As of the survey published 2025-11-20, more than 600,000 people across 41 states were on waiting or interest lists for Medicaid home care that federal waiver rules permit states not to provide

What is happening?

More than 600,000 people sat on Medicaid home-care waiting or interest lists across 41 states in the state survey KFF published on 2025-11-20, and the average reported wait was 32 months. The total rose 14 percent over the prior year, with 29 states reporting an increase and 12 reporting a decrease, and the average wait fell from 40 months in the 2024 round.

Almost three quarters of the people counted in that round — 74 percent — have an intellectual or developmental disability. Older adults and people with physical disabilities are 23 percent, and everyone else, including medically fragile children and people with brain or spinal cord injury, makes up the remaining 4 percent. The reported wait splits sharply by group in the same survey: 37 months for intellectual and developmental disability waivers, 63 months for autism-specific waivers, and 15 months for older adults and people with physical disabilities.

Over a longer window the national total moves the other way. The same survey series counted 819,886 people in FY2018 and more than 692,000 in 2023. Part of that decline is a change in counting rather than a change in service — between 2018 and 2020 the total fell by 155,000, and nearly half of that came from one state introducing an eligibility screen that removed close to 70,000 people from its list at once.

None of this is a malfunction of Medicaid. It is a permitted feature of one Medicaid authority. Section 1915(c) lets a state forego certain Medicaid requirements in order to target home and community-based services to specific populations, cap the number of beneficiaries who receive them, and hold a waiting list for people who cannot be served under the cap. State plan authorities carry no such permission, which is why MACPAC describes state plan HCBS as putting a state at greater financial risk than a waiver.

Whose problem is this?

The body that could end the queue outright is Congress, and the body that moves it in practice is a state legislature writing a one-year appropriation. Neither is the body that publishes the number.

Where does this problem end?

A person who qualifies for nursing facility care receives it because no queue is permitted there. That asymmetry is the whole boundary of this document, and it is written into the authority a state chooses when it decides to deliver care at home instead.

What is the state now, and what should it be?

Now

Needs a new measurementthe target state: no source opened here names a number of slots that would clear the lists, a maximum acceptable wait, or a date by which either should happen. The Olmstead opinion requires only that a list move at a pace it calls reasonable and declines to quantify that pace. CMS waiver instructions require only that selection policies be objective and applied consistently across the service area of the waiver. KFF and MACPAC both publish the wait and neither proposes a benchmark to measure it against.

How big is it?

Between 275,000 and 600,000 people, as of the KFF state survey published 2025-11-20.

The high bound is everyone a state has written down: more than 600,000 across 41 states. The low bound removes the block that no state has checked. Six states — Florida, Iowa, Oklahoma, Oregon, South Carolina and Texas — add people to a list without first screening them for Medicaid eligibility, and those six hold 325,000 people, 54 percent of the national total. Some unknown share of them would not qualify if screened. The only calibration available is Ohio, where introducing an eligibility screen between 2018 and 2020 removed close to 70,000 people from a list without serving any of them.

The gap between the two bounds is not measurement noise but a difference between states about when eligibility gets checked. That boundary has itself moved twice in seven years: eight states did not screen in FY2018, holding 61 percent of the national total, and six did not in 2025, holding 54 percent. The floor is soft in a further way worth stating rather than hiding. KFF describes the screening states as checking eligibility before or while a person is on the list, whereas the earlier MACPAC wording is before placement. Those are different guarantees, so 275,000 is a floor on listed people in states that check, not a certified count of eligible people denied service. No source reports the latter.

Neither bound should be read as people receiving no help. More than 80 percent of people on waiver waiting lists are eligible for personal care through the Medicaid state plan of their state, and MACPAC found that people waiting for a waiver slot commonly receive state plan services, EPSDT for children under 21, school-based services, or a lower-tier waiver in the meantime. The honest sentence is that 275,000 to 600,000 people have asked for or qualify for a level of home support their state is legally permitted not to provide, and that no source in existence says how many of them are receiving nothing.

Under what conditions does it arise?

1. The waiver is chosen because it permits the queue. MACPAC describes 1915(c) as the authority that allows a state to forego certain Medicaid requirements in order to target benefits to specific populations, cap the number of beneficiaries served, and create waiting lists for people who cannot be served under the cap. The corollary is stated just as plainly: state plan authorities do not allow waiting lists, because state plan services must be available to every enrollee for whom they are medically necessary, so state plan HCBS puts a state at greater financial risk than a waiver does. The queue is not a defect of the instrument. It is the thing the instrument buys.

2. Home care is optional and nursing facility care is not. KFF states this asymmetry in two separate briefs. A state official told MACPAC that because HCBS are optional they are always more vulnerable to budget cuts than nursing facility services. Budget pressure therefore has one target and not the other.

3. Capacity is a negotiated budget line, not a function of how many people qualify. A 1915(c) waiver is approved by CMS for up to five years and names a maximum capacity for each operating year; changing that number requires an amendment. MACPAC found that state funding was cited as the most important factor in many states for increasing waiver capacity, with explicit backing from a governor or a legislature making the difference.

4. The queue has a legal bypass, reserved for people who are already institutionalized. States hold reserved capacity so that certain people can enroll immediately without joining the list, and MACPAC found this was most often used for people leaving institutions, including transitions under Money Follows the Person. The fastest route to the front of the line runs through the setting the waiver was meant to avoid.

5. The most common ordering rule rewards joining before you need to. Of the 254 waivers reviewed in March 2020, 199 document how the list is managed, and first come first served is the most common approach at 62 waivers. MACPAC records that this can encourage individuals to seek enrollment in anticipation of future needs, and describes families adding young children to intellectual and developmental disability lists years before the need arrives.

What has been tried?

Two of these attempts move the number without moving a person. Adding an eligibility screen and clearing duplicates both shorten a list, and the survey that publishes the national total cannot separate that from service.

What was found?

Why is it still unsolved?

Institutional exemption — the entitlement exists and is fully operative, and the queue rests on a lawful carve-out from it that a federal agency reviews and signs.

The people on these lists are not people the system failed to notice. Every one of them is on a document CMS approved. A 1915(c) waiver names a maximum capacity for each operating year, runs for up to five years, and can only be enlarged by amendment, so the number of people a state may serve is settled in advance of knowing how many will qualify. What the waiver trades away is comparability, freedom of choice and statewideness, and what it obtains in exchange is legal permission to serve fewer people than qualify. A state that delivered the same services under a state plan authority would have no ability to hold a list at all, which is precisely why MACPAC describes that route as the financially riskier one.

The second movement is that the queue is legally sufficient rather than legally deficient. The Olmstead opinion is where a reader expects the remedy to be, and it is instead where the safe harbor is: a state that has a comprehensive, effectively working plan and a list moving at a reasonable pace, not driven by an effort to keep its own institutions full, meets the reasonable-modifications standard. No source opened this round quantifies that pace. A state whose list moves is compliant, and a state whose list moves slowly is, absent litigation, also compliant. MACPAC records litigation moving two individual states and nothing setting a national floor.

The third movement is why nothing registers. There is no adverse determination at the moment a person joins a list — there is a slot count and a date, and the person who qualifies and does not receive is administratively still in process, for 32 months on average and, by the interview evidence MACPAC collected, for as long as 14 years in one state. Nobody is denied, so no adverse determination accumulates in a record that a court or a legislature could later read, and whether the placement itself can be contested is a question no source opened this round answers in either direction. Meanwhile the lever that actually moves capacity is an annual state appropriation, which is why 29 states rose and 12 fell in the same year and why no national trajectory exists to point at. The 2025 reconciliation law now presses on the optional side of that ledger by USD 911 billion over ten years, and eight states had already said they intend to newly cap waiver participants in FY2026 before it was enacted.

What observation would mean it is solved?

Candidates — (a) the reported national list total falls (b) the reported average wait falls (c) home and community-based services become a benefit that cannot carry a waiting list.

(a) alone is the weakest of the three. One state removed close to 70,000 people by introducing an eligibility screen, and states that cut lists sharply told MACPAC they did it partly by clearing duplicates and people who had moved, died or no longer needed services. The set of states that add people without screening changed between FY2018 and 2025, so the FY2018 total of 819,886 and the 2025 total of more than 600,000 do not count the same population. A falling total is compatible with more people being served and equally compatible with fewer people being counted.

(b) alone is worse, because of how the number is made. The 32-month figure comes from 33 of 41 responding states, is not weighted by list size, and the eight states that did not answer are not named. In the same round the list grew 14 percent while the average wait fell 20 percent, and no source explains how both can be true. Behind the average there is no median, no percentile and no state-level figure, against an interview range running from 291 days to 14 years.

(c) alone counts law rather than care. A benefit that cannot hold a list can still be limited in other ways: 34 states limited personal-care hours or visits in FY2025 and 47 used prior authorization for state plan home care. All 50 states and DC report direct-care workforce shortages and 41 report permanent provider closures within a year, so a slot with no worker attached is served by the measure and unmet in fact. The three have to be read together, and (a) has to be read against what each state changed about its counting that year.

What is it connected to?

Fills with researchthe direct-care workforce and immigration policy, the effect of the 2025 reconciliation law on optional Medicaid benefits as states write FY2026 and FY2027 budgets, the enforcement history of Olmstead since 1999, unpaid family caregiving as the absorbing layer, and the HCBS content of the CMS Ensuring Access final rule. Relation type and evidence grade were not confirmed in this round.

What these sources do not say

• No target exists anywhere. Not a number of slots that would clear the lists, not a maximum acceptable wait, not a date. The Olmstead opinion requires a reasonable pace and declines to quantify it, CMS waiver instructions require selection policies to be objective and consistently applied and say nothing about speed, and both KFF and MACPAC report the wait without proposing a benchmark for it.
• Wait time is not reported to the federal government. MACPAC states it flatly, and notes that the survey does not report at the state level either. The single national wait-time figure in American policy comes from a foundation survey of state officials, now in its 23rd round, whose series page gives no response rate, no fieldwork dates and no statement of whether participation is voluntary.
• The drop from 40 months to 32 months is reported and never explained. The list grew 14 percent in the same round in which the average wait fell 20 percent. Both numbers are published on the same page with no account of how they fit — not a change in composition, not a change in the responding set, not faster service. The earlier decline from 2018 to 2020 is explained in detail; this one is not explained at all.
• Nobody defines when the wait starts. Application, screening, slot request or something else — no source says, and because screening happens at different points in different states, MACPAC concludes the lists are not comparable across states. No source proposes a common definition and no body claims authority to impose one.
• The distribution behind the average is never published. There is one 63-month figure for autism waivers and an interview range from 291 days to 14 years, which together imply a long right tail. There is no median, no percentile, no state-level wait time, and no count of how many people have waited more than five years.
• Whether placement on a list can be appealed. The MACPAC review dedicated to how states administer these lists never uses the words appeal, fair hearing or due process. That silence is reportable and it is not the same as establishing that no such right exists — nothing opened this round settles the question in either direction.
• How people meet their needs while waiting, and what becomes of those who leave without a slot. MACPAC states directly that states do not track how people may be getting their needs met as they wait, so no source can say how many of the 600,000 receive nothing at all. Nobody counts deaths on a list, entries into institutions from a list, or people who simply stop waiting; MACPAC interviewees said it was possible that people enter institutions to obtain services while waiting, and left it there.
• Nobody counts people who hold a slot and cannot find a worker, and for 55 of 254 waivers the existence of a list is itself undocumented. All 50 states and DC report shortages and 41 report permanent provider closures within a year, yet no source quantifies authorized service hours that go unfilled — those people are served on paper and appear in no queue. And in the March 2020 review, 55 waiver documents do not mention waiting lists at all, so whether one exists, and how the state manages capacity and unmet need, is unknown. A federal agency approved all 254 of those documents.

See the evidence

No primary federal document was read this round. Every attempt at one failed: Medicaid.gov, Congress.gov and CMS.gov returned HTTP 403, and the eCFR and Federal Register redirected to a bot-block interstitial. The consequence is specific and worth naming — the central legal claim of this document, that 1915(c) permits a state to cap enrollment and hold a list while state plan authorities do not, rests on MACPAC stating it twice in one brief and once more on a separate topic page, with KFF stating it independently in the same terms. It is well corroborated among secondary authorities and unconfirmed against the statute or the regulation. One primary source was read directly, the Olmstead opinion, and it is the source of the safe harbor rather than of a remedy. Where the sources disagree the disagreement is left standing rather than resolved. The national total reads 819,886 for FY2018, more than 692,000 for 2023 and more than 600,000 for 2025, but the set of states that add people without screening turned over in between, so the endpoints do not count the same population; the average wait reads 39, then 40, then 32 months across responding sets that change every round and are never named; the share held by non-screening states reads 61 percent across eight states and then 54 percent across six; the HCBS share of Medicaid LTSS dollars reads 59 percent for FY2016 from MACPAC and 56 percent for 2018 from KFF, and neither acknowledges the other. Two figures are deliberately not treated as independent confirmation: the count of 41 states reporting permanent provider closures appears identically in two survey rounds with different fieldwork years, and it is cited here from one round only. Two counts of family caregivers, 63 million as of 2025 and 38 million as of 2021, use different definitions and are not reconciled by either publisher, so neither is used to support a claim here. This is a Path A output, research-based rather than observation-linked, so observation_refs is empty and provenance_mode: press-derived.

People affected

Estimated range 275,000–600,000 As of 2025 state survey published 2025-11-20

Sensitivity The width of the range, 325,000 people, is not measurement noise. It is a difference between states about when eligibility is checked, and it is more than half the headline figure. That boundary itself moves: eight states did not screen in FY2018, holding 61 percent of a national total of 819,886, and six did not in 2025, holding 54 percent of a total above 600,000, with three states leaving the set and one joining. The FY2018 and 2025 totals therefore do not count the same population. Neither bound counts unmet need that produces no list at all: people in states that do not cover the applicable service, people subject to per-participant limits rather than an enrollment cap, with 34 states limiting personal-care hours or visits in FY2025, people who hold a waiver slot and cannot find a worker, with all 50 states and DC reporting direct-care workforce shortages and 41 reporting permanent provider closures in the round published 2026-01-05, and people in the roughly ten states that keep no list. In the opposite direction, neither bound is a count of people receiving nothing. More than 80 percent of people on waiver waiting lists are eligible for personal care through the Medicaid state plan of their state, and MACPAC found that people waiting commonly receive state plan services, EPSDT for children under 21, school-based services or a lower-tier waiver. No source in existence reports how many receive nothing at all, because MACPAC states that states do not track how people meet their needs while waiting.

Regional breakdown The only national source, the KFF survey of state HCBS officials, did not publish waiting-list figures at the state level in the round published 2025-11-20, and MACPAC states that this information is not reported to CMS. The six non-screening states are named with a combined total of 325,000 but no per-state values were opened this round. MACPAC further concludes that the lists are not comparable across states, because screening happens at different points in different states and no source defines when the wait begins, so a state-level split could not be constructed even from the fragments available.