What is the difference between a medical device and a drug?

A drug achieves its principal intended action pharmacologically, immunologically, or metabolically. A device achieves its principal action by physical, mechanical, or informational means. Products combining both are handled through specific combination-product rules.

Does FDA clearance mean the same as FDA approval?

No. "Clearance" generally refers to a 510(k) determination of substantial equivalence to a predicate device. "Approval" refers to premarket approval of a Class III device following review of clinical evidence. The two involve substantially different evidentiary standards.

What does CE marking mean for a medical device?

CE marking indicates that the manufacturer has demonstrated conformity with the applicable European requirements, assessed by a notified body for all but the lowest-risk class. It permits the device to be placed on the market in the European Union.

Is a fitness tracker a medical device?

It depends on the claimed intended purpose. A device marketed for general wellness is generally not regulated as a medical device. If the manufacturer claims it detects, diagnoses, or monitors a medical condition, it may fall within the definition and require regulatory clearance for that function.

Are medical devices tested in clinical trials like drugs?

Sometimes, but not always, and the designs differ. Blinding is often impossible, outcomes depend on operator experience, and devices evolve iteratively. Evidence therefore frequently combines bench testing, clinical investigation where warranted, and long-term registry data.

What Are Medical Devices? Types, Classes, Regulation

Medical devices explained: what counts as a device, risk classes, how FDA and EU MDR regulation works, software as a medical device, and post-market surveillance.

A medical device is any instrument, apparatus, implant, in vitro reagent, software, or related article intended for a medical purpose — diagnosis, prevention, monitoring, treatment, or alleviation of disease or injury — that does not achieve its principal action by pharmacological, immunological, or metabolic means. That final clause is what separates a device from a medicine: a device works primarily through physical, mechanical, or informational means.

The category is extraordinarily broad. It spans tongue depressors and surgical gloves at one end and implantable defibrillators, robotic surgical systems, and diagnostic algorithms at the other. Regulation therefore works by risk classification rather than by applying a single standard to everything.

What Counts as a Medical Device

Regulatory definitions differ in wording between jurisdictions but share a common structure: an article intended by its manufacturer for a medical purpose in humans, whose principal intended action is not achieved by pharmacological, immunological, or metabolic means, though such means may support its function.

Two elements do most of the work. Intended purpose is decisive: the same physical object may or may not be a regulated device depending on what the manufacturer claims it does. A wearable marketed for general fitness is not a medical device; the same hardware marketed to detect a cardiac arrhythmia may be. Mode of action separates devices from medicines and determines how borderline products — drug-eluting stents, prefilled injectors, antimicrobial dressings — are classified, sometimes through specific combination-product procedures.

Categories of Medical Device

Diagnostic devices — imaging systems, electrocardiographs, blood pressure monitors, and endoscopes.
In vitro diagnostics — tests performed on samples taken from the body, including laboratory analysers, molecular assays, and point-of-care tests. These are regulated under dedicated frameworks in most jurisdictions.
Therapeutic and surgical devices — surgical instruments, ablation systems, dialysis machines, ventilators, and infusion pumps.
Implantable devices — pacemakers, defibrillators, orthopaedic joints, stents, and cochlear implants.
Assistive and rehabilitation devices — prostheses, orthoses, hearing aids, and mobility aids.
Monitoring devices — continuous glucose monitors, cardiac telemetry, and remote monitoring systems.
Software as a medical device — standalone software performing a medical function without being part of a hardware device.
Consumables and single-use items — catheters, syringes, dressings, and surgical drapes.

Risk Classification

Classification systems assign devices to classes according to the risk they pose, considering how invasive the device is, how long it remains in contact with the body, whether it is active, and the consequences of failure. Higher class means greater regulatory scrutiny.

United States

Class I — low risk, subject to general controls such as registration, labelling requirements, and quality system obligations. Many are exempt from premarket submission.
Class II — moderate risk, requiring general controls plus special controls such as performance standards or specific labelling. Most reach market through a premarket notification, known as a 510(k).
Class III — highest risk, typically life-sustaining, life-supporting, or implanted devices. These generally require premarket approval supported by valid scientific evidence, usually including clinical data.

European Union

Under the Medical Device Regulation, devices fall into Class I, IIa, IIb, or III, with in vitro diagnostics classified separately as Class A to D under the In Vitro Diagnostic Regulation. Except for the lowest-risk category, conformity assessment involves a notified body — an independent organization designated to audit quality systems and technical documentation. Successful assessment permits CE marking, which allows the device to be placed on the market.

Classification rules vary between systems, so the same device may fall into different classes in different jurisdictions, and manufacturers frequently pursue parallel routes with different evidence packages.

How Devices Reach the Market

The 510(k) Route

The most common route to the US market for moderate-risk devices, a 510(k) submission argues that the device is substantially equivalent in intended use and technological characteristics to a legally marketed predicate device. Clinical data are not always required. The pathway enables incremental improvement to reach patients efficiently; it has also been criticised on the grounds that long chains of predicates can distance a current device from any original clinical evidence base.

Premarket Approval

For Class III devices, premarket approval requires a comprehensive submission including clinical investigation results, manufacturing information, and labelling. Review is substantially more demanding than 510(k) notification, and advisory panel review and facility inspection may be involved.

De Novo Classification

Where a novel device is low or moderate risk but has no suitable predicate, the De Novo route allows classification into Class I or II with special controls, and the resulting device may then serve as a predicate for later submissions.

EU Conformity Assessment

Manufacturers compile technical documentation demonstrating conformity with general safety and performance requirements, supported by a clinical evaluation. A notified body assesses the documentation and the quality management system for higher classes. The Medical Device Regulation strengthened clinical evidence expectations, tightened rules on equivalence claims to other manufacturers' devices, and increased post-market obligations relative to the earlier directives.

Clinical Evidence for Devices

Device evidence differs from medicines evidence in important respects. Randomized blinded trials are often impractical: procedures cannot easily be blinded, sham comparators raise ethical difficulties, devices are frequently iterated during their commercial life, and outcomes depend substantially on operator skill and the learning curve associated with a new technique.

Consequently, device evidence often combines bench and simulated-use testing, biocompatibility assessment, animal studies, clinical investigations where warranted, and systematically collected registry data. Implant registries in orthopaedics and cardiology have proved particularly valuable for detecting late failures that pre-market studies could not.

Software as a Medical Device

Software intended for a medical purpose can itself be a regulated device. This includes image analysis tools, algorithms that triage or flag findings, monitoring applications that generate clinical alerts, and some clinical decision support systems. Whether a particular product is regulated depends on its intended purpose and, in some frameworks, on whether a clinician can independently review the basis of its recommendation.

Software raises regulatory questions that hardware does not: cybersecurity, interoperability, version control and update management, and — for machine learning systems — how to handle models that may be retrained over time. Regulators have developed frameworks for predetermined change control so that anticipated modifications can be managed without a new submission for every update, and expect transparency about training data, intended population, and performance limitations.

After Market Entry

Post-market surveillance. Manufacturers must systematically collect and analyse data on device performance in use, with periodic safety reporting for higher-risk classes.
Vigilance and adverse event reporting. Serious incidents and device malfunctions must be reported to regulators within defined timelines.
Field safety actions and recalls. Manufacturers may issue safety notices, corrections, or recalls, which regulators monitor and can require.
Unique Device Identification. Standardised identifiers on devices and packaging support traceability, recall management, and registry linkage.
Registries. Implant and procedure registries track long-term outcomes across large populations and have repeatedly identified problems earlier than passive reporting alone.

Ongoing Challenges

Evidence adequacy for devices cleared on the basis of equivalence rather than new clinical data.
Notified body capacity and certification timelines under strengthened EU requirements, with implications for availability of niche and legacy devices.
Cybersecurity of connected and implantable devices.
Regulating adaptive algorithms whose performance may change after deployment.
Global access, where regulatory capacity, servicing, consumables, and power infrastructure limit which devices are usable in lower-resource settings.
Environmental impact of single-use devices and the reprocessing debate.

Sources

U.S. Food and Drug Administration — device classification; 510(k), De Novo, and premarket approval pathways; software as a medical device guidance; Unique Device Identification
European Commission and European Medicines Agency — Medical Device Regulation and In Vitro Diagnostic Regulation frameworks
International Medical Device Regulators Forum — software as a medical device definitions and risk framework
World Health Organization — medical devices and health technology guidance
International Organization for Standardization — quality management standards for medical devices