Provenance
Coverage
Cited Record publishes from primary sources only: documents issued by the agencies themselves. This page states what is live, what is being built, in what order, and what we have chosen to leave out.
1. Live today
The first source is U.S. FDA warning letters. A warning letter is the Food and Drug Administration’s formal notice to a company that it has violated federal law, and that enforcement may follow if the matter is not corrected. Letters are issued by the agency’s centers and its Office of Regulatory Affairs, and published on fda.gov.
Every letter is fetched from fda.gov, archived byte-for-byte at ingest, and hashed. Every field we publish is extracted from — and cited to — that archived copy, not the live page. If fda.gov changes or removes a letter, our citation still resolves.
- Regulator
- U.S. Food and Drug Administration
- Document type
- Warning letters, with close-out letters where issued
- Obtained by
- Agency index and per-letter pages, archived at ingest
- Re-checked
- Every six hours against the agency index
Fields published
- Company
- The recipient entity, as named in the letter.
- Date issued
- The date on the letter.
- Issuing office
- The center or office that signed the letter.
- Reference
- The agency’s case or CMS reference number.
- Subject
- The matter the letter concerns.
- Cited authorities
- The statutes and regulations the letter cites.
- Status
- Open, closed out, or withdrawn, per the agency.
- Excerpts
- Verbatim passages, each cited to character offsets in the archived source.
Close-outs and withdrawals
When the agency issues a close-out letter, the record’s status changes and the close-out is archived the same way as the original. When the agency withdraws a letter, the record says so. Neither event deletes the record.
2. The coverage programme
The corpus is being built out to eighty-one regulators across four jurisdictions. The full list is on the front page. Warning letters are one publication surface of one agency; the FDA alone also publishes inspection observations, import alerts, recalls and debarment lists, and each is treated as a separate source with its own verification.
- United States
- 28 regulators
- Canada
- 13 regulators
- European Union
- 27 regulators, including EEA and Switzerland
- United Kingdom
- 13 regulators
A regulator is described as live only once its records are complete and verified. Until then it appears on the programme, not in the corpus.
3. The order sources are built in
Sources are sequenced by how their documents are published rather than by how interesting they are, because the publication format determines how confidently a field can be extracted and cited.
- First
- Structured APIs and feeds, where the agency publishes machine-readable records.
- Then
- HTML indexes and document pages, which need a parser per source and break when the agency redesigns.
- Last
- PDF and scanned paper, which carry the lowest extraction confidence and the highest hold-back rate.
This is deliberately the opposite of the order that would produce the most impressive coverage numbers soonest.
4. How records are verified
Every record passes through deterministic extraction, triangulation against the agency’s index, an independent audit pass, and anchored citation to character offsets in the archived source. The method is described in full on the front page. A record that fails any step is not published.
5. What we deliberately exclude
Cited Record covers regulatory and administrative enforcement against organisations. It does not cover criminal prosecutions, and it does not publish records about identified individuals.
Sources whose output is principally criminal — prosecutions, indictments and convictions — are outside the corpus. So are records whose subject is a natural person rather than a company. Where a source mixes the two, the individual records are held back rather than the source being dropped.
This narrows what we can offer, and it is a choice rather than an oversight. Criminal offence data carries obligations that a commercial data set is poorly placed to meet, and the value of this corpus is in what regulators do to companies.
6. Corrections
Every record shows the source it was read from. If we have misread a source, we want to know. Write to contact@citedrecord.com with the reference number, and we will re-check the record against the archived document.