Shifting care patterns: How Medicaid policies shape family and formal care use

Abstract

INTRODUCTION

Medicaid programs can support aging in place by paying for care provided by direct care workers (i.e., formal care) and providing financial support, education, and training to otherwise unpaid family members and friends (i.e., family care).

METHODS

We used nationally representative data of adults 70 years and older with functional limitation(s)—and the subsamples with dementia, enrolled in Medicaid, or rural residents—and multinomial regressions to estimate the relationship between Medicaid waivers paying family caregivers and the types of care received.

RESULTS

Adjusting for demographics and family structure, living in a state with waiver payments for family caregivers decreased the probability of receiving formal care alone and increased the probability of receiving family care among persons living with dementia, Medicaid beneficiaries, and rural residents.

DISCUSSION

Understanding how policies incentivize caregiving has implications for broader discussions to (1) address workforce supply concerns and (2) mitigate economic consequences of family caregiving.

Highlights

• Medicaid Home and Community‐Based Services (HCBS) play a critical role in addressing the care needs of community‐dwelling older adults.

• Medicaid programs paying family members for care reduced use of formal care.

• The relationship was strongest among persons with dementia, Medicaid, and rural residents.

• These findings have implications for workforce shortages and supports for caregivers.

1. BACKGROUND

Growing numbers of older persons are aging in place due, in part, to the aging of the US society and policy changes reflecting preferences to age in the community. As the population ages, an increasing share of older adults with limitations on daily activities rely on unpaid family members or friends (i.e., family caregivers). Nearly 75% of older adults with limitations living in the community rely on family caregivers and this share is growing over time. ¹ Historically, spouses and daughters provided care for aging loved ones. However, studies point to demographic changes, such as reduced fertility, delayed retirement, and increases in stepfamilies, as driving changes in who provides care and how much care is provided. ² Caregiving causes financial burden on working caregivers, including wage penalties, as well as negative physical and mental health effects, which compound over time. ³ , ⁴ , ⁵ , ⁶ , ⁷ , ⁸ Consequently, in 2019, the Centers for Disease Control and Prevention (CDC) declared family caregiving to be a public health issue. ⁹ Some negative consequences of caregiving may be mitigated through support like financial payments and education; evaluations of some caregiver interventions revealed supportive services can reduce caregiver burden and improve mental and physical health. ¹⁰ , ¹¹ , ¹²

However, the consequences of caregiving can vary given the heterogeneity in who becomes a caregiver, and the type and amount of care provided. ¹³ Persons in rural areas are more likely to receive care by family members than their urban counterparts. ¹⁴ , ¹⁵ For persons living with dementia (PLWD), caregiving needs are especially high. As the disease progresses, PLWD require extensive care for daily activities, an estimated 83% of which comes from family, friends or other unpaid caregivers, many of whom (25%) are also caring for a child and/or have low incomes (41%). ¹⁶ In addition, caregiver burden and negative health consequences have been shown to be higher among dementia caregivers than other caregivers. ⁶ , ¹⁷ , ¹⁸ Understanding the impact policies and interventions have on disparities in providing care and caregiver outcomes is particularly important.

Programs that support aging in place by providing financial payments, education, training to family caregivers, and access to trained, professional caregivers (i.e., direct care workers) may increase the likelihood that older persons receive care in the home or community. Yet, no national long‐term care insurance system exists in the United States. Rather, most home‐ and community‐based services (Home and Community‐Based Services [HCBS]) are paid for out‐of‐pocket or through Medicaid HCBS waivers. For older adults and individuals under 65 with disabilities who meet income‐based eligibility requirements, Medicaid provides coverage and benefits that Medicare alone does not, including long‐term care. ¹⁹ , ²⁰ Historically, Medicaid covered care in institutional settings exclusively. As a result of the Olmstead Decision in 1999 and myriad state and federal policy initiatives beginning in the 1980s to shift care to the home and community, coverage for HCBS has expanded over the last few decades. ²⁰ , ²¹ , ²² States can provide HCBS through waivers [e.g., 1915(c), 1115] or through amendments to the state plan. ²³ Medicaid HCBS is primarily funded by 1915(c) waivers (hereafter referred to as waivers), which allow states flexibility to experiment with innovative models, such as paying family caregivers for providing care. This is particularly notable given the challenges of ensuring an adequate supply of direct care workers to support adults with functional limitations living in the community, particularly in rural areas and to patients with dementia who have more substantial care needs. ²⁴ However, the impact of access to services through these waivers on whether eligible individuals receive care, or the types of care, is unknown.

We used National Health and Aging Trends Study (NHATS) data to estimate the association of the availability of payments to family members through Medicaid 1915(c) waivers on whether older adults with functional limitation(s) receive formal or family help. We investigated differential associations between Medicaid waivers providing payments to family caregivers and type of care received by dementia status, enrollment in Medicaid, and rurality. Collectively, these findings elucidate how Medicaid policies can shape patterns of care to help older adults age in place.

RESEARCH IN CONTEXT

1. Systematic review: The authors reviewed the existing literature using traditional sources, primarily Google Scholar. Several studies have explored the relationship between family structure and caregiving and described use of Medicaid 1915(c) waivers and their use over time. These relevant articles have been cited.

2. Interpretation: Our findings indicate access to Medicaid 1915(c) home‐ and community‐based waivers offering payment to spousal and adult children caregivers of older adults decreased reliance on formal caregivers, particularly among persons living with dementia, Medicaid beneficiaries, and rural residents.

3. Future directions: Future studies should collect data on Medicaid waivers 1915(j)(k) and states to generate a comprehensive understanding of all available home‐ and community‐based services potentially providing financial support to family caregivers. Understanding the effect of these services on the health and wellbeing of persons living with dementia and their caregivers is needed.

2. METHODS

2.1. Data and sample

To complete our objectives, we used three data sources. First, we used data from nine waves of the NHATS (2011–2019). ²⁵ NHATS is an annual, panel survey of a nationally representative sample of Medicare beneficiaries providing longitudinal information on demographic characteristics, family structure, self‐reported health status/conditions, functional status, cognitive functioning, economic status, receipt of caregiving from both formal and family caregivers, and state of residence of respondent. Among NHATS respondents, we identified a sample of community‐dwelling respondents who are 70+ who reported needing help with at least one instrumental activity of daily living (IADLs, e.g., shopping, medication management) or activities of daily living (ADLs, e.g., bathing, eating, toileting) in each wave.

Second, we created a state‐year level dataset to capture which states allowed for payments to relatives and legally responsible persons (hereafter referred to as family caregivers) through Medicaid 1915(c) waivers for personal care services. To create this dataset, we used current and historical Medicaid 1915(c) waiver applications, renewals and amendments, providing care to adults ages 65 or older between 2010 and 2019 acquired from the Centers for Medicare & Medicaid Services (CMS) website and collected by Skira et al. ²⁶ From these waiver applications (n = 431), we extracted the indicator of whether payments may be made to relatives. We then used approval and termination dates of waivers as reported by each state through the 372 Form, that is, required reporting for 1915(c) waivers, to indicate when a state had an active waiver allowing payments to caregivers as this may vary over time within state. See Appendix and Table A1 for additional information on the data that was extracted. We reported results from the data extraction and construction process in a recently published study. ²⁷ Briefly, we found that between 2010 and 2019, waivers providing payments to family members increased from 66.7% to 73.4% of all state waivers.

Third, we augmented our state‐year dataset to include additional policies supporting family caregivers and older adults included in the American Association of Retired Persons (AARP) State Scorecard (e.g., paid family leave) and state‐level funding for the National Family Caregiver Support Program through the Older Americans Act which was collected from the Administration for Community Living's website. ²⁸ , ²⁹ , ³⁰

Finally, we linked the state‐year dataset indicating allowance of payments to caregivers with the state of residence and corresponding year from the NHATS data. Our final sample included 19,087 respondent‐year observations.

2.2. Outcomes

The primary outcome was the receipt of help with ADLs or IADLs with or without payments. Among our sample of individuals with any ADL or IADL care needs, we constructed a mutually exclusive, categorical outcome indicating whether a respondent received: (1) no care, (2) only family care, (3) only formal care, or (4) both family and formal care. Family care includes care provided by family members and non‐related helpers (e.g., friends, neighbors) whether they are paid or not. Formal care includes any care provided by direct care workers, housekeepers, aides, or other paid helpers.

2.3. Key measures

The primary predictor of interest is a measure of waiver availability. Specifically, we identified whether residents resided in a state with 1915(c) waiver(s) allowing payments to family caregivers by examining whether payments to legally responsible individuals and family members for personal care services or other waiver services were allowed in the waiver. We examined the eligibility criteria of family caregivers to receive payment for each waiver document technically allowing payments to legally responsible individuals and/or family members and denoted waivers as effectively excluding spouses if the waiver: (1) explicitly excluded spouses or (2) allowed payments to spouses only in the case of “extraordinary care” or “extraordinary circumstances.” We included waivers with these restrictions because, while spouses can only be paid in extraordinary circumstances/care under 1915(c) regulations, states often concurrently run 1915(j) waivers which allow for payments to spouses. Thus, this approach was more conservative and underestimates the potential number of people impacted by payments to family members as some of the included waivers may exclude certain family caregivers from payment (e.g., spouses). Each of the three authors independently verified the criteria of each waiver to ensure consistent identification of payments to family caregivers.

2.3.1. Dementia

We defined possible or probable dementia using an algorithm created by NHATS investigators. ³¹ The measure is based on three main sources: (1) report by NHATS respondent or proxy of a doctor telling them they had dementia or Alzheimer's disease, (2) a score of probable dementia on the AD8 Dementia Screening Interview of proxy respondents, and (3) a series of cognitive tests that evaluate NHATS respondents’ memory, orientation, and executive function.

2.3.2. Additional explanatory variables

Individual‐level characteristics included age, sex, race, education level, Medicaid enrollment, income above 300% of the federal poverty level (FPL), rural residence, and number of functional limitations. We further included partnered status and whether the respondent has daughters to account for traditional sources of family care. We also included time‐varying, state‐level policies which could also influence reliance on formal versus family caregiving: exceeding federal standards of Family Medical Leave Act, mandatory paid family leave, and mandatory paid sick leave. Finally, we included the state‐year specific allocation of Older Americans Act funding for the National Family Caregiver Support Program.

2.4. Statistical analysis

Our primary analysis leveraged multinomial logistic regression models to estimate the relationship between living in a state with a waiver allowing payments to family caregivers and whether the respondents received family care only, formal care only, or both formal and family care, relative to no care received. Our main model was a multinomial logistic regression including our key explanatory and additional explanatory variables described above and year fixed effects (see Equation 1). In Equation 1, Y_ij indicates the categorical outcome of interest and j indicates the category of the outcome of interest (e.g., no care, formal care, etc.) and $x_i^{\prime }$ is a vector of case‐specific covariates (e.g., age, sex). We conducted the Small‐Hsiao test to check for violations of the Independence of Irrelevant Alternatives assumption and found no evidence of violation.

$$Y_{ij}=\frac{\exp \left(x_i^{\prime }{\beta }_j\right)}{{\sum }_{l=1}^m\exp \left(x_i^{\prime }{\beta }_l\right)}\mathrm{where}j=1,\cdots ,m$$ (1)

Because the reliance on family caregivers and the consequences of caregiving can vary across dementia status of the care‐recipient, income, and rurality, we conducted analyses to examine differential associations of residing in a state with 1915(c) waiver(s) allowing for payments to family caregivers among subsamples of (1) people living with dementia, (2) enrolled in Medicaid, and (3) rural residents. As a falsification test, we conducted a negative control analysis by examining outcomes only among the sample not enrolled in Medicaid. ³² If our results are driven by the 1915(c) waivers, there should be no effect for the subsample of persons who are not enrolled in Medicaid.

For all models, we used survey weights to yield nationally representative estimates and corrected standard errors reflecting the complex sampling frame. We calculated marginal/differential effects using the method of recycled predictions which allows us to interpret findings as a percentage point change in the predicted probability of the outcome occurring. Among the subsamples of Medicaid enrollees and rural respondents, we tested for differential effects among those with and without dementia using an adjusted Wald test. Because we were not using a causal inference study design, we hereafter refer to marginal/differential effects as marginal/differential associations. Full model results are available upon request.

3. RESULTS

3.1. Sample description

Table 1 displays sample characteristics of our sample for the first year observed in our sample for the full sample and, separately, for the subsamples with dementia, on Medicaid, or living in rural settings. On average, the full sample was majority female (58.3%) and non‐Hispanic White individuals (75.5%). Approximately 16% were on Medicaid and 50% lived in a state with waiver(s) paying family caregivers. Compared to the full sample, persons with dementia were generally older (54.1% ages 80 and older vs. 34.4%) and less likely to be non‐Hispanic White individuals (62.4% vs. 75.5%). PLWD were also more likely to be on Medicaid (28.3% vs. 15.8%). Compared to the full sample, Medicaid enrollees were less likely to be non‐Hispanic White individuals (40.5% vs. 75.5%) and partnered (32.8% vs. 54.7%). They were also less educated—54.2% had less than a high school education, compared to 22.3% in the full sample. Medicaid enrollees had higher rates of dementia than the full sample (42.7% vs 23.8%). Persons living in rural settings in our sample were more likely than the full sample to be non‐Hispanic White individuals (85.8% vs. 75.5%) and were less likely to have more than a high school education (39.3% vs. 48.7%).

TABLE 1.

Sample description at baseline: Full sample, dementia, Medicaid, rural only.

Parameter	Full sample	Dementia	Medicaid	Rural
Outcome—care type (%)
No care	24.8	15.0	17.7	25.0
Family care only	62.7	63.7	56.2	63.0
Paid care only	3.7	4.8	8.7	3.6
Both family and paid care	8.8	16.5	17.4	8.4
Predictor—waivers with payments (%)	50.2	48.5	40.0	75.7
Subgroups
Dementia	23.8	100.0	42.7	22.1
Medicaid (%)	15.8	28.3	100.0	14.5
Rural (%)	19.1	17.8	17.5	100.0
Other covariates
Age group (%)
70–74	43.0	24.9	46.5	42.3
75–79	22.6	21.1	23.0	23.4
80–84	17.3	23	15.8	18.5
85–89	11.7	19.7	9.9	10.8
90+	5.4	11.4	5.0	4.5
Female (%)	58.3	56	63.1	46.6
Race/ethnicity (%)
White, non‐Hispanic	75.5	62.4	40.5	85.8
Black, non‐Hispanic	9.0	13.7	20.0	6.0
Other	7.0	9.7	13.6	4.4
Hispanic	8.5	14.3	26.0	3.9
Partnered (%)	54.7	44.5	32.8	59.2
Has daughter (%)	76.3	75.8	75.2	81.7
Highest level of education (%)
Less than high school/missing	22.3	24.6	54.2	24.7
High school	26.6	26.6	21.8	34.1
Greater than high school	48.7	46.5	21.9	39.3
Missing	2.5	2.2	2.1	1.9
N	6665	2136	1245	1315

Note: Dementia, Medicaid, and rural samples are sub populations of the full sample included in our models.

In terms of our outcome of interest, type of care received, use of family care only was consistently the most common type of care in the full sample (62.7%) and all three subsamples. About 25% of the full sample and the rural subsample received no care, compared to 15% of persons with dementia and 18% with Medicaid. Reliance on formal care alone was least common across all samples, but highest among Medicaid beneficiaries in our sample (8.7%). About 50% of the full sample lived in states with waiver(s) paying family members for care. Notably, the subsample with the fewest people living in states with waiver(s) was the Medicaid sample (40.0%); the rural sample had the most (75.7%).

3.2. Waiver availability association with care type received by dementia status

Figure 1 displays the results from our multinomial logistic regression predicting type of care received by dementia status in the full sample. Marginal associations of waiver availability are reported and represent percentage point changes in likelihood of each type of care for persons with and without dementia. Among our sample of individuals aged 70 years and older requiring assistance with at least one ADL or IADL, availability of payments to family caregivers via 1915(c) Medicaid waivers negatively predicted use of formal care only, but only among persons with dementia [‐2.2 percentage points, 95% confidence interval (95% CI), ‐3.6 to ‐1.0]. Waiver availability was positively associated with receiving family care only and negatively associated with the combination of formal and family care for people with dementia, but these estimates are only significant at p < 0.10. The association of waivers and types of care received did not differ significantly by dementia status.

FIGURE 1.

Marginal associations of waiver availability by dementia status (2011–2019). Model adjusted for age, sex, race, education, Medicaid, > 300% above Federal poverty line, number of function limitations, dementia status, rural residence, survey round, any daughter, partnered, exceeding federal standards of Family Medical Leave Act, mandatory paid family leave, mandatory paid sick leave, and state‐year specific allocation of Older Americans Act funding for the National Family Caregiver Support Program state policies.

3.3. Waiver availability association with care type received by Medicaid status

The results from our model testing differences across Medicaid status are displayed in Figure 2. The patterns were generally the same as the model by dementia status. Waivers were negatively associated with use of formal care only, but only statistically significant among Medicaid beneficiaries (‐5.3 percentage points, 95% CI, ‐8.6 to ‐2.1). Like the previous model, waivers were positively associated with family care only and negatively associated with formal and family care in combination, but the estimates were not significant at p < 0.05. In the negative control model stratified by Medicaid enrollment status (Table 2), waiver availability was not associated with any of the outcomes among persons who were not enrolled in Medicaid. Marginal effects were close to zero and not statistically significant.

FIGURE 2.

Marginal associations of waiver availability by Medicaid status (2011–2019). Model adjusted for age, sex, race, education, Medicaid, > 300% above Federal poverty line, number of function limitations, dementia status, rural residence, survey round, any daughter, partnered, exceeding federal standards of Family Medical Leave Act, mandatory paid family leave, mandatory paid sick leave, and state‐year specific allocation of Older Americans Act funding for the National Family Caregiver Support Program state policies.

TABLE 2.

Marginal effect of waivers in a subsample of person not enrolled in Medicaid.

Parameter	Estimate (95% CI)	p‐value
No care	0.42 (−1.48 ‐ 2.32)	0.66
Family care only	0.38 (−2.42 ‐ 3.19)	0.79
Paid care only	−0.27 (−1.46 ‐ 0.93)	0.66
Both paid and family care	−0.54 (−2.76 ‐ 1.69)	0.63
N = 14,305

Note: Model adjusted for age, sex, race, education, > 300% above Federal poverty line, number of function limitations, dementia status, and rural residence.

3.4. Waiver availability association with care type received by rurality

In Figure 3, the marginal association between living in a state with a waiver paying family caregivers is shown by rural and urban status. In this model, waivers positively predicted family care only (12.7 percentage points, 95% CI 6.2–19.1) among those living in rural settings. Waivers were negatively associated with formal care alone (‐6.5 percentage points, 95% CI ‐11.0 to ‐2.0) or in combination with family care (‐8.2 percentage points, 95% CI ‐13.1 to ‐3.3) for rural residents. We found no evidence of an association among non‐rural residents.

FIGURE 3.

Marginal associations of waiver availability by rural status (2011–2019). Model adjusted for age, sex, race, education, Medicaid, > 300% above Federal poverty line, number of function limitations, dementia status, rural residence, survey round, any daughter, partnered, exceeding federal standards of Family Medical Leave Act, mandatory paid family leave, mandatory paid sick leave, and state‐year specific allocation of Older Americans Act funding for the National Family Caregiver Support Program state policies.

3.5. Waiver availability effects by dementia status—for Medicaid and rural subsamples

Finally, we examined differences across dementia status in the subsample of Medicaid beneficiaries (N = 4125) and the subsample of those in rural settings (N = 3756). Table 3 displays estimates of the marginal effect of waivers by dementia status in these two subsamples. In the Medicaid subsample, waivers were positively associated with family care only and negatively associated with paid care only for both those with and without dementia. Waiver availability was associated with a decreased likelihood of receiving paid care only by 6.2 percentage points (95% CI ‐10.9 to ‐1.6) among persons without dementia and 8.3 percentage points (95% CI ‐12.1 to ‐4.5) among those with dementia. We found no evidence that the differential associations of the waiver payments are significantly different by dementia status. ,  

TABLE 3.

Marginal associations of waivers by dementia status among Medicaid and rural samples.

Parameter	No care Est. (95% CI)	Family only Est. (95% CI)	Paid only Est. (95% CI)	Family + paid Est. (95% CI)
Medicaid sample (N = 4125)
No dementia	3.01 (−1.26 to 7.27)	8.50 (0.62 to 16.38)	−6.23 (−10.88 to −1.58)	−5.28 (−11.24 to 0.68)
Dementia	−1.35 (−6.65 to 3.96)	9.74 (0.14–19.34)	−8.33 (−12.14 to −4.52)	−0.07 (−8.16 to 8.02)
Rural sample (N = 3756)
No dementia	1.68 (−3.82 to 7.19)	11.22 (5.89–16.55)	−7.29 (−11.76 to −2.82)	−5.61 (−8.19 to −3.03)
Dementia	0.64 (−5.44 to 6.73)	8.29 (0.03–16.56)	−5.36 (−9.10 to −1.62)	−3.57 (−10.30 to 3.16)

Note: Models adjusted for age, sex, race, education, Medicaid, > 300% above Federal poverty line, number of function limitations, dementia status, and rural residence.

In the rural subsample, waiver availability was associated with an increased likelihood of family care only (11.2 percentage points, 95% CI 5.9–16.5) and reduced likelihood of paid care alone (‐7.3 percentage points, 95% CI ‐11.8 to ‐2.8) or in combination with family care among persons without dementia (‐5.6 percentage points, 95% CI ‐8.2 to ‐3.0). Among persons with dementia in the rural sample, waivers were associated with an increased likelihood of family care only (8.3 percentage points, 95% CI 0.02–16.6) and reduced likelihood of paid care alone (‐5.4 percentage points, 95% CI ‐9.1 to ‐1.6). We found no evidence that the differential associations of the waiver payments were significantly different by dementia status. Due to small cell sizes, we were unable to report differences by dementia status among Medicaid beneficiaries in rural settings. 

4. DISCUSSION

This study used nationally representative data on older persons in the United States, the National Health and Aging Trends Study, linked to original data on availability of Medicaid 1915(c) waivers providing payment family caregivers of older, community‐dwelling adults to estimate the relationship between waiver availability and care receipt and differences across dementia status, Medicaid receipt, and rurality. Living in a state with waiver access decreased likelihood of receiving formal care alone among PLWD, Medicaid beneficiaries, and those in rural settings. The association of waivers and types of care received were largest among rural respondents. For persons in rural settings, waiver availability additionally increased likelihood of family care only and decreased use of formal care (provided with or without family care). By contrast, our analysis found no significant associations between waivers and type of care received by persons without dementia, not on Medicaid, or those in urban areas. Together, these findings suggest HCBS waivers paying family members for care reduce reliance on formal care for groups that may lack access to the direct care workforce—that is, persons living in rural settings. Collectively, our findings indicate that the association of waivers with care type was primarily driven by Medicaid and rurality and was similar across persons with and without dementia. Our finding that state‐level policies can predict type of care received has implications: (1) for ensuring an adequate caregiving workforce to reduce unmet needs, and (2) to advance the identification and rigor of policy evaluation to support caregivers.

State‐level policies play an important role in reducing reliance on formal care, even after adjusting for the historically strongest predictors of receiving family care. Our finding of the association between waivers and caregiving in Medicaid beneficiaries is notable in the context of the HCBS in the United States, because Medicaid is the primary payer of HCBS, spending over $116 billion in 2020 alone. ³³ Similarly, the strong association between waivers and all types of care in rural settings is important given that persons in rural settings are much more likely to rely on family caregivers than their urban counterparts. ¹⁴ Understanding the impact of this spending on whether community‐dwelling persons with care needs receive necessary care is important for evaluating effectiveness of these types of policies and their impact on both care recipients and family caregivers. Access to these waivers decreases the likelihood that persons with functional limitations receive care by paid caregivers, particularly among vulnerable populations. In our sample, PLWD were much less likely to be married or living with a partner than those without dementia. Thus, they may rely more heavily on adult children to provide care who may receive supports through Medicaid waivers. Medicaid waivers providing payment for that care would likely be particularly important given that their adult children may substitute caregiving for some or all of their labor supply. Payments from Medicaid waivers would also help offset documented wage penalties among working caregivers. ³ , ³⁴ Thus, while family structure is an important predictor of the type of care received, we observed that state‐level policies can also move the needle on incentivizing caregiving.

Exploiting state‐level variation in policies to predict uptake of family caregiving offers the potential to advance the identification and rigor of causal policy evaluation to support caregivers. Historically, researchers have used attributes of family structure (e.g., number, gender, and distance living of adult children or siblings from parents) as instrumental variables to address endogeneity of family care to elucidate causal effects of family caregiving on outcomes related to the care recipients and caregivers. We demonstrated the potential for state‐level policies to predict selection into family caregiving which may offer new research opportunities to estimate the effects of caregiving among data sources with state‐level indicators but in which family structure may not be readily available to predict selection into caregiving.

The strength of our study was our use of novel data on state‐year waiver characteristics and nationally representative data to estimate, for the first time, waivers’ payment to family members and type of care received by community‐dwelling persons with functional limitations. Our study also has some limitations. First, the majority of the 1915(c) waivers require beneficiaries need nursing home‐level care to be eligible for services. While we limited our sample to community‐dwelling persons with at least one IADL or ADL, our sample may not be restrictive enough to match the population eligible for these waivers. As such, our estimates are likely smaller than the true impact of these waivers on caregiving. In addition, although our main findings suggest there might be interesting differences in the impact of waivers on care by dementia status for rural, Medicaid beneficiaries, our sample size did not allow us to estimate this relationship. It is also possible that states providing waivers differ in unobservable ways that provide a supportive environment for family caregivers. One method to account for these differences would be to use state fixed effects; however, fixed effects were not supported by our data after accounting for survey weights, sampling, and strata. We included time‐varying measures of state‐level policies supportive of family caregiving—exceeding federal standards of Family Medical Leave Act, mandatory paid family leave, and mandatory paid sick leave—as well as state‐year allocation of Older Americans Act funding for the National Family Caregiver Support Program to account for state‐year differences in other caregiver supports. Additionally, our negative control model found no relationship between waivers and care in those not enrolled in Medicaid, suggesting the associations found in our main analyses were driven by the waivers, rather than the state environment. Finally, we did not estimate the causal effects of 1915(c) waivers or examine changes in the intensity of types of care provided (e.g., hours of care). Future work should consider causal study designs to elucidate effects such as changes in intensity of care and unmet needs.

Our study demonstrates the positive impact of access to 1915(c) waivers supporting family caregivers on whether community‐dwelling persons with at least one functional limitation receive care and from whom. Identifying policy levers for improving health and wellbeing of vulnerable persons (e.g., PLWD, low income) and their care partners is of critical importance as the number of persons aging in place grows. While our study indicates waivers decrease reliance on formal care, and an increase in use of family care for some, their impact on the health wellbeing of care recipients and caregivers remains unknown. This is especially salient for PLWD and their caregivers, who often experience worse physical and mental health outcomes and greater caregiver burden. Future research exploring the relationship between waiver services, types of services, and the health and wellbeing of PLWD and their caregivers is needed.

CONFLICT OF INTEREST STATEMENT

None. Author disclosures are available in the supporting information.

Supporting information

Supporting information

Supporting information

Thunell J, Parrish EM, Miller KEM. Shifting care patterns: How Medicaid policies shape family and formal care use. Alzheimer's Dement. 2025;21:e70715. 10.1002/alz.70715