Measuring Consumer-Reported Quality of Life Among Recipients of Publicly Funded Home- and Community-Based Services: Implications for Health Equity

Abstract

Objectives:

Despite an increased policy focused on home- and community-based services (HCBS), little is known about their quality of life (QoL)—a key measure of person-centered care. This paper addresses this gap by measuring consumers’ self-reported QoL and identifying factors associated with disparities in QoL.

Methods:

We analyzed the 2015–2016 National Core Indicators–Aging and Disability survey for 3426 respondents in Minnesota, using factor analyses to identify latent QoL domains. Multivariable regression models identified predictors of QoL domains.

Results:

Factor analyses identified three valid and reliable latent QoL domains: security, self-determination, and care experiences. Younger consumers with disabilities (versus consumers ≥65 years of age), minoritized racial/ethnic groups, consumers with hearing loss, without a spouse/domestic partner, and not living in consumer’s own/family home reported significantly lower QoL in various domains (p < .001).

Discussion:

Disparities in HCBS consumer-reported QoL exist, necessitating equitable reforms to improve HCBS quality for its increasingly diversified consumer base.

Introduction

In 2020, an estimated 5.8 million Americans used Medicaid long-term services and supports (LTSS) in home and community environments, while 1.9 million received LTSS in institutional settings, with a continued shift away from institutional LTSS (Chidambaram & Burns, 2023). Medicaid is the primary payer of LTSS (Murray et al., 2023). HCBS consumers are a diverse subset of the LTSS population with considerable variation according to need; HCBS consumers include older adults (65 years and older), younger adults with physical disabilities, and adults of any age with intellectual or developmental disabilities (ID/DD) or severe mental health conditions (AHRQ, 2012; Musumeci, 2019). Medicaid-funded HCBS are offered through Medicaid state plan options or Section 1915(c) HCBS waivers (Murray et al., 2023) and are available to consumers who both qualify for Medicaid and are deemed to require an institutional level of care (Reaves & Musumeci, 2015). Under Section 1915(c) HCBS waivers, states can waive certain Medicaid program requirements, instead opting to set their own parameters regarding eligibility, service criteria and service areas, or eligible groups of beneficiaries (Centers for Medicare & Medicaid Services, 2019).

The LTSS population is becoming increasingly heterogeneous due to increasing diversity in the consumer base as well as increasing complexity of medical and social needs of consumers with intersecting identities and unique comorbidity profiles or case-mix (AHRQ, 2012). Additionally, quality varies considerably within and across markets (MACPAC, 2020). A growing proportion of consumers need long-term services and supports that address their complex medical and social needs, which has prompted an increase in the intensity and/or frequency of HCBS (Konetzka et al., 2020). However, it is unclear whether efforts in HCBS expansion have been applied equitably, to account for the increasing heterogeneity in the HCBS consumer population; or the efforts have been unresponsive to the growing diversity in the HCBS consumer base, potentially exacerbating inequities in HCBS quality and access (Chong et al., 2022). Furthermore, despite the growing recognition by policy makers and consumers of the need for person-reported measures of quality of life (QoL) and their use for assisted living and nursing home residents, there exists a paucity of research evaluating HCBS consumer’s self-reported QoL (Bucy et al., 2023; Shippee et al., 2015, 2023). This gap in knowledge hinders our understanding of the unmet needs of HCBS users and any related inequities, which are crucial metrics of delivering person-centered care in LTSS (Shippee et al., 2022). This paper aims to address this gap by measuring self-reported QoL among younger and older users of HCBS in Minnesota (MN) and identifying factors associated with differences in QoL.

Conceptual Framework

Quality measurement is key to driving systems change. As it relates to HCBS, quality measurement connects performance metrics to outcomes such as consumer safety, assists consumers in making informed choices, and enables policy makers and researchers to assess the effectiveness of care delivery models (National Quality Forum, 2016). A key component of quality measures involves evaluating person-centered measures of quality, such as QoL, which refers to individuals’ subjective assessments of their lives. In their 2016 report, the National Quality Forum (NQF) recommended the inclusion of person-centered metrics, including QoL, in the performance measurement of HCBS, aiming to improve consumer outcomes and promote community living. Additionally, the NQF emphasized the importance of assessing quality at three distinct levels—the person-level, service-level, and systems-level (National Quality Forum, 2016).

However, significant gaps persist in quality measurement across publicly funded HCBS programs. For example, performance measures often concentrate on a limited set of quality domains, standardized measures of quality are scarce across programs and states, and there is a notable absence of emphasis on the person-centered needs and experiences of the people receiving services (National Quality Forum, 2016).

In our efforts to address this gap, we draw upon Zubritsky et al. conceptual model of QoL for people who receive LTSS. Zubritsky et al. model advocates for a multidimensional approach to QoL measurement, encompassing different domains to assess person-centered quality for LTSS recipients (Zubritsky et al., 2013). The model underscores the importance of considering both individual characteristics (e.g., functional health and age) and environmental factors, which may include features of the program or service delivery. This approach aligns with NQF recommendation to account for person-, service-, and system-level factors in measures of quality.

We advance this work by using NCI-AD consumer survey data, an underutilized resource for HCBS quality in MN. Minnesota has participated in NCI-AD data collection since 2015. In Minnesota, the Department of Human Services (DHS) manages and provides publicly funded HCBS for older adults via Medical Assistance (Minnesota’s Medicaid program), waiver programs, and other publicly funded programs such as the Older Americans Act. Minnesota spent nearly 77% of its total Medicaid LTSS expenditure on HCBS in 2015, which was among the highest in the nation (Eiken et al., 2017). Using the 2015–2016 NCI-AD survey data, we aim to (1) describe NCI-AD-derived consumer-reported quality of life (QoL) measures for both young (<65) adult consumers with disabilities and older adult consumers of publicly funded HCBS and (2) examine individual correlates of self-reported QoL.

Tailoring QoL measures for two distinct groups of HCBS users—younger adults with disabilities and older adults—is important, given variation in demographic characteristics and functional needs. We examined differences in perceived QoL based on age as suggested by Sprangers and Schwartz’s (1999) Response Shift Theory. Response shift denotes a modification in the interpretation of an individual’s self-assessment of a particular concept due to (a) a shift in the respondent’s internal measurement standards (scale recalibration), (b) a shift in the respondent’s priorities and values (reprioritization), or (c) a redefinition of the target concept (reconceptualization). With aging, changes in health status and care needs can involve alterations in internal standards (requiring scale recalibration) and core values and preferences (requiring reprioritization and reconceptualization of QoL) (Vanier et al., 2021). Older recipients, typically those aged 65 or above, are usually enrolled in programs or waivers designed for aging individuals who require various levels of functional support. In contrast, younger adults with disabilities often receive services from programs or waivers designed for populations with specific disability diagnoses such as traumatic brain injury, physical disability, and/or ID/DD (ADvancing States & Human Services Research Institute, 2017). These groups represent distinct populations with different care needs. Analysis of the NCI-AD data from prior research indicates that younger adults with disabilities and frail older adults weigh community inclusion differently while assessing their self-reported health (Duan et al., 2021). Younger HCBS users value community inclusion more than their older counterparts. Age differences in the determinants of well-being have also been observed in other studies; for example, Idler et al. (2018) found that the negative impacts of poor physical health and activity limitation on self-rated health diminished in older adults, while they remained significant for younger adults.

Methods

Study Population and Data Sources

The study sample consists of older adults (>65 years of age) and younger adults with disabilities who participated in the 2015–2016 wave of the NCI-AD consumer survey in Minnesota (n = 3426). NCI-AD consumer surveys: To address these gaps in person-centered performance measures in HCBS, the National Core Indicators-Aging and Disabilities (NCI-AD^™) consumer survey serves as a standardized way to assess QoL and outcomes of older adults and adults with physical disabilities receiving publicly funded HCBS. NCI-AD is a collaborative effort between ADvancing States (formerly the National Association of States United for Aging and Disabilities) and the Human Services Research Institute (HSRI), and state participation is voluntary. The NCI-AD consumer survey measures a broad range of quality domains, including employment, respect and rights, service/care coordination, choice, and health and safety, via in-person interviews with consumers or their family members. To date, the NCI-AD consumer survey has not undergone psychometric validation and has not been widely used by health service researchers.

This study excluded data from the proxy component (13.2%) as most QoL-related items are not available in the proxy version. A third-party agency was contracted to implement the survey through face-to-face interviews with a sample of publicly funded HCBS waiver recipients. The interviewers, some of whom were bilingual, were hired and trained by the third party. Interviews were conducted at the time and location of participants’ choosing. More information regarding the sample distribution by program and the data collection process can be found in the state report (Minnesota Department of Human Services, 2016). Of the survey sample, the aging subsample (n = 1849) included participants 65+ who were enrolled in programs including the Elderly Waiver, State Plan Funded Home Care, Alternative Care (a state-funded program for people with low income and assets who were not eligible for Minnesota Medicaid), and the Older Americans Act. The disability subsample (n = 1577) included adult participants between 18 and 64 years old enrolled in the State Plan Funded Home Care program and a variety of targeted waiver programs for populations with traumatic brain injury, physical disability, and/or ID/DD. Table 1 shows characteristics of the sample.

Table 1.

Characteristics of the 2015–2016 National Core Indicators—Aging and Disability Survey Respondents From Minnesota Included in the Study (n = 3426).

	Younger adults with disabilities (n = 1577)	Older adults aged 65 or older (n = 1849)	t Test/Chi-square test	P
Age (mean±SD)	46.45 ± 13.05	78.22 ± 8.42	-85.77	<0.001
Female (n [%])	820 (52.00%)	1384 (74.89%)	194.41	<0.001
Race/ethnicity (n [%])
White	1227 (78.35%)	1337 (73.75%)	78.37	<0.001
Black or African American	260 (16.60%)	230 (12.69%)
Asian	47 (3.00%)	111 (6.12%)
Hispanic or Latino	32 (2.04%)	135 (7.45%)
Marital status (n [%])
Single, never married	409 (26.00%)	201 (11.57%)	1300	<0.001
Married or has a domestic partner	977 (62.11%)	282 (16.23%)
Separated or divorced	144 (9.15%)	501 (28.84%)
Widowed	43 (2.73%)	753 (43.35%)
Place of residence (n [%])
Living in group settings	468 (30.12%)	387 (21.16%)	35.69	<0.001
Own or family house or apartment	1086 (69.88%)	1442 (78.84%)
Geographic area of residence (n [%])
Metropolitan	795 (51.06%)	1174 (64.08%)	72.68	<0.001
Micropolitan	338 (21.71%)	228 (12.45%)
Small town	248 (15.93%)	257 (14.03%)
Rural	176 (11.3%)	173 (9.44%)
Moderate or complete hearing impairment (n [%])	218 (14.98%)	681 (40.68%)	251.05	<0.001
Moderate or complete visual impairment (n [%])	478 (33.13%)	1119 (66.85%)	352.70	<0.001
Diagnosis—physical disabilities (n [%])	1577 (100.00%)	501 (27.10%)	-	-
Diagnosis—intellectual or developmental disability (n [%])	51 (3.23%)	8 (0.43%)	6.28	<0.001
Diagnosis—Alzheimer’s disease or other dementia (n [%])	92 (5.83%)	213 (11.52%)	215.90	<0.001
Diagnosis—acquired brain injury or traumatic brain injury (n [%])	131 (8.31%)	31 (1.68%)	945.33	<0.001
Diagnosis—mental illness (n [%])	627 (39.76%)	673 (36.40%)	252.25	<0.001
Medicare (n [%])	1048 (66.46%)	1693 (96.58%)	517.37	<0.001

QoL Measurement

We conducted both exploratory and confirmatory factor analyses in order to develop QoL measures specific to the NCI-AD consumer survey. This process included (1) initial item screening, (2) an iterative process of domain identification based on an exploratory factor analysis and a comprehensive examination of item property, (3) domain confirmation based on a confirmatory factor analysis, and (4) psychometric property test of QoL domains. We performed the above-mentioned process separately for the older adult and younger adult subsamples, assuming the manifestation of QoL may differ for HCBS users of different age groups.

The 2015–2016 NCI-AD survey consisted of 88 items of which 36 items were initially selected for the subsequent factor analyses. Each item was measured using a 3-level categorical scale (0 = no, 1 = maybe, and 2 = yes). Items were excluded (see list in Supplemental Table 1) because they were (1) downstream items belonging to a branching or conditional logic (40 items such as “can you see or talk to your friends and family who do not live with you when you want to? answer only when the answer to the preceding question about family involvement is ‘yes’.”); (2) items that asked to check all applied options (12 items such as “what don’t you like about where you live? check all that apply”).

Identification of QoL domains based on selected items was an iterative process. An exploratory factor analysis (EFA) was conducted to examine the factor structure of selected items. The polychoric correlation matrix of selected items was used in the EFA to handle categorical variables. Oblique rotation was used as we assumed that factors would be correlated with one another. Based on EFA results, items were excluded if they cross-loaded (difference in factor loadings on any two factors <0.1) or had low factor loadings (factor loading on any factor <0.3). We also conducted a comprehensive examination on individual item properties to decide whether or not to retain an item, including item missing data analysis, examination of item discrimination ability, item reliability test (if adding the item will decrease Cronbach’s alpha of the domain), and discussion with stakeholders on content relevance of individual items and potential domain names. The examination of item properties, the discussion on item conceptual content, and EFA were conducted in an iterative manner until a final factor structure was agreed upon that had demonstrated content validity.

Next, a confirmatory factor analysis (CFA) was used to test data model fit. Model fit indices including root mean squared error of approximation (RMSEA) and comparative fit index (CFI) were examined. To examine psychometric properties of individual domains, we examined internal consistency of each QoL domain using Cronbach’s alpha and their concurrent validity by testing the association between each QoL domain and several global measures of well-being and satisfaction including sense of control, self-rated health, negative mood, and overall satisfaction with services. Additionally, we sought feedback from technical experts for variable selection and face validity, which included a stakeholder panel of representatives from Minnesota DHS and administrative officials from programs and facilities.

Sense of control, self-rated health, negative mood, and overall satisfaction with services were used to test the concurrent validity of identified domains of QoL. Sense of control was measured by asking participants “do you feel in control of life.” The responses were 0 = no, 1 = in-between, and 2 = yes. Self-rated health was measured by asking participants to rate their overall health based on a 5-level scale (1 = poor, 2 = fair, 3 = good, 4 = very good, and 5 = excellent). Negative mood was measured with the question “how often do you feel lonely, sad, or depressed” with responses being 0 = never or almost never, 1 = not often, 2 = sometimes, or 3 = often. Overall satisfaction with services was measured with question “Does the service you receive meet your needs.” The responses were 0 = no, not at all, needs or goals are not met, 1 = somewhat, some needs and goals, 2 = mostly, most needs and goals, or 3 = yes, completely, all needs and goals.

Descriptive and Regression Analyses

We constructed a composite score for each QoL domain by taking the arithmetic mean of the constituent items to make it more comparable across domains and age subsamples. Means and standard deviations were used to describe composite scores of each QoL domain. We conducted generalized ordinary least squares (OLS) regressions based on the distributions of the QoL scores in each domain, to examine association between QoL domain and demographic characteristics including age group, marital status, race/ethnicity, place of residence, geographic area of residents, and enrollment in Medicare. In the regression models, each QoL domain, stratified by age, was specified as the outcome separately. In the regression analysis, we applied sample weights to adjust for over-sampling or under-sampling of certain groups of HCBS recipients by race and waiver/program-type. The proportion of cases with missing values for at least one variable was 13%–16% across models for QOL domains. Under the assumption of missing at random, we used multiple imputation to handle missing values. Logistic regression was used to examine the correlates of high versus low levels of security, self-determination, and care experience. Considering the left-skewed distribution of the composite scores for each QoL domain, indicating that the majority of respondents reported relatively high values, we categorized the scores of each QoL domain using the 75th percentile as the threshold. Scores falling within the top three quartiles were labeled as “1,” indicating high QoL, while scores in the bottom quartile were marked as “0,” indicating low QoL. Analyses were conducted in Mplus Version 8 and Stata 15.0.

Results

Table 1 shows the sample characteristics of Minnesota NCI-AD respondents stratified by age. Sociodemographic and health-related characteristics differed significantly across the two age subsamples. Younger adults with disabilities were 46.45 (SD, ±13.05) years old, 52% were female, and 78% were White. Just under two-thirds of younger HCBS recipients were married or had a domestic partner, 70% lived in their own or family house or apartment, and more than half lived in metropolitan areas. Older HCBS recipients were on average 78.22 (SD, ±8.42) years old, 75% were female, and 75% were white. Approximately 16% of older adults were married or had a domestic partner, 79% lived in their own or family house or apartment, and 64% lived in metropolitan areas. Approximately 15% of younger HCBS recipients had hearing impairment, and 33% had visual impairment. In contrast, 40% of older HCBS recipients had hearing impairment, and 66% had visual impairment. Types of diagnoses significantly varied by age. All younger HCBS recipients had physical disabilities and about 8% had acquired brain injury (ABI) or traumatic brain injury (TBI), whereas less than one-third of older HCBS recipients were diagnosed with physical disabilities and 2% of HCBS recipients had ABI or TBI. The prevalence of Alzheimer’s disease or other dementia was 6% and 12%, respectively, among younger adults with disabilities and older adults. Nearly 40% of younger adults with disabilities and 36% of older adults had a mental illness. Most older adults were enrolled in Medicare (97%), while about two-thirds of younger adults with disabilities were enrolled in Medicare.

Latent QoL Domains

Older Adults.

Sixteen items were retained for older adults. Our factor analyses identified three latent factors. We labeled these three factors based on the NQF report, reflecting common person-centered QOL domains including security (5 items), care experience (4 items), and self-determination (7 items). We labeled the domains with reference to the NQF report and with input by stakeholders at the state to represent content validity (National Quality Forum, 2016). Table 2 shows item composition for each latent factor and factor loadings for each item. A CFA of this three-factor model demonstrated a good fit (RMSEA = 0.03; CFI = 0.93). Reliability was somewhat moderate for some of the factors (Cronbach’s alpha, 0.52–0.70). Evidence for concurrent validity was present in 13/15 statistically significant zero-order correlations between each QoL domain and global measures of well-being in the predicted directions. Each QoL domain was positively associated with sense of control, self-rated health, functional status, and overall satisfaction with services, suggesting concurrent validity. Likewise, each QoL domain was negatively related to negative mood. The exception was a positive correlation of care experience with self-rated health and functional status did not reach statistical significance (see Table 2). Moderate correlations (r = 0.25–0.31) were found for self-determination with sense of control and overall satisfaction with services and for care experience with overall satisfaction with services.

Table 2.

Results From Factor Analyses Identifying QOL Domains for Older Adult Respondents ≥65 Years of Age in the 2015–2016 National Core Indicators—Aging and Disability Survey From Minnesota (n = 1849).

		Security	Self-determination	Care experience
Item	Mean (SD)	Factor loading	Factor loading	Factor loading
Like where I am living right now	1.86 (0.45)	0.92	−0.07	−0.03
Prefer to live somewhere elsea	1.68 (0.69)	0.83	0.04	−0.08
Feel safe	1.92 (0.30)	0.76	−0.13	0.05
Worry about belongingsa	1.81 (0.51)	0.52	0.00	0.11
Money has been taken without permissiona	1.87 (0.47)	0.34	0.01	0.04
Like how I spend a day	1.67 (0.56)	0.26	0.32	0.02
Can access healthy food	1.85 (0.43)	0.20	0.38	0.04
Can eat meals when I want tob	1.81 (0.55)	−0.04	0.73	−0.09
People ask my permission before coming into my home/roomb	1.88 (0.40)	0.09	0.41	0.01
Can get up and go to bed at the time when I want tob	1.95 (0.26)	−0.14	0.76	−0.02
Have transportation going outside	1.74 (0.59)	0.00	0.56	0.09
Can do things I enjoy outside of my home	1.65 (0.66)	0.06	0.48	−0.07
Know whom to call when I have a complaint about the services	1.58 (0.76)	0.03	−0.17	0.85
Can choose or change types of services and determine how often and when to get them	1.64 (0.68)	−0.03	0.23	0.68
Know whom to call when I need different types of services	1.64 (0.70)	−0.03	−0.10	0.89
Can choose or change who provides my services	1.66 (0.69)	−0.05	0.30	0.60
Self-identified disabilitya	0.69 (0.95)	0.10	−0.04	−0.03
Need assistance with self-carea	1.15 (0.79)	−0.07	0.04	0.00
Need assistance in daily life activitiesa	0.70 (0.67)	−0.08	0.12	0.06
Domain	Security	Self-determination	Care experience
Number of items	5	7	4
Average score of domain items, mean (SD)	1.83 (0.31)	1.79 (0.26)	1.59 (0.57)
Reliability
Cronbach’s alpha	0.60	0.52	0.70
Concurrent validity
Correlation with sense of control	0.18*	0.27*	0.14*
Correlation with self-rated health	0.09*	0.11*	0.04
Correlation with functional status	0.05*	0.10*	0.05
Correlation with negative mood	−0.20*	−0.19*	−0.10*
Correlation with overall satisfaction with services	0.21*	0.29*	0.25*
Correlation with other QOL domains
Self-determination	0.27*	-	-
Care experience	0.17*	0.24*	-

^*p < .05.

^aitems with reversed score.

^bitems not included for the disability subsample.

Younger Adults.

Fifteen items were retained for younger adults with disabilities. Again, three latent factors were identified that had conceptual meanings aligned with the three factors for older adults: security, care experience, and self-determination. Factor composition was quite similar for the domains of security (5 items), care experience (4 items), and self-determination (6 items). Table 3 shows item composition for each latent factor and factor loadings for each item in younger adults with disabilities. A CFA of this three-factor model again demonstrated a good fit (RMSEA = 0.04; CFI = 0.93). Reliability was less variable with younger adults (Cronbach = .63–.67). Evidence for concurrent validity was present in 14/15 statistically significant zero-order correlations between each QoL domain and global measures of well-being in the predicted directions. Each QoL domain had a significant positive correlation with sense of control, self-rated health, functional status, and overall satisfaction with services, as well as a significant negative correlation with negative mood (see Table 3). One exception was that care experience was not statistically correlated with functional status. Self-determination had a moderate correlation with four global measures of well-being including sense of control, self-rated health, negative mood, and overall satisfaction with services (the absolute value of r = 0.33–0.48), and security likewise had a moderate correlation with sense of control, self-rated health, negative mood, and overall satisfaction with services (the absolute value of r = 0.26–0.30). Care experience was moderately correlated with overall satisfaction with services (r = 0.26). In our secondary analyses for external validation, we repeated CFA using MN NCI-AD 2017–2018 data. The reliability and validity metrics in these analyses were very similar to our primary analyses (supplemental table 2).

Table 3.

Results From Factor Analyses Identifying QOL Domains for Younger Consumers With Disabilities (<65 Years of Age) in the 2015–2016 National Core Indicators—Aging and Disability Survey From Minnesota (n = 1577).

		Security	Self-determination	Care experience
Item	Mean (SD)	Factor loading	Factor loading	Factor loading
Like where I am living right now	1.74 (0.61)	0.94	0.00	−0.05
Prefer to live somewhere elsea	1.43 (0.87)	0.93	−0.13	−0.07
Feel safe	1.86 (0.43)	0.68	0.09	0.06
Worry about belongingsa	1.70 (0.63)	0.48	0.05	0.12
Money has been taken without permissiona	1.79 (0.6)	0.33	0.09	0.01
Like how I spend a day	1.64 (0.6)	0.24	0.40	0.05
Can access healthy food	1.77 (0.51)	0.07	0.61	0.13
Have transportation going outside	1.73 (0.6)	−0.07	0.82	0.01
Can do things I enjoy outside of my home	1.67 (0.64)	−0.03	0.65	−0.05
Have transportation going to medical appointmentsb	1.95 (0.28)	−0.08	0.75	−0.05
Have to skip a meal due to financial constraintsa,b	1.81 (0.5)	0.11	0.53	0.01
Know whom to call when I have a complaint about the services	1.61 (0.74)	−0.04	0.00	0.74
Can choose or change types of services and determine how often and when to get them	1.57 (0.73)	0.04	0.06	0.71
Know whom to call when I need different types of services	1.59 (0.75)	−0.06	−0.04	0.83
Can choose or change who provides my services	1.68 (0.65)	−0.03	0.01	0.65
Self-identified disabilitya	0.56 (0.9)	−0.04	0.22	−0.07
Need assistance with self-carea	0.58 (0.61)	0.00	−0.19	0.11
Need assistance in daily life activitiesa	1.09 (0.8)	−0.04	−0.03	−0.03
Domain	Security	Self-determination	Care experience
Number of items	5	6	4
Average score of domain items, mean (SD)	1.70 (0.41)	1.76 (0.33)	1.60 (0.54)
Reliability
Cronbach’s alpha	0.63	0.66	0.67
Concurrent validity
Correlation with sense of control	0.26*	0.39*	0.23*
Correlation with self-rated health	0.13*	0.41*	0.07*
Correlation with functional status	0.05*	0.10*	0.05
Correlation with negative mood	−0.26*	−0.33*	−0.15*
Correlation with overall satisfaction with services	0.30*	0.48*	0.26*
Correlation with other QOL domains
Security	—	0.27*	—
Care experience	0.17*	0.24*	—

^*p < .05.

^aitems with reversed score.

^bitems not included for the aging subsample.

Associations of Demographic Characteristics With QoL Domains

The distributions of security, self-determination, and care experience were highly negatively skewed (skewness was between −1.89 and −1.37); therefore, we dichotomized the scores by the bottom quartile such that 1 = high QOL (scores in the top three quartiles) and 0 = low QOL (scores in the bottom quartile). Table 4 shows the regression results with sample weights being adjusted for and multiple imputations being used to handle missing data. Older adults were more likely to be in the top three quartiles of security (OR = 1.92, p < .001).

Table 4.

Regression Analysis: Associations Between Characteristics of Survey Respondents and QOL Domain Scores (n = 3426).

	Security (top three quartiles)a		Self-determination (top three quartiles)a		Care experiences (top three quartiles)a
	OR	95% CI	OR	95% CI	OR	95% CI
Older adults (vs younger with disabilities)	1.92***	(1.4, 2.63)	1.25	(0.91, 1.72)	1.06	(0.79, 1.43)
Female (vs male)	0.91	(0.73, 1.14)	0.72**	(0.57, 0.90)	1.03	(0.83, 1.28)
Race/ethnicity (Ref = White)
Black or African American	0.73*	(0.53, 0.98)	0.70*	(0.52, 0.96)	0.57***	(0.42, 0.76)
Asian	2.06*	(1.12, 3.79)	1.44	(0.89, 2.31)	0.32***	(0.22, 0.48)
Hispanic or Latino	0.71	(0.36, 1.38)	1.19	(0.55, 2.58)	0.39**	(0.22, 0.69)
Marital status (Ref = married or has domestic partner)
Single, never married	0.64**	(0.49, 0.84)	0.3***	(0.23, 0.4)	0.76*	(0.58, 0.99)
Separated or divorced	1.04	(0.73, 1.47)	0 49***	(0.35, 0.69)	0.94	(0.66, 1.33)
Widowed	1.01	(0.66, 1.56)	0.58**	(0.4, 0.86)	0.66*	(0.46, 0.94)
Geographic area of residence (Ref = Metropolitan)
Micropolitan	1.10	(0.81, 1.49)	1.36*	(1.00, 1.85)	1.00	(0.74, 1.36)
Small town	1.27	(0.90, 1.79)	1.46*	(1.04, 2.06)	0.82	(0.59, 1.14)
Rural	1.30	(0.88, 1.91)	1.31	(0.90, 1.91)	1.29	(0.87, 1.91)
Living in own or family house or apartment (vs living in group settings)	1.58***	(1.24, 2.01)	1.02	(0.79, 1.31)	1 62***	(1.28, 2.06)
Moderate or complete hearing impairment	1.13	(0.84, 1.51)	0.57***	(0.45, 0.74)	0.89	(0.69, 1.15)
Moderate or complete visual impairment	0.84	(0.66, 1.06)	0.83	(0.66, 1.05)	0.87	(0.69, 1.09)
Have Medicare	1.34*	(1.04, 1.72)	1.2	(0.92, 1.56)	1 59***	(1.23, 2.06)
Intercept	2 22***	(1.59, 3.1)	6.28***	(4.42, 8.92)	2.33***	(1.67, 3.25)

Note. Regression results based on multiple imputations. Sample weights were applied. OR, odds ratio; CI, confidence interval.

^*p < .05

^**p < .01

^***p < .001.

^aLogistic regression was used for security, self-determination, and care experience with 1 = high QOL (scores in the top three quartiles) and 0 = low QOL (scores in the bottom quartiles).

Consumers living in their own or family house or apartment versus those living in group settings (OR = 1.58, p < .001) and recipients with Medicare (OR = 1.34, p < .001) were more likely to report a high level of security. Compared to White recipients, Black recipients were less likely to report a high level of security (OR = 0.73, p < .05), whereas Asian recipients were more likely to report a high level of security (OR = 2.06, p < .05). Recipients who were single or never married were less likely to report a high level of security (OR = 0.64, p < .01) than those who were married or had a domestic partner.

Recipients living in micropolitan areas or small towns were more likely to report a high level of self-determination than those living in metropolitan areas (OR = 1.36, 1.46, respectively, p < .05). Self-determination differed by sex, race/ethnicity, and marital status. Female versus male recipients (OR = 0.72, p < .01), Black versus White recipients (OR = 0.70, p < .05), and recipients who were unmarried (including single or never married, separated or divorced, or widowed) versus who were married or had a domestic partner (OR = 0.30–0.58, p-values <0.01) were less likely to report a high level of self-determination. Moderate or complete hearing impairment was also significantly associated with a low level of self-determination (OR = 0.57, p < .001).

Compared to their White counterparts, Black, Asian, and Hispanic or Latino recipients were less likely to report better care experiences (OR = 0.32–0.57, p-values <0.01). Recipients living in their own or family house or apartment versus those living in group settings (OR = 1.62, p < .001) and recipients with Medicare (OR = 1.59, p < .001) were more likely to report a high level of care experiences. Recipients who were single or never married were less likely to report a high level of care experiences (OR = 0.76, p < .05) than those who were married or had a domestic partner.

For sensitivity check, we conducted logistic and linear regression analyses for each QOL domain using substituting listwise deletion for imputation to deal with missing data. The results, presented in supplementary Tables 3 and 4, indicated similar patterns of findings concerning the predictors of QoL domains.

Discussion

Community-based LTSS, including HCBS, are favored over institutional LTSS by both consumers and policymakers alike (Konetzka et al., 2020). Although HCBS is of increasing importance in the LTSS continuum, there is a lack of standardized measurement of self-reported QoL and a paucity of research describing variation in consumer-reported QoL (Chong et al., 2022). To help fill the gap in this space, we leveraged data from the 2015–2016 NCI-AD survey in Minnesota to develop measures of self-reported QoL among consumers of publicly funded HCBS. Furthermore, we examined sociodemographic factors associated with self-reported QoL and found several differences related to race/ethnicity, sex, and partnered status. Black consumers were less likely to report high levels of security, self-determination, and care experiences compared to White consumers, as were consumers who were single/never married versus those who were married or partnered. Further work is needed to identify and measure contributing factors—such as systemic racism or availability of family/social support—and how they influence care experience and outcomes.

We identified three domains of QOL including self-determination, security, and care experience from the NCI-AD consumer survey for both younger adults with disabilities and older adults who received publicly funded HCBS in Minnesota. The importance of these domains has been previously described in detail in the 2016 NQF consensus report on HCBS quality (National Quality Forum, 2016). Our analyses were informed by the Response Shift Theory, which suggests differences in perception of QoL between younger and older adults (Duan et al., 2021; Idler et al., 2018; Sprangers & Schwartz, 1999; Spuling et al., 2015). Contrary to our expectations, item composition was similar between the younger and older adult subsamples for the domains of security and care experience. Similarity between these factors is surprising given the use of age-based population parameters to define many LTSS programs and is a finding that may signal the importance of evaluating populations primarily with respect to functional needs and chronological age secondarily. More nuanced exploration of commonalities and differences between younger and older adult populations of HCBS consumers in describing QoL is warranted, especially with respect to state-level differences in program and service definitions.

Slight differences in indicators between the aging and disabilities subsamples were observed for the domain of self-determination. While autonomy in choosing time for meals or sleep and allowing others entering rooms was reflective of self-determination for older adults, having transportation for medical appointments and meal security (not having to skip a meal due to financial constraints) were significant indicators of control and self-determination for younger adults with disabilities. The factor structure and concurrent validity of the three QOL domains were supported with acceptable CFA model fit and significant correlations with global measures of well-being. However, the three QOL measures identified had low-to-moderate internal consistency. This may partially be since the NCI-AD custom survey was not initially developed to measure discrete latent constructs of QoL such that conceptual relatedness among items and empirical inter-item correlations should be key considerations in the item selection phase (Tavakol & Dennick, 2011).

Both older adults and younger adults with disabilities reported a high level of QoL with regards to security, self-determination, and care experience. Yet, we identified systematic differences by sociodemographic factors (gender, race/ethnicity, living arrangement, and Medicare enrollment) for the three QoL domains, including self-determination, security, and care experience. These three domains essentially reflected critical aspects of HCBS quality concerning individual psychosocial needs. It’s important to acknowledge that adults receiving HCBS support generally have lower functional capacity and socioeconomic status than non-institutionalized, community-dwelling adult populations due to restrictive eligibility requirements (Khatutsky et al., 2006). However, White older adults in the United States regularly report their health more positively than Black, Asian, and Hispanic older adults, a finding that persists even after accounting for social, demographic, health, and socioeconomic factors (Shippee et al., 2020). Previous studies of community-dwelling Black and Asian older adults have shown that these groups regularly report a lower sense of control (Fabius et al., 2019; Shippee et al., 2020). However, prior research using the NCI-AD dataset showed that Black older adults exhibited a higher sense of control when White older adults receiving publicly funded HCBS, when controlling for functional status (Chong et al., 2022; Shippee et al., 2020).

In the past few decades, the consumer population of publicly funded HCBS has become increasingly heterogeneous both, demographically and in medical and social needs (AHRQ, 2012), leading to the expansion of HCBS services. However, expansion efforts made so far may not adequately accommodate the increasing diversity of the HCBS consumer population (Konetzka et al., 2020). Our study leverages a unique consumer-reported data source to report several inequities in HCBS consumers’ self-reported QoL. These findings highlight the necessity of directing future efforts toward equitable HCBS expansion to meet the needs of all consumer subgroups, especially historically marginalized groups (AHRQ, 2012; Konetzka et al., 2020; Shippee et al., 2022). This work also presents several policy recommendations. First, policies should ensure equitable access to HCBS services for all demographic groups, with a particular focus on those historically marginalized. This may involve targeted outreach programs, promoting culturally sensitive service delivery, and reducing eligibility barriers. Second, funding for diversity training for HCBS providers is needed to ensure they can deliver culturally sensitive care that meets the needs of diverse populations. Finally, policies should aim to improve national data collection and monitoring across states and service waivers, aligning with recommendations from the Zubritsky et al. (2013) model that emphasizes the role of individual, program, and other environmental factors for quality improvement.

Limitations

This study leveraged data from the NCI-AD consumer survey in one state, which may limit the generalizability of findings because (1) not all states have the same type of HCBS programs or programs for non-Medicaid residents, (2) Minnesota is primarily a rural state with predominantly White older adult population of Medicaid HCBS users, and (3) environment may differ between states—isolation might be different in the winter in Minnesota than in states like Texas or Florida. Our study uses survey data that is inherently vulnerable to recall bias and selection bias. The items used in this study are systematically collected in the NCI-AD survey but are not systematically collected in other surveys of HCBS or collected by states to evaluate HCBS consumer-reported QoL. For pay for performance to work in HCBS, these consumer-reported QoL measures will need to be systematically collected for all HCBS recipients. Another limitation was our focus on responses from HCBS users while excluding 13.2% of cases where proxy respondents, such as family members or close friends, provided information on behalf of service recipients unable to respond themselves. This exclusion was due to differences in the proxy version of the NCI-AD survey, which lacks some items included in the consumer version. Focusing solely on HCBS users’ responses limited the generalizability of our findings, particularly for those with severe cognitive impairments who cannot personally report their care experiences. However, this approach emphasizes the importance of person-reported outcomes.

Future research needs to explore the comparative reliability and quality of reports between direct service HCBS users and proxy respondents. If significant discrepancies are found, it will highlight the need for survey designers to either adapt existing instruments to accommodate respondents with varied cognitive abilities or develop proxy questionnaires that more accurately reflect the experiences of service users. Additionally, future studies should address the identified limitations through ongoing national data collection efforts.

Conclusion

In a sample of HCBS consumers in Minnesota, we leveraged the NCI-AD survey data to identify latent measures of HCBS consumers’ self-reported QoL and evaluated inequities in QoL. In factor analyses of 36 self-reported survey items, we identified three latent QoL domains, including security, care experience, and self-determination. Further, we evaluated correlates of these three distinct QoL domains using multivariable regression analyses. We found that younger consumers with disabilities (compared to older adults ≥65 years of age), minoritized racial/ethnic groups, females, consumers with hearing loss, not having Medicare, not having a spouse/domestic partner, and not living in one’s own/family home reported significantly lower scores in various QoL domains. These findings highlight several disparities in HCBS consumer-reported QoL and suggest the need for equitable reforms to improve HCBS quality for its increasingly diversified consumer base.

Data Availability Statement

The data use agreement between University of Minnesota and the Health Services Research Institute (HSRI) does not allow sharing of NCI-AD data with the general public.