What we watch. And what we don’t.
Every item on FDA Radar links to its primary FDA source — an item whose source link is dead is treated as a defect, not a cosmetic issue. Here is where everything comes from, and the gaps most vendors don’t mention.
Monitored continuously
| Source | What it gives you | Access | Cadence |
|---|---|---|---|
| Federal Register | Guidance availability notices, device classification rules, proposed and final rules, comment deadlines | federalregister.gov API — structured, official | Every business day |
| FDA Warning Letters | CDRH letters: company, issue and posting dates, violation themes | FDA's public letters database | ~Weekly (typically Tuesdays) |
| Device recalls / enforcement | Recalls with class, firm, and root-cause narrative. Class I queued for expert review; Class II/III kept searchable | openFDA enforcement API | Irregular; FDA lags ~1 week |
| 510(k) clearances | Every clearance: K-number, applicant, product code, decision. Feeds product-code matching and the competitor corpus — never pushed as a feed | openFDA 510(k) API | Daily |
| Device classification database | Class, regulation, pathway, review panel, device flags — powers product-profile derivation at signup | openFDA classification API | On demand |
Acquisition runs on weekday mornings with a rolling seven-day window, so a failed run or an FDA outage self-heals on the next pass instead of leaving a silent hole.
Deliberately not covered
Being comprehensive is FDA’s job, and FDA does it better. Being honest about the boundary is ours.
FDA Form 483s
Not systematically public — FOIA-gated. Services that sell 483 coverage obtain them at cost through FOIA requests. We do not have them, so we say so rather than listing the source and shipping silence.
ISO / IEC standards content
Standards texts are copyrighted and sold by the standards bodies. What actually affects a submission is FDA recognition of an edition — we are evaluating a reliable feed for the Recognized Consensus Standards database and will add it when one exists, not before.
Pending 510(k) submissions
FDA does not publish them. Any product claiming to monitor competitor submissions before clearance is describing data that does not exist publicly.
MAUDE adverse events
Enormous and noisy; per-product adverse-event analytics need statistical care we haven't built yet. Deferred, not forgotten.