"Will we need a clinical trial?" is one of the most financially consequential questions in medical device planning, because the answer can move a program by years and by an order of magnitude in budget. The reassuring part: most devices cleared in the United States never run one. The large majority of devices reach market through a 510(k) built on bench testing, biocompatibility, electrical safety, and a demonstration of substantial equivalence to a legally marketed predicate. Clinical data becomes necessary mainly when the device is genuinely novel, when the risk class is high, or when the claim you want to make cannot be supported any other way.

What Actually Triggers a Clinical Study

  • No suitable predicate. If nothing legally marketed does substantially what your device does, there is no equivalence argument to make — see how to choose a predicate device.
  • Class III and the PMA pathway. Life-supporting, life-sustaining, and implantable devices generally require clinical evidence of safety and effectiveness.
  • The De Novo route. A novel low-to-moderate-risk device with no predicate often needs clinical data to establish that the risks are mitigable — see the De Novo pathway.
  • A clinical claim. If your labeling says the device improves an outcome, diagnoses a condition, or replaces an existing test, you have to show it.
  • A new intended use or population. Even a familiar technology aimed at a new patient group can require data.

Which of these applies is determined by classification and pathway, so the decision is inseparable from the work described in the FDA approval process and FDA device classes.

The IDE: Significant vs Non-Significant Risk

In the US, a clinical investigation of a device that is not yet cleared or approved runs under the Investigational Device Exemption framework. There are three practical buckets:

  • Exempt investigations. Certain studies — for example, some uses of already-cleared devices consistent with their labeling, or non-invasive diagnostic specimen studies — fall outside the requirements entirely.
  • Non-significant risk (NSR) studies. The sponsor proposes the NSR determination, the reviewing institutional review board makes the call, and if the IRB agrees the study proceeds under abbreviated requirements without a separate FDA application. Labeling, IRB approval, informed consent, monitoring, and records still apply.
  • Significant risk (SR) studies. Implants, life-supporting devices, and anything presenting a serious risk to health require an IDE application submitted to and approved by FDA before enrollment, in addition to IRB approval.

Getting the SR/NSR determination wrong is expensive in both directions: assume NSR incorrectly and your data may be unusable, assume SR unnecessarily and you have added months for no benefit. This is exactly the point at which experienced regulatory advice earns its fee.

Study Types, Smallest to Largest

  • Early feasibility / first-in-human. A handful to a few dozen participants, aimed at basic safety and design direction. Findings from this stage routinely reshape the device, which is the point.
  • Traditional feasibility. A larger confirmation that the design and procedure work as intended before you commit to the expensive study.
  • Pivotal study. The study designed to support the marketing submission: pre-specified endpoints, powered sample size, statistical analysis plan agreed in advance, frequently multi-site.
  • Post-market studies. Condition-of-approval studies and postmarket surveillance orders that continue collecting data after clearance — increasingly common, and part of the long-term cost of ownership.

Do not confuse any of these with human factors validation, which nearly every device needs and which measures whether people can operate the device safely rather than whether it produces a clinical outcome. That is a separate discipline — see usability engineering for medical devices.

How a Study Actually Runs

  • Protocol first. Population, inclusion and exclusion criteria, endpoints, sample size, and statistical analysis are all fixed in advance. Changing the analysis after seeing the data invalidates the result.
  • IRB approval at every participating institution, plus FDA approval of the IDE for significant-risk studies.
  • Sites and investigators. Hospitals or clinics with a principal investigator at each, documented informed consent from every participant, and ongoing monitoring of data quality.
  • Registration and reporting. Applicable studies are registered publicly, adverse events are reported on defined timelines, and everything flows into the marketing submission.
  • Good clinical practice. Device trials follow their own GCP expectations, including the international standard for clinical investigation of medical devices.

The Devices Used in the Study Are Not Prototypes

This surprises first-time teams. Investigational units must be built to a locked design under design controls, with documented manufacturing, inspection records, unit-level traceability, and investigational labeling. A lab-bench build with hand-tuned firmware does not qualify. Changing the design mid-study is a serious regulatory event that can force you to treat the data as two separate populations — which is why design freeze discipline and configuration control matter so much before enrollment opens, and why the risk file described in ISO 14971 risk management has to be current before the first participant is consented.

Time, Money, and the Cost Drivers

Even a modest feasibility study is measured in many months and in the low-to-mid six figures once everything is counted. A multi-site pivotal study runs into years and into the millions. The line items that dominate:

  • Contract research organization fees for operational management.
  • Per-participant site payments and site start-up costs.
  • Clinical trial insurance and indemnification.
  • Monitoring, data management, and statistical work.
  • Manufacturing the investigational units under controlled conditions.
  • Regulatory preparation of the IDE and the protocol itself.

Investors in this sector know these numbers and will ask about them directly, so a clinical plan belongs in the fundraising narrative from the start rather than as a surprise in diligence.

The Cheapest Study Is the One You Design Away

Because the requirement flows from classification, intended use, and pathway, the highest-return work happens at the concept stage. A carefully bounded intended-use statement, a defensible predicate, or a claim expressed in terms you can support with bench data can remove the need for a study entirely — legitimately, not by cutting corners. Conversely, an over-broad claim written casually in a pitch deck can commit you to a pivotal trial nobody budgeted for.

The Upside Nobody Mentions

A successful study is not only a regulatory ticket. Published clinical data is the strongest commercial asset a device company can hold: it moves hospital value-analysis committees, supports reimbursement conversations, persuades clinicians, and separates you from every competitor asserting that their device works. In a market where everyone claims performance, the company with controlled evidence holds the proof.

Educational Information Only

Projects House is an engineering firm, not a regulatory consultancy or a clinical research organization. This article is general educational information about how device clinical studies are structured in the US; it is not regulatory advice. Classification, SR/NSR determinations, and study design should be reviewed with qualified regulatory and clinical professionals.

Planning a device and want the engineering, design controls, and controlled builds handled by a team that understands what a study will demand of them? Get in touch through our contact form. Broader background is collected in our medical device development guide.