These exams decide whether a doctor is allowed to practise. A wrong explanation is not a lost mark on a mock — it is something a doctor learned and carried onto a ward. So every claim on this site is traceable to a named document, from a body entitled to make that claim, with the date someone last read it.
DocPasser works to a published evidence hierarchy. A question's explanation is settled by the highest tier that addresses the point, and the tier is recorded against the item.
Another question bank. A revision course's notes. An unattributed web page, blog, forum post or Wikipedia article. Lecture slides with no primary reference. Any machine-generated text used as its own authority. These are excluded by policy and rejected automatically — a citation that does not resolve to an organisation on the register below fails the build and the item cannot be published.
An exam marks you against the practice of its own jurisdiction, so DocPasser cites accordingly. NICE does not settle an AMC question and Therapeutic Guidelines does not settle a PLAB question. Where a jurisdiction genuinely works from another's guidance — Irish hospitals routinely use UK and European guidelines — the citation is permitted but logged as editorial debt, so it gets replaced with a local source as one becomes available. US and Australian content accepts nothing from outside its own jurisdiction, because those exams test local pathways, local formularies and local law.
Exam formats move. Pass marks are reset. Guidelines are superseded. So a check is not permanent: every verified fact carries the date it was read, who read it, the exact document, and the line they read — and it expires after 6 months, at which point the page reverts to saying it needs re-checking. A two-year-old verification is not a verification, and a site that never revisits its numbers is just a site that stopped looking.
Where a figure has not been checked, DocPasser does not print it as fact. The page says it is unverified, or withholds the number entirely and tells you where to look. This is deliberately worse-looking than what competitors do, and it is the entire point: a number with no provenance is a guess wearing a suit.
Published live from the same data the app runs on, so it cannot drift from reality.
| Items in the banks | 9334 |
|---|---|
| Items citing an approved source | 9334 of 9334 |
| Citations from a regulator or national body (tier 0–1) | 3545 |
| Citations from a specialty society or college (tier 2) | 3065 |
| Citations from a national formulary (tier 3) | 2724 |
| Cross-jurisdiction citations awaiting a local replacement | 17 |
| Exam facts with a live verification | 48 of 48 |
| Exam facts never checked | 0 |
| Verifications expired and queued for re-check | 0 |
Last generated 2026-09-13.
Separately from sourcing, every item goes to a credentialed physician reviewer before it is published to subscribers. Review status is enforced in the database, not in the interface: an item that has not been signed off cannot be read by the app at all. Reviewer credentials shown to subscribers are independently verifiable registration numbers — and until reviewers are appointed, no reviewer name appears anywhere on this site, because a badge nobody earned is worse than no badge.
Every question has a report control. If you spot an error, one tap sends it to us and we correct anything that turns out to be wrong. We're not going to promise a large review team or a fixed turnaround we can't guarantee. What we will say is this: the most repeated complaint about the incumbent products is that reported errors sit untouched for years, and that is the thing we are determined not to do. A real mistake gets fixed, and the correction reaches everyone.
77 organisations. Adding one is an editorial decision, not a convenience: it says this body is entitled to settle a clinical question for candidates in that jurisdiction.