Correct Health Information Can Still Lead Patients to Bad Decisions

John Whyte does not tell people to stop looking up health information. His premise is that everyone already does it, often reflexively, and that online medical information can be useful when it helps patients become informed and empowered. The failure point is what happens when a symptom search supplies information without context.

He opens with the familiar escalation: a search for whether a headache can be a symptom of something serious leads to results listing stroke, aneurysm, meningitis, encephalitis, brain tumor and carbon monoxide poisoning. The next search asks whether a brain tumor can be fatal; another asks whether an aneurysm can kill you. The screen fills with mortality statistics, disease descriptions and survival caveats. The implied joke is a last will and testament template. The point is not subtle, but it is precise: a search can move from a common symptom to catastrophic possibilities faster than it can establish what is likely, urgent or personally relevant.

Whyte says that health searches now happen at enormous scale.

He breaks that figure down on screen as 41,666,666 searches per hour and 694,444 per minute. The scale matters because the behavior is no longer marginal. Whyte says he used to ask audiences whether they had searched health conditions online; now he asks whether they have searched for health issues today.

As the former chief medical officer at WebMD, Whyte says he believed the company’s motto: better information leads to better health. As a physician, he still wants patients to be informed. But he draws a line between information and knowledge. “TMI,” in his framing, applies to medicine too. Too much information about symptoms can produce anxiety, and in some cases can become dangerous.

Whyte’s central example is a patient he calls Mary, who came in with cerumen impaction — a clogged ear from wax. The condition is common. Mary had searched online and found, correctly, that hearing loss can be associated with dementia. She was alarmed because she had decreased hearing, though Whyte says it had been present only for a day. She also read that too much earwax could signal infection, which Whyte says was not true.

Mary then found that hydrogen peroxide, mixed equally with water, can help with earwax. Her mistake was not that every piece of information was fabricated. Her mistake was applying fragments without clinical judgment. She microwaved the mixture and put it in her ear. She came to the office with a burned, red, irritated ear canal.

Whyte says he asked her, “Why?” Her answer was: “I read it online, and I thought it wouldn’t hurt.” It did.

That anecdote carries much of Whyte’s argument. Online medical information is often presented as if the missing ingredient is access. But Mary had access. What she lacked was a way to rank relevance, assess risk and decide what action was appropriate. The injury came from the common health-search assumption that a low-effort home intervention is harmless enough to try.

Whyte links this to the older idea of the hypochondriac: someone who interprets every symptom as the most serious possibility. Online, he says, that becomes the “cyberchondriac” — a person who searches every detail, magnifies it and lands on the worst-case scenario.

Whyte is careful not to dismiss online information as simply false. His own family examples are designed to preserve the tension.

His 12-year-old son developed a blister from oversized Kobe cleats. After a few days it looked infected, his wife took him to urgent care, and the clinicians cultured it and prescribed an antibiotic. That evening, the boy told Whyte it could be MRSA and that the antibiotic did not cover it. Whyte thought that was “pretty clever for a 12-year-old,” but also thought MRSA was unusual. The next day the doctor’s office called: it was MRSA, and the antibiotic had to be changed.

Whyte’s point is that the child’s searched information was correct. Often, he says, the information people find is correct. But correctness is not the same as context. A tool can identify a possibility without knowing the patient’s history, examining the body or weighing probabilities the way a clinician would.

The same son later told Whyte’s wife that her knee pain might be “jumper’s knee,” or patellar tendinitis, after she restarted workouts at the gym. Whyte says it was not; she had simply started exercising again and her knee was bothering her. The example is ordinary by design. The same search habit that can correctly flag MRSA can also overfit a benign complaint.

Whyte includes generative AI, ChatGPT, Google and symptom checkers in the same category: powerful tools that can help people understand health, but that do not know the full person. The distinction he wants is “clarity versus confusion.” If the tools produce more possible diagnoses but no useful hierarchy, they have not improved the patient’s situation.

Whyte’s critique extends beyond search results to algorithmic feeds and health influencers. He says misinformation is widespread, and algorithms can drive users toward inaccurate information by reinforcing what they already search or watch.

He illustrates this with his own Instagram feed. Because his younger son wants a golden retriever, Whyte says his feed is full of “Goldies,” to the point that it seems everyone has one. Because he is a physician and worked at WebMD, he also sees a lot of wellness content. He is on a statin for primary prevention because of high lipids, and he follows doctors’ feeds. He began seeing cardiologists argue that people do not really need statins, that diet and exercise should be enough, and that statins can increase dementia rather than decrease it.

Whyte says he has been a physician for more than 20 years and knows the statin data, yet repeated exposure made him wonder whether he needed the medication. More similar feeds appeared. He dug deeper and found, he says, that many of the physicians making these claims were selling a particular product, and that some had issues with their medical licenses.

The mechanism he is describing is not a single bad post. It is the way algorithmic repetition can simulate consensus. Seeing the same type of claim again and again can make it feel more trustworthy, even to someone trained to evaluate medical evidence. That makes the problem less about whether patients are smart enough and more about whether the information environment is built to clarify or to keep attention.

Whyte’s practical advice is to treat health information with at least the discrimination people often apply to financial advice. If an influencer with a million or two million followers recommends a cryptocurrency or an investment product, he says, most people do not simply act. They check credentials, try to corroborate the claim, ask whether it makes sense and consult someone they trust. The reason is obvious: it is their money and their future.

He argues that health deserves more, not less, scrutiny. Yet people often approach health with a “why not try it?” attitude, especially when someone else reports a good experience. That, for Whyte, is the dangerous gap: people can be more careful with a speculative investment than with a supplement, treatment idea or medical interpretation that affects their body.

His checklist is simple: know who the messenger is; check the information; corroborate it elsewhere; ask whether it sounds plausible; and talk it through with someone trustworthy. The trusted person may be a doctor, nurse or pharmacist. The central contrast is human relationship versus chatbot or influencer.

Whyte is not anti-technology. He says technology has a major role, and he wants patients to be educated and empowered. But the final step, especially when the stakes are health, is not another search. It is context supplied by someone who can connect general information to the person in front of them.