We ran 1,028 scans of real healthcare websites for WCAG 2.1 AA. Here's what we found.
Most accessibility statistics you'll see cite the general web. We scanned real websites belonging to real US healthcare providers — the exact population the HHS Section 504 rule covers — against WCAG 2.1 AA, the standard the rule names. Here's the honest breakdown.
Who and how
Every organization in this dataset comes from the HRSA Data Warehouse — the official, public directory of health centers that receive HHS federal financial assistance. That matters because it means every single one of them is, by definition, in scope for the Section 504 rule. There's no "does this apply to them" guesswork in this data the way there would be for a generic list of businesses.
Each site was scanned with a real headless browser running axe-core against WCAG 2.1 Level AA — the homepage plus the highest-risk pages a patient would actually need (appointment booking, patient portal login, contact, intake forms), prioritized in that order. No sampling tricks, no synthetic data: 1,028 real crawl-and-audit runs on 9 and 10 September 2026. On 25 September we re-ran the two headline groups below and archived every per-site result, so each figure on this page can be checked.
The headline number
A random sample of 100 organizations drawn from the full pool of 1,441 (94 scanned successfully) came back with 91.5% showing at least one failure type that shows up routinely in filed US web-accessibility complaints — missing alt text, unlabeled form fields, insufficient color contrast, and similar. Only 6.4% had no automatically detectable failures. This is not a cherry-picked worst case. It's the baseline across the population the rule actually covers.
An earlier random sample on 9 September came back at 85.6%. Its per-site results were not kept, so on 25 September we drew a new sample with a fixed seed, re-scanned it and archived the list and every result. The figure above is from that re-scan.
The most common finding, by a wide margin
Color contrast was the single most common failure type across every segment we scanned in the first run, appearing on 43.6% to 74.5% of sites depending on the cohort — more common than missing alt text, which is usually the headline finding in general-web studies. It's also the easiest one for a non-technical reader to picture: text that's genuinely hard to read for patients with low vision, color blindness, or anyone checking a website on their phone in daylight.
Does location make a difference?
We tested a specific hypothesis: does litigation exposure concentrate in small organizations in states with heavy ADA Title III filing activity? We took New York, Florida and Illinois (New York and Florida alone accounted for 63% of federal web accessibility filings in 2025) and scanned every small health center there, meaning one to three delivery sites: a full census of 28 organizations, not a sample. In the first run this group came back at 92.9%, above the 85.6% baseline of that day, which suggested these clinics were more exposed than most. The 25 September re-scan does not bear that out.
| Metric | Small clinics in NY, FL and IL census of 28, 25 Sep | Random baseline 100 sites, 25 Sep | Small clinics elsewhere 100 sites, 9 Sep |
|---|---|---|---|
| Common-complaint failure present | 92.6% | 91.5% | 81.1% |
| No automatically detectable failures | 7.4% | 6.4% | 11.6% |
| Scan completed successfully | 96% | 94% | 95% |
Scanned the same day, small clinics in New York, Florida and Illinois (92.6%) and the random baseline (91.5%) are level. With only 28 organizations, a single site moves the rate by about 3.6 points, so the earlier gap was within the noise. The control column comes from the first run and has not been re-scanned. The honest reading: these failures are close to universal among health centers, wherever they are and whatever their size.
What this doesn't mean
This is a scan result, not a legal conclusion
The scans in this study used rule checks (axe-core), which reach only part of WCAG: 12 to 15 of the 50 success criteria on the clinic websites we measured. Rule checks cannot judge whether alt text is meaningful, whether keyboard focus order is logical, or whether a patient using a screen reader can actually complete a booking. A finding rate this high says a lot about where the industry stands technically. It says nothing about any specific organization's legal exposure, which depends on facts no automated scan can see. If you want to know where your own site stands, the fastest way is to look directly.
See where your own site stands
Free WCAG 2.1 AA scan of your homepage plus your four highest-risk pages. Report emailed as soon as the scan finishes — no account, no card, no obligation.
Scan my website free