The investigator (the study's principal investigator at the site) holds the legal responsibility for making the initial risk determination for a device being used in a study. That means the investigator decides first whether each device is a significant-risk or nonsignificant-risk device, and whether an investigational device exemption (IDE) is required, before enrollment begins. Sponsors, IRBs, and the FDA all review or override that call, but none of them makes it first. The initial call sits with the person who will actually use the device on participants.
That single sentence hides a layered system. The investigator's judgment is constrained by the sponsor's written protocol, the IRB's authority to disagree, and the FDA's power to regulate the underlying device regardless of what anyone at the site concludes. Understanding how those layers interact is what separates a defensible determination from a documentation gap that surfaces in an audit two years later.
The mechanism: how the initial determination actually gets made
The legal architecture comes from the FDA's IDE regulations, which define a significant-risk device as one that presents a potential for serious risk to the health, safety, or welfare of a subject. That definition covers three prongs: the device is an implant, it is used to support or sustain life, or it is of substantial importance in diagnosing, curing, mitigating, or treating disease. A device meeting any prong is significant-risk. A device meeting none is nonsignificant-risk, provided it does not otherwise present a serious-risk potential.
The investigator applies that test to the specific use in the specific protocol, not to the device's label in general. A legally marketed catheter used off-label for a new indication can become significant-risk in that study context. A novel sensor used only to log ambient temperature in a clinic may be nonsignificant-risk even though it has never been cleared. The determination is use-specific, which is why a blanket statement like "this device is 510(k)-cleared, so it is nonsignificant-risk" is wrong on its face.
Mechanically, the process runs in a fixed order. The sponsor writes a protocol describing the device and its intended study use. The investigator reviews that description and makes the initial significant-risk call. For nonsignificant-risk devices, the investigator signs an abbreviated IDE and the IRB reviews the study under streamlined rules. For significant-risk devices, a full IDE application goes to the FDA, which has 30 days to approve, disapprove, or approve with conditions. The IRB cannot approve a significant-risk study until FDA approval exists or the device is exempt.
The deeper reason the investigator owns the first call is epistemic. The investigator is the only party with direct knowledge of how the device will be deployed, on which patients, with what concomitant care. Sponsors design studies but do not run them. IRBs review paperwork but do not touch the device. FDA regulates products but does not see the bedside. The regulation places the initial judgment where the operational knowledge lives.
Quantified context: timelines, thresholds, and review clocks
The system runs on fixed clocks, and missing one reshuffles responsibility in practice. The table below shows the standard federal time course for a significant-risk device study in the United States.
| Stage | Responsible party | Standard clock | What triggers the next step |
|---|---|---|---|
| Protocol and device description drafted | Sponsor | No federal deadline | Investigator receives written use description |
| Initial risk determination | Investigator | Before IRB submission | Signed determination documented |
| Abbreviated IDE submitted | Sponsor or investigator | With IRB submission | IRB begins review |
| Full IDE submitted to FDA | Sponsor | Before IRB approval of significant-risk study | FDA 30-day review clock starts |
| FDA decision | FDA | 30 calendar days | Approval, conditional approval, or disapproval |
| IRB approval | IRB | Convened or expedited per local schedule | Enrollment may begin |
| Continuing review | Investigator and IRB | At least annually for significant-risk studies | Renewed approval |
The 30-day FDA clock is statutory and runs from receipt, not from a courtesy acknowledgment. A study that reaches the IRB before FDA action on a significant-risk device is not approvable, and an IRB that approves it anyway has created a compliance defect independent of the investigator's original call.
Volume context matters for calibration. The FDA's Center for Devices and Radiological Health oversees the majority of the more than 6,000 medical device submissions it receives annually across premarket pathways, and IDE applications are a small fraction of that total. Most device studies therefore run under nonsignificant-risk determinations and abbreviated IDEs, which is precisely why the investigator's initial call carries so much weight: it routes the entire study into one of two very different regulatory tracks, and the vast majority of studies take the lighter one.
Edge cases: where the initial call gets contested
Three situations routinely scramble the clean division of responsibility. Each has a defensible resolution, but only if someone documents the reasoning before enrollment.
The first is the marketed device studied off-label. Suppose an investigator wants to use a cleared infusion pump to deliver a drug at a flow rate outside the cleared labeling. The pump is legally marketed, so the sponsor may argue it is nonsignificant-risk. The FDA's position is that the risk determination tracks the proposed use, not the clearance status. If the new use raises the potential for serious harm, the device is significant-risk in that study and a full IDE is required. The investigator's initial call must reflect the study use, and the sponsor cannot override it downward.
The second is the device that changes mid-study. A protocol amendment adds a new sensor, a new implantation site, or a new patient population. The initial determination does not automatically carry forward. Each substantive change requires a re-determination, and a change that pushes the device across the significant-risk line requires a new or supplemental IDE before the change is implemented. Audit findings frequently trace to a mid-study amendment that was never re-routed through the risk-determination step.
The third is the combination product gray zone. When a device is used to deliver a drug, or a drug is packaged with a device, jurisdiction can shift between CDER and CDER's device counterpart. The lead center is assigned by a formal request-for-designation process, and the investigator's initial determination should identify which review pathway the combination falls under. Getting this wrong does not usually harm participants immediately, but it can invalidate the regulatory basis for the entire study.
Misconceptions: what the initial determination is not
The most common error is treating the initial risk determination as a signature on a form rather than a reasoned judgment. The regulation expects the investigator to weigh the three significant-risk prongs against the protocol's specific use and to document why each prong was or was not met. A checked box with no rationale is a finding waiting to happen, because an auditor cannot reconstruct the reasoning that produced the answer.
A second misconception is that FDA clearance of the device settles the question. It does not. Clearance addresses the device's intended use as labeled, not the study's use, not the patient population, and not the concomitant interventions. A device can be perfectly safe as cleared and significant-risk as studied.
A third is that the IRB makes the initial call. IRBs have authority to disagree with the investigator and to require a full IDE, but their review is downstream. An IRB that is asked to make the initial determination is being asked to do the investigator's job, and the resulting record will not show who owned the judgment.
A counterintuitive detail worth flagging: a nonsignificant-risk determination is not a low-scrutiny determination. Nonsignificant-risk studies still require IRB approval, informed consent, and abbreviated IDE documentation. The lighter track concerns FDA submission, not ethical oversight. Investigators who equate "nonsignificant-risk" with "no oversight" are the ones most likely to skip the documentation step.
Practical actions: documenting a defensible initial call
The investigator's determination should be written, dated, signed, and filed with the protocol before the IRB submission. The following steps make that record audit-ready.
State the device's intended use in the study in one sentence, distinct from its cleared labeling. Walk through each significant-risk prong and record why it is or is not met for this use. Identify the regulatory pathway the determination routes the study into: exempt, abbreviated IDE, or full IDE. Note any off-label, combination-product, or vulnerable-population features that could shift the call. Obtain the sponsor's written concurrence or documented disagreement, and route disagreements to the IRB. Re-run the determination for every substantive protocol amendment before implementation.
Red-flag warnings that should stop a study until resolved:
The device is an implant, supports life, or is substantially important to treatment, and no full IDE exists. The study use differs materially from the cleared labeling and no re-determination was documented. An IRB approved a significant-risk study before FDA action. A mid-study amendment added a device or population without a new risk determination. The determination form is signed but contains no rationale for the prongs.
Reconciling the thresholds is straightforward once the sequence is clear. The investigator's prong-by-prong call is the initial determination. The IRB's approval is the ethical gate, and it cannot open for a significant-risk device until the FDA's 30-day clock has run. The FDA's IDE decision is the regulatory gate. All three must align, and the investigator's call is the first domino.
Limitations: where the initial call ends and others begin
The investigator's initial determination is authoritative but not final. The IRB can disagree and require a full IDE. The FDA can determine that a device the investigator called nonsignificant-risk is in fact significant-risk, and that determination binds the study going forward. Sponsors can withdraw support if they disagree with the call, which in practice ends the study regardless of who was legally correct.
The system also assumes the investigator has enough information to make the call. When a sponsor provides a thin device description, or when the device is so novel that no precedent exists, the initial determination becomes a judgment under uncertainty. In those cases the defensible move is to document the uncertainty, consult the FDA through the pre-submission process, and treat the call as provisional until the agency responds.
This article is general educational information about research regulatory roles, not a legal opinion or a compliance determination for any specific study. Device risk classification depends on facts unique to each protocol, population, and use. Consult your institution's research compliance office, IRB, or a qualified regulatory professional before making a determination for an actual study.