Skip to main content
NCBI home page
The Gerontologist logoLink to The Gerontologist
. 2025 Jun 17;65(9):gnaf156. doi: 10.1093/geront/gnaf156

Caregiving network precarity among community-living, dually enrolled persons with dementia

Julia G Burgdorf 1,✉, David Russell 2, Chanee D Fabius 3, Katherine E M Miller 4, Jennifer M Reckrey 5
PMCID: PMC12453545  PMID: 40574658

Abstract

Background and Objectives

Most persons with dementia live in the community, relying on caregiving networks to meet their needs. Caregiving network precarity refers to insecurity or instability in these networks, defined as caregiver(s) being unable or unwilling to continue in their role. Given the importance of caregiving to aging in place and the high burden associated with dementia caregiving, we sought to identify factors associated with caregiving network precarity for community-living, dually enrolled persons with dementia.

Research Design and Methods

We linked 2021–2022 administrative, claims, and clinical assessment (baseline and follow-up) data for a diverse sample of community-living, dually enrolled persons with dementia in New York State. We operationalized caregiving network precarity as having one or more caregivers report being unable or unwilling to continue providing care. We modelled incident caregiving network precarity as a function of care recipient characteristics, health services utilization, and caregiving network factors using multivariable logistic regression.

Results

Declines in health status among care recipients were strongly associated with caregiving network precarity. Individuals were more likely to experience caregiving network precarity if they had recently experienced an increase in cognitive impairment (aOR: 2.98; 95% CI: 1.97–4.51), functional impairment (aOR: 1.71; 95% CI: 1.07–2.74), or bowel incontinence frequency (aOR: 2.33; 95% CI: 1.38–3.93), or began resisting care (aOR: 2.47; 95% CI: 1.69–3.61).

Discussion and Implications

Findings highlight the importance of identifying and addressing shifts in care recipient status and offering targeted supports to caregivers during key inflection points in the care recipient’s disease trajectory.

Keywords: caregiver, dementia, Alzheimer disease, Medicare, Medicaid

Background and objectives

The majority (70%) of older persons with dementia in the United States live in the community (Alzheimer’s Association, 2024). This reflects a commonly held preference for aging in place but is also associated with heightened risk for unmet care needs due to the difficulty of accessing affordable, home-based clinical care and supportive services (Black et al., 2019; National Academies of Sciences, Engineering, and Medicine [NASEM], 2021). Family and unpaid caregivers (hereafter, “caregivers”) provide the majority of care for persons with dementia. Caregivers assist with a range of tasks including co-managing health care and finances, helping with self-care activities (e.g., bathing, dressing, eating), and performing household chores (e.g., making meals, cleaning) (Assistant Secretary for Planning and Evaluation, 2018; Kasper et al., 2015; Spillman et al., 2020). Although much of the existing caregiving literature focuses on a single “primary” caregiver, persons with dementia are more likely to rely on caregiving networks—groups of individuals who help meet their needs (Spillman et al., 2020). Yet, little is known about how these caregiving networks change over time.

Caring for someone with dementia can be simultaneously a uniquely rewarding and highly challenging experience. Caregivers may reap benefits including becoming closer with the care recipient, developing new skills, and feeling pride in their ability to deliver care (Quinn & Toms, 2019). However, caregivers for persons with dementia are at high risk of experiencing role-related strain and negative consequences of caregiving due to the significant physical and emotional demands of providing care (Adelman et al., 2014; Kasper et al., 2015; NASEM, 2021; Riffin et al., 2019). This risk is exacerbated by a lack of structural supports—e.g., access to caregiving training and support resources, respite care, and affordability of paid personal care—to mitigate these stressors (Burgdorf et al., 2025; Johnson & Wang, 2019; Wolff et al., 2016). Additionally, these demands unfold over a long and often unpredictable disease trajectory. Persons with dementia require more intense care for longer time periods, compared to those without dementia (Spillman et al., 2020). Among a national sample of caregivers for older adults, caregiving-related strain was highest among those caring for a community-living person with dementia (Kasper et al., 2015). Given the differential caregiving-related strain among caregivers of persons with dementia, it is important to understand willingness of these caregivers to continue to provide care.

Caregivers face additional complexity when caring for a ­person who is dually enrolled in Medicare and Medicaid. (In the United States, Medicare is a government-sponsored health insurance for older adults and Medicaid is government-­sponsored health insurance for low-income individuals.) Compared to Medicare-only enrollees, dual enrollees have fewer financial resources, higher underlying care needs and social vulnerability, and are at greater risk for adverse outcomes (ATI Advisory, 2022; Gorges et al., 2019). Medicaid coverage for long-term care delivered in the home varies by state and can provide access to a critical additional source of care, but it often introduces added responsibilities for caregivers. Especially in the context of dementia, caregivers may need to take on new tasks including navigating insurance coverage details, finding and managing paid aides, and negotiating task-sharing with paid care providers (Fabius et al., 2024; Reckrey et al., 2022b).

Researchers have applied the concept of precarity to contexts in which older adults and their caregivers experience insecurity and/or uncertainty in trying to marshal resources to meet growing care needs during the aging process, especially when exacerbated by “failing social and economic networks of support” (Butler, 2009; Grenier et al., 2020). Precarity is particularly relevant for persons with dementia, who commonly experience variable and often unpredictable symptom progression and difficulties in securing adequate resources to meet resultant growing care needs (e.g., hiring paid aides). These difficulties are often heightened within the context of lifelong, structural disadvantages and social vulnerabilities such as economic insecurity and other characteristics associated with being dually enrolled in Medicaid (Grenier et al., 2017; Grenier et al., 2020; Hillman et al., 2023; Portacolone, 2020). Precarity, which refers to a state of uncertainty, is distinct from the related concept of care poverty that identifies a “deprivation of adequate coverage of care needs” for an older adult (Kröger, 2022, p. 26). While care poverty may be an endpoint following a stage of precarity, the two will not always coincide if individuals and/or their social and caregiving networks are able to address and mitigate the causes of precarity.

Notably, precarity may be experienced by older adults/care recipients, caregivers, and/or caregiving networks. For example, the care recipient may struggle to find community-based services early in the disease trajectory, a caregiver may have difficulty balancing caregiving alongside other family/job responsibilities, and a caregiving network may have to supplement paid caregiving with an over-reliance on family members who have not received necessary training. Precarity in the caregiving networks of persons with dementia may be reflected in changes in network size, component members, or functions and tasks of the networks. However, such changes may alternatively reflect beneficial network reorganization to meet changing needs. Caregiving network precarity may more meaningfully be operationalized as instances in which caregiver(s) indicate that they cannot or will not continue in their caregiving role.

Preliminary evidence describes the dynamic nature of caregiving networks for persons with dementia, indicating that the number of caregivers changes over time (Spillman et al., 2020). Prior work not limited to those with dementia has identified individual caregiver-level factors associated with cessation of caregiving, including male sex, more distal relationship to the care recipient, and symptoms of depression (Allen et al., 2012; Wang et al., 2022). However, to our knowledge, no previous study investigates how care recipient-level factors (e.g., changes in health or function) or structural factors (e.g., Medicaid policy and program benefits governing intensity and type of paid care) may impact caregiving network precarity for persons with dementia. The present study leverages longitudinal 2021–2022 linked health plan, Medicaid claims, and clinical assessment data for a diverse sample of community-living, dually enrolled persons with dementia to identify factors associated with caregiving network precarity. By focusing on dually enrolled individuals who are receiving personal care services through the New York State Medicaid long-term care program, we are able to analyze a population with greater homogeneity in regards to personal care access, and to measure the intensity of paid/formal personal care provided (not often possible when considering privately funded personal care). We examine incident caregiving network precarity and consider its associations with the care recipient’s baseline characteristics and changes in their status over time. We also investigate relationships between use of other sources of care (including Medicaid-funded personal care) and caregiving network precarity. Findings are relevant to ongoing efforts to better meet the care needs of community-­living persons with dementia and to enhance supports for their caregivers.

Research design and methods

Conceptual framework

We adopted two conceptual models to guide our work: the Convoys of Care and Stress Process models (Kemp et al., 2013; Pearlin et al., 1990), which together allow us to recognize the integral role of caregivers within the social networks of persons with dementia as well as see caregiving network precarity as a response by caregivers to external stressors and supports. The Convoys of Care model illustrates the fact that high-need individuals, such as those with dementia, tend to rely upon a dynamic and evolving network of both paid and unpaid caregivers working together to provide support (Kemp et al., 2013). In particular, Convoys of Care emphasizes the intersecting impacts of formal and “informal” sources of care. This framing underscores the importance of accounting for use of other sources of care when examining caregiving network precarity among dual enrollees; in our sample, Medicaid-funded personal care is especially relevant. The Stress Process model illustrates the interactive and multifactorial process by which individual caregivers experience stressors and use social and personal resources to cope. Following this model, caregiving network precarity can be viewed as a response to overlapping sources of caregiving-related stress without sufficient resources for coping. These may include primary stressors related to changes in the needs of care recipients (e.g., assistance with toileting and bathing) as well as changes in their behaviors (e.g., resistance to care) as caregivers may have limited available supportive resources (e.g., skills training or respite care) to help them manage those changes (Pearlin et al., 1990).

Data and sample

This study draws on administrative data from a large Managed Long-Term Care (MLTC) plan in New York State, with linked Uniform Assessment System (UAS) clinical assessments and Medicaid claims from 2021 to 2022. MLTC plans operate under contract with the state Medicaid program to manage Medicaid-covered long-term care services for their clients (New York State Department of Health (NYSDOH), 2024). Health plan administrative data include information on ­clients’ sociodemographic characteristics and enrollment status. The UAS is a standardized clinical assessment instrument for New York State Medicaid enrollees receiving home- and community-based long-term care, based on the interRAI Home Care Assessment tool (NYSDOH, 2019). Assessments are completed during home visits by a Registered Nurse or Licensed Social Worker with questions posed to the Medicaid enrollee and their caregiver(s), and include information on health and function, living environment, cognition, social supports, and behavioral symptoms. Medicaid claims include information on types and intensity of Medicaid-funded care. We focus on measuring use of personal care given its relevance to family caregiver experiences and task demands (Fabius et al., 2024; Reckrey et al., 2022b).

We identified MLTC plan enrollees in 2021 or 2022 who were dually enrolled in Medicare and Medicaid, were 65 years or older, had at least two clinical assessments with a gap between assessments of 90–425 days, had non-missing data on caregiver ability/willingness to continue caregiving, and had a dementia diagnosis reported in either Medicaid claims or on the UAS assessment (n = 3,880). To determine dementia status using Medicaid claims, we applied the logic of the Medicare Chronic Conditions Warehouse indicator for Alzheimer’s Disease or Senile Dementia, identifying claims with a matching ICD-10 code over a 3-year lookback period (Centers for Medicare & Medicaid Services, 2024a). All individuals were “full” Medicaid enrollees, meaning eligible for the full suite of benefits, including personal care, available under New York State Medicaid. This, along with the fact that hours of personal care are determined by the Medicaid program via algorithm based on clinical assessment results, leads to greater homogeneity among our sample in terms of access to paid/formal personal care.

We then identified the most recent pair of UAS assessments (a “baseline” and a “follow-up” assessment) in the study timeframe for each individual. Assessments are required annually but may be performed more frequently in response to changing health status.

The primary objective of this work is to identify factors associated with incident caregiving network precarity, in an effort to recognize key periods of vulnerability in care trajectories that may be ripe for intervention. Given our interest in incident caregiving network precarity, we limited the analytic sample to those at risk for this outcome by excluding those with no caregiver reported at baseline (20.5%; n = 794) or whose caregiver(s) reported being unable or unwilling to continue providing care at baseline (6.3%; n = 245). The final analytic sample included 2,841 community-living, dually enrolled persons with dementia receiving caregiver assistance. Follow-up assessments for these individuals included reports on 4,161 individual caregivers. The mean time between baseline and follow-up assessment was 254 days (SD =75 days).

Measures

Outcomes

Incident caregiving network precarity is determined by a response of “yes” to one of two statements on the follow-up assessment: “Informal helper is unable to continue caring activities” or “Informal helper is unwilling to assist with care”. We combined these questions as they are conceptually interrelated, both measure a single underlying construct of network precarity. Additionally, as many family caregivers are motivated by cultural/societal values and expectation, there is stigma related to being “unwilling” to continue providing care (Zarzycki et al., 2023). Therefore, caregivers may be more willing to report that they are “unable” to continue caring, especially when discussing this issue in the context of a clinical assessment with the care recipient present. The UAS captures information on up to two caregivers for each health plan member; we considered a member to experience caregiving network precarity if one or both caregiver(s) answered “yes” to one of the statements above.

Baseline care recipient explanatory variables

From the baseline assessment, we drew measures of care recipient sociodemographic characteristics (age, sex, race/ethnicity), whether they lived alone, and clinical characteristics (levels of cognitive and functional impairment, frequency of bowel incontinence, and number of chronic conditions). Cognitive impairment was operationalized as a categorical variable denoting low, moderate, or high impairment based on assessor reports of whether the individual exhibited independence or minimal impairment, moderate impairment, or severe impairment, respectively. Functional impairment refers to a summary score of 10 activities of daily living (ADL) impairments scored by assessors on a scale from 0 (independence) to 6 (total dependence). Following an approach reported by Reckrey et al (2024), we calculate the mean score across these 10 items for an overall score of 0–6, with higher scores denoting greater impairment. Frequent bowel incontinence refers to assessor reports of daily incontinence with some control present or total incontinence with no control present.

Dynamic care recipient explanatory variables (measured between baseline and follow-up)

From the baseline and follow-up assessments, we construct measures capturing changes in care recipient status. These include increases in functional impairment, cognitive impairment, or bowel incontinence frequency, hospitalization, and beginning to resist care. Increased functional impairment is defined as an increase of one full point or more on the ADL impairment scale between assessments. Increased cognitive impairment is defined as any of the following transitions from baseline to follow-up: from “low” to “moderate” impairment, from “low” to “high” impairment, or from “moderate” to “high” impairment. Increased bowel incontinence refers to a shift from having infrequent/no incontinence at baseline to frequent incontinence at follow-up. Hospitalization refers to a report of hospitalization within 90 days preceding the follow-up assessment. Beginning to resist care is defined as a care recipient with “resists care” noted as a behavioral symptom on the follow-up assessment but not on the baseline assessment.

From Medicaid claims, we measured indicators of structural access to personal care, including intensity and type of personal care between baseline and follow-up assessments. Intensity refers to hours of Medicaid-funded personal care received per week. Type refers to whether personal care was delivered by an agency-employed or consumer-directed aide. Consumer-directed aides are designated by the Medicaid enrollee or someone acting on their behalf and then paid to deliver Medicaid-covered personal care services to the enrollee (NYSDOH, 2025). In New York, Medicaid enrollees who qualify for personal care can choose between agency-employed or consumer-directed aide services with few eligibility restrictions.

Caregiver explanatory variables

Limited caregiver-level information is included in the UAS. However, for each of the 4,161 caregivers who assisted one of the 2,841 persons with dementia in our sample, we characterized their relationship to the care recipient, whether they lived with the care recipient, and whether they helped with ADLs and/or instrumental activities of daily living (IADLs). We categorized relationship to the care recipient as child/child-in-law, spouse, or other.

Statistical analyses

We present descriptive statistics, including frequencies and proportions, means and SD, for the sample overall and by the presence or absence of incident caregiver network precarity. We compared caregiver-level characteristics by whether that caregiver reported being unable/unwilling to continue providing care on the baseline assessment. We tested for between-group differences using Pearson’s x2 and t-tests.

We estimated adjusted odds of incident network precarity using multivariable Generalized Linear Models with a logit link, binomial family, and an exposure term to account for time under observation (days between baseline and follow-up assessments). Model 1 adjusts for baseline indicators only. Model 2 adds dynamic indicators capturing change from baseline to follow-up. This study was approved by the VNS Health Institutional Review Board.

We completed two sensitivity analyses. First, given that less than 10% of our sample experienced incident network precarity, we fit modified Poisson models with robust error variance to estimate relative risk ratios (Zou, 2004). Second, we ran models with continuous rather than binary indicators of change in functional impairment, cognitive impairment, and bowel incontinence frequency.

Results

Nearly one-third (27%) of community-living, dually enrolled persons with dementia were experiencing potentially insufficient caregiver support at baseline, due to having no reported caregiver or a caregiver who reported being unable/unwilling to continue caregiving (Figure 1). Among the analytic sample of 2,841 who were potentially at risk of incident caregiving network precarity (i.e., those who were not experiencing potentially insufficient caregiver support at baseline), 5.5% (n = 157) experienced incident caregiving network precarity during the observation period (Table 1). The overall mean age of persons with dementia was 83.8 years (SD = 8.4), most (78.0%) were female, 28.4% were Non-Hispanic White, 32.4% were Hispanic, 18.7% were Non-Hispanic Black, and 12.8% were Asian. Between baseline and follow-up, one-third (32.1%) experienced increased cognitive impairment and 11.9% experienced increased functional impairment. Half (53.0%) had only one caregiver reported at baseline.

Figure 1.

Pie chart showing that 27% of community-living persons with dementia experience insufficient caregiver support at baseline.Pie chart showing that 27% of community-living persons with dementia experience insufficient caregiver support at baseline.

Open in a new tab

Potentially insufficient caregiver support at baseline assessment, among community-living, dually enrolled persons with dementia (n = 3,880).

Table 1.

Characteristics of community-living, dually enrolled persons with dementia (n = 2,841).

Characteristics Full sample (n = 2,841) Experiencing caregiving network precarity
No (n = 2,684; 94.5%) Yes (n = 157; 5.5%) p-value
Mean (standard deviation) or frequency (%)
Baseline indicators
Age 83.8 (8.4) 83.9 (8.4) 83.1 (7.9) 0.25
Female sex 2,216 (78.0%) 2,109 (78.6%) 107 (68.2%) 0.002
Race/ethnicity
Non-Hispanic White 808 (28.4%) 757 (28.2%) 51 (32.5%)
Non-Hispanic Black 920 (32.4%) 869 (32.4%) 51 (32.5%)
Hispanic 530 (18.7%) 502 (18.7%) 28 (17.8%) 0.74
Asian 363 (12.8%) 347 (12.9%) 16 (10.2%)
Other 220 (7.7%) 209 (7.8%) 11 (7.0%)
Lives alone 1,073 (37.8%) 1,020 (38.0%) 53 (33.8%) 0.29
Cognitive impairment
Low 1,269 (44.7%) 1,202 (44.8%) 67 (42.7%)
Moderate 1,171 (41.2%) 1,106 (41.2%) 65 (41.4%) 0.19
High 401 (14.1%) 376 (14.0%) 25 (15.9%)
Functional impairmenta 4.0 (1.0) 4.0 (1.0) 3.9 (1.0) 0.70
Number of chronic conditions 2.1 (1.4) 2.1 (1.4) 2.1 (1.4) 0.75
Frequent bowel incontinence 718 (25.3%) 675 (25.2%) 43 (27.4%) 0.53
More than one caregiver 1,123 (39.5%) 1,035 (38.6%) 88 (56.1%) <.001
Dynamic indicators (measured from baseline to follow-up)
Increased cognitive impairment 913 (32.1%) 825 (30.7%) 88 (56.1%) <.001
Increased functional impairment 339 (11.9%) 296 (11.0%) 43 (27.4%) <0.001
Increased bowel incontinence frequency 243 (8.6%) 210 (7.8%) 33 (21.0%) <.001
Hospitalization 398 (14.0%) 364 (13.6%) 34 (21.7%) 0.005
Begins resisting care 431 (15.2%) 381 (14.2%) 50 (31.9%) <.001
Hours of family caregiving (weekly) 26.6 (24.0) 26.6 (23.9) 27.5 (24.7) 0.67
Use of consumer-directed personal care 1,246 (43.9%) 1,180 (44.0%) 66 (41.0%) 0.64
Hours of personal care (weekly) 54.9 (34.4) 55.2 (34.7) 49.2 (27.5) 0.03
Open in a new tab

Note. Linked Medicaid managed long-term care health plan data, Medicaid claims, and assessment data from 2021 to 2022.

a

Functional impairment measured as composite score of activities of daily living impairments (0–6).

Compared to those who did not experience caregiving network precarity, a greater proportion of those who did had more than one caregiver reported on their initial assessment (56.1% v 38.6%; p < .001) and had greater cognitive impairment (56.1% v 30.7%; p < .001), functional impairment (27.4% v 11.0%; p < .001), and bowel incontinence frequency (21.0% v 7.8%; p < .001) (Table 1). Those who experienced network precarity received fewer hours of Medicaid-funded personal care per week, on average (27.5 v 34.7; p = .03).

A greater proportion of caregivers who reported being unable or unwilling to continue providing care were spouses of the care recipient (15.9% v 6.5%; p < .001) and did not help with ADLs (34.7% v 16.7%; p < .001) (Figure 2). There were no significant differences by living with the care recipient or helping with IADLs. Most (79.6%) caregivers were children assisting a parent.

Figure 2.

Bar chart showing differences in characteristics of caregivers who are willing/able to continue in their role compared to those who are not. Characteristics presented include relationship to care recipient, living with care recipient, helping with Activities of Daily Living (ADL), and helping with instrumental ADL.Bar chart showing differences in characteristics of caregivers who are willing/able to continue in their role compared to those who are not. Characteristics presented include relationship to care recipient, living with care recipient, helping with Activities of Daily Living (ADL), and helping with instrumental ADL.

Open in a new tab

Characteristics of caregivers for community-living, dually enrolled persons with dementia, by caregiver ability/willingness to continue caregiving (n = 4,161).

In a first multivariable model adjusting only for baseline indicators, females had half the odds of experiencing network precarity compared to males (adjusted odds ratio (aOR): 0.60; 95% Confidence Interval (95% CI): 0.42–0.87), and those with more than one caregiver reported on the assessment had two times greater odds of experiencing network precarity (aOR: 2.04; 95% CI: 1.47–2.84) (Table 2).

Table 2.

Adjusted odds of incident caregiving network precarity among community-living, dually enrolled persons with dementia (n = 2,841).

Characteristics Model 1: Baseline indicators only
 Model 2: Baseline & dynamic indicators
aORb (95% CIc) p-Value aOR (95% CI) p-Value
Baseline indicators
Age 0.99 (0.97–1.01) .56 0.98 (0.96–1.01) .15
Female sex 0.60 (0.42–0.87) .004 0.57 (0.39–0.83) .003
Race/ethnicity
Non-Hispanic White REF REF
Non-Hispanic Black 0.88 (0.59–1.33) .56 1.04 (0.68–1.59) .86
Hispanic 0.88 (0.54–1.43) .60 1.00 (0.61–1.66) .99
Asian 0.59 (0.33–1.06) .08 0.68 (0.37–1.24) .21
Other 0.69 (0.35–1.36) .29 0.77 (0.38–1.56) .48
Lives alone 0.97 (0.68–1.38) .84 1.08 (0.74–1.59) .69
Cognitive impairment
Low REF REF
Moderate 1.10 (0.75–1.62) .64 1.38 (0.90–2.11) .14
High 1.38 (0.74–2.54) .31 0.95 (0.50–1.81) .88
Functional impairmenta 0.87 (0.69–1.09) .22 0.93 (0.71–1.22) .62
Number of chronic conditions 1.02 (0.91–1.14) .73 1.02 (0.91–1.15) .74
Frequent bowel incontinence 1.17 (0.73–1.89) .52 1.45 (0.86–2.44) .16
More than one caregiver 2.04 (1.47–2.84) <.001 2.33 (1.65–3.29) <.001
Dynamic indicators (measured from baseline to follow-up)
Increased cognitive impairment 2.98 (1.97–4.51) <.001
Increased functional impairment 1.71 (1.07–2.74) .03
Increased bowel incontinence frequency 2.33 (1.38–3.93) .002
Hospitalization 1.59 (1.03–2.46) .04
Begins resisting care 2.47 (1.69–3.61) <.001
Hours of family caregiving (weekly) 0.99 (0.99–1.00) .30
Use of consumer-directed personal care 0.79 (0.55–1.13) .20
Hours of personal care (weekly) 0.99 (0.98–0.99) .01
Open in a new tab

Note. Linked Medicaid managed long-term care health plan data, Medicaid claims, and assessment data from 2021 to 2022. Generalized Linear Models with logit link and exposure term adjusting for days between baseline and follow-up assessments.

a

Functional impairment measured as composite score of activities of daily living impairments (0–6).

b

Adjusted odds ratio.

c

95% confidence interval.

A second model adding dynamic indicators capturing change between baseline and follow-up reveals that declines in health status were strongly associated with greater odds of caregiving network precarity. Persons with dementia were also more likely to experience caregiving network precarity if they had increased cognitive impairment (aOR: 2.98; 95% CI: 1.97–4.51), increased functional impairment (aOR: 1.71; 95% CI: 1.07–2.74), increased bowel incontinence frequency (aOR: 2.33; 95% CI: 1.38–3.93), had been hospitalized (aOR: 1.59; 95% CI: 1.03–2.46), or began resisting care (aOR: 2.47; 95% CI: 1.69–3.61). In addition, each additional hour/week of paid personal care was associated with lower odds of network precarity (aOR: 0.98; 95% CI: 0.98–0.99).

Sensitivity analyses

Findings were robust to a change in model specification (fitting modified Poisson models with robust error variance) with no changes observed in significance or direction (see online supplementary material, Table S1). In models using continuous, rather than binary, dynamic indicators of changes in cognition, function, and frequency of bowel incontinence, hospitalization was no longer statistically significant and increased functional impairment showed a stronger relationship with network precarity (aOR: 1.74; 95% CI: 1.23–2.45) (see online supplementary material, Table S2).

Discussion and implications

Among a diverse sample of community-living persons with dementia dually enrolled in Medicare and Medicaid, 5.5% experienced incident caregiving network precarity, as evidenced by one or more caregivers reporting being unable or unwilling to continue providing care. Changes in care recipient health and behavior during the observation period—including increases in cognitive impairment, increases in frequency of bowel incontinence, and beginning to resist care—were strongly associated with incident caregiving network precarity. Findings highlight the importance of identifying and addressing shifts in care recipient status and offering targeted supports to caregivers during these key inflection points.

Our finding that caregivers may be especially likely to step away from their roles when care recipients experience a deterioration in health and function is consistent with the Stress Process Model. The model posits that increased care needs represent primary stressors that can cause secondary role strains which may ultimately lead to yielding of the caregiving role (Pearlin et al., 1990). In addition to increased role-related stress, this finding could also be related to caregiver’s own physical limitations and health conditions that may make meeting emerging care needs difficult or impossible. This is supported by the greater proportion of spousal caregivers who were unable/unwilling to continue providing care. Spousal caregivers are older and thus more likely to be experiencing health issues and functional limitations of their own, perhaps necessitating that they relinquish their caregiving role as care recipient needs intensify.

Findings also suggest the important ways that behavioral symptoms of dementia (ie, resisting care) may impact caregivers’ ability and willingness to provide care. Managing behavioral symptoms of dementia is well-established as a significant and consequential source of role-related stress for caregivers (Ornstein & Gaugler, 2012). These symptoms are consistently associated with nursing home placement across multiple studies and are the target of a wide range of interventions to support persons with dementia living in the community (Gaugler et al., 2009; Toot et al., 2017). Future qualitative research should explore the underlying reasons why caregivers of community-living persons with dementia may feel unable/unwilling to continue during times of heightened care needs and how specific aspects of the dementia caregiving experience and underlying personal relationships impact these reasons. For example, the observed relationship between female sex and lower odds of caregiving network precarity could be a function of mothers maintaining closer bonds with their adult children, particularly amid rising divorce rates, or potentially a reflection of gender roles and expectations related to providing intimate care. These issues deserve further examination.

Finally, models also indicate that receiving greater hours of paid personal care was associated with lower odds of caregiving network precarity. This highlights how policy structures shape receipt of paid/formal care and potentially impact larger caregiving networks for community-living persons with dementia. Further, these findings support a conceptualization of these networks based on the Convoys of Care model. Existing evidence confirms that paid personal care is a particularly important source of support for persons with dementia, particularly in later disease stages, and demonstrates that family/unpaid and paid caregivers frequently collaborate closely to ensure the complex care needs of persons with dementia are met at home (Fabius et al., 2024; Reckrey et al., 2022a; Reckrey et al., 2024). Future work should specifically investigate ways to ­foster greater relational coordination (Bolton et al., 2021) and mutual support between paid and unpaid/family caregivers and examine how these relationship dynamics influence caregiving network precarity.

Given that persons with dementia appear vulnerable to caregiving network precarity during periods of decline, there is a need to: (1) identify these inflection points and (2) take the opportunity to assess and address areas of need (including need for additional paid personal care or targeted family caregiver supports) within the caregiving network. Routine assessments are an important way to monitor for changes in need, function, or symptoms, but assessments must be sufficiently frequent to detect changes before negative consequences occur. Specific care settings or health events that commonly accompany shifts in health and function should prompt additional assessments of care needs—these events may include a fall at home, hospitalization, or referral to home health care. Leveraging expanded data mining techniques to analyze real-time clinical data and proactively identify those at greatest risk for caregiver stress and/or caregiving network precarity may also be helpful (Burgdorf et al., 2021; Wolff et al., 2023). In addition, more needs to be done to promote proactive communication between dementia caregivers and the broader care team; meaningful challenges to caregiver/provider information sharing have been found in primary care, hospital, home health, and personal care (Burgdorf et al., 2023; Burgdorf et al., 2021; Fabius et al., 2023; Russell et al., 2021; Wolff et al., 2020).

Once declines in health and function are identified, those enrolled in health plans may benefit from mobilization of care management supports and community resource linkages to better support caregivers and facilitate caregiving network stability (Skaria, 2022). There is growing awareness of the need to design policies and interventions to better serve community-living persons with dementia and their families among researchers, advocates, and policymakers (Alzheimer’s Association, 2023; Centers for Medicare & Medicaid Services, 2024b). These types of support may be especially impactful in the context of the dually enrolled population given their overlapping clinical and social needs (Sorbero et al., 2018). Dually enrolled persons with dementia may also be able to access critical, Medicaid-funded paid personal care to help bolster faltering care networks.

The current study’s focus on older adults who are dually enrolled in Medicare and Medicaid, and receiving personal care through the New York State Medicaid benefit, strengthens ­internal validity but limits our ability to examine variability in larger structural factors that may be critical influences on caregiving network stability. These include federal and state policies regarding coverage of health care and personal care services, labor market forces affecting the supply of personal care workers, and social trends shaping access to family caregivers such as changes in family structure—e.g., declining birth rates and/or increasing divorce rates. In the case of low-income older adults (e.g., those qualifying for Medicaid), economic vulnerability among individuals and their families may contribute to further marginalization and/or precarity in caregiving networks (Grenier et al., 2017). For example, limited access to paid leave or diminished savings due to reducing work hours in order to provide care. All of these stressors may contribute to vulnerability in caregiving arrangements; our findings regarding Medicaid-funded paid care suggest the value of state policies that aim to directly support these family networks in times of need.

It is also critical to consider the structural supports, or lack thereof, for family caregivers. Caregivers face many entrenched barriers to accessing role-related support and education, including misaligned financial incentives for providers, lack of integrated caregiver assessment tools at the point of care, and difficulty identifying existing services (Macleod et al., 2017; Wolff et al., 2016; Wolff et al., 2020). A recent study of a nationally representative sample of caregivers for community-living persons with dementia found low levels of supportive service use (including 11% receiving any role-related training and 18% using respite care) and indicated that access to support does not vary with caregiver stress-level (Burgdorf et al., 2025). It is notable that, within the initial sample of 3,880 community-living persons with dementia, nearly one-third (27%) were excluded from the analytic sample due to experiencing potentially insufficient caregiver support at baseline (i.e., no reported caregiver or a caregiver reporting being unable/unwilling to continue). While our primary focus was understanding periods of heightened jeopardy due to incident caregiving network precarity, this figure illustrates the even more widespread vulnerability experienced by many of this population, who may not have the social resources (in the form of family caregivers) available to fully meet their care needs. This may stem, in part, from the relative lack of systematic supports for those in a caregiving role. Better supporting dementia caregivers, particularly as they are faced with increasing or changing demands related to their caregiving role, is a critical component to facilitating stable and effective dementia care for community-living persons with dementia.

Limitations

Notable strengths of this study include use of a large and racially/ethnically diverse sample, examination of both baseline and dynamic factors potentially related to caregiving network precarity, and inclusion of information on both family caregiving and Medicaid-funded personal care. Limitations include lack of information on why a caregiver reported being unable/unwilling to continue providing care, inability to track caregivers longitudinally in this sample (thus, inability to determine whether caregiver(s) reported at follow-up were the same individual(s) reported at baseline), and limited observation periods. Our analytic sample focused exclusively on dually enrolled older adults in New York state, potentially limiting generalizability to other populations and precluding us from examining variability in structural factors (e.g., access to paid care) that may play a pivotal role in caregiving network precarity for the broader population. We did not have information on other potentially relevant sociodemographic characteristics such as gender or sexual orientation. Additionally, the UAS includes reports on a maximum of two caregivers, so may overlook changes in caregiving network size above this threshold. Greater information on the number of caregivers within the older adult’s caregiving network, their relationships to the older adult, and their caregiving tasks, would allow for a more in-depth exploration of structural network factors.

Conclusions

Among a diverse sample of community-living, dually enrolled persons with dementia, declines in health and functional status were strongly associated with greater odds of caregiving network precarity (defined as one or more caregivers reporting being unable or unwilling to continue providing care). Findings highlight the importance of offering targeted supports to caregivers during key inflection points in the care recipient’s disease trajectory.

Supplementary Material

gnaf156_Supplementary_Data
gnaf156_supplementary_data.zip (28KB, zip)

Contributor Information

Julia G Burgdorf, Center for Home Care Policy & Research, VNS Health, New York, New York, United States.

David Russell, Center for Home Care Policy & Research, VNS Health, New York, New York, United States.

Chanee D Fabius, Department of Health Policy & Management, Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland, United States.

Katherine E M Miller, Department of Health Policy & Management, Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland, United States.

Jennifer M Reckrey, Icahn School of Medicine at Mt Sinai, New York, New York, United States.

Supplementary material

Supplementary material is available online at The Gerontologist.

Data availability

Data are not publicly available as they were obtained through agreement with a partner health insurance plan and Data Use Agreement with the Centers for Medicare & Medicaid Services. Those interested in the statistical code used to clean the data, create measures, or perform statistical analyses described in this manuscript may contact the corresponding author.

Funding

This work was supported by funding from the Alzheimer’s Association (AARG-23-1074651) and the National Institute on Aging (K01AG081502 to J.G.B.).

Conflicts of interest: None declared.

References

  1. Adelman R. D., Tmanova L. L., Delgado D., Dion S., Lachs M. S. (2014). Caregiver burden: a clinical review. JAMA, 311, 1052–1060. 10.1001/jama.2014.304 [DOI] [PubMed] [Google Scholar]
  2. Allen S. M., Lima J. C., Goldscheider F. K., Roy J. (2012). Primary caregiver characteristics and transitions in community-based care. The Journals of Gerontology. Series B, Psychological Sciences and Social Sciences, 67, 362–371. 10.1093/geronb/gbs032 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Alzheimer’s Association (2023). March). House reintroduces critical legislation to improve dementia care. https://www.alz.org/news/2023/house-comprehensive-care-act. Accessed December 10, 2024.
  4. Alzheimer’s Association (2024). 2024 Alzheimer’s disease facts and figures.https://www.alz.org/media/Documents/alzheimers-facts-and-figures.pdf. Accessed July 25, 2024.
  5. Assistant Secretary for Planning and Evaluation (2018). A profile of older adults with dementia and their caregivers. https://aspe.hhs.gov/sites/default/files/migrated_legacy_files//186646/DemChartIB.pdf. Accessed December 10, 2024.
  6. ATI Advisory (2022). A profile of Medicare-Medicaid dual beneficiaries. https://atiadvisory.com/wp-content/uploads/2022/06/A-Profile-of-Medicare-Medicaid-Dual-Beneficiaries.pdf. Accessed December 10, 2024.
  7. Black B. S., Johnston D., Leoutsakos J., Reuland M., Kelly J., Amjad H., Davis K., Willink A., Sloan D., Lyketsos C., Samus Q. M.(2019). Unmet needs in community-living persons with dementia are common, often non-medical and related to patient and caregiver characteristics. International Psychogeriatrics, 31, 1643–1654. 10.1017/S1041610218002296 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Bolton R., Logan C., Gittell J. H. (2021). Revisiting relational coordination: A systematic review. The Journal of Applied Behavioral Science, 57, 290–322. 10.1177/0021886321991597 [DOI] [Google Scholar]
  9. Burgdorf J. G., Fabius C. D., Riffin C., Wolff J. L. (2021). Receipt of posthospitalization care training among medicare beneficiaries’ family caregivers. JAMA Network Open, 4, e211806. 10.1001/jamanetworkopen.2021.1806 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Burgdorf J. G., Reckrey J., Russell D. (2023). “Care for me, too”: A novel framework for improved communication and support between dementia caregivers and the home health care team. Gerontologist, 63, 874–886. 10.1093/geront/gnac165 [DOI] [PubMed] [Google Scholar]
  11. Burgdorf J. G., Freedman V., Wolff J. (2025). A look at caregivers for community-living persons with dementia: Implications for the GUIDE model. Alzheimer’s & Dementia: the journal of the Alzheimer’s Association, 21, e70013.   10.1002/alz.70013 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Butler J. (2009). Frames of War: When is Live Grievable?. Verso. [Google Scholar]
  13. Centers for Medicare & Medicaid Services (2024. a). Chronic conditions data warehouse condition categories. https://www2.ccwdata.org/web/guest/condition-categories.Accessed December 10, 2024.
  14. Centers for Medicare & Medicaid Services (2024. b). Guiding an Improved Dementia Experience (GUIDE) model. https://www.cms.gov/priorities/innovation/innovation-models/guide. Accessed December 10, 2024.
  15. Fabius C. D., Wec A., Saylor M. A., Smith J. M., Gallo J. J., Wolff J. L. (2023). “Caregiving is teamwork…” Information sharing in home care for older adults with disabilities living in the community. Geriatric Nursing (New York, N.Y.), 54, 171–177. 10.1016/j.gerinurse.2023.09.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Fabius C. D., Gallo J. J., Burgdorf J., Samus Q. M., Skehan M., Stockwell I., Wolff J. L. (2024). Family care partners and paid caregivers: National estimates of role-sharing in home care. Gerontologist, 65, gnae177. 10.1093/geront/gnae177 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Gaugler J. E., Yu F., Krichbaum K., Wyman J. F. (2009). Predictors of nursing home admission for persons with dementia. Medical Care, 47, 191–198. 10.1097/MLR.0b013e31818457ce [DOI] [PubMed] [Google Scholar]
  18. Gorges R. J., Sanghavi P., Konetzka R. T. (2019). A national examination of long-term care setting, outcomes, and disparities among elderly dual eligibles. Health Affairs (Project Hope), 38, 1110–1118. 10.1377/hlthaff.2018.05409 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Grenier A., Lloyd L., Phillipson C. (2017). Precarity in late life: Rethinking dementia as a ‘frailed’ old age. Sociology of Health & Illness, 39, 318–330. 10.1111/1467-9566.12476 [DOI] [PubMed] [Google Scholar]
  20. Grenier A., Hatzifilalithis S., Laliberte-Rudman D., Kobayashi K., Marier P., Phillipson C. (2020). Precarity and aging: A scoping review. Gerontologist, 60, 620–632. 10.1093/geront/gnz135 [DOI] [PubMed] [Google Scholar]
  21. Hillman A., Jones I. R., Quinn C., Pentecost C., Stapley S., Charlwood C., Clare L. (2023). The precariousness of living with, and caring for people with, dementia: Insights from the IDEAL programme. Social Science & Medicine (1982), 331, 116098. 10.1016/j.socscimed.2023.116098 [DOI] [PubMed] [Google Scholar]
  22. Johnson R. W., Wang C. X. (2019). The financial burden of paid home care on older adults: Oldest and sickest are least likely to have enough income. Health Affairs (Project Hope), 38, 994–1002. 10.1377/hlthaff.2019.00025 [DOI] [PubMed] [Google Scholar]
  23. Kasper J. D., Freedman V. A., Spillman B. C., Wolff J. L. (2015). The disproportionate impact of dementia on family and unpaid caregiving to older adults. Health Affairs (Project Hope), 34, 1642–1649. 10.1377/hlthaff.2015.0536 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Kemp C. L., Ball M. M., Perkins M. M. (2013). Convoys of care: Theorizing intersections of formal and informal care. Journal of Aging Studies, 27, 15–29. 10.1016/j.jaging.2012.10.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Kröger T. (2022). Care poverty: When older people’s needs remain unmet. Palgrave Macmillan Cham. 10.1007/978-3-030-97243-1 [DOI] [Google Scholar]
  26. Macleod A., Tatangelo G., McCabe M., You E. (2017). “There isn’t an easy way of finding the help that’s available.” Barriers and facilitators of service use among dementia family caregivers: a qualitative study. International Psychogeriatrics, 29, 765–776. 10.1017/S1041610216002532 [DOI] [PubMed] [Google Scholar]
  27. National Academies of Sciences, Engineering, and Medicine; Health and Medicine Division (2021). Meeting the challenge of caring for persons living with dementia and their care partners and caregivers: A way forward. C Stroud, EB, Larson Eds.; Board on Health Care Services; Board on Health Sciences Policy; Committee on Care Interventions for Individuals with Dementia and Their Caregivers. National Academies Press. [PubMed]
  28. New York State Department of Health (NYSDOH). (2019). UAS-NY Reporting. https://www.health.ny.gov/health_care/medicaid/redesign/uniform_assessment_system/archives/uasny_reporting02.htm Accessed January 27, 2025.
  29. New York State Department of Health (NYSDOH). (2024). Managed Long-Term Care (MLTC). https://www.health.ny.gov/health_care/managed_care/mltc/ Accessed January 27, 2025.
  30. New York State Department of Health (NYSDOH). (2025). Consumer Directed Personal Assistance Program (CDPAP). https://www.health.ny.gov/health_care/medicaid/program/longterm/cdpap.htm Accessed January 27, 2025.
  31. Ornstein K., Gaugler J. E. (2012). The problem with “problem behaviors”: A systematic review of the association between individual patient behavioral and psychological symptoms and caregiver depression and burden within the dementia patient-caregiver dyad. International Psychogeriatrics, 24, 1536–1552. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Pearlin L. I., Mullan J. T., Semple S. J., Skaff M. M. (1990). Caregiving and the stress process: An overview of concepts and their measures. Gerontologist, 30, 583–594. 10.1093/geront/30.5.583 [DOI] [PubMed] [Google Scholar]
  33. Portacolone E. (2020). A framework to identify precarity in the social sciences: Insights from qualitative research. In A.Grenier, C. Phillipson, & R. A. Settersten (Eds). Precarity and ageing (147–166). Policy Press. [Google Scholar]
  34. Quinn C., Toms G. (2019). Influence of Positive Aspects of Dementia Caregiving on caregivers’ well-being: A systematic review. Gerontologist, 59, e584–e596. 10.1093/geront/gny168 [DOI] [PubMed] [Google Scholar]
  35. Reckrey J. M., Li L., Zhan S., Wolff J., Yee C., Ornstein K. A. (2022. a). Caring together: Trajectories of paid and family caregiving support to those living in the community with dementia. The Journals of Gerontology. Series B, Psychological Sciences and Social Sciences, 77, S11–S20. 10.1093/geronb/gbac006 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Reckrey J. M., Watman D., Tsui E. K., Franzosa E., Perez S., Fabius C. D., Ornstein K. A. (2022. b). “I am the home care agency”: The dementia family caregiver experience managing paid care in the home. International Journal of Environmental Research and Public Health, 19, 1311. 10.3390/ijerph19031311 [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Reckrey J. M., Russell D., Fong M. C., Burgdorf J. G., Franzosa E. C., Travers J. L., Ornstein K. A. (2024). Home care worker continuity in home-based long-term care: Associated factors and relationships with client health and well-being. Innovation in Aging, 8, igae024. 10.1093/geroni/igae024 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Riffin C., Van Ness P. H., Wolff J. L., Fried T. (2019). Multifactorial examination of caregiver burden in a national sample of family and unpaid caregivers. Journal of the American Geriatrics Society, 67, 277–283. 10.1111/jgs.15664 [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Russell D., Burgdorf J. G., Kramer C., Chase J. D. (2021). Family caregivers’ conceptions of trust in home health care providers. Research in Gerontological Nursing, 14, 200–210. 10.3928/19404921-20210526-01 [DOI] [PubMed] [Google Scholar]
  40. Skaria A. P. (2022). The economic and societal burden of Alzheimer disease: managed care considerations. The American Journal of Managed Care, 28, S188–S196. 10.37765/ajmc.2022.89236 [DOI] [PubMed] [Google Scholar]
  41. Sorbero M., Kranz A., Bouskill K., Ross R., Palimaru A., Meyer A. (2018). Addressing social determinants of health needs of dually enrolled beneficiaries in medicare advantage plans. https://aspe.hhs.gov/sites/default/files/private/pdf/259896/MAStudy_Phase2_RR2634-final.pdf Accessed December 10, 2024.
  42. Spillman B. C., Freedman V. A., Kasper J. D., Wolff J. L. (2020). Change over time in caregiving networks for older adults with and without dementia. The Journals of Gerontology. Series B, Psychological Sciences and Social Sciences, 75, 1563–1572. 10.1093/geronb/gbz065 [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Toot S., Swinson T., Devine M., Challis D., Orrell M. (2017). Causes of nursing home placement for older people with dementia: a systematic review and meta-analysis. International Psychogeriatrics, 29, 195–208. 10.1017/S1041610216001654 [DOI] [PubMed] [Google Scholar]
  44. Wang S., de Almeida Mello J., Declercq A. (2022). Factors associated with informal caregiver’s ability to continue care: A subgroup analysis. Age and Ageing, 51, afac275. 10.1093/ageing/afac275 [DOI] [PubMed] [Google Scholar]
  45. Wolff J. L., Feder J., Schulz R. (2016). Supporting family caregivers of older Americans. The New England Journal of Medicine, 375, 2513–2515. 10.1056/NEJMp1612351 [DOI] [PubMed] [Google Scholar]
  46. Wolff J. L., Freedman V. A., Mulcahy J. F., Kasper J. D. (2020). Family caregivers’ experiences with health care workers in the care of older adults with activity limitations. JAMA Network Open, 3, e1919866. 10.1001/jamanetworkopen.2019.19866 [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Wolff J. L., DesRoches C. M., Amjad H., Burgdorf J. G., Caffrey M., Fabius C. D., Gleason K. T., Green A. R., Lin C. T., Nothelle S. K., Peereboom D., Powell D. S., Riffin C. A., Lum H. D. (2023). Catalyzing dementia care through the learning health system and consumer health information technology. Alzheimer’s & Dementia: The Journal of the Alzheimer’s Association, 19, 2197–2207. 10.1002/alz.12918 [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Zarzycki M., Morrison V., Bei E., Seddon D. (2023). Cultural and societal motivations for being informal caregivers: A qualitative systematic review and meta-synthesis. Health Psychology Review, 17, 247–276. 10.1080/17437199.2022.2032259 [DOI] [PubMed] [Google Scholar]
  49. Zou G. (2004). A modified Poisson regression approach to prospective studies with binary data. American Journal of Epidemiology, 159, 702–706. 10.1093/aje/kwh090 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

gnaf156_Supplementary_Data
gnaf156_supplementary_data.zip (28KB, zip)

Data Availability Statement

Data are not publicly available as they were obtained through agreement with a partner health insurance plan and Data Use Agreement with the Centers for Medicare & Medicaid Services. Those interested in the statistical code used to clean the data, create measures, or perform statistical analyses described in this manuscript may contact the corresponding author.

Articles from The Gerontologist are provided here courtesy of Oxford University Press

RESOURCES