Treatment use and emergency use open different doors

A study file that logs “emergency use” and “expanded access” as two names for the same event has collapsed two provisions of 21 CFR Part 812 that run in opposite directions. One lets an investigator deviate from an approved investigational plan, after the fact, to save a subject already enrolled in a trial. The other lets a sponsor open access to an investigational device, before the fact, for patients who were never going to be trial subjects at all. This blog has already covered how most IDE studies never trigger the IDE application in the first place; emergency use and treatment use sit on the other side of that line entirely — both apply only where an IDE, or an approved investigational plan, already exists.

Emergency use is a deviation, reported afterward

Section 812.35(a)(2) covers a deviation from the investigational plan made to protect the life or physical well-being of a subject in an emergency. It doesn't require the supplemental-application approval that 812.35(a)(1) otherwise demands for a plan change — the whole point is that a subject in front of a physician in a genuine emergency can't wait for a supplement to clear. What it does require is a report to FDA within 5 working days after the sponsor learns of the deviation. The investigator's own obligation runs on a parallel clock: 812.150(a)(4) requires notifying the sponsor and the reviewing IRB as soon as possible, and no later than 5 working days after the emergency occurred. Two five-day clocks, running from two different trigger events — the sponsor's own awareness, and the emergency itself — and a study file that only tracks one of them is missing half the obligation.

Treatment use is a submission, cleared before use starts

Treatment use is a different mechanism built for a different moment: 21 CFR 812.36 lets a sponsor request access to an investigational device, outside the controlled trial, for patients with a serious or immediately life-threatening disease who aren't enrolled subjects at all. Eligibility runs on four conditions together — the device treats a serious or immediately life-threatening disease or condition; no comparable or satisfactory alternative exists for that patient population; the device is under investigation in a controlled clinical trial under an approved IDE, or such trials are already complete; and the sponsor is actively pursuing marketing approval with due diligence. Where the request qualifies, 812.36(d)(1) sets its own clock: treatment use may begin 30 days after FDA receives the submission, unless FDA notifies the sponsor sooner that it may, or may not, proceed. That's a waiting period built around FDA's own review, not an emergency exception at all — the opposite structure from 812.35's deviate-first, report-after pattern.

Neither one is the informed-consent emergency exception

A third provision gets folded into this same conversation and shouldn't be. 21 CFR 50.23 provides a narrow exception from general informed-consent requirements for a single patient in a life-threatening situation where consent can't be obtained and no time remains to find a legally authorized representative, certified in writing by a physician. That's a consent mechanism, not an IDE mechanism, and it's also not the same rule as 21 CFR 50.24, which governs a completely different situation — planned emergency research conducted under a community-consultation protocol, the kind used for trauma or resuscitation trials designed in advance. A study file that cites 50.24 for a single unplanned emergency, or 50.23 for a pre-planned trial, has cited the wrong half of Part 50. The eight required elements of informed consent still apply everywhere none of these exceptions is in play.

Why the distinction is worth a dedicated field

These three mechanisms answer to three different reviewing bodies on three different timelines: emergency use reports to the sponsor, the IRB, and FDA after the fact; treatment use waits on FDA's own review before anything happens; the consent exceptions run through the IRB's own oversight of the physician's certification, not through FDA at all. A study-conduct tracker that logs all three under one “exception used” column can't show which body has to be notified, by when, or whether the notification is even still outstanding.

Where this meets the file

An IDE deviation log built around this split needs separate rows for an emergency deviation under 812.35(a)(2), tagged with the sponsor's awareness date and the investigator's emergency date, and a treatment-use request under 812.36, tagged with FDA's receipt date and the 30-day clock it starts — not one undifferentiated exceptions log. An IDE reporting tracker built around these separate obligations is previewed in the launch catalog. If your program has run a treatment-use request and found gaps this one misses, the shelf takes that correction directly.