What makes a test a point-of-care test?

Three properties define the category operationally. The test must be simple enough for a non-laboratorian to run, fast enough to inform the same clinical encounter, and self-contained enough not to need laboratory infrastructure for sample processing. Lateral-flow antigen tests are the familiar case, but the frontier of the category is molecular: Roche's cleared Bordetella assay delivers PCR-technology results in about 15 minutes on the cobas liat system in general practitioner practices and emergency rooms, according to the company's December 2, 2025 announcement.

Sample preparation is frequently the hard part, not detection. Co-Diagnostics announced in October 2025 that it had developed a proprietary sample preparation instrument for its point-of-care tuberculosis test, engineered for low-cost, user-friendly processing in resource-limited settings, supporting both sputum and tongue swab collection, with single-button operation and a built-in safety feature that inactivates live organisms to protect operators, per the company. The design goal the company stated is decentralizing PCR out of roughly 1,000 district hospitals and toward nearly 30,000 primary health centers in India that currently perform microscopy.

The category also includes instruments masquerading as appliances: handheld meters, cartridge-based analyzers, and benchtop systems that walk the user through steps a technician once performed. What unifies them is the regulatory question of who is allowed to run them, which CLIA answers by category.

How does a test earn a CLIA waiver?

The pathway is documentary and evidence-based, and it runs through the FDA:

1. Clearance or approval. The test is cleared through 510(k) or approved through a premarket approval application on its analytical and clinical performance.
2. Complexity categorization. Tests waived by regulation under 42 CFR 493.15(c), or cleared or approved for home use, are automatically categorized as waived; others are categorized moderate or high complexity under the criteria of 42 CFR 493.17, per the FDA.
3. CLIA Waiver by Application. A manufacturer of a moderate-complexity test may submit a CW application asking FDA to categorize the test as waived, providing evidence that it meets the statutory criteria.
4. Statutory standard. The statute, which the FDA quotes directly, requires that waived examinations be simple laboratory procedures with an insignificant risk of an erroneous result, employing methodologies so simple and accurate as to render the likelihood of erroneous results by the user negligible.
5. Waived categorization. If granted, the test may be performed under a Certificate of Waiver or Certificate for Provider Performed Microscopy, in addition to other CLIA certificate types.

Manufacturers sometimes file the 510(k) and the waiver application together, a dual submission, so that clearance and waived categorization arrive together and the launch can target near-patient settings from day one. The waiver evidence burden is what pushes developers toward forgiving workflows: locked cartridges, automated steps, and internal controls that catch operator error rather than trusting technique.

Why is the waiver the commercial hinge?

Venue volume. Central laboratories number in the thousands; Certificate of Waiver sites number in the hundreds of thousands across U.S. physician offices, clinics, and other settings. A test limited to moderate-complexity settings sells instruments to labs; a waived test sells consumables to every examination room with a certificate. That asymmetry explains why companies invest in waiver-grade evidence packages and why Roche paired its Bordetella clearance with a waiver and CE IVDR certification in a single announcement.

The clinical argument runs alongside the commercial one. Roche framed its test against a diagnostic gap: early pertussis symptoms are often indistinguishable from other respiratory illnesses, and the lack of rapid, accessible diagnostics causes clinicians to treat based on symptoms, with the company citing an estimated 24.1 million cases and 170,000 deaths annually from pertussis. A same-visit molecular answer changes that decision, and only a waived test reaches the settings where most of those decisions are made.

For infectious disease specifically, point-of-care molecular testing also generates geographically dense signal. Co-Diagnostics has described its platform as integrating cloud-based infrastructure with data aggregation intended to help track localized outbreaks as they occur, a surveillance byproduct of distributed testing that central labs, with their transport delays, provide more slowly.

What must a waived site do to stay compliant?

The obligations of the setting itself are lighter than a certified laboratory's, but they are not zero, and they explain part of the category's design. A site performing only waived testing operates under a Certificate of Waiver, which requires enrollment in CLIA, payment of certificate fees on the renewal cycle, and performance of waived tests only, following the manufacturer's instructions exactly as written. The instructions-for-use requirement is the operational heart of the regime: modifications to sample types, timing, or procedure take the test outside its waived categorization, which is why waived products are engineered with locked workflows that make deviation difficult rather than merely discouraged.

The tradeoff embedded in the system is deliberate. Certificate of Waiver sites are not subject to routine proficiency testing or the personnel qualifications of higher-complexity laboratories, which is what allows physician offices to test at all; in exchange, the tests themselves must carry the quality controls internally, and the FDA's waiver standard, insignificant risk of an erroneous result, is the regulatory mechanism that keeps that bargain honest. Every design choice in a waived molecular product, from cartridges to internal controls to single-button operation, exists to satisfy that allocation of responsibility.

For manufacturers, the compliance picture continues after the waiver is granted: labeling changes, new sites of manufacture, and instrument modifications can each raise questions about whether the waived categorization still holds, which is one more reason point-of-care platforms evolve slowly and deliberately compared with laboratory systems that can iterate through software and protocol updates more freely.

What are the limits readers should keep in mind?

Waived does not mean trivial. The category tolerates simple methodologies precisely because the risk of erroneous results must be insignificant, and the FDA's standard is written to keep it that way; complex multiplex panels reach the category only with evidence that untrained operation does not degrade accuracy. Quality management in waived settings is also lighter than in certified laboratories, which is the trade the system accepted in exchange for access.

The second limit is menu breadth. Respiratory and infectious targets dominate waived molecular testing because the clinical urgency justifies the evidence cost; chronic disease markers, complex panels, and quantitative assays remain mostly laboratory territory. The third is cost discipline: near-patient instruments and cartridges must be inexpensive enough for settings without laboratory budgets, which is why companies advertise low-cost instrument and consumable models as a design goal rather than an afterthought.

None of these limits is static. Each newly waived molecular panel lowers the evidentiary and precedential barrier for the next, and the direction of the industry, visible in every clearance-plus-waiver announcement, is toward the examination room.

What Is Total Laboratory Automation?

The Annals of Laboratory Medicine review defines TLA as a solution addressing growing demands for operational efficiency, accuracy, and rapid turnaround times in patient care, integrating advanced technologies across the pre-analytical, analytical, and post-analytical phases to streamline workflows, reduce manual intervention, and enhance quality control. The review attributes adoption to increasing test volumes, cost reduction and regulatory compliance pressure, and labor shortages. In physical terms, a TLA installation is a track system that carries sample tubes from sorting through centrifugation and analysis to storage, with analyzers from one or more vendors docked along the line. The device category is the track, the interfaced analyzers, and the middleware that routes every tube.

What Did a Real Implementation Measure?

A 2026 study in Clinical Laboratory evaluated a hematology and coagulation TLA integration at Taichung Veterans General Hospital, where analyzers have been integrated with an Abbott GLP SYSTEMS TRACK since October 2023. The published study reports that turnaround-time compliance improved from 82.0% to 95.2% for hematology and from 76.9% to 93.0% for coagulation one year after implementation. Monthly blood collection tube consumption decreased by 2,915 units, which the authors price at 437 US dollars of savings and 32 kg CO2e of avoided emissions, and the system eliminated manual specimen transport, saving one full-time-equivalent staff position. The study's authors note that future work should focus on long-term sustainability and clinical outcomes.

What Are the Device Layers in a TLA System?

An automation line is assembled from distinct device classes, each with its own vendors and validation burden.

The review emphasizes that integration across these layers, rather than any single instrument, is what produces the efficiency gains. System integration complexity is likewise listed among the technology's open challenges.

What Does Implementation Require?

The review is candid about the cost side, and the implementation path is a project with laboratory-wide blast radius.

1. Map current workflow volumes and turnaround targets against automation capacity.
2. Select track architecture and confirm analyzer interface compatibility.
3. Rework the physical laboratory around the track footprint.
4. Validate routing rules, autoverification limits, and failure modes.
5. Retrain staff for exception handling rather than routine handling.
6. Monitor turnaround compliance, error rates, and downtime against baseline.

The 2025 review lists high implementation costs, workforce training needs, cybersecurity concerns, and system integration complexities as the standing challenges. The measured benefits arrive only after the validation work is complete.

Where Is the Category Heading?

The review describes recent developments incorporating artificial intelligence, machine learning, robotics, and Internet of Things technologies that enable predictive analytics and automated data management, with future trends pointing toward enhanced AI integration, sustainable practices, and big data analytics. The named market leaders and the technology directions come from the review's synthesis, and vendor-specific claims beyond it are not yet disclosed in comparable form. For hospital laboratories, the decision framework is the one the Taichung study models: measure baseline turnaround and cost, implement, and report the delta with its time window. The device category earns its capital case one measured workflow at a time.

What Changed in the Pre-Analytical Phase?

Most of automation's measured value is won before any assay runs. The pre-analytical phase covers everything from tube receipt through sorting, centrifugation, and aliquoting, and it is where manual handling generates mislabeling, missed spins, and temperature excursions. A track system enforces process physically: tubes are identified, routed, and prepared by the same machinery every time, with exceptions diverted to defined stations rather than improvised. The review's framing of error minimization as a core TLA benefit refers to exactly this stretch of workflow. The Taichung implementation, which eliminated manual specimen transport entirely, demonstrates the endpoint of that logic: one full-time-equivalent position reassigned because the hands-on step no longer exists. Post-analytically, automated storage makes add-on testing a retrieval operation rather than a hunt.

How Do Laboratories Justify the Capital Outlay?

The business case is built from the same measurements the Taichung study reports, and its honesty depends on the baseline. Turnaround-time compliance gains, 82.0% to 95.2% for hematology and 76.9% to 93.0% for coagulation in that study, translate into capacity only if the laboratory can act on faster results. Consumable and staffing deltas, 2,915 fewer tubes per month and one full-time-equivalent position saved in the same implementation, are counted against installation, validation, and service contracts over the system's life. The review is explicit that high implementation costs are among the technology's standing challenges, alongside training and integration. Laboratories that model downtime and middleware validation in the total cost reach more defensible decisions. The measured study is the template precisely because it publishes its baseline, its deltas, and its time window.

Vendor service models absorb some of this risk, with response-time guarantees and spare-part logistics priced into contracts. Laboratories weigh those terms as carefully as throughput specifications, because a track is only as available as its support chain.

What Are the Failure Modes?

Automation concentrates risk as it concentrates workflow, and the failure modes are well characterized. A track stoppage idles every docked analyzer at once, so downtime and recovery procedures carry more weight than in standalone-instrument laboratories. Middleware misconfiguration can route tubes to the wrong assay or autoverify results outside intended rules, which is why validation covers the logic layer and not only the hardware. Cybersecurity appears on the review's challenge list, a consequence of connected systems sitting inside hospital networks. Integration complexity compounds when analyzers from multiple vendors dock one line, each with its own interface behavior. None of these are arguments against the technology; they are the maintenance obligations that come with it.

Regulatory treatment follows the device, not the installation. Track components and interfaced analyzers each carry their own clearances and classifications, and laboratories assembling a line from multiple vendors hold the integration evidence that ties the system together for inspection purposes.

How Does the Category Sit Among Other Lab Devices?

Total automation is one point on a spectrum of laboratory devices, and most laboratories occupy points between. Standalone analyzers with autoverification deliver much of the error reduction without a track. Modular pre-analytical自动化 systems automate sorting and centrifugation only, fitting laboratories whose constraint is front-end handling. Point-of-care instruments sit at the opposite end, trading central volume for proximity to the patient. A track makes sense where volume is high and sustained, which is why the large reference laboratories and big hospitals anchor the market the review describes. The device decision follows the workflow arithmetic, not the technology ceiling.

What Counts as Software as a Medical Device?

The regulatory question is whether the software performs a medical purpose on its own. FDA's Digital Health Center of Excellence explains that SaMD can run across medical device platforms, commercial off-the-shelf platforms, and virtual networks, and that use of the category continues to increase. The definition deliberately excludes software embedded in a hardware device and software used to manufacture or maintain devices, which follow the pathway of the hardware product itself. It also excludes general-purpose wellness software, which is not a device at all. The boundary cases occupy most of the category's regulatory attention, particularly software that interprets data rather than merely displays it.

Which Software Functions Does FDA Actually Regulate?

FDA maintains a public examples page that maps regulated functions to classification regulations. The agency's examples list includes a mobile platform used to produce controlled test tones for diagnostic hearing evaluations, an audiometer function tied to 21 CFR 874.1050, and a sensor-based function measuring tremor caused by certain diseases, tied to 21 CFR 882.1950. The page also lists functions for which FDA exercises enforcement discretion and functions that are not medical devices, which makes it a practical first screen for product teams. The common thread in the regulated examples is interpretation for a diagnostic purpose in specific patients. Display-only and general health functions generally fall outside the device framework.

How Does a SaMD Product Reach the U.S. Market?

The pathway question reduces to risk class, and the process follows a recognizable sequence.

1. Determine whether the software performs a medical purpose; if not, it is outside device regulation.
2. Identify an existing classification regulation and product code for comparable software.
3. If a predicate exists, file a 510(k) premarket notification demonstrating substantial equivalence.
4. If the function is low to moderate risk with no predicate, use the De Novo classification request.
5. For higher-risk functions that support or sustain life, file a premarket approval application.
6. Comply with quality system regulation and postmarket obligations, including software maintenance and updates.

Each step is documentary: the agency acts on the file a manufacturer submits, and the classification follows the intended use statement, not the underlying technology. Changes to a cleared SaMD, including algorithm changes, can require a new submission depending on their effect on safety and effectiveness.

How Does SaMD Differ From Other Device Software?

The three buckets are easily confused, and the comparison determines both the pathway and the evidence burden.

FDA's examples page anchors each row in specific classification citations, which is what makes the framework usable in a product plan. Companies that misclassify early discover the error at submission, which is the most expensive place to find it.

Why Does the SaMD Framework Matter Now?

The category is where diagnostic algorithms, imaging software, and digital biomarkers land when they seek U.S. marketing, and its definitions predate the current wave of AI-enabled products. FDA's device framework applies to AI-enabled medical devices regardless of the underlying model architecture, and the IMDRF-harmonized SaMD definitions give regulators across jurisdictions a shared vocabulary. For developers, the practical discipline is unchanged: state the intended use precisely, find the predicate, and expect the evidence package to scale with risk class. The framework does not pre-judge any pending filing, and each submission is decided on its own record.

Where Do Clinical Decision Support Functions Fit?

Decision support is the busiest boundary in the category, because much of it is intended for clinicians rather than patients and much of it displays rather than interprets. FDA's examples framework treats functions that acquire, process, or interpret patient data for diagnostic or treatment purposes as device functions, while display-only and workflow tools generally are not. The practical test a product team applies is whether the software's output would substitute for clinical judgment about a specific patient. If it would, the function is inside the device framework and needs a classification basis. If it merely organizes information the clinician already has, it likely sits in the enforcement-discretion or non-device rows of the same table. The examples page is the reference for drawing that line before committing to a submission strategy.

What Evidence Does a SaMD Submission Contain?

Whatever the pathway, the file has to demonstrate that the software does what it claims, on the population it claims, under the conditions it will actually meet. Validation datasets with defined ground truth, sensitivity and specificity with their intervals, and a description of failure modes form the analytical core. Software documentation covers architecture, version control, and the testing that ties each release to the validated configuration. For connected products, cybersecurity documentation is part of the file rather than an afterthought. The evidence burden scales with risk class, which is the entire logic of splitting 510(k), De Novo, and premarket approval into separate tracks. Developers who scope the evidence package to the intended use statement early avoid the most common cycle of deficiency letters.

How Is SaMD Regulated Outside the United States?

The category's shared vocabulary is international by construction. The definition FDA uses comes from the International Medical Device Regulators Forum, a body in which multiple regulators work toward harmonized principles, and the same framework underpins how other jurisdictions approach standalone medical software. Harmonization does not mean a single global filing: classification rules, reviewing authorities, and local representation requirements still differ by market. What harmonization buys is a common definitional layer, so a manufacturer can describe one product consistently across dossiers. Companies planning multi-market launches still sequence submissions market by market, on each regulator's documented timeline. The framework aligns vocabulary, not calendars.

What Obligations Continue After Clearance?

Marketing authorization is the midpoint of a SaMD product's regulatory life, not the end of it. Manufacturers operate a quality system that governs how the software is built, released, and changed, and significant changes can require a new submission depending on their effect on safety and effectiveness. Complaint handling, adverse event reporting, and field actions apply to software exactly as they apply to hardware. Because SaMD ships as updateable code, the change-control discipline is more visible to regulators than it is for physical devices, and version history is part of the inspection record. Postmarket evidence can also reshape the intended use over time. The durable rule is simple: the regulatory story of a software product continues for as long as the product runs.

Why is FDA clearance not enough to get paid?

Regulatory clearance answers whether a device is safe and effective for its intended use. Coverage answers a different question: whether a payer will pay for it in a defined population. A device can hold 510(k) clearance and still lack any coverage pathway, because CMS evaluates whether the evidence supports the item being reasonable and necessary for the Medicare population — typically meaning evidence in the population, indication and setting of expected use, not merely in registry-enriched cohorts or younger patients.

The second gate is the benefit category. Requests must identify the benefit category the requester believes matches the item. Certain items are statutorily excluded regardless of evidence — CMS's guidance cites hearing aids, certain dental services and cosmetic surgery — so a technically successful device outside any category has no coverage path without legislation.

What is the difference between an NCD and an LCD?

Per CMS's coverage determination process page, in the absence of a national coverage policy, an item or service may be covered at the discretion of Medicare administrative contractors based on a local coverage determination, which applies only within that contractor's jurisdiction. An NCD, by contrast, binds the program nationally. The Medicare Prescription Drug, Improvement, and Modernization Act of 2003 amended several portions of the NCD development process with an effective date of January 1, 2004, and CMS's current process follows the Federal Register notice of August 7, 2013 (78 FR 48164).

How does the NCD process run step by step?

A complete, formal request can be initiated either by an outside party or internally by CMS staff, and requests are accepted on a rolling basis, prioritized by the magnitude of potential impact on the program and its beneficiaries and by staffing resources, per the agency's process and timeline guide:

1. A requester submits a complete formal request, including the benefit category, supporting evidence and documentation, information on usefulness and benefit to the Medicare population, and a complete explanation of the design, purpose and method of using the item.
2. CMS accepts or declines the request and opens a national coverage analysis if accepted.
3. CMS gathers evidence, at times commissioning an outside technology assessment and consulting the Medicare Evidence Development and Coverage Advisory Committee.
4. CMS publishes a proposed decision for public comment.
5. CMS finalizes the determination, and the policy takes effect on a stated date.

What role does evidence development play?

Where evidence is promising but incomplete, CMS can link coverage to additional data collection — coverage with evidence development — so that payment proceeds alongside registries or studies that answer the open questions. For investigational devices, coverage is also tied to the framework for IDE studies. These routes recognize that device evidence matures after market entry, but they add design obligations that companies must plan for years ahead of a launch.

What should a device company do with this map?

Treat reimbursement as a second development program running in parallel with regulatory work. That means generating coverage-grade evidence — relevant populations, clinically meaningful endpoints, adequate follow-up — and choosing deliberately between a national determination, which suits technologies with broad expected use, and contractor-by-contractor coverage, which can be faster but produces a patchwork. The formal request itself is a regulatory document: incomplete requests are declined on completeness grounds before science is ever discussed.

How do old policies get revisited?

The coverage landscape also ages in place. CMS's process notice of August 7, 2013 outlined an expedited administrative process, using specific criteria, to remove certain NCDs older than ten years since their most recent review — freeing Medicare administrative contractors to determine coverage locally under the Act. The effect is a deliberate devolution: national policies that no longer reflect current evidence can be retired to local discretion rather than rewritten centrally, which creates both opportunity and inconsistency for device companies whose products were covered — or restricted — under an aging national policy.

For a device manufacturer, that mechanism is a strategic lever. A technology restricted by a national policy written a decade or more ago, in an era of thinner evidence, may find that the fastest route to broader coverage runs through reconsideration or through the retirement of the old determination rather than through a fresh evidence program. Reading the date on the governing NCD is therefore as important as reading its text.

What are the common failure modes for device companies?

Four recur. Companies pursue regulatory clearance with populations and endpoints chosen for the agency, then discover that the Medicare population — older, more comorbid, treated in different settings — was never studied, leaving the coverage file thin where it matters. Companies assume that a CPT code implies payment at a workable amount, when coding, payment rates and coverage are separate decisions. Companies underestimate the completeness bar for a formal NCD request and lose months to a declined filing. And companies plan launches around clearance dates with no reimbursement calendar at all, so the device reaches the market and stalls in procurement while the coverage process runs.

Each failure mode has the same root: treating reimbursement as a commercial afterthought rather than a parallel regulatory pathway with its own evidence standard, its own decision documents and its own calendar. The device companies that scale reliably are the ones that write the coverage plan before the pivotal trial is designed — because the trial that satisfies the agency and the trial that satisfies the payer are not always the same trial.

What happens after coverage is secured?

Coverage is the first of three gates, not the last. Coding determines whether a claims pathway exists for the service; payment rules determine what the item or service actually earns; and contractor discretion within local coverage determines how consistently claims are paid in practice. A device with national coverage but no straightforward coding can spend its first commercial year in manual claims review, and a device with coding but restrictive local policies can see materially different utilization across contractor jurisdictions for the same technology.

This is why reimbursement planning documents usually track a pathway matrix: for each target setting and payer, the benefit category, the governing coverage policy — national or local — the codes to be used, and the expected payment. Building the matrix early exposes the gaps while they can still be fixed with evidence, rather than after launch when they can only be fixed with lobbying and time.

How does the public participate?

Both coverage tracks are formally open to participation, and the mechanism differs by level. National coverage analyses run through published proposed decisions with comment periods, and the MEDCAC advisory committee meetings are public, with agendas, presentations and voting questions posted in advance. Local coverage determinations carry their own contractor comment processes. Manufacturers, clinical societies and patient organizations all file comments in meaningful numbers on consequential decisions — which is why the readable record of a technology's coverage fight is usually longer than its clinical literature.

What role does the IDE play in device trial design?

The IDE is the device-world counterpart of the drug IND, but its logic differs. A drug trial's IND mainly opens a pathway to dose humans with an investigational molecule; an IDE frames an entire investigational plan — the device, the protocol, the labeling, the monitoring, and the risk classification. FDA's page describes the requirements: clinical evaluation of devices not cleared for marketing requires an IRB-approved investigational plan, plus FDA approval for significant-risk devices, patient informed consent, investigational-use-only labeling, study monitoring, and required records and reports.

The significant-risk threshold is a design fork. A significant-risk device — one that presents a potential for serious risk to health — needs both FDA and IRB approval before enrollment; non-significant-risk studies can proceed with IRB oversight alone. That classification shapes everything downstream: how much preclinical bench data the package needs, how quickly a first-in-human study can start, and how much the pivotal design will be negotiated with the agency rather than simply filed.

The evidentiary target also differs by pathway. Studies under an IDE are typically performed to support a PMA, the premarket approval application for high-risk devices, while only a small share of 510(k) submissions need clinical data at all. Device trial design is therefore not one discipline but several, calibrated to the risk class and submission type the manufacturer intends to file.

Why do device pivotal trials look so different from drug trials?

Blinded, placebo-controlled, randomized designs — the default grammar of drug development — often translate poorly to devices. A surgeon cannot be blinded to which stent is implanted; a device's effect is frequently mechanical and visible; and sham procedures raise ethical questions that a sham tablet does not. Device pivotal trials therefore lean on alternatives: active comparators against a predicate device, objective performance criteria drawn from prior submissions, historical controls, and single-arm studies judged against performance goals.

FDA's guidance "Design Considerations for Pivotal Clinical Investigations for Medical Devices" makes the design-first posture explicit. The document is "intended to provide guidance to those involved in designing clinical studies intended to support pre-market submissions for medical devices," and it "describes different study design principles relevant to the development of medical device clinical studies" fulfilling pre-market clinical data requirements — while stating it is not "a comprehensive tutorial on the best clinical and statistical practices." The guidance walks through choosing objectives, endpoints, controls, and sample size in that device-specific frame.

Endpoints are the other divergence. Drug endpoints are frequently event rates or survival metrics measured over years; device endpoints often pair a procedural or technical success measure — did the device perform as engineered — with a clinical outcome at a defined follow-up. For an implant, durability data at multiple time points is part of the pivotal question, because the device remains in the body long after the procedure ends.

What does the path from concept to pivotal data look like?

The sequence a device sponsor typically follows:

1. Bench and animal testing: engineering validation and preclinical safety data sized to the risk classification.
2. Risk determination: the study is classified significant-risk or non-significant-risk, which sets the approval path.
3. IDE submission: the investigational plan, protocol, consent materials, and labeling go to FDA and the IRB.
4. First-in-human feasibility study: a small study to refine the procedure, endpoints, and safety profile.
5. Pivotal investigation: the adequately powered study designed to support the marketing submission, negotiated with FDA in advance.
6. Submission: the clinical data set supports a PMA or another premarket pathway, depending on device class.

The negotiation step deserves emphasis. Under the IDE framework, sponsors routinely agree on the pivotal design with FDA before the study launches, because a device PMA lives or dies on whether the agency accepts the chosen comparator and endpoint. FDA's IDE page notes that such studies are typically performed to support PMA, which is precisely why the agency's design guidance exists — the pivotal device trial is a regulatory instrument as much as a scientific one, and its design is part of the approval strategy itself.

How are device endpoints and sample sizes actually set?

Device endpoints usually come in pairs: a performance endpoint measuring whether the device did what its engineering says it should do, and a clinical endpoint measuring what that performance means for the patient. A cardiac ablation catheter, for instance, may be judged on acute electrical isolation as well as on arrhythmia recurrence at twelve months. Because device effect sizes are often large relative to drug effects, sample sizes can be smaller — sometimes tens of patients in a single-arm pivotal study judged against an objective performance goal, rather than the thousands a survival-endpoint drug trial enrolls.

Follow-up duration is set by the device's risk profile. An implant expected to remain in the body for a decade cannot be approved on thirty-day data alone; premarket studies typically specify follow-up windows negotiated with FDA, with longer-term surveillance continuing in post-approval studies. The design question the agency cares about is whether the study will detect the failure modes the bench testing could not rule out.

Where is device trial design heading?

Two currents are visible in the design conversation. One is broader use of real-world evidence: registry data and electronic health records increasingly support — though rarely replace — premarket clinical data, particularly for device modifications and expanding indications. The other is statistical flexibility: adaptive designs that let a study be resized on interim data are more tractable for devices than for drugs, because device iterations are frequent and development cycles short.

The constant across all of it is the negotiation posture established under the IDE framework. Because FDA reviews the investigational plan before the study starts, device trial design is a dialogue with the agency rather than a unilateral bet. Sponsors that treat the pivotal design as a regulatory instrument — agreed, documented, and followed — reach the filing with data the agency has already contextually accepted. Sponsors that discover the design question at submission time pay for it in review time.

When can a device go to trial without one?

Not every device needs a clinical trial at all, and knowing when data is unnecessary is as strategic as knowing how to design it. The 510(k) pathway clears devices that are substantially equivalent to a predicate, and the majority of those submissions rest on bench testing alone. Clinical data enters when the technology is new, when the intended use differs from any predicate, or when the risk classification pushes the device into the PMA or De Novo categories.

That boundary moves over time. As a technology class matures — as happened with some imaging algorithms and established implant categories — the clinical evidence expectations can settle into recognized standards and objective performance goals, letting later entrants run smaller, more standardized studies. The first mover in a category carries the heaviest evidentiary burden; followers inherit the framework that reviewer experience and guidance documents codify.

What is the AI-Enabled Medical Device List?

The list is a curated resource maintained by FDA's Digital Health Center of Excellence. Per the agency's AI-enabled devices page, the devices in the list have met FDA's applicable premarket requirements, including a focused review of overall safety and effectiveness, which includes an evaluation of study appropriateness for the device's intended use and technological characteristics. Each entry links to the FDA database record, which contains releasable information such as summaries of safety and effectiveness.

Two caveats come straight from the agency. The summaries are not all-inclusive and do not include most of what a sponsor may have submitted, and the list is not a comprehensive resource of every AI-enabled device; it was built primarily by identifying AI-related terms in marketing authorization summaries. Imaging dominates the list, which reflects where algorithms reached clinical maturity first: triage of CT and MRI studies, lesion detection and measurement, and image reconstruction.

How do imaging algorithms usually reach the market?

Most AI-enabled imaging devices enter through the 510(k) pathway, demonstrating substantial equivalence to a predicate device. Under the premarket notification framework, a submitter must receive an order finding the device substantially equivalent before marketing, and the comparison runs on intended use and technological characteristics. For algorithm products, predicates are frequently earlier-cleared versions of the same software, or a device of the same type, with performance bench data standing in for clinical trials.

The evidence expectations scale with claimed function. A device that flags a suspected large-vessel occlusion for prioritized review is validated on retrospective reader studies against reference standards; a device that quantifies a measurement must show agreement with the accepted method. What clearance does not confer is autonomy: most cleared algorithms are labeled as aids to the interpreting physician, whose read remains the diagnosis.

What makes AI devices different from other software devices?

Three properties distinguish them in regulatory terms. First, the model is trained on data, so the submission must describe the training, tuning, and test datasets and their separation. Second, performance is statistical, which is why summaries report sensitivity and specificity by use case rather than a single accuracy figure. Third, some models are designed to change after authorization, which engages FDA's discussion of predetermined change control plans: a sponsor specifies up front what parts of the model may update and under what limits, so that changes can occur within the cleared boundaries.

None of this required a new statutory pathway. FDA has so far handled AI devices through existing classifications, special controls where a De Novo created a category, and guidance on changing algorithms. The practical consequence is that an imaging algorithm's regulatory status is readable from its authorization letter and database summary, exactly as for any other device.

What should a buyer or hospital evaluator check?

The regulatory trail answers a short list of questions.

1. The authorization pathway and date, from the FDA database entry.
2. The stated intended use: triage aid, detection aid, quantification, or reconstruction.
3. The population and imaging conditions in the validation studies.
4. Whether the model is locked or operates under a predetermined change control plan.
5. The labeling's statement of the clinician's role in the final interpretation.

The AI list makes the first item a five-minute check, and the database summaries carry most of the rest. Devices absent from the list are not necessarily unauthorized, given the list's own stated limits, but absence plus no database record is a warning sign worth resolving before procurement.

What kinds of imaging algorithms are on the market?

The category is broader than detection tools. Cleared AI-enabled imaging devices span several functional families: triage and notification software that reprioritizes worklists for suspected findings, computer-aided detection and diagnosis tools that mark lesions for the radiologist, quantification software that measures structures or change over time, and reconstruction algorithms that use learned models to produce diagnostic images from less acquired data.

Each family carries its own validation logic. Triage tools are validated on sensitivity to the critical finding and time-to-notification; detection tools on reader performance with and without the algorithm, often in multi-reader multi-case studies; quantification tools on agreement with reference measurements; reconstruction tools on image quality and low-dose performance versus conventional pipelines. The intended use statement in the database summary tells the reader which logic applies.

The common thread is that the algorithm operates inside a clinical workflow that retains a human interpreter. Most authorizations are written as aids, and the summary documents state the studied use conditions, which is what a procurement evaluation should test against local practice.

What is a predetermined change control plan?

Some machine-learning models are designed to improve after deployment, retrained on new data. A predetermined change control plan is the mechanism by which a sponsor and FDA agree in advance on what may change: which parts of the model, trained on what data, validated by what method, with what limits. Changes inside the plan's boundaries can proceed without a new submission; changes outside them cannot.

The plan matters for imaging because model drift is real. Scanner fleets, protocols, and patient populations differ across sites, and a model that quietly degrades at the edge of its training distribution is a safety question, not just a performance question. The plan converts that risk into a documented, auditable schedule.

For evaluators, the practical question is whether a vendor's model is locked, adaptive under a plan, or unclassified on the point. A locked model behaves predictably but freezes its performance; an adaptive model under a plan carries governance obligations; an unspecified answer is a reason to pause the procurement conversation.

How should a hospital evaluate a cleared algorithm?

A disciplined evaluation separates the regulatory record from the local validation question. The sequence below reflects what the authorization documents can and cannot answer.

1. Pull the FDA database entry from the AI-enabled device list and read the intended use and study description in the summary.
2. Map the studied population, scanner platforms, and acquisition parameters against local practice.
3. Confirm the claimed function, triage, detection, quantification, or reconstruction, matches the clinical need.
4. Check whether the model is locked or operates under a predetermined change control plan.
5. Run a local silent-mode evaluation on the institution's own case mix before clinical reliance.

The last step is the one the regulatory record cannot supply. Clearance means the device met FDA's requirements for its stated intended use as studied, not that it will generalize to every scanner fleet and population. Institutions that treat the authorization as the beginning of their evidence, rather than the end, get the value these tools promise and avoid the failures that make headlines.

The regulatory posture also shapes what vendors can build next. Because authorization attaches to an intended use and studied conditions, expansion to a new imaging modality, a new patient population, or a new scanner platform is a new regulatory question, however similar the underlying model. The device list shows this pattern plainly: authorizations cluster by family, with each cluster grown through successive, individually cleared steps rather than a single sweeping approval. That granularity is slow by design, and it is what keeps the market auditable.

What does a CGM actually measure?

A CGM does not measure blood sugar directly. A small filament sensor inserted just under the skin sits in interstitial fluid, where glucose concentration tracks blood glucose with a lag of several minutes. An enzyme-based electrochemical reaction on the sensor generates a current proportional to glucose concentration, and the transmitter sends readings on a fixed schedule to a display device. For the Stelo system, FDA's clearance announcement states that the device presents blood glucose measurements and trends every 15 minutes in the accompanying app, and that users can wear each sensor up to 15 days before replacement.

The signal chain is where the regulatory risk lives. Sensor accuracy, calibration strategy, and the algorithm that smooths raw current into a glucose value all determine whether the system meets accuracy criteria for its category. Interstitial lag also means CGMs are not equivalents of blood glucose meters during rapid glucose changes, which is why labeling constrains who should use any given system.

What is an integrated CGM, and why does the category matter?

The integrated CGM, or iCGM, is a device category FDA created for systems designed to connect safely with other devices, such as insulin pumps and dosing algorithms. iCGM systems must meet special controls on accuracy and reliability so that downstream devices can rely on their signal. The Stelo clearance was for an iCGM, which means the category now spans both prescription-connected systems and consumer wearables.

The iCGM category itself was established through FDA's De Novo classification process, the pathway for novel devices of low-to-moderate risk that have no legally marketed predicate. Under the De Novo classification request framework, a granted request classifies the device into class I or II and creates a predicate that later 510(k) submissions can cite. That is the standard pattern in medtech: one novel device establishes a category, and followers enter through predicate-based 510(k) clearances.

Who is the over-the-counter version for?

The OTC indication is deliberately narrow. FDA's announcement states the Stelo system is intended for adults 18 and older who do not use insulin, including people managing diabetes with oral medications and people without diabetes who want to understand how diet and exercise affect glucose levels. The agency was explicit that the system is not for individuals with problematic hypoglycemia, because it is not designed to alert the user to that condition.

That boundary reflects what an OCGM can and cannot do. A wellness-oriented wearable without hypoglycemia alarms cannot substitute for a prescription CGM in insulin-using patients, where low-glucose alerts are a safety function. FDA's then-CDER-and-CDRH framing in the announcement emphasized access: the clearance lets individuals purchase a CGM without involving a health care provider, which the agency described as a step forward in health equity.

How do these devices get cleared?

Most CGM line extensions follow a short, predictable sequence.

1. Design inputs: accuracy targets, wear duration, and intended user population are fixed against the chosen category's controls.
2. Bench and clinical performance studies against a reference method, typically YSI laboratory glucose analysis.
3. Submission, either a 510(k) demonstrating substantial equivalence to a predicate or a De Novo request if the category is new.
4. FDA review and clearance with labeling that defines the indicated population.
5. Postmarket surveillance, including adverse event reporting through the manufacturer's obligations.

What should professional readers take away?

The CGM story is a case study in how device categories evolve. A prescription-only monitoring tool became a platform category, and the category then split into clinical and consumer branches, each with its own labeling. The numbers that define the branch points are in the FDA documents, not in marketing claims: 15-day wear, readings every 15 minutes, adults not on insulin, no problematic hypoglycemia. For anyone evaluating a new entrant, the clearance letter and indicated population are the first documents to read.

How do prescription and OTC monitors differ in practice?

The split now runs through the middle of the category. Prescription iCGMs are indicated for insulin-using patients and integrated with alarms and, in some systems, automated insulin delivery. OTC systems are indicated for adults not on insulin, and their labeling excludes people with problematic hypoglycemia because the hardware does not promise alerts for dangerous lows. The Stelo clearance language draws that line explicitly, which means the two branches are not substitutes.

The user interface follows the indication. A prescription CGM is a disease-management tool whose data feeds clinical decisions. A consumer CGM is a behavior-feedback device: it shows glucose responses to meals, exercise, and sleep so users can adjust lifestyle. Regulators treat the difference seriously because the populations differ in consequence, not just in intention.

For manufacturers, the branch point defines the evidence package. A consumer submission leans on usability and clear labeling for an untrained population; a prescription submission leans on clinical accuracy in the indicated disease population. Getting the population wrong is the most expensive mistake in the category, because labeling fixes who may buy the product at all.

What do the cleared specs actually say?

The cleared specifications tell more than marketing materials do.

Read as a set, the specs describe a wellness monitoring device with clinical-grade measurement at its core. The 15-minute cadence is slower than alarm-capable prescription systems, which stream continuously to support alerts, and that difference is one more expression of the same population boundary.

What should readers watch in this category?

Three developments define the near term. First, wear time and accuracy targets continue to move, and each clearance that extends duration resets the predicate conversation for followers. Second, the OTC branch is expanding indication by indication, and each new population is a separate regulatory decision rather than an extension of the last. Third, connectivity matters: iCGM classification exists precisely so other devices can build on the glucose signal, and interoperability obligations are part of what the category buys.

The documentary trail remains the fastest way to evaluate any claim in this market: the clearance letter states the population, the specs, and the limitations in FDA's own words. Anything not in that letter is marketing until shown otherwise, and the gap between the two is where most category confusion originates.

The category's trajectory is also a lesson in how regulation follows engineering. Sensors improved first, alarms and connectivity followed, and only after prescription systems matured did a consumer branch open with its own population and its own labeling. Each step left a documentary trail: a De Novo classification that created the iCGM category, predicate clearances that widened it, and an OTC clearance that defined its consumer edge. Readers who follow those documents, rather than launch announcements, will see the next branch point before the press release arrives.

How did the field actually get started?

The documented lineage runs through defense and space research, not hospitals. The concept originated from robotics research funded by NASA and the Defense Advanced Research Projects Agency during the 1970s, with the objective of enabling procedures to be remotely controlled in hazardous or hard-to-reach environments such as battlefields and spacecraft, per the Frontiers in Surgery review. The engineering goal was teleoperation at a distance; surgery on civilians was the peacetime application.

The first milestone came in 1985, when the PUMA 560 robot was used for a neurosurgical biopsy in Pittsburgh, the first-ever surgical robot application documented in the literature. In 1988, Imperial College in London developed ProBot to assist in transurethral prostatectomies, with four axes of movement and a high-speed rotating resection blade. Commercialization followed in the 1990s through Computer Motion, whose AESOP arm received FDA approval in 1994 as the first telepresence surgical robot and was later developed into the three-armed ZEUS system.

What does a modern system consist of?

Three components, consistently across vendors. A surgeon console provides the hand controls and a magnified three-dimensional view of the operative field. A patient-side cart carries the articulated arms holding instruments and camera. A vision system renders the view. The surgeon's movements are translated, scaled, and filtered into instrument motions inside the patient, with tremor reduction that handheld instruments cannot offer.

The architecture has been stable for two decades because it works: the review literature describes these machines as fully controlled remote extensions of the surgeon, and no cleared system operates independently of human command. Generational change has concentrated in the console, the software, and the instrument tips rather than in the fundamental design.

How are these systems regulated?

Almost entirely through 510(k), the FDA's substantial-equivalence pathway, because new systems are cleared against predicate devices already on the market. The documented pattern:

1. A manufacturer develops a new system or a substantial modification of an existing one.
2. The company submits a 510(k) premarket notification demonstrating substantial equivalence to a cleared predicate.
3. The FDA clears the system for specific procedure families.
4. New indications and new software features are added through subsequent 510(k) submissions.

What this pathway does not require is a randomized trial comparing robotic surgery with open or conventional laparoscopic surgery before marketing. Equivalence is to a device, not necessarily to a clinical outcome, and that distinction shapes how the evidence base accumulates: after clearance, in the published literature, rather than before it, in a registrational program.

What did the latest generation add?

The current edge of the dominant platform illustrates the software-led pattern. Intuitive's fifth-generation da Vinci 5 received FDA clearance in March 2024, and the company began a limited rollout of the system thereafter, as MedTech Dive reported. The system brought force feedback sensing, better console ergonomics, a smaller physical footprint, and greater computing power than earlier generations.

In September 2025, three new software capabilities for da Vinci 5 received 510(k) clearance. One gives surgeons a replay of key moments in a procedure, reviewable without removing their head from the console. Another is a gauge displaying measurements of the force applied to the patient's tissue by instruments, working like a speedometer on top of the existing force-feedback sensing. The updates are the first in a planned series of capabilities, per the company's statement to the trade press.

What does the evidence show about outcomes?

What the literature supports is narrower than the marketing. The Frontiers in Surgery review reports that, in comparison to open surgery, use of the da Vinci robot has shown significant improvement in clinical outcomes, often demonstrating less blood loss and shorter recovery times. Those findings come largely from comparative studies across surgical specialties, not from a single uniform trial program, and the review's own framing is cautious about generalizing across procedures.

The honest statement of the gap: robotic approaches compete against laparoscopy, not just against open surgery, and advantages over laparoscopy are smaller and more procedure-dependent than advantages over open surgery. Operative time, cost, and conversion rates vary by procedure and by surgeon experience, and the learning curve is documented in the literature. Early-user observations, such as a robotic surgery medical director's report that surgeons using force feedback consistently apply less force to tissue, are practitioner reports rather than controlled findings.

What does the operating-room economics look like?

Three cost layers define a robotics program for a hospital procurement and finance team. The capital system, console, patient-side cart, and vision tower, is a seven-figure-range acquisition amortized over years. The instruments are constrained consumables, limited in number of uses, which makes per-procedure cost track procedure volume directly. And service coverage ties uptime to a maintenance contract, because a robotic case cancelled mid-procedure for equipment reasons loses the hospital both the case itself and the schedule slot it occupied on the surgical calendar.

Against that stands the revenue and utilization logic. Robotic cases can support shorter stays than open surgery, which the review literature documents as less blood loss and shorter recovery, and a busy program spreads the capital cost across a high case volume. The honest accounting, though, is procedure-specific: where the clinical advantage over laparoscopy is small, the cost difference still has to be justified, and hospitals that cannot fill a program's schedule carry the capital cost without the volume to absorb it.

The training layer is a real cost too, and not only in money. Surgeon and team proficiency is acquired through structured training programs and supervised case volume, and the peer-reviewed literature treats the learning curve as a variable in outcome comparisons, which means early program results are not the steady-state results by which the platform should ultimately be judged.

Where is the field going?

Three documented directions. Single-port platforms, which pass multiple instruments through one incision, are expanding the procedure set toward operations where cosmesis and recovery time dominate. Competing multiport systems from new entrants are reaching markets as the dominant installed base ages and as procurement teams gain alternatives to evaluate. And software features, force feedback and case analytics among them, are where announced differentiation is concentrated, sold as capability updates to an installed base rather than as new towers.

In each direction the adoption question is the same one the field has carried since PUMA 560: not whether the arm can move, but which patients measurably benefit, at what cost, compared with the alternative already in the room next door. That answer continues to accumulate procedure by procedure, in the published record, one specialty and one system update at a time. The honest read, three decades into the robotic era, is that it is still accumulating, and that the burden of proof sits exactly where it always has, with the clinical evidence rather than with the arm.

What does the general wellness policy actually cover?

FDA's guidance, docketed FDA-2014-N-1039, states its purpose as providing "clarity to industry and FDA staff on the Center for Devices and Radiological Health's (CDRH's) compliance policy for low risk products that promote a healthy lifestyle (general wellness products)," and notes it does not apply to products regulated by other FDA Centers, per the guidance page. The policy rests on the 21st Century Cures Act provision that removed certain healthy-lifestyle software functions from the device definition. In practice, that covers step counters, sleep scores, and exercise coaching — claims about staying well. The moment a claim references a disease state, the product crosses into device territory and the claim, not the hardware, carries it there.

Which pathway applies once a wearable is a device?

Device regulation for wearables is a ladder, and the rungs are determined by risk and precedent.

1. General wellness (no premarket review) — low-risk lifestyle claims only, per the Cures Act carve-out and FDA's compliance policy.
2. 510(k) clearance — the manufacturer shows substantial equivalence to a legally marketed predicate; most ECG-equipped wearables take this route.
3. De Novo classification — for novel devices of low-to-moderate risk without a predicate; it creates a new classification and a future predicate for others.

The ladder explains a pattern readers will recognize: each new sensor category — heart rhythm, blood pressure, temperature — initially needs a heavier pathway, then later products of the same type clear against earlier ones more quickly.

How does a clearance read in practice?

The December 2025 HeartBeam decision shows the mechanics. On December 10, 2025, the company announced FDA 510(k) clearance for its cable-free synthesized 12-lead ECG for at-home arrhythmia assessment — a credit-card-sized device delivering clinical-grade insights directly to patients — after a successful appeal overturning a prior Not Substantially Equivalent outcome, per HeartBeam's press release. The notable regulatory fact is procedural: 510(k) decisions are reviewable through the appeal process, and an NSE determination is not the end of the road when the equivalence argument is restated. The clearance is the company's announcement of the agency's decision; the device's exact cleared claims live in the agency's 510(k) database.

Where is the boundary most often tested?

Two places. First, consumer copy drifting toward clinical claims — marketing that says a feature detects a condition can convert a wellness product into an unclassified device, which is a compliance question, not a software question. Second, software updates: a cleared algorithm's performance claims are fixed to the version reviewed, and meaningful changes generally require a new submission or a new 510(k). The comparison that matters for readers:

What does software change about the review?

Three things, all procedural. First, software functions can be regulated as devices in their own right — Software as a Medical Device — so a wearable's regulated component may be the algorithm rather than the strap or sensor housing it runs on. Second, the review evaluates the algorithm's validation population: for which patients, against which reference standard, with what agreement, in the labeling the sponsor proposed. Third, change control: a cleared algorithm's claims attach to the reviewed version, and sponsors navigate predetermined change protocols or new submissions to ship performance-affecting updates. For readers of clearance announcements, the durable facts are the cleared indication, the predicate where one exists, and the date — all of which live in FDA's public databases rather than the press release.

The working rule: read the claim on the box, not the sensor on the wrist. FDA regulates the sentence, and every pathway question in wearables resolves back to it.

Where is the market pushing the line next?

Toward features that used to be clinic-only. Blood-pressure estimation, temperature trends, and continuous glucose access for non-diabetic wellness uses each sit differently on the boundary: some have predicates and clear through 510(k), some would need new classifications, and some remain wellness claims so long as their labeling avoids disease language. The consistent regulatory fact is that the boundary moves by guidance and clearance precedent, published and dated, rather than by device category. Readers tracking the space should watch three public records — the guidance docket, the De Novo grant list, and the 510(k) database — because every future wearable's regulatory identity will be decided in one of them.

When a new claim appears whose pathway is unclear, the fastest professional check is the labeling: if the claim references a disease or a clinical action, a pathway applies; if it references fitness and lifestyle only, the wellness policy governs. That single reading discipline, applied consistently, resolves most of the confusion that surrounds wearable regulation in industry discussion.

Everything else in the wearable regulatory conversation — category debates, sensor novelty, consumer enthusiasm — is downstream of that sentence-level decision, made first by the sponsor and confirmed or challenged by the agency.

What counts as an in vitro diagnostic in the US?

An IVD is a device, under the Federal Food, Drug, and Cosmetic Act, that consists of reagents, instruments or systems intended for use in the diagnosis of disease or other conditions. The regulatory definition in 21 CFR 809.3(a) became the fulcrum of the LDT fight: on May 6, 2024, the FDA issued a final rule amending that definition to add the words including when the manufacturer of these products is a laboratory, per the agency's Laboratory Developed Tests page. That single clause would have pulled tests made and used inside clinical laboratories into FDA device regulation.

The attempt did not survive judicial review. On March 31, 2025, a federal district court vacated that final rule, and on September 19, 2025, the FDA issued a final rule reverting to the text of the regulation as it existed prior to the effective date of the May 2024 final rule, per the same page. The practical result is that laboratory-developed tests sit outside FDA premarket review under the current text, while IVDs manufactured by companies for sale remain fully within it.

Which FDA pathway applies to a marketed IVD?

For tests sold as products, the route to market follows device classification, and the decision sequence is deterministic:

1. Classify the IVD by risk and special controls; most routine chemistry and immunology assays fall in class II.
2. For an IVD with a valid predicate, submit a 510(k) premarket notification demonstrating substantial equivalence.
3. For a novel low-to-moderate-risk IVD without a predicate, request De Novo classification.
4. For high-risk IVDs, including many companion diagnostics, proceed through premarket approval.
5. Sustain compliance with quality-system and postmarket requirements across the product life.

Companion diagnostics illustrate the stakes of pathway choice, because a drug's approved use can be conditioned on a specific test result, tying the test's regulatory status to the drug's label. The quality-system layer changed for everyone on February 2, 2026, when the QMSR took effect, amending 21 CFR Part 820 by incorporating ISO 13485:2016, per the FDA's device program pages. IVD manufacturers are subject to the same quality-system rule as therapeutic-device makers, and the FDA began using its updated inspection process the same day.

What does the EU IVDR require?

Europe runs its own regime under Regulation (EU) 2017/746, which replaced the older IVD directive and has been applicable since May 26, 2022. The regulation imposes conditions on laboratories that develop and use in-house IVDs, a European counterpart to the LDT question that ended differently: rather than exempting in-house tests wholesale, the IVDR allows them only where no equivalent commercial device is available and a set of quality, justification and documentation conditions is met, as analyzed in the peer-reviewed pathology literature.

The EU system also classifies IVDs into risk classes with notified-body involvement for most categories, a structural contrast with the US where the majority of IVDs clear through predicate-based review. A 2023 analysis in the journal Pathologie, examining practical implementation of the regulation's requirements in pathology institutes, describes the conditions the IVDR imposes on in-house IVD development and use; that literature is the reference point for how European laboratories have experienced the transition.

How do the two systems differ where it matters?

The sharpest divergence is now the laboratory test question. The EU chose conditional inclusion: in-house tests are permitted inside a regulated envelope. The US, after the 2024 rule was vacated in court, has returned to a text under which laboratory-developed tests are not FDA-regulated devices, per the agency's September 2025 reversion. Manufacturers selling products in both regions therefore face a notified-body conformity regime in Europe and a classification-based FDA pathway in the US, with separate quality-system obligations that the QMSR has now partially harmonized with the ISO standard European authorities already use.

Timing asymmetry compounds the structural one. A US 510(k) review runs on a statutory decision clock measured in months, while European certification under the IVDR depends on notified-body slot availability as much as on document quality, a constraint laboratories and manufacturers have documented since the regulation became applicable. Programs that sequence the US filing first and EU certification second, or that limit early launches to selected member states, are responding to that calendar reality rather than to any difference in scientific standard.

Postmarket obligations also diverge in emphasis. The FDA's device system leans on inspection, recall authority and the quality-system rule now expressed through the QMSR, while the IVDR builds in periodic safety update reporting, performance follow-up studies for certain devices and economic-operator registration duties across the supply chain. For a diagnostics company, the compliance function is effectively two disciplines run in parallel, and the cost of treating one jurisdiction's habits as universal shows up in review correspondence.

What should manufacturers watch from here?

Three threads are live. First, the LDT question is unresolved rather than settled: the FDA's reversion is a regulatory text change, and any durable framework for laboratory test oversight would now require either new rulemaking on a different legal basis or legislation, neither of which has been enacted. Second, QMSR inspection practice is new: the FDA retired its legacy inspection technique on the February 2026 effective date, per the agency's FAQ page, and early inspection outcomes under Compliance Program 7382.850 will define what compliance looks like in practice. Third, the European transition continues to consume notified-body capacity, which is a scheduling variable for any IVD requiring certification.

For teams planning programs, the checklist is straightforward: classify early in both regions, sequence US pathway selection against EU conformity assessment timelines, and treat the FDA's LDT page and the peer-reviewed IVDR analyses as the checkable anchors for the two boundary questions that moved most recently.

What changed on February 2?

The QMSR harmonizes the FDA's CGMP framework with the one used by other regulatory authorities, per the agency's program page. Two operational changes took effect the same day: the FDA stopped using the Quality System Inspection Technique (QSIT) for device inspections and began using the inspection process described in Compliance Program 7382.850, and the agency retired two legacy inspection documents (7382.845 and 7383.001), per the FDA's QMSR frequently asked questions page.

The rule was finalized in 2024 after a long comment process, giving manufacturers close to two years to close gaps between the old part 820 text and ISO 13485:2016 clauses. For established device makers already certified to ISO 13485, the burden is largely mapping and documentation; for smaller shops built on the legacy US text, the transition touches design controls, supplier controls and record-keeping.

Does an ISO 13485 certificate now prove compliance?

No. The regulation incorporates ISO 13485:2016 by reference, but the FDA has been explicit that the agency keeps its own requirements layered on top of the standard. The revised part 820 is titled the Quality Management System Regulation, effective February 2, 2026, and it gives the FDA authority to inspect areas such as management review, quality audits and supplier audit reports, per the agency's FAQ page.

That inspection authority is the practical teeth of the change. A certificate from an accredited certification body is evidence a quality system exists, but it is not the legal standard; the FDA's own inspection outcomes under Compliance Program 7382.850 are. Manufacturers that treated the effective date as a paperwork milestone face the difference in their next inspection, since the QMSR program page states the agency began utilizing the updated inspection process on February 2, 2026.

How does the QMSR interact with clearance pathways?

The QMSR sits in the quality-system layer of device regulation, separate from the premarket pathways that get a device on the market. For context, the route a device takes depends on its classification and predicate history, and the general sequence runs:

1. Determine the device's classification and applicable controls.
2. For a device with a lawful predicate, submit a 510(k) premarket notification before marketing.
3. For a novel low-to-moderate-risk device without a predicate, request De Novo classification.
4. For high-risk class III devices, proceed through premarket approval.
5. Maintain QMSR compliance across the product's life, verified by FDA inspection.

The February 2026 change did not alter which pathway a device follows or its review timeline; it changed what a compliant manufacturing operation looks like once the device is cleared or approved. IVD manufacturers are affected alongside therapeutic-device makers, since the quality-system rule reaches all regulated device categories. The agency's pages describe the effective-date mechanics; company-specific remediation status is not yet disclosed by most manufacturers.