The clock stops. The calendar doesn't.

A 510(k)’s public record shows two dates: received, decided. Subtract one from the other and the result looks like a review-speed number. It isn’t the number FDA’s own performance goal measures, and reading it as one is a mistake worth naming before it shapes a benchmarking exercise or a filing strategy.

FDA Days are not calendar days

Under the current MDUFA V performance goals, FDA aims to issue a decision on 95% of 510(k) submissions within 90 FDA Days. FDA Days is a defined term, not a synonym for elapsed time — FDA’s own guidance on the review clock excludes from that count every day a submission spends on hold waiting on the submitter. A 510(k) can sit open far longer than 90 days on the calendar and still have used only 90 FDA Days, if the rest of the span was hold time the performance goal was never built to charge against the agency.

Two holds, two different clocks

The first hold comes early. Under FDA’s Refuse to Accept policy, the agency has 15 calendar days after receipt to decide whether a submission is administratively complete against its acceptance checklist. Fail that check and the file goes on hold before the review clock starts at all — day one of substantive review is whenever the missing pieces arrive, not the original receipt date. The second hold comes mid-review: an Additional Information request pauses the clock the moment FDA sends it, and the submitter has 180 calendar days to answer in full. Miss that window and the submission is considered withdrawn — not decided, not denied, gone, with a new 510(k) required to try again.

FDA Radar’s own ingestion of the 510(k) feed keeps exactly the two dates the public record publishes — date received, date decided — the same two dates that produced the elapsed-days number in the first place. No RTA hold, no Additional Information timestamp, and no FDA-Days total appears anywhere in what the pipeline ingests, so no downstream tool, including this one, can reconstruct a submission’s real review clock from public data alone. A firm’s own submission correspondence is the only place that reconstruction is even possible.

Elapsed calendar days is the honest label for what the public record shows. Review time is not.

What this doesn’t change

None of this says a specific K-number’s long calendar span was, or wasn’t, hold time — that answer lives in the sponsor’s own correspondence file, not in a dataset that never recorded the holds at all. This is regulatory intelligence about what a public timestamp can and cannot support, never regulatory advice about a specific submission’s timeline.