The dataset behind ChinaMedBase

This site holds a structured record of 124 Grade III-A hospitals across 44 Chinese cities: 3,145 departments, 6,463 named clinicians, 93 published fee lines, and 51 hospitals whose international department could be confirmed on their own website. Every field names the page it was read from and the date it was checked. The figures below are counted at build time from the same records the hospital pages render.

What the numbers count

Each figure, what it does and does not include, and where it comes from.

MeasureCountWhat it countsRead from
Hospitals profiled124Hospitals with a full record page. Every one is Grade III-A or above, in a tier-1, new-tier-1 or tier-2 city.The hospital's own site, its health-authority listing, or its official announcements.
Cities covered44Cities with at least one profiled hospital.Administrative classification, not a ranking of the cities.
Departments recorded3,145Department entries read from hospital sites and mapped onto this site's specialty taxonomy. A hospital that lists 40 departments contributes 40 rows, not 40 unique specialties.The hospital's own department pages.
Named clinicians6,463928 names published by the hospital itself and 5,535 romanised by this site from the Chinese name. A name in this count means the hospital lists the person — not that the person is available to international patients.Department rosters on hospital sites.
Clinicians individually confirmed for international patients111The subset where a hospital page states, in its own words, that this person sees international or foreign patients. Deliberately small: being on a roster is not the same claim.International-department pages and clinician profiles.
Hospitals with a confirmed international department51Hospitals publishing an international department, international clinic or foreign-patient office that this site could confirm on their own website.The hospital's international-department page.
Hospitals with a verified official website123Where the domain could be tied to the hospital rather than to an agency or a directory.Certificate, footer, or an official page linking to it.
Published fee records93Individual price lines a hospital or a provincial health authority published. These are list prices, not what a patient paid, and not a quote.Published fee schedules and price-disclosure pages.
Distinct source documents59Separate pages the records were read from. One document commonly supports many fields, so this is documents, not citations.—

What is not in the dataset

The absences are part of the record. Each one is a thing hospitals rarely publish, or a thing this site refuses to estimate.

  • Clinical outcomes. No success rates, no mortality figures, no complication rates, no volumes. Chinese hospitals do not publish them in comparable form, and an outcome number this site invented would be the most dangerous thing on it.
  • Quality scores or rankings of our own. External rankings are reproduced as their publisher ranked them. The site publishes no “best hospital” score.
  • What a patient actually paid. Fee lines are published list prices. The bill depends on the case, and no page here quotes one.
  • Patient reviews. Not collected, not aggregated, not summarised.
  • Anything unconfirmed. Where a field could not be tied to a public source, it is empty and labelled. The blank is deliberate and is not an error state.

How to cite this

Free to cite, with attribution. Attribution is how a reader checks the number — it is not a permission fee.

Suggested form

ChinaMedBase, "The dataset behind ChinaMedBase", https://chinamedbase.com/data (accessed YYYY-MM-DD)

When citing a single hospital rather than the dataset, link that hospital’s page instead — it carries the specific source and check date, which a count of the whole site does not. If a figure here is wrong, tell us and the correction is published on the corrections list.

For machines

The same information in the formats an agent or crawler would rather read.

  • /llms.txt — what the site covers, what it will not claim, and where the entries are.
  • /llms-full.txt — the full record set as plain text.
  • /pricing.md — access terms: everything is free, no account, no paywall.
  • /sitemap.xml — every indexable page.

How this page is kept true. The counts are compiled at build time from the same snapshot the hospital pages render from, and a build that ships a snapshot older than the database fails its own checks. A figure on this page therefore cannot drift from the pages it describes without the deploy stopping first.