Your IDE Risk Determination Isn't Final Until the IRB — or FDA — Says So

A sponsor's significant risk (SR) or nonsignificant risk (NSR) call under 21 CFR 812.3(m) is a proposal, not a determination. The IRB reviews it independently and can overturn it. If FDA has made its own determination — through a submitted IDE or a Q-Submission — that determination is final and controls over both. Teams that treat the sponsor's initial classification as settled find that out at the worst possible time: after sites have already started screening subjects.

What 812.3(m) actually tests

A device is significant risk if it meets any one of four conditions: it is intended as an implant and presents a potential for serious risk to a subject's health, safety, or welfare; it is purported or represented to support or sustain human life and presents that same potential for serious risk; it is of substantial importance in diagnosing, curing, mitigating, or treating disease, or in preventing impairment of human health, and presents that potential for serious risk; or it otherwise presents a potential for serious risk. Nothing in the test asks how invasive the device is, how novel it feels, or how the last similar device on the market was classified. It asks what happens to the subject if the device fails or performs as intended in a way that causes harm — a narrower, more specific question than most first-pass determinations actually answer.

The sponsor's call is a proposal, not a determination

For a study the sponsor believes is nonsignificant risk, the abbreviated requirements at 812.2(b) apply: labeling, IRB approval, informed consent, monitoring, records, and reporting — but no IDE application to FDA before the study starts. That is a real efficiency, and it is also the reason the determination gets rushed. The sponsor submits its classification and supporting justification to the IRB along with the protocol. Under 812.66, the IRB does not rubber-stamp that call — it independently reviews the determination against the same 812.3(m) test and can modify it if it disagrees, in which case it notifies the investigator and, where appropriate, the sponsor.

• The sponsor proposes. The initial SR/NSR determination, with device-specific justification, goes to the IRB with the protocol — not to FDA first.
• The IRB reviews independently. It applies 812.3(m) itself and can overturn a sponsor's NSR call; a bare assertion without reasoning is the easiest overturn to earn.
• FDA's call, wherever it comes from, is final. Through a submitted IDE application or a Q-Submission risk determination request, FDA's answer controls over both the sponsor's and the IRB's.
• Silence isn't a safe default. An unresolved disagreement between sponsor and IRB doesn't default to NSR — it defaults to a study that cannot proceed on the sponsor's original classification.

The sponsor's determination is a starting position the IRB is required to test, not a classification the IRB is expected to accept. Why an IRB reversal mid-study is expensive

What happens when the IRB disagrees

If the IRB determines that a study the sponsor proposed as nonsignificant risk is in fact significant risk, 812.150(a)(9) requires the sponsor to report that determination to FDA within 5 working days of first learning of it. That study cannot continue to enroll under the abbreviated NSR pathway while the reclassification is unresolved — it needs an approved IDE, which means the sponsor is now filing under the timeline pressure of an active study rather than a planning one. Sites that have already begun screening under the original classification are left holding a protocol whose regulatory basis just changed under them, which is a harder conversation with a clinical investigation team than it needs to be.

None of this is a reason to default every ambiguous device to significant risk out of caution — that trades a real efficiency for an unnecessary IDE filing and the review timeline that comes with it. It is a reason to treat the determination as what the regulation actually makes it: a reasoned, documented, defensible position, not a box checked once at protocol drafting and never revisited. Programs that build the IDE strategy around a Q-Sub-confirmed risk determination, rather than an internal judgment call, are the ones that don't discover their classification was wrong from an IRB letter three months into enrollment. Our Q-Submission strategy work exists for exactly this decision point — get the harder cases resolved with FDA before the study clock starts, not after.