Key takeaways
- An analysis of 5,472 citations across 4 AI services ranked these hospitals among all sources: Cleveland Clinic 2nd, Mayo Clinic 3rd, and Johns Hopkins Medicine 7th.
- The report cited Cleveland Clinic and Mayo Clinic as examples of “institutional health media,” and found that summaries and explainers accounted for 59%, compared with 41% for peer-reviewed research.
- All three hospitals run health information pages for the public. This does not explain their rankings, but it gives us a useful point of comparison for our own hospital.
Study findings - Which hospitals and materials were cited?
Outcomes Rocket, a U.S. healthcare marketing company, analyzed 5,472 citations collected from health answers by ChatGPT, Gemini, Claude, and Perplexity in August 2025. Becker's Hospital Review reported on the hospital rankings in October 2025.
- Hospital rankings among all sources: 2nd Cleveland Clinic · 3rd Mayo Clinic · 7th Johns Hopkins Medicine · 28th Nemours Children's Health · 57th Mount Sinai Health System
- Most-cited source: PubMed Central (385 citations)
- Source categories: Health media 30.7% · Commercial and affiliate sites 23.1% · Academic and research sources 22.9% (the report cited Cleveland Clinic and Mayo Clinic as examples of “institutional health media”)
- Types of material: Summaries and explainers 59% · Peer-reviewed research 41%
- Site metric: 62.4% of citations came from sites with a domain rating (DR) of 81–100
- Separately, the Institute checked (October 2026):The Cleveland Clinic Health Library, Mayo Clinic Diseases & Conditions, and Johns Hopkins Medicine Health pages are publicly available.
The Institute’s interpretation - What can hospitals learn from the top-ranked systems?
This is a marketing company’s own analysis; it did not disclose the number of questions, and it reflects AI models as they were in 2025. The rankings of large U.S. hospitals cannot be applied directly to hospitals in Korea. The analysis does not establish that health information pages increase citations; this additional perspective comes from the Institute’s review of the sites.
Hospitals in Korea do not need to match their scale. The essential work is to create content around patients’ questions and check whether our hospital’s information is cited in AI answers.
Rather than chasing rankings, first consider whether our hospital has enough health information for patients to read.
- Hospitals are cited as publishers of health information, too.It is worth noting that the report used the two hospitals as examples of “institutional health media.” Their role includes publishing health information as well as explaining care.
- Accessible explanations can also inform answers.In this analysis, summaries and explainers outnumbered peer-reviewed research. This does not mean simpler language leads to more citations, but it suggests that hospitals without research papers can still contribute through explanatory articles.
- A large library is not the only place to start.Although 62.4% of citations came from sites in the top domain-rating band, these data do not show that page count determines citations. A practical first step is to organize clear explanations of conditions and treatments in the hospital’s main areas of care.
Does our hospital website have health information pages?
- Are there pages explaining conditions and treatments separately from descriptions of clinical services?
- Are they written in language patients can understand without specialist jargon?
- Are explanations of our main areas of care connected rather than scattered?
Open the health information pages of the three hospitals and the difference from a general hospital profile becomes clear. Reading our own main service pages through the same lens is a useful comparison.
Sources
Outcomes Rocket, "AI's Sources of Truth: How Chatbots Cite Health Information" (2025, company analysis)
Original source: Outcomes Rocket · Related coverage: Becker's Hospital Review
Health information pages: Cleveland Clinic · Mayo Clinic · Johns Hopkins Medicine