Signals

Operator notes.

How we watch FDA, and how to act on what the watching returns — the method behind the product, written down. Regulatory intelligence and practice notes, never regulatory advice.

Method·Sep 15, 2026 · 4 min

The PMA clock starts at the last module

FDA lets a PMA or HDE applicant file in FDA-agreed modules instead of one complete application — but the statutory 180-day review clock, and the acceptance and filing decisions that start it, all run from the date the last module arrives, not the first.

Read the post →
Practice·Sep 15, 2026 · 4 min

The Q-Sub that doesn't buy you more time

FDA's Q-Submission Program runs five distinct request types, not one — and the Submission Issue Request, built for exactly the moment a deficiency letter lands, does not pause or substitute for the formal response the letter already put on a clock.

Read the post →
Method·Sep 14, 2026 · 4 min

Lapsed and revoked end an EUA differently

An Emergency Use Authorization depends on a declaration underneath it — and the declaration can lapse on its own calendar, ending every EUA built on it, without FDA ever revoking a single one for anything the product did.

Read the post →
Practice·Sep 14, 2026 · 3 min

A recall strategy has a grade. openFDA doesn't.

21 CFR 7.42 makes a firm commit to a depth of recall and an effectiveness-check grade before FDA accepts a recall strategy — a document distinct from the health-hazard classification, and one the public enforcement record never carries.

Read the post →
Method·Sep 13, 2026 · 3 min

A granted De Novo becomes a predicate.

De Novo used to require a rejection first — file a 510(k), collect a Not Substantially Equivalent letter, then ask to be reclassified. FDASIA made the direct route law in 2012, and the device that comes out the other end can become someone else's predicate.

Read the post →
Practice·Sep 13, 2026 · 3 min

Least burdensome changes the ask, not the bar.

Sponsors invoke ‘least burdensome’ as leverage for a lighter review. The statute defines it the other way — a limit on what FDA may ask for, with its own line, written into the law, that the underlying standard does not move.

Read the post →
Method·Sep 12, 2026 · 3 min

The lab is accredited. Not the device.

ASCA lets a submission use an FDA-accredited testing lab and a shorter Summary Test Report instead of a full one. That is a paperwork change to how a submission gets reviewed, never a lower bar for what a device has to meet.

Read the post →
Practice·Sep 12, 2026 · 3 min

Breakthrough is law. STeP is a guidance.

Breakthrough Devices sits in a section of the FD&C Act Congress wrote in 2016. The Safer Technologies Program sits entirely in FDA guidance — and their nearly identical 60-day clocks carry different legal weight because of it.

Read the post →
Practice·Sep 11, 2026 · 3 min

MDSAP replaces routine. Not for-cause.

MDSAP lets one audit satisfy five regulators' routine surveillance at once. It changes nothing about the inspection FDA opens for its own reasons — for-cause, pre-approval, or a combination product's radiation-emitting component.

Read the post →
Method·Sep 11, 2026 · 3 min

No field on a 510(k) says what it treats

Intended use and indications for use are legally distinct questions under FDA's own rules, and neither is a field in the 510(k) record the API actually publishes — only a product code stands in for what a cleared device is for.

Read the post →
Method·Sep 10, 2026 · 4 min

Discretion is a policy. Exclusion is a statute.

FDA's January 2026 guidance revisions put a statutory device-definition carve-out and an enforcement-discretion policy side by side on the same digital-health products. One took an act of Congress to create. The other is a guidance FDA can rewrite on its own schedule.

Read the post →
Practice·Sep 10, 2026 · 4 min

Designated by one office. Approved by another.

A Humanitarian Use Device designation is FDA's Office of Orphan Products Development acting on a 45-day clock — a real, dated determination that precedes the HDE application CDRH later reviews, and that never becomes a Federal Register document at all.

Read the post →
Method·Sep 9, 2026 · 4 min

MedSun sits beside MAUDE, not inside it

MAUDE is the adverse-event database every device monitor reads. MedSun is a second, older CDRH network of roughly 300 hospitals and clinics reporting the close calls MAUDE's mandatory threshold was never built to catch.

Read the post →
Practice·Sep 9, 2026 · 4 min

The five-unit device never gets a K-number

FD&C Act §520(b) exempts a device from 510(k) and PMA review entirely once a firm caps a device type at five units a year, built to a physician's order for a named patient — and the only paper trail left behind is an annual notice to FDA that never becomes a public record.

Read the post →
Practice·Sep 8, 2026 · 4 min

Meeting the criteria isn't a tracking order

21 CFR 821.1 sets three criteria that make a Class II or III device eligible for FDA's device-tracking program. 21 CFR 821.20 makes the eligibility irrelevant on its own — nothing is a tracked device until FDA actually issues the order.

Read the post →
Method·Sep 8, 2026 · 4 min

The PMA clock doesn't stop at approval

21 CFR 814.84 makes continued approval of a PMA contingent on filing a periodic report, once a year unless FDA sets another interval, for as long as the approval stands — a filing with no supplement number, no decision date, and no row in openFDA's PMA record.

Read the post →
Practice·Sep 7, 2026 · 4 min

The K-number doesn't track today's device

21 CFR 807.81(a)(3) lets a manufacturer decide, in its own quality records, that a design or manufacturing change didn't rise to needing a new 510(k) — a judgment FDA never reviews before an inspection asks for it, and the public clearance record never shows either way.

Read the post →
Method·Sep 7, 2026 · 4 min

A MAUDE report proves it was filed

openFDA's device-event data carries a source_type field marking who submitted each report — a manufacturer or user facility working under a legal duty defined at 21 CFR 803, or a health professional, patient, or consumer reporting voluntarily, under none. A search result renders both the same way.

Read the post →
Practice·Sep 6, 2026 · 3 min

Not every MDR starts with the manufacturer

Medical Device Reporting is usually read as a manufacturer's thirty-day obligation. Hospitals, ambulatory surgical facilities, and nursing homes carry their own independent MDR duty under 21 CFR 803 — one that can put a death report in front of FDA before the manufacturer even knows the event happened.

Read the post →
Method·Sep 6, 2026 · 3 min

Completed. Not terminated.

A device recall's public status field carries three words, not two — Ongoing, Completed, and Terminated — and only the last one means FDA itself determined the correction actually worked, not just that the firm said it finished.

Read the post →
Practice·Sep 5, 2026 · 4 min

The device-ban list has three entries

Section 516 of the FD&C Act lets FDA remove a device from the market outright, not just recall it — a bar so high the agency has cleared it three times since 1976, and a court erased the most recent one before it ever took full effect.

Read the post →
Method·Sep 5, 2026 · 4 min

The UDI exception has its own deadline

A device manufactured and labeled before its Unique Device Identification compliance date doesn't need a UDI added retroactively — 21 CFR 801.30 says so. The same regulation also says that exception runs out three years later, for every device it was ever granted to.

Read the post →
Method·Sep 4, 2026 · 3 min

MDUFA VI is being negotiated in public

Section 738A of the FD&C Act requires FDA to consult publicly on the next device user-fee program years before it takes effect — monthly meetings with patient and consumer groups, a public meeting, a comment period, and a docket that opened this summer for a program that won't start until fiscal year 2028.

Read the post →
Practice·Sep 4, 2026 · 3 min

The inspection manual changed, not the rule

FDA replaced QSIT with Compliance Program 7382.850 the same day the Quality Management System Regulation took effect — the document that tells an investigator what to check during a device inspection, posted with no Federal Register notice and no advance comment period, because a compliance program isn't a rule.

Read the post →
Method·Sep 2, 2026 · 3 min

The count Congress had to legislate

FDA has issued Medical Device Safety Communications for years with no CFR trigger, no compliance deadline, and no openFDA endpoint returning them — until a 2023 appropriations act forced FDA to start counting them itself, once every two years.

Read the post →
Practice·Sep 2, 2026 · 3 min

The small-business rate isn't retroactive

FDA's medical device user fee program runs three separate gross-receipts tests, each requiring its own annual Small Business Determination. File a 510(k) or PMA before FDA issues that number and the standard fee is owed — with no refund once the determination clears.

Read the post →
Method·Sep 1, 2026 · 3 min

A detention order has no docket to read

Under 21 U.S.C. § 334(g) and 21 CFR 800.55, an FDA inspector can order a device held in place — unusable, unmoved — the same day an inspection turns up a suspected violation, with no warning letter first and no public record of it ever having happened.

Read the post →
Practice·Sep 1, 2026 · 3 min

PMA approvals moved off the Federal Register

FDA Radar's own taxonomy carries a category called PMA Approval, built like every other category on the assumption that a Federal Register document announces the event. Since January 2022, under FDA's own rule, none does — the approvals moved to FDA's website and the category stayed behind, empty.

Read the post →
Method·Aug 31, 2026 · 3 min

A citizen petition names names

Any person can ask FDA's Commissioner, under 21 CFR 10.30, to act on a single named product or competitor — filed as a public docket, answered on a statutory 180-day clock, through a process none of FDA Radar's four sources ever reads.

Read the post →
Practice·Aug 31, 2026 · 3 min

A market withdrawal isn't a recall

FDA's own definitions split a firm pulling product back into three distinct actions — recall, market withdrawal, stock recovery — by whether a violation exists at all, not by severity. Only the first is built to reach the enforcement database FDA Radar reads.

Read the post →
Method·Aug 30, 2026 · 4 min

Not every device letter comes from CDRH

FDA Radar's warning-letter pipeline keeps a letter only when its issuing office reads "Center for Devices and Radiological Health." A drug-led combination product's manufacturing site can draw a letter over the same design-control and CAPA failures a device-led one would — issued by CDER, and filtered out before it ever becomes an item.

Read the post →
Practice·Aug 30, 2026 · 4 min

Closed on one device. Open on the rest.

FDA Radar tracks a recommended action's state once per matched product, not once per document. Close it out against your lead device and the identical recommendation stays open — correctly — everywhere else the same warning letter or recall reached.

Read the post →
Method·Aug 29, 2026 · 4 min

No IDE is confirmed until the PMA is

An investigational device exemption's existence is confidential by FDA's own rule until the device wins premarket approval or a product development protocol notice takes effect. Until then, the sponsor and the reviewing IRB, not FDA, decide whether an IDE even has to exist.

Read the post →
Practice·Aug 29, 2026 · 4 min

The remedial-action call sets the clock

Two identical device malfunctions can land on different FDA reporting clocks — thirty days under 21 CFR 803.50, five working days under 803.53. The difference isn't the malfunction. It's the manufacturer's own conclusion about whether remedial action is needed to prevent unreasonable risk of substantial harm.

Read the post →
Method·Aug 28, 2026 · 4 min

Jurisdiction is a letter, not a field

Which FDA center reviews a combination product — device, drug, or biologic lead — turns on primary mode of action under 21 CFR Part 3, settled by a Request for Designation letter that never becomes part of any public classification or clearance record. Miss the 60-day deadline, and FDA's own rule hands the sponsor's own recommendation the win by default.

Read the post →
Practice·Aug 28, 2026 · 3 min

An address is not a footprint

The openFDA 510(k) record keeps an applicant's city, state, and country — a mailing address, not a manufacturing site. The establishment registration number 21 CFR 807.87 requires every submission to carry never reaches that record, or FDA Radar's own ingestion of it, at all.

Read the post →
Method·Aug 27, 2026 · 4 min

The final rule that isn't final yet

A Direct Final Rule publishes as a finished action — an effective date, a CFR citation, done. FDA's own procedure ships every one with a second, inert document next to it: a companion proposed rule that only activates if someone files an adverse comment, and FDA Radar's own categorizer reads both documents the same way it reads an ordinary rule.

Read the post →
Practice·Aug 27, 2026 · 4 min

The special controls aren't in the record

A Class II product code isn't low-risk by trust — it comes with special controls, the conditions FDA requires in place of full premarket approval. Those conditions often take the written form of a named guidance document, tied to the product code by the classification order that created it, not by anything the classification database itself returns.

Read the post →
Method·Aug 26, 2026 · 3 min

Exempt from GMP. Not from complaints.

FDA's classification database marks a product code gmp_exempt_flag: Y and stops there. The regulation behind that single letter carves the exemption back out for exactly two things — records and complaint files — for the codes that carry it.

Read the post →
Practice·Aug 26, 2026 · 4 min

The notify list isn't the shortage list

Section 506J's product-code list names roughly 285 codes whose manufacturers must warn FDA before a shortage happens. FDA's separate Medical Device Shortages List names devices FDA has already determined are short. A code on one says nothing about whether it sits on the other.

Read the post →
Method·Aug 25, 2026 · 4 min

A constraint can't be skipped. A script can.

FDA Radar's database refuses to approve an item without a human-written urgency and recommendation — a rule the classifier cannot touch, because the schema it returns has no field for either one. The same bright line applied to the model's own prose has no such guarantee. It has a script, and a script only runs when someone remembers to.

Read the post →
Practice·Aug 25, 2026 · 3 min

No cron sends this. A person does.

FDA Radar's digest job writes an HTML file and a text file to disk. It does not have a send function. The last step — pasting the result into a mail client and pressing send — belongs to a person, on purpose, because that is the same step FDA's own complaint-handling rule refuses to let happen silently.

Read the post →
Method·Aug 24, 2026 · 3 min

Corrected, not rewritten

FDA Radar's correction record has one write path — a new, dated row next to the item it corrects — and no code path in the product ever updates or deletes one. That is the same shape 21 CFR 11.10(e) requires of a manufacturer's own electronic-record changes: a new entry, not an edit to the old one.

Read the post →
Practice·Aug 24, 2026 · 3 min

Quiet is a finding, not a default

A monitor that goes silent because nothing happened and one that goes silent because nobody checked produce the identical blank inbox. FDA Radar's personalized digest states the count of documents it read before calling a week quiet — because that number is the only thing that tells the two apart.

Read the post →
Method·Aug 23, 2026 · 3 min

The summary is public. The statement isn't.

A 510(k) applicant satisfies the same disclosure rule two different ways — a summary FDA posts for anyone to read, or a statement promising the same safety-and-effectiveness data to whoever asks within 30 days. The clearance record doesn't say which one a given filing chose.

Read the post →
Practice·Aug 23, 2026 · 4 min

A competitor's 513(g) leaves no trace

Section 513(g) lets any person ask FDA, in writing, exactly how a device is classified and what applies to it — and get a statutory answer within 60 days. FDA's own procedures keep both the question and the answer out of every public record, including the ones FDA Radar reads.

Read the post →
Practice·Aug 22, 2026 · 3 min

An empty response field proves nothing

FDA's warning-letter search page shows, on every row, whether a response letter and a closeout letter have been posted. A blank response-letter indicator means the firm chose not to publish one — not that none exists, and not that FDA found it wanting.

Read the post →
Method·Aug 22, 2026 · 4 min

A petition is public. Its outcome isn't.

A competitor can ask FDA, under 21 CFR 860.130, to reclassify your product code — a filing FDA must make public the day it arrives. The Federal Register notice that would put it in front of a classification watch doesn't have to follow for 180 days, or ever.

Read the post →
Practice·Aug 21, 2026 · 3 min

The panel is a category, not a meeting

Every 510(k) FDA Radar ingests carries a line reading "Panel: Cardiovascular Devices Panel of the Medical Devices Advisory Committee" or similar — openFDA's own field, on every record. It names which specialty reviews that device type administratively. It does not mean that panel, or any panel, ever actually met to discuss this device — most clearances, and most approvals, never see one.

Read the post →
Method·Aug 21, 2026 · 3 min

An accessory carries its own classification

For decades, FDA classified a device accessory the same as the device it was built for — a delivery catheter for a Class III valve went through PMA review right alongside it. A 2017 law changed the default: FDA now classifies an accessory by its own risk, which can land it on its own product code, in its own class, entirely apart from the parent device a portfolio watch was actually built to follow.

Read the post →
Method·Aug 20, 2026 · 4 min

A TPLC report is a category, not a device

FDA's Total Product Life Cycle database stitches a device's premarket and postmarket record into one report — 510(k)s, PMAs, adverse events, and recalls in a single view. The key it stitches on is the product code, a generic device type shared by every maker in it, so the counts on "your" report belong to a category, not to your device.

Read the post →
Practice·Aug 20, 2026 · 4 min

A 522 order is a question, not a verdict

Recalls and warning letters are the signals every monitor watches. Section 522 of the FD&C Act carries an earlier, quieter one: FDA can order a manufacturer to study a device already on the market, because the agency has a safety question it cannot yet answer. The order is not a finding of harm. It is a question, filed in a public database updated every Sunday.

Read the post →
Method·Aug 19, 2026 · 4 min

A PCCP change leaves no new record

FD&C Act Section 515C, added by statute in December 2022 and given final guidance in December 2024, lets FDA pre-authorize an AI-enabled device's own future software changes inside its original clearance. Implement one, and no new 510(k) gets filed — nothing in the public clearance record marks that the device changed at all.

Read the post →
Practice·Aug 19, 2026 · 3 min

Not every signal becomes a recall

FDA's Emerging Signals policy lets CDRH tell the public about a device safety concern it is still evaluating — before the evidence is confirmed, and often before any company has taken, or ever takes, a corrective action a recall record would capture.

Read the post →
Method·Aug 18, 2026 · 3 min

eSTAR did not remove the acceptance check

FDA's Refuse to Accept review used to run after a 510(k) reached the agency. For an eSTAR submission, mandatory since October 2023, the same completeness check runs inside the template before it is ever filed — and no field in the public 510(k) record marks which way a given clearance went through.

Read the post →
Practice·Aug 18, 2026 · 3 min

An early alert has no recall number yet

CDRH's Early Alert pilot, expanded to every device type in September 2025, posts a likely high-risk removal or correction before FDA classifies it — before the event has a recall number, and before it reaches the Recall Enterprise System feed FDA Radar's own pipeline actually ingests.

Read the post →
Practice·Aug 17, 2026 · 3 min

A consent decree isn't a database row

FDA Radar's warning-letter adapter reads a structured, public database every day. A Warning Letter that goes unanswered can escalate into a federal consent decree of permanent injunction — a court order announced case by case, in a record no adapter here, or in most monitoring built like it, ever touches.

Read the post →
Method·Aug 17, 2026 · 3 min

The record has no cybersecurity field

Since March 2023, a 510(k), PMA, or De Novo for a device FDA calls a "cyber device" cannot even be accepted without a software bill of materials and a vulnerability-management plan. The public clearance record that comes back afterward — FDA's own, and FDA Radar's copy of it — carries neither one.

Read the post →
Practice·Aug 16, 2026 · 4 min

Classification Change is not one event

FDA Radar's Federal Register categorizer sorts a device notice into Classification Change the moment its title says "classification of" or "reclassification" — the same label for a De Novo order creating a product code from nothing and a reclassification order moving an existing one to a different class.

Read the post →
Method·Aug 16, 2026 · 4 min

Publication date is not effective date

Every Federal Register document carries two dates that answer different questions — when it was published, and when it takes effect. FDA Radar's own record keeps both, as two different fields. For a device rule that phases in by class over several years, neither date is the one a specific device actually has to answer to.

Read the post →
Method·Aug 15, 2026 · 3 min

One event. Many recall numbers.

FDA defines a recall event as a firm's action that can cover more than one product. openFDA keeps a field for that grouping, and FDA Radar's own record keeps it too — but every count downstream is keyed to recall_number, one row per product line, not per event.

Read the post →
Practice·Aug 15, 2026 · 3 min

Quantity is a sentence, not a number

openFDA's device recall record carries a field FDA itself defines as the amount of defective product subject to recall. The field's type is text, written by whoever filed the report — and a quarter's worth of them cannot be added together to get a total.

Read the post →
Practice·Aug 14, 2026 · 3 min

How a firm told you is a field too

openFDA's recall data carries a field for how a firm first notified the public — telephone, letter, press release — kept separate from the field that says why it recalled at all. FDA Radar's own ingestion reads it and doesn't carry it into the record yet.

Read the post →
Method·Aug 14, 2026 · 3 min

A RIN outlives the document number

Every Federal Register notice carries three identifiers doing three different jobs — a document number, a docket ID, and a Regulation Identifier Number. Only one of them is built to survive past the document that first used it, and FDA's own correction to its 2024 device-quality rule shows exactly which.

Read the post →
Method·Aug 13, 2026 · 3 min

The docket ID is not the comments

FDA Radar's Federal Register ingestion captures a docket_ids field on every device-relevant notice. A docket ID is a pointer, not a comment count — what a rulemaking or draft guidance actually drew lives on regulations.gov, a separate system the notice only points to.

Read the post →
Practice·Aug 13, 2026 · 3 min

The 510(k) record doesn't name the predicate

Every 510(k) has to name its predicate device under 21 CFR 807.92. The openFDA clearance record FDA Radar ingests daily — like the public database it comes from — carries no field for it. The name is required. It just isn't queryable.

Read the post →
Practice·Aug 12, 2026 · 3 min

The 510(k) FDA didn't review first

The public 510(k) record can carry a field marking that an accredited third party, not FDA staff, conducted the initial review. FDA still signs the clearance. The field is real, FDA Radar already ingests it, and the record everyone reads past does not point it out.

Read the post →
Method·Aug 12, 2026 · 3 min

The letter FDA doesn't have to post

FDA runs two advisory-action letters, not one. Warning Letters post to a structured database on a rule-bound cadence. Untitled Letters, issued for the same kinds of violations minus the enforcement threat, post only when FDA decides three named criteria are met — and the database a monitoring pipeline actually watches was never built to carry them.

Read the post →
Practice·Aug 11, 2026 · 3 min

Cleared doesn't say which track it took

FDA runs three separate 510(k) programs — Traditional, Special, and Abbreviated — each reviewing a different amount of new evidence. The public clearance record, and FDA Radar's own, keep the outcome and drop the path.

Read the post →
Method·Aug 11, 2026 · 3 min

The clock stops. The calendar doesn't.

FDA's 90-day review goal is measured in FDA Days — a count that excludes both an acceptance-review hold and an Additional Information hold. The public 510(k) record keeps only the two calendar dates that started and ended the clock, not the holds in between.

Read the post →
Method·Aug 10, 2026 · 3 min

Why a guessed match never reaches you

Every match FDA Radar computes between an FDA document and your product carries a provenance field: source, human, or model. The database lets only two of the three ever fire an alert — a rule poured into the schema, not remembered by hand.

Read the post →
Practice·Aug 10, 2026 · 3 min

The standards list never names a device

FDA revises its list of recognized consensus standards roughly twice a year, in a single Federal Register notice indexed by standard number and edition — never by product code. No device-level watch, including this one, turns that into a personal alert.

Read the post →
Method·Aug 9, 2026 · 3 min

The guidance isn't in the Federal Register.

FDA's own rule requires two separate acts before a guidance document counts as issued — a notice in the Federal Register, and the guidance itself posted separately on fda.gov. The record that's easy to watch is the first act, not the second.

Read the post →
Practice·Aug 9, 2026 · 3 min

Filtered. Not discarded.

FDA Radar ingests every device recall and every 510(k) clearance. It pushes almost none of them to a human — the boundary between the corpus and the review queue is a single classification field, not a judgment call.

Read the post →
Method·Aug 8, 2026 · 3 min

There are two recall databases, not one

openFDA publishes two separate device recall datasets, sourced and shaped differently — one carries the K-number and PMA number a recall traces back to, the other carries the classification that says how serious it was. No record in either has both.

Read the post →
Practice·Aug 8, 2026 · 3 min

Breakthrough status is confidential by law

Federal law lets FDA withhold whether a sponsor even requested Breakthrough Device designation, let alone whether it was granted — right up until the device wins marketing authorization. A public database has nothing to show in the meantime, by statute, not by omission.

Read the post →
Practice·Aug 7, 2026 · 3 min

One MAUDE report is not one malfunction

For product codes FDA has enrolled in Voluntary Malfunction Summary Reporting, a single row in MAUDE is a quarter's worth of malfunctions bundled by brand, model, and problem code — not the one event a search result implies.

Read the post →
Method·Aug 7, 2026 · 3 min

A supplement number is not a track

openFDA's PMA record logs a supplement number and a free-text reason for every change to an approved device — not which of FDA's four review tracks, from a same-day call to a full panel, that change actually took.

Read the post →
Method·Aug 6, 2026 · 4 min

Cleared is not approved

A 510(k) clearance and a PMA approval are findings from two different review standards, on two different forms of words — and FDA's own misbranding rule names the specific act of calling the first one the second.

Read the post →
Practice·Aug 6, 2026 · 4 min

Nationwide doesn't mean your building

openFDA's recall records carry a distribution_pattern field describing where a firm first shipped a device — and the field's own definition says plainly that what happens after that first shipment isn't in it.

Read the post →
Practice·Aug 5, 2026 · 3 min

Not every correction reaches FDA

21 CFR 806.10 requires a device correction or removal to reach FDA within ten working days — but only when it reduces a risk to health. A correction that only improves quality is exempt by the agency's own rule, recorded internally, and never becomes a public recall.

Read the post →
Method·Aug 5, 2026 · 3 min

Registered. Not reviewed.

FDA's device registration and listing record can carry a 510(k) or PMA number — typed in by the company itself, once a year, on a filing FDA's own rule says does not double as a review of that number.

Read the post →
Method·Aug 4, 2026 · 4 min

Contact ODE is not a pathway

FDA's own field reference for device classification documents three submission pathways and a fourth answer that isn't one — "Contact ODE," the database's own way of saying it doesn't know. A product-code lookup that hits it isn't broken; it has reached a category FDA itself left open.

Read the post →
Practice·Aug 4, 2026 · 3 min

Almost no recall is FDA-mandated

openFDA marks every device recall Voluntary: Firm Initiated, Voluntary: FDA Requested, or the rare FDA Mandated — three different stories about who decided, usually read as one word. The mandated label requires FDA to make a formal, written finding first, which is exactly why it almost never appears.

Read the post →
Method·Aug 3, 2026 · 3 min

FDA's device names read like a card catalog

FDA's device classification database names each device category noun-first — Catheter, Continuous Flush, not continuous flush catheter — a filing convention nowhere explained and everywhere load-bearing the moment a search is typed the way a person would say it.

Read the post →
Practice·Aug 3, 2026 · 3 min

openFDA is free. It is not unlimited.

openFDA is free and keyless, and FDA publishes exactly how far that goes: 1,000 requests a day, 240 a minute, before the ceiling ever shows up in a script's output. A free key raises it to 120,000 — the difference between a monitor that keeps working and one that goes quiet the week a recall needs reading.

Read the post →
Method·Aug 2, 2026 · 3 min

GUDID stores the model. Not the box.

A Unique Device Identifier splits by rule into a fixed device identifier and a variable production identifier. FDA's public device database keeps only the first — the fact capable of naming which specific unit shipped was never asked to be part of the record.

Read the post →
Practice·Aug 2, 2026 · 3 min

Detention doesn't need a warning letter

FDA's device import alerts authorize border detention from a documented inspection finding alone, no warning letter has to exist before or after. The list that runs on is a set of static web pages with no API behind them at all.

Read the post →
Method·Aug 1, 2026 · 3 min

A Class I device, a Class I recall

FDA assigns a device its regulatory class once, as a permanent attribute of the device type. FDA assigns a recall its class per event, as a snapshot of one defect's severity. Both scales use the identical three labels, and a device's own class predicts nothing about the class of a recall against it.

Read the post →
Practice·Aug 1, 2026 · 4 min

21 CFR 820 didn't move. Its contents did.

FDA's Quality Management System Regulation, effective February 2, 2026, left the address unchanged and emptied most of what used to live there. A citation to Part 820 written last year and one written this year can point at the identical number and mean two different things.

Read the post →
Method·Jul 31, 2026 · 4 min

The product code that doesn't exist yet

FDA's classification database has no row for a device type until a De Novo request is granted — the product code, the regulation number, even the classification itself are created by the grant order, not looked up before it. A watch keyed to "your product code" cannot fire on a category that is still being invented.

Read the post →
Practice·Jul 31, 2026 · 4 min

The classification posts. The 483 behind it doesn't.

FDA scores every closed device inspection No Action, Voluntary Action, or Official Action Indicated — a free, weekly, non-FOIA database. The Form 483 narrative behind that score is a different document, gated behind a records request FDA answers on its own schedule. A firm watch that conflates the two is missing the earlier, public half of the signal.

Read the post →
Method·Jul 30, 2026 · 3 min

A recall has three dates, and the pipeline had to pick one

A device recall record carries three separate date fields — when the firm told the public, when FDA finished judging the hazard, and when the report posted — and they can be months apart. Read the wrong one as "when this happened" and a portfolio timeline is quietly off by however long the classification took.

Read the post →
Practice·Jul 30, 2026 · 3 min

The count is real. The rate is not.

A MAUDE search returns a number that looks like a fact you can compare — this many adverse event reports, this device, this window. FDA's own documentation says that number cannot become a rate, because nothing in the reporting system tells you how often the device was used.

Read the post →
Method·Jul 29, 2026 · 4 min

The firm on the record is not the firm on the box

A device label is allowed by rule to carry a brand's name while a different company actually made the thing. FDA's enforcement records follow the maker, not the brand — so a firm watch keyed only to your own name has a blind spot built into it by design.

Read the post →
Practice·Jul 29, 2026 · 4 min

Low confidence is an instruction, not a hedge

Our classifier can say it doesn't know. That single field changes what happens to an item next — and it's a habit worth borrowing the day you're reading a draft this product, or anyone else's AI, handed you.

Read the post →
Method·Jul 28, 2026 · 3 min

The same company, spelled three ways

FDA's own filings can't agree on how to spell a firm's name — a comma here, no comma there, Inc. versus INC. FDA Radar's matcher corrects for that, then stops on purpose, because the one join in this product that names a company is the one it refuses to make clever.

Read the post →
Practice·Jul 28, 2026 · 3 min

A device that was never a medical device

FDA published a Federal Register notice with the word Device, capitalized, in its own title. It has nothing to do with CDRH. Word-boundary matching fixes accidental substrings — it does nothing for a real word doing its ordinary job in the wrong industry.

Read the post →
Method·Jul 27, 2026 · 3 min

FDA's warning letters, addressed by column number

The data behind fda.gov's warning-letter search has no field names — company, date, and subject are positions in an array, and nothing FDA publishes promises they'll stay in order. What a monitoring pipeline does about a contract nobody signed.

Read the post →
Practice·Jul 27, 2026 · 3 min

Your regulation has a decimal. FDA's watch doesn't.

A device classification regulation reads down to the section — 21 CFR 890.5740, powered heating pad. The Federal Register indexes rule changes only down to the part — 21 CFR 890. Watch at the wrong granularity and a real change to your regulation never matches.

Read the post →
Method·Jul 26, 2026 · 3 min

Three letters is too short to search for

FDA Radar's own Federal Register filter once matched 'udi' inside 'studies' and nearly filed a drug guidance about antibodies as a device document. The failure mode behind it, and why the fix is a stricter filter, not a smarter one.

Read the post →
Practice·Jul 26, 2026 · 3 min

Most warning letters never close

FDA issues a close-out letter only after it has verified — usually by reinspection — that a firm's corrective actions actually worked. Violations that can't be corrected, and firms that no longer exist, get none. What the absence of one does and doesn't tell you.

Read the post →
Method·Jul 25, 2026 · 3 min

The guidance that skipped the draft

FDA's Good Guidance Practices rule lets a significant guidance document skip the draft entirely — issued final, in effect, on the day it posts. The record's title doesn't tell you which path a document took to get there.

Read the post →
Practice·Jul 25, 2026 · 3 min

Ongoing, as of when?

openFDA's recall status field reads like a live indicator — Ongoing, Completed, Terminated. FDA's own documentation says it isn't one. What gets corrected on a recall record after it posts, what doesn't, and the practice for telling the difference.

Read the post →
Method·Jul 24, 2026 · 3 min

The least interesting fact in a 510(k)

Cleared or not is a one-bit answer. The record underneath encodes the pathway, the conditions, and a disclosure choice the applicant locked in before submitting — and those are the fields worth reading across a product code.

Read the post →
Practice·Jul 24, 2026 · 3 min

No recall arrives with a product code attached

openFDA's recall data has a field for it. Across the last 300 device recalls we ingested, that field came back empty every time. The practice for finding your portfolio in a feed that isn't indexed by the thing you'd search for.

Read the post →
Practice·Jul 23, 2026 · 4 min

The deadline that doesn't email you

A Federal Register comment period can run ten days or nine months, and nothing on the page warns you which. The field that has the real answer, and the habit that checks it every week.

Read the post →
Method·Jul 23, 2026 · 4 min

The warning letter has two dates

One starts your compliance clock. The other is what the public sees, and it can lag by weeks. Confuse them and the subject line's real signal goes unread too.

Read the post →
Method·Jul 23, 2026 · 8 min

How to monitor FDA without drowning

Four official sources cover the material risk to a device portfolio. The hard part is not finding them — it is surviving their volume.

Read the post →
Practice·Jul 23, 2026 · 9 min

Your predicate was recalled. Now what?

A recall on the device your clearance compares against is not automatically your problem — and is never automatically not your problem. The assessment that tells you which.

Read the post →
Method·Jul 23, 2026 · 6 min

Why no item reaches you without a human

The model drafts. It is structurally unable to publish. The database constraint that makes our review rule a fact rather than a promise — and why this audience should demand one.

Read the post →

The posts explain the method. The product runs it.

Enter your products once; FDA Radar watches the same four sources daily and joins every item against your portfolio — reviewed by a person before it reaches you.

Open portfolioHow it works