The Workbench · Craft

A clinical hold can stop a study FDA already approved

A device file that already tracks how 812.30 handles an IDE application — approval, approval with conditions, or disapproval — can start to treat FDA's authority over a study as fully spent once that first decision lands. This blog has already covered how 812.30's own clock runs on FDA's silence at the outset, and the appeals rules that follow a disapproval. Section 520(g)(8) of the FD&C Act gives FDA a separate, later-arriving power that neither of those posts reaches: the authority to stop a study already running, at any point, on a finding that has nothing to do with the paperwork the sponsor filed.

A hold is not a disapproval, and not a withdrawal either

812.30(b) already gives FDA two ways to stop a study: disapprove the application before it starts, or withdraw an approval already granted. Withdrawal runs on a compliance standard — 812.30(b)(2) lets FDA act on a failure to comply with Part 812, the Act, another applicable regulation or statute, or a condition an IRB or FDA itself imposed. A clinical hold is a third instrument, added to the FD&C Act's device provisions by the FDA Safety and Innovation Act of 2012 as section 520(g)(8) and codified at 21 U.S.C. 360j(g)(8), and it doesn't run on that same standard at all. Nothing about compliance has to be broken for a hold to issue.

The standard is risk, weighed five ways

520(g)(8)(B) sets the finding FDA has to make: that the device represents an unreasonable risk to the safety of the persons who are subjects of the clinical investigation. The statute names five things that finding has to weigh together — the qualifications of the investigators, the information available about the device, the design of the investigation, the condition the device is meant to treat, and the health status of the subjects involved. A study can be running exactly as its approved protocol describes, every case report filed on time, every consent form on file, and still draw a hold if that five-part picture turns against it — a device underperforming in the subjects already enrolled, a design weakness only visible once real patients are in the study, an investigator's own record raising a concern the original application never surfaced.

It reaches a study 812.30 already cleared

The authority isn't limited to the 30-day window 812.30 sets for FDA's first look at a fresh application. 520(g)(8)(A) lets FDA impose a hold at any time, on a study FDA has already approved and that may be actively enrolling subjects. When FDA does, the statute requires the agency to specify the basis for the hold in writing, including the specific information that served as its basis, and to confirm the determination in writing. A study-conduct file that only has one column for “FDA action” — filled in once, at approval, and never revisited — has no place to record a stop that can arrive well after enrollment has already started.

The clock runs from the sponsor's request, not FDA's application review

812.30's 30-day clock runs from FDA's receipt of the application, and favors the sponsor on FDA's silence — no notice by day 30 and the study may proceed. A clinical hold's clock runs the other direction and starts from a different event: 520(g)(8) requires that a written request from the sponsor asking FDA to remove the hold receive a decision, in writing and with reasons stated, within 30 days of FDA receiving that request. The hold doesn't lift on its own if that 30 days passes in silence — the sponsor has to ask, and FDA owes a written answer to the specific request, not a default outcome the calendar produces by itself.

It sits on the fast appeal track, not the slow one

Section 517A's own list of significant decisions names clinical hold determinations under 520(g)(8) directly, alongside 510(k) and PMA decisions, which puts a disputed hold on the 30-day filing window with FDA's guaranteed 30-or-45-day answer — not the slower, open-ended track most other CDRH decisions get. A sponsor disputing a hold has a faster, better-defined appeal available than the general 800.75 process would otherwise offer, provided the request is framed as what it actually is.

Where this meets the file

An IDE study tracker needs a row for a clinical hold that's separate from disapproval and withdrawal — carrying the five-factor basis FDA cited, the date of FDA's written determination, and the date any sponsor removal request was filed, since that date is what starts the only clock 520(g)(8) actually promises. A hold-tracking worksheet built around this structure is previewed in the launch catalog. If your program has been through a clinical hold and tracks it differently, the shelf takes that correction directly.

The Regulatory Toolkit launches soon — a free shelf of source-mapped templates, checklists and browser-only tools for regulatory teams. Get one email when it opens, or contribute a template.

All Workbench notes