New Study Reveals Most Doctor-Made YouTube Health Videos Lack Strong Scientific Proof

Breaking Down the Landmark JAMA Network Open Study

A groundbreaking quality improvement study published in JAMA Network Open has exposed a critical issue in digital health communication: the vast majority of physician-created YouTube videos on major topics like cancer and diabetes fail to back their claims with robust scientific evidence. Researchers led by EunKyo Kang, MD, from South Korea's National Cancer Center, analyzed 309 popular videos and found that only 19.7 percent earned a top grade for high-quality evidence, while a staggering 62.5 percent relied on very low or no evidence at all. 131 78 This revelation challenges the trust patients place in 'Dr. YouTube' and underscores the urgent need for better standards in online medical content creation.

The study, titled 'The Quality of Evidence of and Engagement With Video Medical Claims,' used a novel Evidence-GRADE (E-GRADE) framework adapted from the established Grading of Recommendations Assessment, Development, and Evaluation (GRADE) system widely used in evidence-based medicine (EBM). E-GRADE categorizes claims into four levels: Grade A for high certainty from systematic reviews or major guidelines like those from the World Health Organization (WHO); Grade B for moderate certainty from randomized controlled trials (RCTs) or high-quality observational studies; Grade C for low certainty from limited data; and Grade D for anecdotal evidence or none at all. Two independent reviewers assessed each video's claims, achieving high interrater reliability (Cohen's kappa 0.97-0.98), with disputes resolved by a third. 131

Videos were sourced via YouTube searches in English and Korean using terms related to cancer and diabetes, focusing on content from verified health professionals (75 percent physicians), with at least 10,000 views and over one minute in length. The dataset, frozen on June 20-21, 2025, captured a median of 164,454 views per video and 19 minutes duration, reflecting real-world popularity. 131

Shocking Distribution of Evidence Quality

Of the 309 videos, 164 addressed cancer and 145 diabetes, mirroring high-stakes conditions affecting millions in the United States. The results painted a dismal picture: just 61 claims (19.7 percent) hit Grade A, 45 (14.6 percent) Grade B, a mere 10 (3.2 percent) Grade C, and 193 (62.5 percent) Grade D. Traditional quality tools like DISCERN, JAMA benchmark criteria, and Global Quality Scale (GQS) correlated weakly with E-GRADE (Spearman coefficients 0.11-0.23), highlighting their inadequacy for detecting evidence gaps. 131

• Grade A examples: Videos citing NCCN guidelines for cancer treatment or ADA standards for diabetes management.
• Grade D pitfalls: Reliance on personal anecdotes like 'In my 20 years of practice, I've seen this diet cure diabetes' without supporting trials.

This gap persists despite producers' credentials, creating a 'halo effect' where MD credentials lend undue credibility to unproven advice.

The Algorithmic Bias Amplifying Weak Claims

Even more alarming, low-evidence content thrives. Multivariate negative binomial regression showed Grade D videos garnered 34.6 percent more views than Grade A (incidence rate ratio [IRR] 1.35, 95% CI 1.00-1.81, P=0.047), adjusted for channel size, upload recency, and length. YouTube's engagement-driven algorithms favor sensationalism over rigor, as longer, anecdote-rich videos (median 19 minutes) hook viewers but often skimp on citations. 131 79

In the US, where 70 percent of adults use YouTube for health info per Pew Research, this dynamic risks widespread misinformation. Patients delaying evidence-based care for unproven remedies could exacerbate chronic disease burdens, costing billions annually in avoidable treatments.

Expert Perspectives: From Lead Author to US Academics

Lead author Dr. Kang stated, 'This reveals a substantial credibility-evidence gap... where physician authority frequently legitimizes claims lacking robust empirical support.' She advocates for 'evidence-based content-creation guidelines' and 'algorithmic reforms.' 78

Accompanying editorialist Richard Saver, JD, professor at the University of North Carolina School of Law, Chapel Hill, notes physician misinformation predates the internet, rooted in resistance to EBM's data-over-intuition emphasis. He praises E-GRADE as a 'useful framework' for assessing social media claims and calls for intervention by 'medical education institutions, professional organizations, and regulatory bodies.' 79

US medical bodies like the AMA echo this, urging faculty to model EBM in public comms.

Historical Context and Prior Research

This isn't new. A 2021 systematic review found academic-physician YouTube videos varied widely in quality, while 2025 LLM assessments confirmed inconsistent medical content reliability. 0 21 Prostate cancer videos showed 77 percent biased per NYU Langone, and COVID-era analyses revealed persistent flaws. 20 Yet, physician-led content often outperforms lay influencers, amplifying risks.

Implications for Higher Education and Medical Training

Medical schools bear responsibility. The study urges 'professional schools including medical, pharmacy, nursing... to integrate training on evidence hierarchy assessment, science communication principles, and ethical responsibilities of public health communication.' 131 US programs like UMN's misinformation course and AMA wellness tools are steps forward, but most lack mandatory social media literacy. 83

Universities could certify faculty videos with E-GRADE badges, positioning grads for roles blending clinical expertise and digital outreach. Explore academic CV tips for highlighting science comm skills or faculty positions in medical education.

Public Health Consequences in the United States

61 percent of US physicians report patients swayed by misinformation, per Physicians Foundation, leading to delayed care and worse outcomes. For cancer/diabetes, this means rejecting proven therapies for untested diets or supplements, with CDC noting prediabetes in 1 in 3 youth amid online noise. 123 Vulnerable groups—low health literacy adults, minorities—suffer most, eroding trust in institutions.

Pathways Forward: Guidelines, Training, and Platform Changes

• Develop E-GRADE-like standards for all med content.
• Integrate digital ethics into med school curricula, as 80 percent of schools have policies but vary widely. 82
• Platforms prioritize cited sources via AI fact-checks.
• Pros orgs like AMA enforce disclosure of evidence levels.
• Consumers: Check creators' citations, cross-reference PubMed.

Check Rate My Professor for faculty science comm feedback or career advice on ethical digital presence.

Photo by Brett Jordan on Unsplash

Future Outlook: Toward Credible Digital Health Education

By 2026, expect med schools to mandate EBM training for social media, with AI tools aiding verification. Faculty leading this shift will shape tomorrow's physicians. For US higher ed, this study spotlights opportunities in interdisciplinary programs blending medicine, comms, and data science. Stay informed via university jobs in health education.