Methodology

Most facility data here is reproduced from public U.S. government sources: the CMS Provider Data Catalog (nursing homes, home health, hospice), HUD and USDA (affordable senior housing), and state licensing and regulator records (assisted living, CCRCs). Cost figures come from the CareScout (Genworth) 2025 Cost of Care Survey. One cost figure is not theirs: CareScout does not survey memory care, so where we show a memory-care median we derive it — the state's assisted-living median plus $1,200/month, the typical secured-dementia-unit premium. It is labelled as our estimate wherever it appears, because it is one. We add no opinions to the data; we organize it and translate the regulatory language.

Where the Medicaid numbers come from — and why that matters

The per-state Medicaid figures on this site (income and asset limits, the Community Spouse Resource Allowance, home-equity limits, the look-back period) are not from a government source. They are reproduced from the American Council on Aging's annual compilation at medicaidplanningassistance.org, which gathers them from the 50 state Medicaid agencies. We use it because no single federal table publishes all 50 states' long-term-care limits in one place. Two things you should know. First, it is a private compilation, so a transcription error or a mid-year state change can reach this site without a government record contradicting it. Second, the organization that publishes it earns referral income from Medicaid-planning services — we take none, but you are entitled to know the provenance of the numbers, especially these ones. Treat every Medicaid figure here as a planning estimate and confirm it with your state Medicaid agency or a certified elder-law attorney before you act on it. Where a state page or calculator shows one of these numbers, it names this source and its year on the page.

Assisted-living coverage varies by state — a lot

Assisted living is licensed by states, not by CMS, and there is no federal equivalent of the nursing-home data. What we can publish therefore depends entirely on what each state publishes. We list licensed communities in 49 states and Washington, DC — every state except Idaho, whose licensed-facility roster sits behind a bot check that blocks bulk access, so rather than publish a stale or partial Idaho list we publish none. Of the 50 states, 19 publish a downloadable enforcement file and 31 do not, so a large share of communities show a roster entry with no inspection or violation record at all. An empty record is not a clean record, and there are two different reasons one can be empty: the state publishes no bulk file at all, or it publishes one and this community is not in it. Those are not the same thing — the first tells you nothing about the community, the second tells you it drew no posted finding in a file that would have listed it — and each community’s page says which of the two applies to it, rather than leaving you to guess. And because each state grades, counts, and posts violations under its own rules, AL flag counts are not comparable across state lines; a community in a state that posts every complaint will look worse than an identical one in a state that posts only final actions. Each state’s AL page states what we have for that state.

We un-blend the star rating

The CMS overall 5-star rating combines three things: an independent health-inspection score assigned by state surveyors; a staffing score built from payroll-based (PBJ) hours drawn from facilities' payroll records; and a quality-measures score that leans heavily on clinical data facilities report about themselves. Investigations have repeatedly shown the staffing and quality components can inflate the overall score relative to what inspectors find. We show all three separately and flag any home whose staffing and quality ratings run two or more stars above its inspection score. One caveat we state on every page: CMS assigns the health-inspection star by ranking each home within its own state's distribution, so a 5-star inspection score in a lenient-survey state is not directly comparable to a 5-star home in a strict one — read the actual citations, not just the star.

The quality-measure benchmark is ours, and it is a median

On each nursing-home page we compare the home's quality measures against a benchmark column labelled typical state. That figure is not a CMS-published national average, and it is not an average at all: it is the median of the 50 state averages, computed by us. We do it this way because a plain average across every home is distorted by a handful of states' reporting practices — on long-stay depressive symptoms, Illinois codes the measure several times higher than most states, which by itself pushed an all-homes average to about twice what a typical state records, so every home in the country looked good against a bar no state actually sits on. A median of state averages resists that and leaves well-behaved measures essentially unmoved. The cost of the choice, which you should know: it weighs every state equally, so California's roughly 9,400 homes count no more than Wyoming's few dozen — it describes the typical state, not the typical home. Where a state's own average diverges from that benchmark by 2x or more we say so on the row and compare the home with its own state instead. We also do not compute confidence intervals: CMS does not publish the measure denominators, and a statistic we cannot compute is not one we will estimate.

Deficiencies, in plain English

We show each health-inspection citation with its official federal description, its date, and its position on the CMS scope-and-severity grid (A–L). We label the four tiers plainly: immediate jeopardy (J–L), actual harm (G–I), potential for harm (D–F), and no harm found (A–C).

Fines and ownership

We total federal fines and payment denials from the CMS penalties file, and we reproduce the ownership records CMS publishes. CMS ownership data is known to under-report private-equity and REIT ties, so we present it as a starting point, not a complete map.

Red flags are defined, not editorialized

Every flag on a facility page is triggered by an objective, disclosed rule from the CMS record (for example, 'one or more immediate-jeopardy citations') with the underlying data shown alongside it. We do not assert why anything happened or who is at fault — the inspection record speaks for itself.

The planner and calculators

The free tools use transparent, editable assumptions and current-year figures: 2026 federal income-tax brackets, standard deductions, and the estate exemption (IRS Rev. Proc. 2025-32); the IRS Uniform Lifetime Table for RMDs and the SECURE 2.0 RMD ages (73, or 75 for those born in 1960 or later); statutory Social Security rules; and CareScout cost-of-care medians. The planner runs a Monte-Carlo simulation and shows its assumptions on screen. These are planning tools, not individualized financial, tax, or legal advice — confirm any real decision with a fee-only planner or an attorney.

Freshness

CMS data is refreshed monthly; HUD, USDA, and state data on their own schedules. Each page cites its own source and vintage. Inspection reports can lag real conditions; the absence of a citation does not guarantee the absence of a problem.

When we cannot verify that a source is current, what we do next depends on what the record holds. A facility with citations or penalties on file still shows them, with a note saying the set may be incomplete — everything shown did happen, and more may have happened since. A facility whose record looks empty is different: we withhold it rather than show it, because an empty record we cannot verify is indistinguishable from a clean one, and showing it would let a gap in our data read as a clean bill of health.

The same rule governs anything built on top of that record. A facility whose inspection record we cannot verify is left out of rankings rather than ranked on what remains, and we say how many we left out — a ‘worst homes’ list assembled from partial data is a list of the ones we could see, which is a different claim. Figures derived from a stale record go with it: the CMS overall star rating is built from the health-inspection score, so when that inspection record is stale we withhold the star too rather than publish a rating whose foundation we cannot stand behind.

Where a fact depends on a state certifying it, we only assert it where that state's own roster does. We describe a community as memory care only where the state licenses or designates it as such; elsewhere we name the designation the state actually issues, or say nothing. We do not infer the category from a facility's own marketing.

Editorial standards & corrections

Eldercare Lens is an independent publication, paid for by display advertising and by nobody it writes about. Every factual claim traces to a named public source with its vintage shown on the page; figures that change yearly (tax, Medicaid, and benefit numbers) carry a ‘reviewed’ date and are checked against the issuing agency. Health and money content is written to be YMYL-careful: we explain, we cite, and we tell you to confirm consequential decisions with the relevant professional or agency — we never present our tools as individualized medical, legal, tax, or placement advice. Found something wrong or out of date? Email us from the contact page — messages are read and errors are corrected. This is a small project rather than a newsroom, so we do not promise a turnaround we might not be able to keep; what we can promise is that a correction gets made and, where the error is in the government record rather than ours, traced to its source. Our editorial policy sets out what we publish and what we withhold, and our corrections policy covers what happens when a facility disputes something.