disease and disability. Journal of Applied Gerontology, 23(3), 266–278. https://doi.org/10.1177/0733464804267579

30.

Kahana

Bhatta

Lovegreen

L. D.

Kahana

Midlarsky

(2013). Altruism, helping, and volunteering: Pathways to well-being in late life. Journal of Aging and Health, 25(1), 159–187. https://doi.org/10.1177/0898264312469665

31.

KFF . (2020). Medicaid Section 1915(c) HCBS waiver participants, by type of waiver. KFF. https://www.kff.org/state-category/medicaid-chip/home-and-community-based-services/

32.

Kreider

A. R.

Werner

R. M.

(2023). The home care workforce has not kept pace with growth in home and community-based services. Health Affairs, 45(5), 1351. https://doi.org/10.1377/hlthaff.2022.01351

33.

Little

Roderick J. A.

(1988). A Test of Missing Completely at Random for Multivariate Data with Missing Values. Journal of the American Statistical Association, 83(404), 1198–1202. https://doi.org/10.2307/2290157.

34.

Mackenzie

C. S.

Abdulrazaq

(2021). Social engagement mediates the relationship between participation in social activities and psychological distress among older adults. Aging and Mental Health, 25(2), 299–305. https://doi.org/10.1080/13607863.2019.1697200

35.

McNamara

T. K.

Gonzales

(2011). Volunteer transitions among older adults: The role of human, social, and cultural capital in later life. The Journals of Gerontology: Series B, 66(4), 490–501. https://doi.org/10.1093/geronb/gbr055

36.

Mendes de Leon

C. F.

Glass

T. A.

Berkman

L. F.

(2003). Social engagement and disability in a community population of older adults: The new haven EPESE. American Journal of Epidemiology, 157(7), 633–642. https://doi.org/10.1093/aje/kwg028

37.

Missouri Department of Health and Senior Services . (2023). Home and community based services manual. Missouri Department of Health & Senior Services. https://health.mo.gov/seniors/hcbs/hcbsmanual/index.php

38.

Morrow-Howell

Backes

O‘Neill

Greenfield

J. C.

(2018). Volunteering in later life: Policies and programs to support a resilient aging society. In Resnick

Gwyther

L. P.

Roberto

K. A.

(Eds.), Resilience in aging: Concepts, research, and outcomes (pp. 171–189). Springer International Publishing. https://doi.org/10.1007/978-3-030-04555-5_9

39.

Morrow-Howell

Gehlert

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

T. R.

Anderson

L. A.

Binstock

R. H.

(Eds.), Public health for an aging society. JHU Press.

40.

Muramatsu

Hoyem

R. L.

Yin

Campbell

R. T.

(2008). Place of death among older Americans. Medical Care, 46(8), 829–838. https://doi.org/10.1097/MLR.0b013e3181791a79

41.

Muramatsu

Yin

Campbell

R. T.

Hoyem

R. L.

Jacob

M. A.

Ross

C. O.

(2007). Risk of nursing home admission among older Americans: Does states’ spending on home- and community-based services matter? Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 62(3), S169–S178. https://doi.org/10.1093/geronb/62.3.s169

42.

Muramatsu

Yin

Hedeker

(2010). Functional declines, social support, and mental health in the elderly: Does living in a state supportive of home and community-based services make a difference? Social Science and Medicine, 70(7), 1050–1058. https://doi.org/10.1016/j.socscimed.2009.12.005

43.

Myers

Ager

Kerr

Myles

(1998). Outside looking in? Studies of the community integration of people with learning disabilities. Disability and Society, 13(3), 389–413. https://doi.org/10.1080/09687599826704

44.

Neyman

Dabrowska

D. M.

Speed

T. P.

(1990). On the application of probability theory to agricultural experiments. Essay on principles. Section 9. Statistical Science, 5(4), 465–472. https://doi.org/10.1214/ss/1177012031

45.

Norman

G. J.

Wade

A. J.

Morris

A. M.

Slaboda

J. C.

(2018). Home and community-based services coordination for homebound older adults in home-based primary care. BMC Geriatrics, 18(1), 241. https://doi.org/10.1186/s12877-018-0931-z

46.

Pepin

Leggett

Sonnega

Assari

(2017). Depressive symptoms in recipients of home- and community-based services in the United States: Are older adults receiving the care they need? American Journal of Geriatric Psychiatry: Official Journal of the American Association for Geriatric Psychiatry, 25(12), 1351–1360. https://doi.org/10.1016/j.jagp.2017.05.021

47.

Pertl

M. M.

Sooknarine-Rajpatty

Brennan

Robertson

I. H.

Lawlor

B. A.

(2019). Caregiver choice and caregiver outcomes: A longitudinal study of Irish spousal dementia caregivers. Frontiers in Psychology, 10, 1801. https://doi.org/10.3389/fpsyg.2019.01801.

48.

Pishgar

Greifer

Leyrat

Stuart

(2021). MatchThem: Matching and weighting after multiple imputation. The R Journal, 13(2), 228. https://doi.org/10.32614/RJ-2021-073

49.

Reynolds

J. R.

Carr

D. C.

(2022). Long-term correlates of racially diverse schooling: Education, wealth, and social engagement in later life. Social Currents, 9(5), 427–458. https://doi.org/10.1177/23294965211060382

50.

Robinson

K. N.

Menne

H. L.

Gaeta

(2020). Use of informal support as a predictor of home- and community-based services utilization. The Journals of Gerontology: Series B, 76(1), 133–140. https://doi.org/10.1093/geronb/gbaa046

51.

Rubin

D. B.

(1976). Inference and missing data. Biometrika, 63(3), 581–592. https://doi.org/10.1093/biomet/63.3.581

52.

Rubin

D. B.

(2006). Matched sampling for causal effects. Cambridge University Press. https://doi.org/10.1017/CBO9780511810725

53.

Shen

H.-W.

(2017). Resource and strategic mobilization model (RSM) of productive aging: Examining older Americans’ participation in various productive activities. Ageing International, 42(1), 23–43. https://doi.org/10.1007/s12126-016-9259-1

54.

Shen

H.-W.

Wang

Wong

Amano

(2020). Resource and network predictors of multiple volunteering cessations: Implications for social policy and practice. Clinical Social Work Journal, 48(2), 189–195. https://doi.org/10.1007/s10615-020-00748-x

55.

Smith

B. M.

Yao

Chen

K. S.

Kirby

E. D.

(2018). A larger social network enhances novel object location memory and reduces hippocampal microgliosis in aged mice. Frontiers in Aging Neuroscience, 10, 142. https://doi.org/10.3389/fnagi.2018.00142.

56.

Smith

Ryan

Fisher

Sonnega

Weir

(2017). HRS Psychosocial and Lifestyle Questionnaire 2006-2016. Institute for Social Research, University of Michigan.

57.

Stuart

E. A.

(2010). Matching methods for causal inference: A review and a look forward. Statistical Science, 25(1), 1–21. https://doi.org/10.1214/09-STS313

58.

Stuart

E. A.

Lee

B. K.

Leacy

F. P.

(2013). Prognostic score–based balance measures can be a useful diagnostic for propensity score methods in comparative effectiveness research. Journal of Clinical Epidemiology, 66(8 Suppl), S84–S90. https://doi.org/10.1016/j.jclinepi.2013.01.013

59.

Sugiyama

Chau

H.-W.

Abe

Kato

Jamei

Veeroja

Mori

Sugiyama

(2023). Third places for older adults’ social engagement: A scoping review and research agenda. The Gerontologist, 63(7), 1149–1161. https://doi.org/10.1093/geront/gnac180

60.

Thomas

K. S.

Gadbois

E. A.

Shield

R. R.

Akobundu

Morris

A. M.

Dosa

D. M.

(2020). It’s not just a simple meal. It’s so much more”: Interactions between meals on wheels clients and drivers. Journal of Applied Gerontology, 39(2), 151–158. https://doi.org/10.1177/0733464818820226

61.

White

I. R.

Royston

Wood

A. M.

(2011). Multiple imputation using chained equations: Issues and guidance for practice. Statistics in Medicine, 30(4), 377–399. https://doi.org/10.1002/sim.4067