When Patients Can't Understand Their Own Care

The average hospital discharge summary is written at a college reading level.

The average American reads at an eighth-grade level. Fourteen percent of adults — roughly 30 million people — score Below Basic in prose literacy. And for the more than 27 million adults living with cognitive disabilities — acquired brain injuries, intellectual disabilities, dementia, or the cognitive effects of stroke — the gap between what they're given and what they can understand isn't a nuisance. It's a safety hazard.

The stakes are clinical, not abstract

Health literacy failures kill people. The Agency for Healthcare Research and Quality has documented for decades that patients who don't understand their discharge instructions are significantly more likely to be readmitted. Medication non-adherence — often driven by patients not understanding dosage, timing, or interactions — costs the U.S. healthcare system an estimated $100-300 billion annually and contributes to 125,000 deaths per year.

When a patient with a cognitive disability leaves the hospital with a three-page discharge summary describing medication schedules, wound care protocols, warning signs, and follow-up requirements — all written in clinical language — and she cannot understand it, the consequences are not theoretical. They are readmissions, complications, and preventable suffering.

Now consider: that same patient can paste her discharge summary into an AI tool and ask, "Can you explain this to me in simple language?" The AI responds: take this pill twice daily with food, change the bandage every morning, call this number if the redness spreads, and see Dr. Martinez on April 15th.

She understands. She follows the instructions. She recovers.

Hospitals are building walls around this information

Patient portals — the primary digital interface between hospitals and their patients — are increasingly deploying anti-scraping measures, copy-paste restrictions, and bot-detection systems. These tools are designed to protect data from unauthorized extraction. But they don't distinguish between a data scraper and a patient who wants to paste her own lab results into an AI to understand them.

The result: patients are locked out of using comprehension tools on their own health information.

CAPTCHAs designed to escalate in difficulty when they detect assistive technology patterns mean that "users who most need accessibility support face the hardest verification tasks." Anti-bot systems that break screen readers also break the copy-paste workflow that AI comprehension depends on.

This isn't a security decision. It's an accessibility decision — and most hospital IT departments don't realize they're making it.

The legal framework is clear

Healthcare providers have some of the strongest effective-communication obligations in American law:

• Title III of the ADA requires private hospitals and healthcare providers to furnish auxiliary aids and services for effective communication with patients who have disabilities.
• Section 504 of the Rehabilitation Act imposes the same obligations on any healthcare facility receiving federal funding — which is nearly all of them, through Medicare and Medicaid.
• Section 1557 of the Affordable Care Act explicitly prohibits discrimination on the basis of disability in health programs receiving federal financial assistance and requires "appropriate auxiliary aids and services" for individuals with disabilities.

The Department of Health and Human Services has stated that these obligations extend to digital platforms and electronic communications. A patient portal is not exempt because it's digital. The obligation to provide effective communication travels with the information, regardless of format.

When a hospital blocks a patient from using a comprehension tool on her own medical records — records she has a legal right to access under HIPAA — the hospital may be creating an accessibility barrier under these statutes. Not intentionally. Not maliciously. But the effect is the same.

What AI comprehension looks like in healthcare

The use case is narrow and specific:

• A patient pastes discharge instructions into an AI to get a plain-language summary
• A patient asks AI to explain what their lab results mean in simple terms
• A patient uses AI to organize a complex medication schedule into a clear daily checklist
• A caregiver uses AI to translate a specialist's report into language they can use to manage a family member's care

In every case, the patient already has the information. They are not requesting access to something restricted. They are trying to understand something they've already been given — the exact scenario the ADA's "effective communication" requirement was designed to address.

This is not AI acting on a patient's behalf. It is not AI making medical decisions. It is a patient exercising their right to understand their own care.

The universal-design case is even stronger

Health literacy is not only a disability issue. Limited health literacy affects an estimated 88 million American adults. It correlates with lower use of preventive services, higher hospitalization rates, higher healthcare costs, and higher mortality.

When hospitals make information more understandable — whether through plain-language summaries, visual aids, or AI comprehension tools — every patient benefits. Elderly patients managing complex medication regimens. Patients whose first language isn't English. Patients navigating a diagnosis they've never encountered before.

The disability-rights framework provides the legal leverage. But the beneficiaries extend far beyond the disability community. This is the curb-cut effect in digital healthcare: accommodations designed for people with specific needs make the system better for everyone.

What needs to change

AI Access Alliance is working on three healthcare-specific interventions:

1. Model institutional policies that ensure patient portal accessibility includes AI comprehension tool compatibility — not as an optional feature, but as an effective-communication requirement.
2. Federal agency guidance asking HHS to clarify that blocking AI comprehension tools on patient-facing platforms may constitute an accessibility barrier under Section 1557 and Title III.
3. Evidence documentation — real patient stories of comprehension barriers in healthcare settings, contributing to the human evidence base that makes the legal argument undeniable.

Healthcare providers are not the enemy. Most hospitals are trying to protect patient data — a legitimate goal. But data protection and comprehension access are not in conflict. A patient pasting her own discharge summary into an AI tool is not a data breach. It is a patient trying to understand her care.

The question is not whether to protect patient data. It is whether to block patients from understanding it.