Medicare and Medicaid provide health insurance coverage to tens of millions of people in the U.S. Medicaid is the nation’s largest public health insurance program for low-income Americans and the primary payer for long-term care. Approximately 12 million people are enrolled in both Medicare and Medicaid, referred to as dual-eligible individuals, receiving their primary health insurance coverage through Medicare and additional assistance from their state Medicaid program. Within this group, nearly 9 million people are “full-benefit” dual-eligible individuals, meaning they are eligible for Medicaid benefits that are not otherwise covered by Medicare, including long-term care, vision, and dental services. The remaining 3 million are “partial-benefit” dual-eligible individuals, and are only eligible for assistance with Medicare premiums and, in many cases, cost sharing through the Medicare Savings Programs.

The 2025 budget reconciliation law made major policy changes resulting in significant reductions in future federal Medicaid spending. Those provisions are expected to reduce the number of low-income Medicare beneficiaries who are also enrolled in Medicaid and could also affect the services that dual-eligible individuals receive if state Medicaid programs respond to increased fiscal pressures by reducing coverage of optional benefits. To better understand the low-income Medicare beneficiaries who may be affected by these changes, this brief examines the demographic, socioeconomic, and health characteristics of dual-eligible individuals compared with Medicare beneficiaries without Medicaid using the 2023 Medicare Current Beneficiary Survey (MCBS).

Key Takeaways

• Compared with Medicare beneficiaries without Medicaid, dual-eligible individuals are more likely to be younger, have lower incomes, and be people of color. Specifically, larger shares of dual-eligible individuals than Medicare beneficiaries without Medicaid are under age 65 (34% vs. 6%), have incomes below $20,000 per person (64% vs. 9%), and are people of color (52% vs. 18%).
• Dual-eligible individuals are more likely to be in poorer health and have greater health needs than Medicare beneficiaries without Medicaid. For example, 44% of dual-eligible individuals report being in fair or poor health, nearly half (48%) have at least one limitation in an activity of daily living (ADL), and more than one-third (36%) report having a cognitive impairment. Among Medicare beneficiaries without Medicaid, 14% report being in fair or poor health, 21% have at least one ADL limitation, and 12% report having a cognitive impairment.
• There are some differences in the demographic and health characteristics of full- and partial-benefit dual-eligible individuals. For example, compared with partial-benefit dual-eligible individuals, a larger share of full-benefit dual-eligible individuals are under the age of 65 (36% vs. 28%), have at least one ADL limitation (50% vs. 39%), and report an intellectual or developmental disability (15% vs. 5%).

Demographics of Dual-Eligible Individuals

More than one-third (34%) of dual-eligible individuals are under age 65 and qualify for Medicare on account of having a long-term disability, compared with fewer than one in ten (6%) Medicare beneficiaries without Medicaid (Figure 1). Additionally, larger shares of full-benefit dual-eligible individuals than partial-benefit dual-eligible individuals are under age 65 (36% vs. 28%). People under age 65 may qualify for Medicare if they are eligible for Social Security Disability Insurance benefits. Medicare beneficiaries younger than age 65 differ from beneficiaries ages 65 and older in terms of having worse health status, higher per capita Medicare spending, and lower incomes. Similar shares of dual-eligible individuals (10%) and Medicare beneficiaries without Medicaid (11%) are over the age of 85.

Nearly two-thirds (64%) of dual-eligible individuals live on an annual income below $20,000 per person compared with one in ten (9%) Medicare beneficiaries without Medicaid coverage. There are multiple pathways through which Medicare beneficiaries become eligible for Medicaid benefits. All of the Medicaid pathways that are specific to adults ages 65 and older and people with disabilities, including the two most common for dual-eligible individuals – Supplemental Security Income and the Medicare Savings Programs – require beneficiaries to have income and assets below certain thresholds (which vary by pathway). Medicare beneficiaries with higher incomes may become eligible for Medicaid if they “spend down” their income due to large medical or long-term care expenses. Some low-income Medicare beneficiaries may not be qualified to enroll in Medicaid because their assets exceed the limit, they are unaware of the programs, or because administrative burdens prevent them from enrolling.

Over one in three (34%) dual-eligible individuals have less than a high school degree compared with 7% of Medicare beneficiaries without Medicaid. Additionally, similar shares of full-benefit dual-eligible individuals (35%) and partial-benefit dual-eligible individuals (32%) have less than a high school degree.

A slightly larger share of dual-eligible individuals are female compared with Medicare beneficiaries without Medicaid. Six in ten (60%) dual-eligible individuals are female, while just over half (53%) of Medicare beneficiaries without Medicaid are female. The same share of full-benefit and partial-benefit dual-eligible individuals (60%) are female.

Dual-eligible individuals are disproportionately people of color compared with Medicare beneficiaries without Medicaid (Figure 2). Higher shares of dual-eligible individuals are Black (22% compared with 8% among Medicare beneficiaries without Medicaid), Hispanic (21% compared with 6% of Medicare beneficiaries without Medicaid), or an other non-White, non-Hispanic race and ethnic group (9% compared with 5% of Medicare beneficiaries without Medicaid).

Dual-eligible individuals are more likely than Medicare beneficiaries without Medicaid to live in a long-term care nursing home or other institutional facility (Figure 3). In particular, more full-benefit dual-eligible individuals (10%) live in an institutional setting relative to other Medicare beneficiaries (1% or less of both partial-benefit dual-eligible individuals and Medicare beneficiaries without Medicaid). This likely reflects Medicaid eligibility provisions and the fact that Medicaid is the nation’s primary payer for long-term care services. The average cost of a private room in a nursing facility was $116,800 in 2023. After spending their income and savings to pay for those costs, Medicare beneficiaries without Medicaid may become eligible for Medicaid coverage of long-term care services. In addition, Medicare beneficiaries who use long-term care and who meet other eligibility criteria may qualify for Medicaid if, for example, they can demonstrate the need for an institutional level of care and meet other income and asset criteria.

A slightly larger share of dual-eligible individuals than Medicare beneficiaries without Medicaid live in a rural area. Nearly one in five (19%) dual-eligible individuals compared with 16% of Medicare beneficiaries without Medicaid live in a rural area. There are similar shares of full- and partial-benefit dual-eligible individuals who live in rural areas.

Health Characteristics of Dual-Eligible Individuals

More than four in ten (44%) dual-eligible individuals report being in fair or poor health compared with 15% of Medicare beneficiaries without Medicaid coverage (Figure 4). Conversely, over half (57%) of Medicare beneficiaries without Medicaid report being in very good or excellent health, which is twice as large as the share of dual-eligible individuals (25%) that report very good or excellent health. A slightly larger share of partial-benefit dual-eligible individuals report being in very good or excellent health (28%) compared with full-benefit dual-eligible individuals (24%).

Dual-eligible individuals are more likely than Medicare beneficiaries without Medicaid to report having difficulty with one or more activities of daily living (such as eating, bathing, and dressing) (Figure 5). Just under half (47%) of dual-eligible individuals report having difficulty performing at least one activity of daily living (ADL), including about one in four (24%) who report difficulty with three or more ADLs. That compares with about one in five (21%) Medicare beneficiaries without Medicaid who report at least one ADL limitation, including just 7% who report limitations with at least three ADLs. A larger share of full-benefit (50%) than partial-benefit (39%) dual-eligible individuals report difficulties with at least one ADL.

Dual-eligible individuals report higher rates of mental health conditions, cognitive impairments, and intellectual and developmental disabilities. Most dual-eligible individuals receive their Medicare and Medicaid benefits through separate coverage arrangements. Difficulty navigating the Medicare and Medicaid programs can lead to fragmented care, lower quality, and higher costs. For people with mental health conditions, cognitive impairments, or intellectual or developmental disabilities, the challenge of navigating coverage under both programs could be even larger.

Dual-eligible individuals are about twice as likely to report having a mental health condition as Medicare beneficiaries without Medicaid coverage (Figure 6). About half (47%) of dual-eligible individuals report having a mental health condition, such as depression, compared with less than one-fourth (24%) of Medicare beneficiaries without Medicaid coverage. The share of full-benefit dual-eligible individuals with a mental health condition (49%) is somewhat higher than the share of partial-benefit dual-eligible individuals with a mental health condition (41%).

The share of beneficiaries who report having a cognitive impairment is three times larger among dual-eligible individuals than among Medicare beneficiaries without Medicaid. Specifically, 36% of dual-eligible individuals report a cognitive impairment, including difficulty concentrating, remembering, or making decisions, compared with 12% of Medicare beneficiaries without Medicaid. The share of full-benefit dual-eligible individuals (39%) that report a cognitive impairment is larger than the share of partial-benefit dual-eligible individuals (27%) reporting a cognitive impairment.

The share of dual-eligible individuals who report having an intellectual or developmental disability is much higher than that of Medicare beneficiaries without Medicaid. Overall, 13% of dual-eligible individuals report having an intellectual or developmental disability compared with only 1% of Medicare beneficiaries without Medicaid coverage. Full-benefit dual-eligible individuals are about three times as likely to report having an intellectual or developmental disability than partial-benefit dual-eligible individuals (15% vs. 5%, respectively).

On July 16, 2022, the federally mandated crisis number, 988, became available to all phone users at no charge. This three-digit number connects users–via phone, text, or chat–to a network of over 200 local and state-funded crisis call centers that provide access to crisis counseling, resources, and referrals through the 988 Suicide & Crisis Lifeline. The introduction of 988 came against the backdrop of rising suicide rates. Between 2014 and 2024, more than half a million lives (516,790) were lost to suicide, and over half of those involved firearms. In 2024, suicide and substance-related deaths, combined, were the third leading cause of death in the U.S., behind heart disease and cancer.     

Data as of March 2026 show that demand for 988 continues to rise, with the volume of calls, texts, and chats 15% higher than a year earlier and nearly 50% higher than two years ago. States are also answering more calls in-state, where counselors are more likely to be familiar with local resources. The broader crisis continuum, including mobile crisis and crisis stabilization services, has expanded across states. Suicide deaths declined modestly in 2024, and remained relatively stable in provisional 2025 data. Research examining  988’s relationship to suicides found lower-than-expected deaths among people ages 15 to 34 after 988 launched, suggesting that 988 may have contributed to improved outcomes for some populations.

Despite some positive signs, sustaining momentum may become more difficult as demand grows and fiscal pressures increase. Relatively few states have adopted telecom fees intended to provide dedicated funding for 988 call centers and other crisis services, leaving many states to rely on other funding mechanisms, which vary widely. Uneven or uncertain funding may affect call center capacity, and staffing is already a challenge, with many call centers reporting staffing shortages. Further, federal changes to Medicaid and the ACA Marketplaces, along with other federal policy changes, are expected to increase the number of people without health coverage. Coverage losses stemming from these changes could disrupt access to behavioral health care and increase demand on 988, just as federal cuts are likely to place greater pressure on state budgets.  

Taken together, 988 is reaching more people and early suicide-related outcome data show some signs of improvement, but sustaining call center capacity and community-based crisis services may become more difficult as demand grows and state resources become more limited. This policy watch examines 988 on its fourth anniversary, drawing from the latest Lifeline data available through May 2026 for state call data and through March 2026 for national call, text, and chat data.

988 received 23.3 million contacts since its launch in July 2022, including 15.8 million calls, 4.2 million texts, and 3.4 million chats. Monthly contact volume has steadily increased, consistently surpassing 600,000 contacts per month over the past year and often approaching or exceeding 700,000 per month since late 2025—more than double the contacts recorded before launch (277,407 in June 2022) (Figure 1). Growing public awareness of  988 likely contributed to this increase, though contact volume may have been slowed by the 2025 elimination of the specialized LGBTQ+ service for young people, which had accounted for 10% of all 988 contacts. Recent reporting suggests that the specialized service may be restored. 988 data excludes calls routed from 988 to the Veteran’s Crisis Line (VCL), which has received an additional 3 million calls since launch, because monthly breakouts and other details are not publicly available. They also exclude suicide hotline contacts from centers operating outside of the 988 network. As of 2024, fewer than half of all hotline call centers (about 200 of 573) participated in the 988 network.

Local 988 response has improved, with 26 states answering at least 90% of calls in-state, but uneven and uncertain funding may pose challenges for sustaining call center capacity. In May 2026, 26 states answered at least 90% of calls in-state, up from just 8 states before 988’s launch (Figure 2).  In-state answer rates ranged from 62% (District of Columbia) to 99% (Mississippi and Rhode Island). 988 is now facilitating more local response by routing callers to the call center nearest to their location, while protecting privacy. Calls not answered in-state are redirected to national backup centers, where counselors can respond to the crisis call but may be less familiar with local resources. Ongoing financing remains a challenge. Federal funding helped launch 988 and continues to support some local call centers, but ongoing center financing largely falls to states. As of 2026, 12 states have passed legislation to help sustainably fund 988 through telecom fees, similar to how 911 is financed, with many assessing flat monthly charges of less than 45 cents per line. Other states rely on mechanisms that may be less stable or vary from year to year. Limited or uncertain funding can affect staffing, quality, and operations. In a 2025 survey of call center leaders, about 3 in 4 call centers reported staffing shortages and 89% reported some level of difficulty acquiring resources to hire staff.  As states enter a period of tighter budgets, questions about the long-term funding to sustain local 988 crisis call centers are likely to persist.

988, along with other investments in the crisis continuum, may have helped to contribute to some recent improvement in suicide-related outcomes. Suicide deaths declined modestly in 2024, and remained relatively stable in provisional 2025 data. Most of the decline has been among non-firearm suicides, which have fallen each year since 2022, while firearm suicides have risen each year since 2019 (Figure 3). Other research also shows declines in suicides or related measures among certain populations. The share of emergency department (ED) visits for suspected suicide attempts decreased by about 7% in 2025 compared with 2021, and a separate study found that, in the years following 988’s launch, suicide deaths among people ages 15 to 34 were about 11% lower than expected if 988 had not launched. These trends have occurred as states continue to build out the broader crisis continuum, which also includes mobile crisis and crisis stabilization services.  According to a survey of state behavioral health agencies, mobile crisis volume increased by 40%, and crisis stabilization center volume more than doubled, relative to 2022. As of 2024, 34 states reported statewide availability of mobile crisis teams, and 25 states reported the same for crisis stabilization services. This expansion may face future financing challenges, as enhanced federal Medicaid funding that supported mobile crisis expansion in some communities expires in 2027, leaving states responsible for more of the cost. States are also making crisis-system data more public, with 21 states now displaying public-facing crisis contact data dashboards.  

In the past decade, nearly 1.8 million people in the United States lost their lives by suicide or alcohol and drug consumption. Collectively, these deaths are sometimes referred to as “deaths of despair”, a term that emphasizes economic and social deterioration as driving factors. However, additional and more complex factors may be linked to these deaths, such as health disparities, challenges with treatment access, reduced public services, and changes in drug supplies. Therefore, the term “substance use and suicide deaths” is used throughout this brief as a composite measure describing deaths due to alcohol, drugs, and/or suicide. These causes are examined together because they share many upstream risk factors and prevention strategies related to mental health and substance use.

This analysis explores trends in substance use and suicide mortality using data from CDC WONDER. Due to overlap in several ICD-10 codes for deaths involving drugs, alcohol and/or suicide, duplicate codes were removed. Therefore, total substance use and suicide deaths are not equivalent to the sum of alcohol-induced, suicide, and drug overdose deaths (see Methods for more information). Key takeaways include:

1. In 2024, substance use and suicide deaths (47.7 per 100,000 population) were the third leading cause of death in the U.S., behind heart disease (157.6) and cancer (139.4).
2. Deaths due to substance use and suicide increased over the past decade (34.6 in 2014) and peaked during the pandemic (59.6 in 2021).
3. Older adults, men, and some communities of color were disproportionately impacted by substance use and suicide deaths compared to their peers.

Despite these high mortality rates, many individuals experience challenges obtaining mental health and substance use services. KFF analyses found that while health insurance is linked to increased utilization of mental health services, many insured individuals do not receive needed mental health treatment. This is associated with multiple factors, including narrow provider networks, coverage limitations, and cost and logistical barriers. A recent KFF survey found that 19% of adults with health insurance rated their insurance negatively based on the availability of mental health providers, and 43% of individuals with poor mental health said they did not receive needed services or medication in the previous year.

The passage and evolution of the Mental Health Parity and Addiction Equity Act (MHPAE) have sought to reduce some access burdens by requiring health insurance plans to provide benefits for mental health and substance use services in alignment with the benefits provided for medical and surgical care. However, enforcement of MHPAE policies is challenging and recent efforts to modify enforcement practices have been slowed under the second Trump administration. Other policy actions under the Trump administration are impacting access to mental health and substance use services. This includes the 2025 reconciliation law which will reduce Medicaid and ACA coverage,  narrowing the scope of federal leadership capacity in mental health and substance use services, and a departure from harm reduction services. Collectively, these changes may disrupt access to and continuity of care at a time when substance use and suicide deaths remain elevated.

1. Substance use and suicide deaths combined were the third leading cause of deaths in the United States in 2024.

In 2024, 170,449 people in the U.S. died from suicide and/or substance use. This translates to an age-adjusted death rate of 47.7 per 100,000 population. Compared to other leading causes of death in 2024, only heart disease and cancer rates (157.6 and 139.4, respectively) have higher age-adjusted rates than suicide and substance use (Figure 1).

2. Over the past decade, substance use and suicide mortality rates increased by nearly 40%.

From 2014 to 2024, substance use and suicide death rates increased from 34.6 per 100,000 population to 47.7 (Figure 2). During this period, death rates increased until 2017 and then held steady before sharply increasing alongside the pandemic (59.6 deaths per 100,000 in 2021). It is likely that the worsening opioid crisis, and specifically the presence of fentanyl in the drug supply, significantly contributed to the spike in substance use and suicide mortality rates during the pandemic. These death rates have declined since peaking in 2021 but remain above pre-pandemic levels (47.7 in 2024 vs. 44.6 in 2019).

3. Substance use and suicide mortality rates have disproportionately affected adults ages 45 to 64, men, and certain communities of color.

From 2014 to 2024, substance use and suicide mortality rates were highest among adults ages 45 to 64 (85.3 per 100,000) (Figure 3). These adults along with adults ages 26 to 44 experienced the sharpest increases in mortality rates during the pandemic. Although these rates have since decreased, they remain above pre-pandemic levels for both age groups. In comparison, young adults (ages 18 to 25) and elderly adults (ages 65 and above) experienced smaller increases in substance use and suicide mortality rates during the pandemic. By 2024, the mortality rate among young adults decreased to 27.2, similar to its rates a decade earlier. However, mortality rates among elderly adults have remained relatively steady since their slight increase during the pandemic.

The rate of substance use and suicide mortality among males grew 40% from 2014 to 2024 and remained nearly three times higher than the rates among females (70.6 per 100,000 vs. 25.4 in 2024, respectively, Figure 3). Over the decade, the gap in these death rates has generally widened between males and females: from 50.3 per 100,000 for males and 19.7 for females in 2014 to 70.6 for males and 25.4 for females in 2024. While both groups experienced peak rates during the onset of the pandemic, the increase impacted males more than females (87.6 vs. 32.2 in 2021, respectively).

Over time, people of color have experienced a faster growth in substance use and suicide death rates compared to their White peers. From 2018 to 2024, death rates increased by 43% for Black people, 30% for Hispanic people, 28% for American Indian and Alaska Native (AIAN) people, 14% for Asian and Pacific Islanders, and 2% for White people (Figure 3). As a result, the gap in substance use and suicide mortality rates has largely narrowed between White and Black people and, to a lesser extent, between White and Hispanic people. Additionally, AIAN people consistently experience the highest rates of substance use and suicide mortality compared to all other racial and ethnic groups (130.1 per 100,000 vs. 53.7 among White people in 2024).

Methods

The Medical Loss Ratio (MLR) provision of the Affordable Care Act (ACA) limits the share of premium income that insurers can keep for administration, marketing, and profits. Insurers that fail to meet the applicable MLR threshold are required to pay back excess profits or margins in the form of rebates to individuals and employers that purchased coverage.

In the individual and small group markets, insurers must spend at least 80% of their premium income on health care claims and quality improvement efforts, leaving the remaining 20% for administration, marketing expenses, and profit. The MLR threshold is higher for large group insurers, which must spend at least 85% of their premium income on health care claims and quality improvement efforts. MLR rebates are based on a three-year average, meaning that rebates issued in 2026 will be calculated using insurers’ financial data in 2023, 2024 and 2025 and will go to people and businesses who bought health coverage in 2025.

This analysis, using preliminary data reported by insurers to state regulators and compiled by Mark Farrah Associates, finds that insurers estimate they will issue a total of just over $759 million in MLR rebates across all commercial markets in 2026. Since the ACA began requiring insurers to issue these rebates in 2012, a total of $14.4 billion in rebates has already been issued to individuals and employers, and this analysis suggests the 2012-2026 total will rise to about $15.1 billion when rebates are issued later this year.

Estimated total rebates across all commercial markets in 2026 ($759 million) are less than total rebates issued in 2024 ($958 million) and in 2025 ($1.6 billion). In 2025, rebates were issued to 5.1 million people with individual coverage and 3.5 million people with employer coverage. In the individual market, the 2025 average rebate per person was $233, while the average rebates per person for the small group market and the large group market were $190 and $91, respectively (though enrollees could receive only a portion of this as rebates may be shared between the employer and employee or be used to offset premiums for the following year).

The estimated $759 million in rebates falls far short of record-high rebate totals of $2.5 billion issued in 2020 and $2.1 billion issued in 2021, driven largely by high margins in the 2018 ACA Marketplace following steep premium increases in response to the elimination of cost-sharing reduction payments, inflating the three-year average on which rebate calculations are based, as well as lower than expected utilization in 2020 due to the pandemic. In the years following 2018, insurers largely held premiums flat or reduced them as claims cost rose and caught up to premium levels, compressing margins and bringing rebate totals down significantly from those peaks. The current rebate environment reflects this period of margin normalization rather than continued insurer profitability stemming from the earlier CSR-related premium increases.

The chart above shows the average “simple loss ratio” in each market. Simple loss ratios differ from the ACA MLR used to calculate rebates because simple loss ratios do not include adjustments for quality improvement expenses or taxes. In 2025, the average simple loss ratio in the individual market was 93%, meaning these insurers spent an average of 93% of their premium income in the form of health claims in 2025. This is significantly higher than the previous year, suggesting that insurers were less profitable. However, rebates issued in 2026 are based not only on 2025 experience, but rather on a three-year average of insurers’ experience in 2023-2025. Consequently, even insurers with high loss ratios in 2025 may expect to owe rebates if they were highly profitable in the prior two years.

Going into 2026, ACA Marketplace premiums increased at the steepest rate (over 20%) since 2018, the last time policy uncertainty fueled premium increases. If it turns out insurers overpriced premiums compared to how much enrollees spend on claims in 2026 and fall short of MLR ratio targets, insurers would likely have to pay their enrollees rebates in the future. 

In the small and large group markets, 2025 average simple loss ratios were 87% and 91%, respectively. Only fully-insured group plans are subject to the ACA MLR rule; about two-thirds of people with insurance through their work are in self-funded plans, to which the MLR threshold does not apply.

Rebate Payment Logistics

The 2026 rebate amounts in this analysis are still preliminary. Rebates or rebate notices are mailed out by the end of September and the federal government will post a summary of the total amount owed by each issuer in each state later in the year.

Insurers in the individual market may either issue rebates in the form of a check or premium credit. For people with employer coverage, the rebate may be shared between the employer and the employee depending on the way in which the employer and employee share premium costs.

If the amount of the rebate is exceptionally small (less than $5 for individual rebates and less than $20 for group rebates), insurers are not required to process the rebate, as it may not warrant the administrative burden required to do so.

Methods

This analysis is based on insurer-reported financial data from Health Coverage Portal TM, a market database maintained by Mark Farrah Associates, which includes information from the National Association of Insurance Commissioners. The Supplemental Health Care Exhibit dataset analyzed in this report does not include data from California HMOs regulated by California’s Department of Managed Health Care. Individual, Small, and Large Group Comprehensive plans were used for this analysis. All individual market figures in this analysis are for major medical insurance plans sold both on and off exchange. Health premiums earned and total incurred claims were set as zero if reported as a negative value. Simple loss ratios are calculated as the ratio of the sum of total incurred claims to the sum of health premiums earned. Only plans with an estimated unpaid current rebate value greater than zero were used to calculate the estimated total 2026 rebates.

Rebates for 2026 are based on preliminary estimates from insurers. In some years, final rebates are higher than expected and in other years, final rebates are lower.

Medicaid covers one-fifth of adults with substance use disorders (SUDs), including many with substantial treatment needs. Inpatient and residential care services are part of the American Society of Addiction Medicine (ASAM) continuum of SUD services. However, long-standing federal law has generally prohibited Medicaid payments for inpatient or residential care services that are provided in Institutions for Mental Disease (IMDs) for adults ages 21 to 64. IMDs include certain psychiatric and substance use treatment facilities with more than 16 beds that are primarily engaged in providing diagnosis, treatment or care for people with mental health or substance use disorders.

Section 1115 demonstration waiver authority for SUD, referred to here as “SUD IMD waivers,” was introduced in 2015 and updated in 2017 to allow states to receive federal Medicaid funds for short-term stays in IMDs to help improve identification of SUD and access to care across the treatment continuum, reduce overdose deaths and avoidable acute care use, build provider capacity, and strengthen care coordination and transitions across levels of care; however, there have also been concerns that expanding Medicaid payment for SUD services provided in IMDs could shift resources toward residential and inpatient care, rather than strengthening community-based SUD services.

As SUD IMD waiver evaluations and renewals are becoming available, states face a shifting Medicaid and policy and waiver landscape. The 2025 Reconciliation Law included historic restrictions in Medicaid financing and coverage that could make it hard to maintain incremental expansions in mental health and SUD benefits. At the same time, other federal efforts, including a recent opinion from the Department of Justice, may affect how care is delivered across settings for some people with SUD, including some experiencing homelessness, following decades of policy aimed at expanding access to community-based care for people with mental health conditions or SUD, when appropriate. This brief examines findings from the first available summative evaluations (due 18 months after the end of the demonstration) for SUD IMD waivers and what they show about progress toward meeting goals and milestones.

Key Takeaways:

• SUD IMD waivers, a common Medicaid pathway for covering short-term residential and inpatient SUD treatment in IMDs, have been adopted by 38 states (including D.C.). Federal SUD IMD authority began in 2015 (updated in 2017)to allow states to use Medicaid funds for short-term residential and inpatient SUD treatment in facilities that were previously excluded from federal reimbursement. As of April 2026, 38 states, including D.C. had approved SUD IMD waivers, three more had applications pending, and at least 20 had received Centers for Medicare and Medicaid Services (CMS) approval for five-year extensions. Most initial waivers were approved between 2016 and 2020, with most demonstrations running five years.
• As of April 2026, six states (CA, IN, NH, WA, UT, and MA) had completed summative evaluations, which are due 18 months after the state’s SUD IMD waiver demonstration period ends. To assess progress, states are required to develop evaluation designs with specific research questions, hypotheses, and methods. Although states follow federal guidance, evaluation plans emphasized different priorities and relied on different measures, data sources, and methods. Most used Medicaid claims data, stakeholder interviews, and survey data.
• Early evaluation findings point to improved SUD treatment access across multiple measures although the pandemic and other intervening issues make isolating waiver findings challenging. Summative evaluations generally suggest increases in SUD residential and inpatient care, along with growth in community-based care or capacity and increased access to medications for opioid use disorder. Most states reporting SUD-related emergency department use found decreases, and some reported improved follow-up after ED visits or discharge from residential or inpatient care, though follow-up rates often remained relatively low. However, evaluation periods often overlapped with major disruptions in SUD treatment need, service delivery, and Medicaid policy, including the COVID-19 pandemic, the fentanyl-driven rise in overdose deaths, and concurrent state Medicaid changes, making it difficult to isolate waiver effects. In addition, evaluations varied in their measures, detail, and focus, limiting the ability to compare across states or draw consistent conclusions about quality and outcomes.
• Evaluations pointed to challenges, including workforce or infrastructure shortages, administrative barriers, and housing instability, that may have affected waiver outcomes. These included workforce shortages, fragmented and uneven treatment systems, administrative barriers, and housing instability, with each of these areas reported across multiple evaluations. Fragmentation and gaps in treatment infrastructure made care transitions harder, while housing instability often made it difficult for patients to sustain recovery after discharge.

What is an IMD and the IMD payment exclusion?

IMDs are specialized behavioral health facilities that provide residential or inpatient treatment (Figure 1).  In Medicaid, an Institution for Mental Diseases, or IMD, is an administrative category for a hospital, nursing facility, or other institution with more than 16 beds that is “primarily engaged in providing diagnosis, treatment, or care of persons with mental diseases, including medical attention, nursing care, and related services” (42 U.S.C. § 1396d(i)). IMD status is based on facility type, the population type served, and its size. The IMD status is determined by the state. Although IMDs tend to overlap with higher levels of care, the clinical intensity of SUD treatment is generally described using frameworks such as the American Society of Addiction Medicine (ASAM) criteria, which do not determine whether a facility is an IMD. Other SUD care settings include outpatient treatment services, specialty outpatient services,  and emergency and crisis services, which are not subject to the IMD payment exclusion (Figure 1). State Medicaid programs cover a range of services and states have expanded these services in recent years.

While all states are required to cover inpatient hospital services, federal Medicaid law generally excludes IMD services for adults 21 to 64, a policy known as the “IMD payment exclusion.” Since Medicaid’s inception, federal law has generally prohibited states from using Medicaid funds for services provided to nonelderly adults in IMDs (§1905(a)(30)(B) of the Social Security Act [SSA]). The IMD payment exclusion was intended to leave states with the primary responsibility for financing inpatient or residential behavioral health services. While all states that choose to participate in the Medicaid program must cover inpatient hospital services, those services specifically exclude care provided in IMDs for adults ages 21 to 64. The payment exclusion applies to services provided within an IMD as well as to services provided outside an IMD while a nonelderly adult is still a patient of an IMD. In recent years, the federal government has provided new mechanisms for states to receive federal matching funds for IMD services for adults 21 to 64 in certain situations. There are now four options for states to cover these services. See Box 1 for exceptions to the IMD payment exclusion.

How many states have SUD IMD waivers, and how are they evaluated?

As of April 2026, 38 states had approved SUD IMD waivers.  Three more states had initial SUD IMD waiver applications pending with CMS (Figure 2). At least 20 states had also received CMS approval for five-year extensions of SUD IMD waivers, and additional states have submitted five-year extension requests (data not shown).  

SUD IMD waivers are intended to expand access to the full SUD treatment continuum, including residential and inpatient services provided in IMDs, and to improve outcomes. CMS established six goals and six milestones for these waivers, covering areas such as treatment access, placement standards, provider capacity, and care coordination (See Box 2). States are required to track progress through independent evaluations, including an interim evaluation report, due one year before the demonstration ends and a summative evaluation report due 18 months after it concludes.

To assess progress, states must develop evaluation designs with specific research questions, hypotheses, and methods. Although all states follow federal guidance in evaluation designs, state evaluation plans can emphasize somewhat different priorities. For example, New Hampshire included a specific focus on adolescent access to residential care networks, and California’s plan factored in its unique county-level behavioral health delivery model. States also organized their evaluation topics differently. Some grouped measures under broad areas such as “access” and “quality,” while others grouped measures more closely to the CMS-enumerated goals and milestones (Box 2) or led with more state-specific questions and priorities. Appendix Table 1 summarizes state-level evaluation topics, hypotheses, and implementation features.

As of April 2026, six states have summative evaluations of SUD IMD waivers: California, Indiana, New Hampshire, Utah, Washington, and Massachusetts. Evaluations rely primarily on Medicaid administrative claims data, stakeholder interviews, and survey data to assess impact. Evaluation methods vary across states, with some relying on descriptive pre-post analyses and others using more rigorous methods designed to control other factors that may contribute to observed changes over time.  

What are the key findings from the SUD IMD waiver summative evaluations?

The six summative evaluations are the most comprehensive assessments available for the SUD 1115 waiver demonstrations, but overlapping factors such as COVID-19, the opioid epidemic, and various state policy changes mean the findings are best read as directional signals within an emerging evidence base rather than isolated estimates of waiver impact. This analysis focuses on a limited set of CMS goals and milestones (CMS goals 1, 3, 4, and 5 and milestones 1, 4, and 6) because they were most consistently reported, most directly tied to SUD/OUD treatment access, utilization, care transitions, and overdose outcomes. The analysis excludes goals and milestones that primarily reflect implementation standards or were not reported consistently enough to support synthesis. Appendix Table 1 summarizes states’ evaluation topics as well as notable waiver features and implementation details.

Cross-state comparisons are limited by differences in evaluation design, measures, baseline SUD treatment capacity, waiver implementation, and broader state context. Though states use a variety of data and analytic methods, this summary does not weigh or assess research methods and draws from both raw and adjusted findings across evaluations. All states used claims data, and most also included survey and/or qualitative findings, but most lacked a comparison group. Timeliness is another challenge, as summative evaluations are due 18 months after the demonstration ends, and some expected demonstrations are not yet readily available through CMS materials. Most evaluation periods coincided with major disruptions in SUD treatment need, service delivery, and Medicaid policy, including the COVID-19 pandemic, the fentanyl-driven escalation in overdose deaths, and concurrent state Medicaid changes, making it difficult to isolate waiver effects. As a result, findings are not interpreted as causal or directly comparable across states; they are best understood as directional signals that contribute to an emerging evidence base on how SUD IMD 1115 waivers may affect access and outcomes.

Evaluations generally showed increased use of inpatient and residential SUD treatment, with some evidence of expanded treatment capacity. SUD IMD waivers expanded access to the full treatment continuum, including coverage for inpatient and residential SUD services provided in an IMD. Measures in this section were reported in some form by all states and were commonly used as indicators of access to IMD care. While average length of stay was not typically identified as a central outcome in state evaluation plans, it is included below as an additional utilization measure.

• Inpatient and Residential SUD Treatment: Inpatient or residential treatment utilization increased in five of the six states reporting this measure (CA, IN, MA, NH, and UT). These findings point to greater use of IMD services under the waivers, which was previously limited by the IMD exclusion. Massachusetts reported large increases in newly covered residential rehabilitation services, with average quarterly utilization substantially higher in the second measurement period than the first. California also reported large increases, with the number of unique patients in waiver counties using residential services rising 400% above non-waiver counties. Though SUD treatment rates in Washington IMDs were unchanged, key informants reported that the waiver helped them connect patients to treatment during their “window of willingness,” when motivation may be higher after first presenting for care. They also noted that the waiver may have been particularly helpful for nondisabled, single men, who were often lower on the priority list for treatment.
• Inpatient and Residential SUD Capacity: Inpatient or residential treatment capacity increased in all four states reporting data (IN, MA, NH, and WA). While the evaluations do not show whether states reached sufficient capacity, the findings suggest progress toward expanding capacity for this type of care. In Indiana, the number of residential SUD providers grew from none at the start of the demonstration to 55 by the end. New Hampshire also reported net growth overall, although some facilities closed in the fourth year of the demonstration. Washington reported growth in the number of SUD treatment facilities and Massachusetts reported growth in residential beds.
• Average Length of Stay (ALOS):  Early SUD waivers approved under 2015 guidance often included maximum length of stay limits of 30-days for residential and 15-days for inpatient IMD stays. CMS guidance updated in 2017 instead established a statewide average length of stay of 30 days as a performance monitoring target for residential SUD services in IMDs, intended in part to ensure that IMD services did not displace community-based care.  Under this approach, longer stays for some individuals could be offset by shorter stays for others.  In the three states reporting this measure (CA, IN, and WA), the ALOS was under or near the 30 days performance target. Indiana’s was well below that target at 4.7 days, though qualitative findings from enrollees raised concerns that stays may be too short to support transition and recovery. California’s average was slightly above the 30-day target, at 31.2 days after excluding extreme outliers. These findings suggest that for states reporting  ALOS, average lengths of stay were generally close to waiver targets, but they do not show whether the length of stay was clinically appropriate for all patients.

Evaluations generally showed increases in SUD diagnosis and treatment access, as well as some evidence of expanded community-based treatment capacity. In addition to expanding access to IMD-based care, the waivers were also intended to strengthen access across the broader SUD treatment continuum, including outpatient treatment and evidence-based care for OUD. Community-based treatment, capacity, and medication for opioid use disorder (MOUD) measures reported in this section were often included in state evaluations and aligned with CMS goals and milestones. This analysis uses MOUD to refer to medication-based OUD treatment measures, including measures that state evaluations described as medication-assisted treatment (MAT). Though diagnosis was not explicitly identified in most evaluation plans, it is included below because it was reported by most states and may suggest progress toward improved identification of SUD.

• Diagnosis: Four of the five states reporting measures of SUD diagnosis (IN, MA, NH, and UT) showed increases. Utah reported the largest growth, with a 67% increase in SUD-only diagnoses, though this increase coincided with the state’s ACA Medicaid expansion. Indiana and New Hampshire also reported increases, though more modest at 23% and 11%, respectively. Massachusetts reported a modest increase in the number of members with SUD diagnoses, though the number with an OUD diagnosis declined, which may be due to data challenges. Diagnosis measures varied across states, with some states, including New Hampshire and Washington, requiring both a diagnosis and service use. Increases in SUD diagnoses may reflect increases in treatment use, since a diagnosis is typically recorded when a treatment claim is submitted. Some of the increases may also reflect the period during the pandemic when Medicaid disenrollments were paused and people retained coverage and access to treatment services.
• Any Treatment: Five of six states (CA, IN, MA, UT, and WA) showed increases in treatment use or initiation. These findings generally point to improved treatment access, but the measures differed across evaluations. Indiana reported a 52% increase in enrollees receiving any treatment, while Utah reported a 160% increase in the number receiving treatment. Utah’s unmet needs for SUD treatment decreased, relative to a synthetic control group. In addition to increases in any SUD treatment, Washington also reported increases in treatment initiation, rising by 8.7 percentage points relative to baseline. Massachusetts reported higher outpatient SUD utilization among members with SUD and OUD that remained above baseline, though use declined somewhat overtime and the OUD growth was weaker than the pre-waiver trend. New Hampshire’s evaluation measured whether treatment was initiated and sustained after a new SUD diagnosis, and findings were more mixed. Treatment initiation, defined as initiating treatment within 14 days of diagnosis, declined in later demonstration years, while treatment engagement (two or more SUD visits within 34 days of initiation visit) increased substantially, ending 30% above baseline in year five.
• Access to Medications for Opioid Use Disorder (MOUD): MOUD is a core part of OUD treatment and is recommended for most people with OUD. Of the three FDA-approved MOUD medications, buprenorphine and naltrexone can be prescribed, while methadone is available only through federally certified Opioid Treatment Programs (OTPs). Three of four states reporting this measure (IN, MA, and WA) showed increases in MOUD treatment. In Massachusetts, the share of members with OUD using medication treatment increased from an average of 32% each quarter during the baseline period to 43% by the final quarter of the evaluation period, though this growth was already underway before the waiver. The number of MOUD users in Indiana increased from about 6,000 before the demonstration started to over 15,000 during it, a 156% increase, though this acceleration was not different from the growth trend already underway before the waiver took effect. California reported a net decline, as decreases in methadone use during the pandemic were not fully offset by increases in buprenorphine use, likely because methadone treatment depended on frequent in-person visits. Though New Hampshire did not include a quantitative MOUD measure, most respondents to its SUD residential provider survey reported that medications were accessible to Medicaid members when needed. Some state metrics include medications used to treat alcohol alongside opioid use disorder medications.
• Community-Based Treatment Capacity: All three states reporting community-based or outpatient treatment capacity measures (IN, MA, and WA) described growth in outpatient treatment infrastructure, although they used different measures. Indiana reported growth in SUD outpatient and MOUD-related providers identified through claims data. Using national survey data, Indiana found that the share of all SUD facilities accepting Medicaid increased from 60% to 78%. Massachusetts also reported growth in MOUD provider capacity compared to baseline, though the growth rate was similar to the baseline period. Washington reported large increases in outpatient SUD providers, from 515 to over 9,500, which coincided with shifts from county-based to managed care delivery. California did not report quantitative measures, but county administrators attributed quality of care improvements to a stronger continuum of care and expanded provider networks, which they said made it easier to match patients to appropriate levels of care. Workforce shortages and uneven geographic access remained common constraints.

Evaluations also point to declines in SUD-related ED use, and some improvement in SUD follow-up after acute care. The measures presented in this section were commonly included in state evaluations, though states framed these measures differently, sometimes included as access measures and others as measures of quality. These measures also align with waiver goals to reduce preventable ED use and preventable readmissions through improved access to other services, along with the CMS milestone focused on care coordination and transitions between levels of care.

• ED Utilization: All five states reporting this measure (IN, MA, NH, UT, and WA) showed declines in SUD-related ED utilization over the demonstration period. These findings suggest that ED utilization decreased in waiver states, which may be related to greater availability of other SUD services and could also reflect overall drops in ED utilization during COVID years. Washington also reported declines in ED visit rates per 1,000 members. Indiana was the only state to assess "potentially preventable" ED visits and found that, although overall SUD-related ED use declined, the share categorized as potentially preventable remained stable.
• Care Transitions After ED Use: Follow-up after ED visits for SUD improved in two of three reporting states (IN and WA). Indiana and Washington showed increases in follow-up care after ED visits for SUD, though rates remained relatively low, at 40% in Washington within 30 days of an ED visit and 10% in Indiana within 7 days. Improved care transitions after an ED visit may support continued treatment engagement. In Indiana, the percentage of enrollees with a follow-up within 7 days increased from 7% to 10%. Washington reported a five-percentage point increase in follow-up care within both 7 and 30 days after an ED visit for SUD. California did not report comparable follow-up measures, though county administrators described stronger ED-to-treatment linkages where EDs used navigators. In Massachusetts, there was little overall change in follow-up after ED visits.
• Care Transitions After Residential or Inpatient Treatment: All three states reporting post-discharge measures (CA, IN, and NH) showed improved transitions or follow-up after residential or inpatient care. Increases in transitions to lower levels of care after discharge may support ongoing treatment. California reported an increase in transitions from residential/inpatient to outpatient settings, from about 7% to 10%. Indiana reported greater use of community-based services and MOUD after discharge, with MOUD increasing from 29% to 44% post-discharge in 2020 — and other community-based services increasing from 82% to 93% post-discharge. New Hampshire reported higher follow-up treatment rates at multiple post-discharge intervals

Evaluations do not support clear cross-state conclusions about overdose outcomes. States used different overdose measures, populations, and time periods, and some did not report overdose outcomes at all, which may reflect differences in evaluation plans. For example, California reported fentanyl-related overdose deaths, Indiana and Massachusetts reported Medicaid-specific overdoses, and Washington reported overall overdose deaths that were not limited to Medicaid, while New Hampshire and Utah did not report overdose findings in their summative evaluations. Among states reporting overdose deaths, deaths generally increased. Massachusetts reported nonfatal opioid overdoses and found that nonfatal opioid overdoses declined somewhat, while fatal overdose death rates remained flat among Medicaid enrollees were mostly flat overall but began to decline by the end of the demonstration period. It is difficult to determine how much of any observed overdose change was due to the SUD IMD waiver versus broader conditions during the evaluation period, including the COVID-19 pandemic and the rapid fentanyl-driven rise in overdose deaths.

What challenges were noted in waiver evaluations?

Evaluations pointed to challenges and barriers that may have affected waiver outcomes. These were drawn primarily from key informant interviews, stakeholder surveys, and focus groups, as well as evaluator observations and provider survey data. Not all challenges were reported from every evaluation, and the depth of qualitative reporting varied across states.

Workforce shortages limited access across the treatment continuum. Across several evaluations, states cited shortages of SUD providers as a key constraint on treatment capacity, particularly for certain specialty services and geographies. California reported gaps in youth treatment services. Key informants in Washington pointed to shortages of licensed behavioral health professionals and cited low reimbursement and complex licensing and credentialing processes as contributing barriers. Utah stakeholders described similar challenges and noted that these barriers may make Medicaid participation more difficult for smaller providers with limited administrative support. Indiana also cited structural limits, noting that state caps on the number of opioid treatment programs constrained access to methadone.

Fragmented systems and uneven treatment infrastructure made care transitions harder. Across multiple evaluations, stakeholders described fragmentation when benefits are split across health plans or systems and when certain types of treatment are in short supply, making it harder to connect people to outpatient treatment after crisis or inpatient care. In California, stakeholders described mental health and SUD treatment as separate systems with different financing and rules that can slow referrals across systems, likely complicating coordination of care across mental health and substance use treatment. Utah stakeholders described similar barriers when physical and behavioral health benefits are split across separate plans. One plan described this fragmentation as creating a  “black hole of care coordination,” where enrollees can lose connection to care as responsibility shifts across plans. Washington cited limited discharge information and regional gaps in outpatient capacity as drivers of uncertain handoffs, and Indiana similarly reported difficulty finding appropriate follow-up care close to home after discharge.

Administrative barriers discouraged provider participation and could delay care. Across evaluations, three issues often surfaced: unclear billing and documentation rules, burdensome provider enrollment and credentialing, and delays tied to prior authorization. In California, administrators estimated that about half of case management services initially went unbilled because requirements were unclear, leading the county to absorb costs rather than navigate the unclear billing processes. Across multiple states, burdensome provider enrollment and credentialing processes were cited as deterrents to provider participation. Indiana and Washington further noted that prior authorization (PA) could delay care, especially when requirements were complex or when interpretations varied across managed care plans. For example, even after Indiana introduced a universal PA form, providers reported continued variation across plans in medical necessity interpretations, documentation requirements, and review processes, increasing burden for providers contracted with multiple plans.

Housing instability made it harder to sustain recovery after discharge. Multiple evaluations described homelessness or unstable housing as common among people in SUD treatment. Providers and other stakeholders noted that patients discharged to unstable environments often regressed quickly. Utah clinicians described feeling "helpless” to prevent negative outcomes without stable housing and reported “downward spirals” that could lead to re-hospitalization or overdose. Evaluations in California, Indiana, and Utah also noted that many people leaving residential care lacked appropriate step-down options, such as recovery residences or transitional housing.

State SUD IMD Waiver Evaluation Topics, Hypotheses, and Implementation Features

Summary of selected state evaluation plan research questions and hypotheses, condensed and grouped by topic, with notable waiver features or implementation details.

This table summarizes selected evaluation topics, hypotheses, and implementation features from state SUD IMD waiver evaluation plans and summative evaluation reports. Information is condensed and organized by topic and may not capture all topics, hypotheses, or implementation details. Cost-related topics and hypotheses are excluded because they are outside the focus of this table. Approval date refers to CMS’s initial approval date for the state’s SUD IMD section 1115 demonstration.

KFF analysis of SUD 1115 Waiver evaluation plans and summative evaluation reports

Long-term care (LTC) encompasses the broad range of paid and unpaid medical and personal care services that assist with activities of daily living (such as eating, bathing, and dressing) and instrumental activities of daily living (such as preparing meals, managing medication, and housekeeping). The Department of Health and Human Services (HHS) reports that after age 65, over half of people will at some point need help with at least two activities of daily living, over half will use paid LTC, and over one-third will use some nursing home care. People under 65 with disabilities also rely on LTC for assistance with activities of daily living and instrumental activities of daily living. Medicaid is the primary payer for long-term care (LTC), and KFF analysis of Medicaid claims data found that over half of people who used any Medicaid LTC were under 65. 

Direct care workers play a pivotal role in providing LTC services. They perform demanding, high-stress work for low wages and often no benefits. This has contributed to the long-standing shortages and high turnover rates among direct care workers in both home care and institutional care settings. Recent federal policy changes could further exacerbate the challenge of retaining and growing the workforce to care for the aging population. Specific policy changes that could exacerbate these challenges include the following.

• Broad changes to Medicaid, including $911 billion in reductions to federal spending and Medicaid work requirements, could have implications for the direct care workforce given Medicaid’s outsized role in LTC spending and relatively high Medicaid coverage rates among direct care workers.
• Direct care workers may also feel the impacts of recent changes in immigration policy. Three in ten direct care workers are immigrants, including naturalized citizens and noncitizens, who include lawfully present and undocumented immigrants. The Trump Administration’s intensified immigration enforcement and restrictive policies are deepening anxiety and fear among immigrants of all statuses and could contribute to reduced immigration in the future, which could exacerbate workforce shortages.
• Two proposed rules from the Department of Labor may also have an impact on direct care workers. One rule would roll back minimum wage and overtime protections for home care workers. The other rule would make it easier for employers to classify direct care workers as independent contractors, which would strip them of some labor protections.
• CMS has also delayed enforcement of a provision in the Medicaid Access Final Rule that would have required states to establish an advisory group to provide recommendations on direct care worker provider rates.

This analysis uses the 2024 American Community Survey (ACS) to provide an overview of demographic and socioeconomic characteristics of the direct care workforce, including home health aides, personal care aides, and nursing assistants who work in nursing facilities, residential care facilities, home health, and settings that provide nonresidential services for older adults and younger adults with disabilities (see Methods). 2024 ACS data was released in late 2025 and is the most recent data available. Other KFF analyses on the direct care workforce in nursing homes include RNs and LPNs (collectively referred to as “nurses”), but this analysis excludes nurses and only includes aides that assist older adults and people with disabilities with essential daily tasks. Nurses provide key services for older adults and younger people with disabilities who use long-term care services but are excluded from this analysis because they are socioeconomically and demographically different than aides. Key takeaways include:

• In 2024, there were 2.3 million direct care workers who provided long-term care to people ages 65 and older and people under 65 with disabilities (Figure 1).
• Direct care workers are significantly more likely to be age 50 years or older, female, Black or Hispanic, or immigrants when compared to all other adult workers in the U.S. (Figure 2).
• Direct care workers are also significantly more likely to have a high school degree or less, work part-time, be low-wage, or be covered by Medicaid or be uninsured when compared to all other adult workers in the U.S. (Figure 2).
• The share of direct care workers who are immigrants varies by state, ranging from 0% in Wyoming to 60% in New York (Figure 4).
• At least one in five immigrant direct care workers is from a country within the 75 countries that are part of the Trump administration’s immigrant visa pause, which could further strain the workforce in future years (Figure 5).

In 2024, there were 2.3 million direct care workers who provided long-term care to people ages 65 and older and people under 65 with disabilities (Figure 1). Direct care workers include personal care aides, nursing assistants, and home health aides.Other sources use a similar definition of direct care workers but use data from the Bureau of Labor Statistics (BLS) to estimate the total number of workers, leading to different counts. The BLS data are more recent than the American Community Survey. See Box 1 for a definition of types of direct care workers included in this analysis.

Direct care workers provide long-term care services across a variety of settings, with 66% providing care in home care settings, 22% in nursing facilities, and 12% in residential care facilities (Figure 1). “Home care settings” include “home health” and “services for the elderly and people with disabilities” from Figure 1 and Box 1. Excluding Figure 1 and Box 1, this analysis presents these categories as one combined “home care” category. See Box 1 for definitions of the direct care worker and setting types. 

Direct care workers are significantly more likely to be age 50 years or older, female, Black or Hispanic, or immigrants when compared to all other adult workers in the U.S. (Figure 2). Direct care workers are more likely to be age 50 years or older when compared to all other adult workers (41% vs 33%), as well as more likely to be female (85% vs 47%). They are also more likely to be Black than all other adult workers in the U.S. (30% vs 11%). They are also more likely to be Hispanic when compared to all other adult workers in the U.S. (23% vs 19%).Immigrants, including naturalized citizens and noncitizens, make up 30% of the overall direct care workforce, higher than the share of all adult workers in the U.S. who are immigrants (18%).

Direct care workers are also significantly more likely to have a high school degree or less, work part-time, be low-wage, and be covered by Medicaid or be uninsured when compared to all other adult workers in the U.S. (Figure 2). Direct care workers are more likely to have a high school degree or less than compared to all other adult workers (55% vs. 31%). They are also more likely to work part-time and make less than $35,000 annually (referred to as “low wage”) when compared to all other adult workers (39% vs 18%; 66% vs 31%). Direct care workers are also much more likely to be covered by Medicaid (32%) than all other adult workers (9%) and more likely to be uninsured when compared to all other adult workers (13% vs. 9%).

Direct care workers in home care settings are different than direct care workers in nursing homes or residential care facilities (Appendix Table 1). A higher share of direct care workers in home care settings are immigrants (33%) when compared to direct care workers in nursing facilities (25%) and residential care facilities (23%) (Figure 3). Additionally, a higher share of direct care workers in home care settings work part-time (42%) when compared to workers in nursing homes (24%) and residential care facilities (25%). Similarly, a higher share of home care workers are low-wage (72%) when compared to those working in nursing homes (56%) or residential care facilities (58%) (Appendix Table 1).

Immigrants make up 30% of the overall direct care workforce providing long-term care services (Figure 3). Naturalized citizens make up a significantly larger share of direct care workers in home care settings (19%) when compared to nursing facilities (16%) and residential care settings (14%). Home care workers are also significantly more likely to be noncitizen immigrants (14%) when compared to workers in nursing homes (9%) and residential care facilities (9%). Noncitizen immigrants include both lawfully present and undocumented immigrants.

The share of direct care workers who are immigrants varies by state, ranging from 0% in Wyoming to 60% in New York, and, in many cases, reflects the share of overall immigrant workers (Figure 4). In 20 states, including many Southern states, the share of direct care workers who are immigrants is under 10%. In 9 states, the share of direct care workers who are immigrants is between 10% and 19%. In 9 states, 20% to 29% of direct care workers are immigrants, and in the remaining 13 states (including DC), 30% or more of direct care workers are immigrants. The share of direct care workers who are immigrants is strongly positively correlated with the overall share of workers in a state that are immigrants.

Immigrants from 13 countries make up nearly two-thirds (65%) of all immigrant direct care workers providing long-term care in the U.S. (Figure 5). The remaining one-third of immigrant direct care workers come primarily from a mix of countries in South and Central America, Africa, and Asia. Six countries each account for 5% or more of immigrant direct care workers: Mexico (14%), Dominican Republic (7%), Philippines (7%), Jamaica (6%), China (6%), and Haiti (6%).

At least one in five immigrant direct care workers is from a country that is part of the Trump administration’s immigrant visa pause (Figure 5). Among the top 13 countries that make up nearly two-thirds of immigrant direct care workers, five of them are part of the Trump administration’s immigrant visa pause (Jamaica, Haiti, Nigeria, Cuba, and Ghana) that impacts 75 countries. While the immigrant visa pause is facing a court challenge, it currently remains in effect. On June 25, 2026, the Supreme Court issued a ruling allowing the Trump administration to proceed with terminating Haiti’s TPS designation, which could impact over 300,000 Haitian TPS holders. These policies could further strain the workforce in future years.

Methods

Data: These findings are based on KFF analysis of the 2024 American Community Survey (ACS) 1-year Public Use Microdata Sample (PUMS) files. The ACS includes a 1% sample of the U.S. population, and the subset of direct care workers used here includes 19,612 observations.

Identifying Direct Care Workers in ACS: Direct care workers are those who fall into the following occupation codes: Home health aides (3601); Personal care aides (3602); and Nursing assistants (3603). This analysis only includes those who work in the following industries: Home Health Care (8170), Nursing Care Facilities (8270), Residential Care Facilities (8290), and Individual and Family Services (8370). “Individual and family services” are also referred to as “services for the elderly and people with disabilities.” Home health care and individual and family services are collapsed into “Home care” for this analysis. These industries capture most workers providing long-term health services. The ACS only includes the primary industry and occupation of a respondent’s current or most recent (in case the respondent is not currently working) job.

We define the direct care workforce as all individuals 18 and older who earned at least $1,000 during the year and indicated that their job was in both the long-term care industry and occupation codes listed above. The comparison group “All Adult Workers in the U.S.” includes all individuals 18 and older who earned at least $1000 during the year.

Health Insurance Coverage in ACS: The ACS asks respondents about their health insurance coverage at the time of the survey. Respondents may report having more than one type of coverage; however, individuals are sorted into only one category of insurance coverage. See notes here for more information on the insurance coverage hierarchy.

Identifying Immigrants in ACS: Immigrants are identified as those who report their citizenship status (variable name: CIT) in ACS as being a “U.S. citizen by naturalization” or as “not a citizen of the U.S.”, with the former being grouped under “naturalized citizens” and the latter being grouped under “noncitizen immigrants” for the purpose of this analysis. Noncitizen immigrants include lawfully present and undocumented immigrants.

Appendix

Characteristics of Direct Care Workers and All Other Adult Workers

*Indicates statistically significant difference between Direct Care Workers and All Other Adult Workers at the p<0.05 level. NH=Non-Hispanic.

KFF analysis of American Community Survey, 2024