A device effect isn't the only unanticipated problem

This blog has already covered how 21 CFR 812.150 keeps a device study's reporting obligations running long after IDE approval, including the ten-working-day clock an investigator owes a sponsor and reviewing IRB once an unanticipated adverse device effect turns up. That clock answers one question well, because 812.3(s) defines a UADE narrowly enough that a study team can build a checklist straight off the regulation's own text. A separate, broader duty sits underneath it. 21 CFR 56.108(b)(1) requires every IRB to have written procedures ensuring prompt reporting of any unanticipated problem involving risk to human subjects or others — a category the regulation never limits to effects the device itself caused, and a study file that tracks UADEs alone has covered only the slice of 56.108 that happens to overlap with 812.3(s)'s own definition.

Two regulations, two different tests

812.3(s) defines an unanticipated adverse device effect as any serious adverse effect on health or safety, any life-threatening problem or death, caused by or associated with a device, if that effect, problem, or death was not previously identified in nature, severity, or degree of incidence in the investigational plan or application — or any other unanticipated serious problem associated with a device that relates to the rights, safety, or welfare of subjects. Every branch of that definition runs back to the device: caused by it, associated with it, or at minimum a problem a device study surfaced. 56.108(b)(1) sets no such anchor. It requires the IRB's own written procedures to ensure prompt reporting of any unanticipated problems involving risks to human subjects or others, full stop, with no clause tying the problem to the device under investigation at all.

What clears 56.108 without ever touching 812.3(s)

The gap isn't hypothetical. A breach of subject confidentiality that exposes participants to social or economic harm involves no device effect whatsoever, yet it plainly involves risk to subjects. A safety signal from another site in a multicenter study, or a change in the standard of care outside the study altogether, can raise the risk profile for subjects still enrolled without any device in the room having malfunctioned or performed unexpectedly. A protocol deviation that unintentionally increases exposure — a missed monitoring visit, a dosing or use error unrelated to device performance — can meet the same bar. None of these has to clear 812.3(s)'s device-effect test to still be exactly what 56.108(b)(1) is asking an IRB's procedures to catch.

FDA's own filter runs on three questions, not one

FDA's guidance for clinical investigators, sponsors, and IRBs on adverse event reporting, finalized January 14, 2009, and the Office for Human Research Protections guidance it draws on, adopt the same practical test for sorting an unanticipated problem from routine study noise: is the incident, experience, or outcome unexpected in nature, severity, or frequency given the protocol and the study population; is it related or possibly related to participation in the research; and does it suggest the research places subjects or others at greater risk of harm than was previously known or recognized. A finding that answers yes to all three is an unanticipated problem requiring prompt report, whether or not a device caused it. A UADE that clears 812.3(s) will almost always clear this three-part test as well — the reverse doesn't hold, and a study team that treats the two tests as the same one is running a narrower filter than 56.108 actually requires.

The same IRB, two different source regulations

Both duties converge on the same reviewing IRB, which makes the gap easy to miss in practice: a sponsor's safety-reporting SOP built entirely around the UADE clock 812.150(a) already requires can look, from the outside, like it's covering the IRB's own reporting needs in full. It isn't. 56.108(b)(1) puts the underlying procedural duty on the IRB itself, not on the sponsor or investigator, and an FDA inspection of an IRB checks whether written procedures for catching the broader category exist and get followed — a question the device study's own UADE log doesn't answer, however complete that log happens to be.

Where this meets the file

An unanticipated-problem determination log needs the general three-part test run as its own explicit field — unexpected, related or possibly related, greater risk than previously known — kept separate from and prior to the device-specific UADE clock this blog has already covered, and keyed to the same reviewing IRB already tracked for continuing review's own interval-setting duty. A determination worksheet built around that structure is previewed in the launch catalog. If your program runs this determination differently, the shelf takes that correction directly.