Chair's Message: The Growing Problem of Medical Misinformation
A few weeks ago, I had the opportunity to attend the International Congress on Academic Medicine (ICAM) in Ottawa. This annual conference was a chance to meet with colleagues from across the country, hear about emerging topics in academic medicine, and leave with fresh ideas to bring back to our department. There was one session in particular, I found fascinating, yet frightening. The opening keynote of the conference, Manufactured Doubt: Health and Science Misinformation and the Fight for Trust, brought together four voices: patient partner Lydia Greene, health law professor Timothy Caulfield, foreign information operations specialist Marcus Kolga, and psychiatrist, health policy expert, and Senator Stan Kutcher. They discussed how health misinformation takes root, how it spreads through political and foreign influence, and what coordinated responses might look like in Canada. The panel's goal was to help participants understand the processes driving misinformation, the respective roles of government and civil society in addressing it, and the practical initiatives healthcare providers can lead. Since the session, I had some time to reflect, read some of the articles that were mentioned, and now want to share some of the information with you, along with my idea that what we do everyday, can help in the fight against medical misinformation.
The scale of health misinformation in Canada is striking. A 2026 Health and Media Tracking survey by the Canadian Medical Association (CMA) found that 89% of Canadians go online for health information, most often because accessing a healthcare provider is not fast or convenient enough.1 Of those, 64% report having encountered health information online that they later discovered was false or misleading. Most strikingly, Canadians who followed health advice from AI tools were found to be five times more likely to experience harms than those who did not.1
I found these numbers staggering. They describe the patients sitting across from us in our offices who have arrived at their appointments after already consulting Google, TikTok, or a chatbot, and who may have received information that is wrong, dangerous, or both.
One of the discussion's recurring themes was how misinformation spreads. Dr. Caufield quoted that approximately 75% of health information is shared on social media without the person ever clicking through to read it. He described that content spreads not because it is accurate, but because it triggers an emotional response. He highlighted rage and fear as significant drivers of both consumption and sharing of health information on social media. A recent example highlighted was a comment made by Donald Trump that stated soda pop kills cancer cells. Despite being widely recognized as lacking any scientific basis, the comment was recirculated and went viral across social media platforms. Dr. Caufield’s point was that absurdity does not protect a claim from spreading, and once shared enough times without context, even the most implausible statements can be construed as fact and have the potential to cause harm.
Among the most powerful moments of the panel was hearing from Lydia Greene, a patient partner and advocate who shared her personal experience as a former anti-vaxxer. Lydia described how she had come to distrust vaccines, not through a single dramatic event, but through a gradual erosion of trust in the healthcare system. The process began when she felt she was not being heard by her providers for an issue unrelated to vaccines. When she brought concerns to her doctor and felt dismissed, she began looking elsewhere for answers. Online, she found communities who were listening, validated her fears, reflected her frustrations, and offered explanations that felt satisfying, even if they were untrue. This led to her taking more advice and using misinformation from these online groups to make healthcare decisions for her and her children.
A growing body of evidence demonstrates that patients who feel heard and respected by their physicians are less susceptible to health misinformation, while those who feel dismissed are more likely to turn to alternative sources.2
What Lydia advised healthcare providers to do was both simple and profound: when a patient brings up something they read online, do not shut it down. Thank them for raising it. Create space for dialogue. Acknowledge their underlying concern before addressing the factual errors. This approach, meeting patients where they are emotionally before attempting to redirect them intellectually is compassionate and more likely to be effective.
The panel also discussed the growing issue of AI chatbots and health misinformation, which was even more sobering than the spread of health misinformation on social media, in my opinion.
The panel cited a recent study published in BMJ Open that audited five popular AI chatbots (Gemini, DeepSeek, Meta AI, ChatGPT, and Grok) by asking them 250 health and medical questions across five categories known to attract misinformation: cancer, vaccines, stem cells, nutrition, and athletic performance.3 The results were alarming. Nearly half, 49.6%, of all responses were rated as problematic by subject-matter experts, with 19.6% rated "highly problematic." The chatbots almost never refused to answer, even when the questions were deliberately designed to push toward misinformation or contraindicated advice. Only two refusals were recorded across 250 questions. Citation accuracy was also poor, with a median reference completeness score of just 40%, and every chatbot produced fabricated or hallucinated citations.
Another study that was cited described an experiment where researchers deliberately invented a fake medical condition, "bixonimania," and planted obviously fake academic papers about it on a preprint server.4 Within weeks of posting, major AI chatbots including Gemini, Copilot, and ChatGPT were confidently describing bixonimania to users as a real, emerging medical condition. The fake papers were even cited in a peer-reviewed journal article, and later retracted, suggesting that some researchers may now be relying on AI-generated references without reading the underlying studies.4
The panel also addressed how the susceptibility to health misinformation is now deeply tied to political identity. For certain topics such as vaccines, unproven therapies, and extreme diets, political affiliation has emerged in research as the single most predictive variable in whether someone accepts or rejects evidence-based information. The discussion also touched on how foreign adversaries have deliberately seeded and amplified health misinformation, with campaigns specifically designed to undermine trust in vaccines, public health institutions, and Western science.
Despite all this evidence for health misinformation being available and often reaching our patients, we can still help. Encouragingly, the CMA's survey found that 85% of Canadians still trust physicians to help them navigate health information. This is more than they trust the internet, social media, or AI.1 This is a significant asset. Even in a fractured information environment, the patient-physician relationship remains a uniquely trusted channel.
It is apparent to me that physicians and health care providers must take an active role in addressing health misinformation. The panel suggested governments have largely proven unwilling or unable to do so at scale often because the general public doesn’t trust government in the first place.
Interestingly, this brings me back to what we at Western have long championed: the patient-centred clinical method.
The approach developed in our Department calls on us to understand not just the disease, but the patient's ideas, feelings, expectations, and context around their illness experience. This framework turns out to be precisely what is needed in the age of misinformation.
When a patient arrives having read that vaccines cause autism, or that a carnivore diet will cure their diabetes, the temptation is to immediately correct the record. But correction without connection tends to harden positions rather than shift them. Research on health misinformation consistently shows that a trusting relationship with a healthcare provider is one of the most robust protective factors against misinformation adoption.2 The patient-centred method, asking what the patient has heard, what they believe, what worries them, and what they hope for, creates the conditions in which accurate information can actually be received.
The session in Ottawa reminded me of why family medicine occupies a unique place in the healthcare system. We see our patients over years, often over decades. We are, for the majority of them, the most trusted source of health information they have, even as that trust is under strain from every direction.
It’s clear the growing health misinformation crisis is real, and the antidote is not fact sheets or a better algorithm. It starts with relationships, just like the patient centred clinical method has taught us all along.
----------------------------------------------------------------------------------------------------------------------------------
As always, if you have comments, questions, feedback, or topics you would like to hear about in future newsletters, please reach out to me – Scott McKay smckay28@uwo.ca
References
- Canadian Medical Association. Doctors Warn: Canadians Are Turning to AI for Health Information and It Is Hurting Them. CMA Press Release, February 10, 2026. https://www.cma.ca/about-us/what-we-do/press-room/doctors-warn-canadians-are-turning-ai-health-information-and-it-hurting-them
- Tandar CE, Lin JC, Stanford FC. Combating medical misinformation and rebuilding trust in the USA. Lancet Digital Health. 2024. PMID: 39379267. https://doi.org/10.1016/S2589-7500(24)00197-3
- Tiller NB, Marcon AR, Zenone M, et al. Generative artificial intelligence-driven chatbots and medical misinformation: an accuracy, referencing and readability audit. BMJ Open. 2026;16:e112695. https://doi.org/10.1136/bmjopen-2025-112695
- Stokel-Walker C. Scientists invented a fake disease. AI told people it was real. Nature. 2026;652:559–561. https://doi.org/10.1038/d41586-026-01100-y
