No IDE is confirmed until the PMA is

A device company’s regulatory history becomes public in stages. A 510(k) clears and the decision posts. A PMA is approved and the summary follows. A recall opens, a warning letter goes out, a classification changes — each one eventually lands in a database FDA Radar or anyone else can read. An Investigational Device Exemption is different. For as long as the underlying study runs, FDA’s own rule keeps its existence confidential — not because no one has built the right database yet, but because the regulation says so.

The rule, plainly

Under 21 CFR 812.38, FDA will not disclose that an IDE exists — unless the sponsor has already publicly disclosed or acknowledged it — until one of two things happens: FDA approves a premarket approval application for the device under investigation, or a notice of completion of a product development protocol for the device becomes effective. Both are downstream events. A study can enroll its first subject years before either one occurs, if it occurs at all.

Even once existence is acknowledged, the rule doesn’t hand over the file. The confidentiality protection lifts for the fact that an IDE exists, not for the data and information inside it, which stays protected under the same section’s narrower disclosure exceptions.

Who decides risk before FDA sees anything

The gap runs deeper than confidentiality. For many device studies, FDA never reviews an application at all before enrollment starts. Under 21 CFR 812.66, the sponsor makes the initial call on whether its own study poses a significant risk or a nonsignificant risk, and presents that determination, with its reasoning, to the reviewing IRB. Unless FDA has already made a risk determination for the study, the IRB reviews the sponsor’s call and can modify it if it disagrees.

If the study lands — sponsor and IRB agreed — as a nonsignificant risk study, the sponsor proceeds under the abbreviated requirements at 21 CFR 812.2(b): IRB approval, labeling under 812.5, informed consent, monitoring, and recordkeeping. No IDE application goes to FDA. No submission exists for FDA to have received, let alone disclose.

If the IRB disagrees with a sponsor’s nonsignificant-risk proposal and calls the study significant risk instead, the sponsor has to tell FDA — but on a short clock and a narrow question. Under 21 CFR 812.150(b)(9), the sponsor reports the IRB’s determination within five working days of learning of it. That report, not a public posting, is the first point FDA is even on notice the study exists.

For a nonsignificant-risk study, there is no application for FDA to have received. There is nothing yet for a confidentiality rule to protect, because there is nothing yet FDA has been told.

What that means for reading a portfolio

Put the two rules together and a specific, structural gap follows: a competitor’s ongoing U.S. clinical device study is not something any FDA record can surface while it runs. Not because FDA Radar, or any other service built on FDA’s public data, hasn’t indexed it yet — because for a nonsignificant-risk study, no FDA record of it exists yet to index, and for a significant-risk study under IDE review, the confidentiality rule keeps its existence unconfirmed regardless. A device that later clears through a 510(k) or wins PMA approval can carry years of investigational history that never surfaces anywhere public until the clearance or approval itself does — and even a PMA approval only confirms that an IDE existed, not what it contained.

The practice

Reading a competitor’s pipeline for signs of an active clinical device program means watching for what a study eventually produces — a PMA approval, a De Novo grant, a 510(k) built on data that had to come from somewhere — not searching for the IDE itself, which is built by rule to stay unconfirmed until one of those events happens. Whether a specific device study qualifies as significant or nonsignificant risk, and what that determination should be for a program of your own, is a call for the sponsor and its reviewing IRB to make under 812.66 — this is a description of how that determination and FDA’s own confidentiality rule work, and what they leave out of the public record, regulatory intelligence and method, never regulatory advice about any specific device’s risk classification.