You get home from a procedure with six pages of discharge instructions. One line tells you to resume a blood thinner “per prior regimen.” Another says to call the office if you notice “significant” swelling. You read both twice. You still do not know what to do tonight.
For most of the last twenty years, the official name for that moment was low health literacy, and the phrase pointed at you. It described your reading skill, your capacity, your ability to cope with the paperwork the system hands out. If a sentence lost you, the deficit was yours.
That is no longer the federal position. Healthy People 2030, the set of ten-year national health objectives run by the U.S. Department of Health and Human Services, split health literacy into two separate things. One is about you. The other is about the hospital, the insurer, the clinic, the drug label and the web page. The second one is the story, and almost nobody has told patients about it.
Two definitions, and the verb that moved the burden
The Office of Disease Prevention and Health Promotion, the HHS office that runs Healthy People, now publishes both definitions on the same page.
Personal health literacy, per the Office of Disease Prevention and Health Promotion, is “the degree to which individuals have the ability to find, understand, and use information and services” to inform health decisions and actions, for yourself and for people you care for.
Organizational health literacy, in the same ODPHP text, is “the degree to which organizations enable individuals to find, understand, and use information and services” toward exactly the same end.
Read those two sentences back to back. They are nearly identical except for the subject and the verb. In the first, individuals have the ability. In the second, organizations enable it. That swap is the entire change, and it is deliberate.
ODPHP names four ways the 2030 definitions depart from the ones used in Healthy People 2010 and 2020. Taken in order of how much they matter to a reader, they go like this.
Organizations now share the responsibility. The old definition had one actor in it, the individual. The new pair has two, and the second one owns the part of the problem it controls: the wording of the form, the design of the portal, the fifteen-minute appointment.
You have to be able to use the information, not just understand it. Understanding a dosing schedule and executing it across a week with three medicines and a work shift are different tasks. The definition now covers the second one.
The goal is a well-informed decision rather than an appropriate one. Appropriate smuggles in a judge. Somebody decides whether your choice was the right one. Well-informed asks only whether you had what you needed in a form you could use. A patient who understands the trade-offs and declines a treatment has met the standard.
The frame is population health, not just the exam room. Health literacy is now something a health department or a state can be measured on, which is how it got national objectives attached to it.
One wrinkle shows how recently this changed. ODPHP’s main health literacy page defines organizational health literacy without the word “equitably.” ODPHP’s own page on the history of the definitions, and the Centers for Disease Control and Prevention’s health literacy page, both render it as organizations that equitably enable individuals. The government has not synchronized its own sentence yet.
Here is the practical consequence. If you did not understand it, the document failed. That is not a consolation prize. It is the stated position of the agency that writes the national health objectives, and it is the reason you can push back on a form without apologizing first.
Why health writing is so hard, measurably
The American Medical Association’s long-standing recommendation, cited across the clinical literature, is that materials written for patients land at a sixth-grade reading level or lower. Almost nothing hits it.
A 2024 systematic review in JAMA Otolaryngology by Maria Armache and colleagues pulled together 17 studies covering 1,124 patient education materials on head and neck cancer, produced by professional societies, hospitals and others. Average reading grade across those studies ranged from 8.8 to 14.8. Not one study reported an average at grade six or below. Across the eight studies that also measured patients, the share with inadequate health literacy ran from 11.9% to 47.0%.
A 2023 study in the Journal of Studies on Alcohol and Drugs makes the gap concrete. Nhu Dang and colleagues pulled the top search results for naloxone, the drug that reverses an opioid overdose, and compared them against the top results for CPR. The naloxone pages averaged grade 11.2. The CPR pages averaged 7.8. None of the naloxone pages met the sixth-grade recommendation, and only 17% were at or below eighth grade, versus 80% of the CPR pages. Two emergency procedures a bystander might have to perform, written four grade levels apart.
Reading grade is a crude measure. It cannot see a sentence that is short and still meaningless. But it catches the usual suspects: medical vocabulary used without definition, passive constructions that hide who does what, and consent forms built from stacked conditions that nobody parses on a clipboard.
The common defense is that precision requires complexity. That trade-off is mostly false. What plain language costs is time, drafting effort and review cycles, which is a budget problem rather than a scientific one. Federal agencies have operated under the Plain Writing Act of 2010 since it passed, which requires content for the public to be written for the audience it is aimed at. Compliance is uneven, and you are allowed to notice.
Numbers are where careful readers get fooled
This is the section that will save you the most trouble, because the problem is not vocabulary. It is arithmetic presented in a shape that misleads people who are paying close attention.
In a 2007 report in Psychological Science in the Public Interest, Gerd Gigerenzer and colleagues called the problem collective statistical illiteracy and documented it in patients, journalists and physicians alike. Their lead example is worth memorizing. Mammography screening has been described as reducing the risk of dying from breast cancer by 25%. What that figure represents is roughly one fewer woman out of 1,000 dying of the disease.
Both statements are true. Only one of them tells you what is at stake.
“Doubles your risk” is almost always the wrong sentence
A relative risk is a ratio between two numbers. It tells you nothing about either one.
Double a risk of 1 in 10,000 and you get 2 in 10,000. Double a risk of 1 in 10 and you get 2 in 10. The headline is identical. One of those changes affects one extra person in ten thousand. The other affects one extra person in ten.
Relative risk is not a trick in itself. Researchers use it because it travels well across populations with different baseline rates. The failure happens when it reaches you stripped of the baseline, which is the part that determines whether you should care. A doubling of a rare thing is still a rare thing. So the question is never how much this raises your risk. It is: raises it from what to what.
Natural frequencies instead of percentages
The fix that holds up best in testing is also the least sophisticated. Stop using percentages and conditional probabilities, and start counting people out of a fixed group.
Rather than a 25% reduction, say: out of 1,000 women screened, about five die of breast cancer; out of 1,000 not screened, about six do. Same group size every time, so the numbers stay comparable. Gigerenzer and colleagues argue for this whole family of changes: absolute risks rather than relative ones, frequency statements rather than single-event probabilities, and counts out of a fixed denominator rather than conditional probabilities.
A subtler point from the same work trips up smart people. Higher survival rates after screening do not mean people lived longer. Screening finds disease earlier, which lengthens the measured interval between diagnosis and death even when the date of death never moves. Mortality rates, which count deaths in a population over time, do not have that flaw. Survival rates do. When a screening program is sold to you on survival statistics, that is the number to question.
Four questions for any risk number you are handed
This is a short list you can carry into an appointment.
- Out of how many people, and over how long? A risk with no denominator and no time window is not a risk, it is a mood.
- What happens to the people who do nothing? That is the baseline. Without it, the benefit figure cannot be interpreted.
- Is this about dying of the disease, or being diagnosed with it? Different outcomes, routinely blurred.
- Can you give it to me as a count out of 100? If the person explaining it cannot, that tells you how well the number is understood on their side of the desk.
Clinicians who do shared decision-making well tend to welcome these questions, because they are the questions the conversation is supposed to be about.
Accessibility is engineering, not courtesy
The second half of organizational responsibility is technical, and it has rules attached.
Section 508 of the Rehabilitation Act requires that information and communication technology developed, procured, maintained or used by federal agencies be accessible to people with disabilities. That covers websites, software, kiosks and electronic documents. The U.S. Access Board writes the standards, issued as a final rule dated January 18, 2017, with a correction in January 2018 restoring provisions on TTY access.
Those standards do not invent their own web rules. They incorporate the Web Content Accessibility Guidelines by reference and require electronic content to meet the Level A and Level AA success criteria. WCAG comes from the World Wide Web Consortium’s Web Accessibility Initiative and breaks accessibility into testable criteria at three conformance levels, covering text alternatives, captions, contrast and keyboard operation.
Here is a gap worth knowing. Section 508 points at WCAG 2.0, published in December 2008. The W3C’s current version is WCAG 2.2, published October 5, 2023 and also issued as an international standard, ISO/IEC 40500:2025. Later versions add criteria rather than replacing them, so meeting 2.2 also satisfies 2.0. But a vendor selling 508 compliance is promising a fifteen-year-old target.
For hospitals, clinics and county health departments, the governing law is the Americans with Disabilities Act rather than Section 508. Department of Justice guidance published in March 2022 names the barriers that turn up most often: images without text alternatives, videos without captions, poor color contrast, and sites that cannot be operated without a mouse. That page carries a caveat noting it predates the state and local government web rule published in the Federal Register in April 2024, so requirements for public entities have moved since.
What those four barriers mean in practice:
Alt text is a short written description attached to an image in the page’s code. A screen reader, the software that reads a page aloud, announces it. Without it, a chart showing how your dose changes over four weeks is announced as nothing at all.
Captions carry a video’s audio as text. They decide whether a post-surgical instruction video is information or decoration, and they also work in a loud waiting room.
Color contrast is the measurable difference in lightness between text and its background. A dose printed in pale gray on white is unreadable to a great many people over fifty, which is most of the people reading dosing instructions.
Keyboard operability means every control can be reached without a pointing device. People who cannot use a mouse include those using switch devices, voice control, or a hand that is currently in a cast.
Plain language belongs in this list rather than in a category of its own. A document a screen reader handles perfectly and a reader cannot parse has failed by the same standard.
Teach-back, and how to ask for it by name
Teach-back is the single tool in this article you can request out loud, and it has a name your clinician will recognize.
The Agency for Healthcare Research and Quality publishes it as Tool 5 of its Health Literacy Universal Precautions Toolkit, third edition, in material created in February 2024. The method is simple: after an explanation, the clinician asks the patient to say back, in their own words, what they need to know or do.
The important framing is that teach-back is a check on the explanation, not a quiz on the patient. AHRQ is explicit that asking whether a patient understands, or whether something makes sense, does not count, because people say yes regardless. If what comes back is wrong, the clinician explains it a different way and checks again. If the patient only parrots the original words, that counts as a miss too.
There is a companion move for anything physical, which AHRQ calls show-me. The agency notes that patients who correctly state when and how much medicine to take still make mistakes when asked to demonstrate the dose. Inhalers, injectable pens, glucose meters and eye drops all belong here. Saying it and doing it are separate skills.
On evidence, be precise about the tier. Teach-back is endorsed as a safety practice for informed consent by the National Quality Forum, the Joint Commission and the Leapfrog Group. The controlled evidence is thinner than the endorsements suggest. A 2023 systematic review in the Journal of Renal Care by Hemamali Jagodage and colleagues found six studies covering 520 patients with chronic kidney disease, too varied to pool, showing some improvement in self-management, self-efficacy and knowledge, with limited evidence on quality of life. That is promising rather than settled, and it still beats nodding.
You can start it yourself without using the word. Try saying: let me repeat that back, and tell me what I got wrong. It takes about twenty seconds, and it relocates the error to where it can be fixed.
Reading a health web page in four minutes
The National Library of Medicine, through MedlinePlus, publishes guidance on evaluating online health information. What follows reorganizes it around how fast each check pays off.
Start at the money. Scroll to the bottom and open the About page. If you cannot establish who runs the site within thirty seconds, stop. NLM’s shortlist of reliable operators is federal agencies, medical schools, and large professional or nonprofit organizations. Ads should be labeled as ads, and anything that looks like neutral content but sits where an ad would sit deserves suspicion.
Check whether the recommendation and the checkout button are the same object. A site can sell things and still be accurate. But when the page recommending a supplement is also the page selling it, you are reading marketing copy formatted as an article.
Look for a date. Not a copyright year at the bottom, which updates automatically, but a written, reviewed or updated date on the page itself. A health page that will not say when it was last touched is telling you something.
Look for sources you can open. A citation that links to a study is a commitment. A sentence that says studies show, with nothing attached, is not.
Watch the verbs. NLM flags dramatic language, promised cures and claims that sound too good to be true as markers of health fraud. One more: a page that never names a downside, a side effect or a group the advice does not apply to is not being careful with you.
Cross-check once. Take the claim to a second independent source, and take anything you plan to act on to your own clinician, who can say whether it applies to someone with your history.
What to do when the document fails you
The useful conclusion is not that you should try harder to read. It is that the standard moved, and you can say so.
Do not sign a consent form you cannot restate in one or two sentences. Ask for the short version out loud, and ask what happens if you decline. That request is routine and it is not rude.
Take ambiguous medicine instructions to a pharmacist before the next dose rather than to a search engine. The conversation is free, and ambiguity about timing or amount is exactly what pharmacists are there for.
Ask for materials in a form you can use. Large print, a translated version, a captioned video or a plain-language summary are things an organization is expected to produce, not favors.
A few situations earn a call before the next appointment. Call your clinician promptly for new or worsening symptoms after a medication change, a procedure or a new treatment, and whenever a discharge sheet lists a warning sign and you are not certain whether you have it. That uncertainty is itself the reason to call. For suspected poisoning or overdose, including a medicine taken twice by mistake, call Poison Help at 1-800-222-1222. For an emergency, call 911.
The sentence worth keeping is short: I read this and I did not understand it. Said to a nurse, a pharmacist, a benefits line or a clinician, it is now a report of a defect, and the defect is not yours.
Sources
- Office of Disease Prevention and Health Promotion, U.S. Department of Health and Human Services. “Health Literacy in Healthy People 2030.” odphp.health.gov
- Office of Disease Prevention and Health Promotion, U.S. Department of Health and Human Services. “History of Health Literacy Definitions.” odphp.health.gov
- Centers for Disease Control and Prevention. “About Health Literacy.” cdc.gov
- U.S. Access Board. “Information and Communication Technology (ICT) Standards and Guidelines.” access-board.gov
- World Wide Web Consortium (W3C), Web Accessibility Initiative. “WCAG 2 Overview.” w3.org
- U.S. Department of Justice, Civil Rights Division. “Guidance on Web Accessibility and the ADA.” ada.gov
- National Library of Medicine, MedlinePlus. “Evaluating Health Information.” medlineplus.gov
- Agency for Healthcare Research and Quality. “Use the Teach-Back Method: Tool #5, Health Literacy Universal Precautions Toolkit, 3rd Edition.” ahrq.gov
- Digital.gov, U.S. General Services Administration. “Plain Language (guide, covering the Plain Writing Act of 2010).” digital.gov
- Psychological Science in the Public Interest (via PubMed). “Gigerenzer G, Gaissmaier W, Kurz-Milcke E, Schwartz LM, Woloshin S. “Helping Doctors and Patients Make Sense of Health Statistics.” 2007;8(2):53-96. PubMed. DOI: 10.1111/j.1539-6053.2008.00033.x.” doi.org
- JAMA Otolaryngology-Head & Neck Surgery (via PubMed). “Armache M, Assi S, Wu R, et al. “Readability of Patient Education Materials in Head and Neck Cancer: A Systematic Review.” 2024;150(8):713-724. PubMed. DOI: 10.1001/jamaoto.2024.1569.” doi.org
- Journal of Studies on Alcohol and Drugs (via PubMed). “Dang N, Khunte M, Zhong A, Chatterjee A. “Naloxone Online Information Exceeds the Recommended Reading Level for Patient Education Materials.” 2023;84(5):680-683. PubMed. DOI: 10.15288/jsad.22-00343.” doi.org
- Journal of Renal Care (via PubMed). “Jagodage HMH, McGuire A, Seib C, Bonner A. “Effectiveness of teach-back for chronic kidney disease patient education: A systematic review.” 2023;50(2):92-103. PubMed. DOI: 10.1111/jorc.12462.” doi.org