FDA Recalls & 510(k) Clearances Scraper (openFDA API) avatar

FDA Recalls & 510(k) Clearances Scraper (openFDA API)

Pricing

from $2.00 / 1,000 fda records

Go to Apify Store
FDA Recalls & 510(k) Clearances Scraper (openFDA API)

FDA Recalls & 510(k) Clearances Scraper (openFDA API)

FDA medical device, drug and food recalls (enforcement reports) and 510(k) device clearances from the public openFDA API. Filter by class, status, keyword, firm, state and date. Newest first. Regulatory monitoring and medtech competitive intelligence. No API key needed.

Pricing

from $2.00 / 1,000 fda records

Rating

0.0

(0)

Developer

Groundplane Labs

Groundplane Labs

Maintained by Community

Actor stats

0

Bookmarked

2

Total users

1

Monthly active users

12 hours ago

Last modified

Categories

Features

Limited permissions

Share

FDA Recalls (Device, Drug, Food) and 510(k) Clearances Scraper, openFDA API

Extract official U.S. Food and Drug Administration (FDA) Enforcement Recall Reports (device_enforcement, drug_enforcement, food_enforcement) and 510(k) Premarket Notification Clearances (device_510k) from the official openFDA JSON APIs (api.fda.gov).

Supports deep cursor pagination via the Link: <...>; rel="next" (search_after) response header to bypass openFDA's 25,000 skip cap, normalizes FDA dates to ISO YYYY-MM-DD, and works out of the box without an API key (or accepts an optional user-supplied openFdaApiKey for a higher quota).

Every run ends with a status starting Complete: or PARTIAL: that states the rows scanned, matched, and emitted. If the source shape changes, dates drift, or rows are skipped, the run says so in the status instead of silently shipping wrong data.

Who it is for

  • Regulatory affairs, quality assurance (QA/RA) and compliance teams monitoring Class I, Class II, Class III, and Not Yet Classified recalls across medical devices, pharmaceuticals, and food products
  • MedTech commercial, competitive intelligence and BD teams tracking newly cleared 510(k) devices (kNumber, deviceName, applicant, contact, productCode, regulationNumber)
  • Supply chain, hospital procurement, legal and risk management analysts auditing vendor recall histories and UDI/lot code notices (codeInfo)

Input

By default, the Actor pulls up to 200 Class I and Class II medical device enforcement recalls reported in the past 90 days.

Canary input (labeled canary)

Use this exact input to verify the Actor is healthy. Expected result: the run succeeds with up to 5 rows, all from device_enforcement, each with classification Class I or Class II, status Ongoing, a state of CA or MN, and a product description or recall reason mentioning catheter. Because maxItems is 5, a healthy canary normally ends PARTIAL: with the reason maxItems limit reached (5) and no Issues CTA: that is the expected user-limit ending, not an error.

{
"dataset": "device_enforcement",
"classifications": ["Class I", "Class II"],
"statuses": ["Ongoing"],
"keywords": ["catheter"],
"states": ["CA", "MN"],
"dateFrom": "2026-08-17",
"dateTo": "2026-11-15",
"maxItems": 5
}

Example input

{
"dataset": "device_enforcement",
"classifications": ["Class I", "Class II"],
"statuses": ["Ongoing"],
"keywords": ["catheter", "infusion pump"],
"firmName": "",
"states": ["CA", "MN"],
"dateFrom": "2026-07-25",
"dateTo": "2026-10-23",
"maxItems": 200
}
FieldTypeDefaultDescription
datasetstring"device_enforcement"device_enforcement, drug_enforcement, food_enforcement, or device_510k.
classificationsarray["Class I", "Class II"]Hazard classifications (Class I, Class II, Class III, Not Yet Classified). Applies to enforcement datasets only. For device_510k this input is ignored, and the run log says so.
statusesarray[] (all)Recall statuses (Ongoing, Completed, Terminated, Pending). Applies to enforcement datasets only. For device_510k this input is ignored, and the run log says so. Note: Pending currently has zero live records in the source, so a Pending-only run returns a healthy zero match.
keywordsarray of strings[] (all)Keywords use OR semantics: a record matching any keyword is included (the list broadens, it does not narrow). Matched in product_description or reason_for_recall (enforcement) or device_name (device_510k).
firmNamestring"" (all)Token or phrase search matched against recalling_firm (enforcement) or applicant (device_510k).
statesarray of strings[] (all)2-letter US state codes (e.g. ["CA", "MN"]).
dateFromstring (YYYY-MM-DD)today - 90 daysStart of report_date (enforcement; minimum 2012-06-20) or decision_date (device_510k).
dateTostring (YYYY-MM-DD)todayEnd of report_date or decision_date.
openFdaApiKeystring (secret)""Optional free openFDA API key supplied by the user (increases that user's quota from 1,000/day to 120,000/day). See the API key note below.
maxItemsinteger200Maximum number of rows to output. Reaching this limit ends the run as PARTIAL: for a user limit, with no Issues CTA.

API key note (disclosed deviation)

openFdaApiKey stays an isSecret input field. This is a disclosed deviation from the usual environment-secret rule, for this reason: the key is a user-supplied optional key for a public API, not an operator secret. Each user may supply their own free openFDA key to raise their own quota; the Actor works without one, no key is hardcoded, and key values are masked in error messages. There is no operator key to protect.

Output

Every output row includes a dataset field identifying the source dataset.

Enforcement datasets (device_enforcement, drug_enforcement, food_enforcement)

FieldTypeDescription
datasetstringdevice_enforcement, drug_enforcement, or food_enforcement
recallNumberstring | nullFDA recall number (e.g. Z-3183-2026)
eventIdstring | nullFDA recall event ID
productTypestring | nullDevices, Drugs, or Food
classificationstring | nullClass I, Class II, Class III, or Not Yet Classified
statusstring | nullOngoing, Completed, Terminated, or Pending
recallingFirmstring | nullRecalling firm name
city / state / countrystring | nullRecalling firm location. Enforcement country is a full country name.
reportDatestring | nullEnforcement report publication date (YYYY-MM-DD)
recallInitiationDatestring | nullRecall initiation date (YYYY-MM-DD)
centerClassificationDatestring | nullFDA classification date (YYYY-MM-DD)
terminationDatestring | nullRecall termination date (YYYY-MM-DD)
voluntaryMandatedstring | nullVoluntary vs. FDA-mandated status
initialFirmNotificationstring | nullInitial firm notification method
distributionPatternstring | nullGeographic distribution pattern
productDescriptionstring | nullRecalled product description
productQuantitystring | nullQuantity distributed
reasonForRecallstring | nullRoot cause / reason for recall
codeInfostring | nullUDI, lot, batch, or serial numbers
openfdaobject | nullCross-referenced openFDA identifiers when present. Only known openFDA keys are kept; unknown keys the source adds later are dropped.

510(k) Clearances (device_510k)

FieldTypeDescription
datasetstringdevice_510k
kNumberstring | null510(k) number (e.g. K262005)
deviceNamestring | nullCleared device trade name
applicantstring | nullApplicant company name
contactstring | nullSubmission contact. Disclosure: this is the official FDA business correspondent name from the federal public 510(k) record. It appears only in device_510k output; the default dataset is enforcement, so it never appears in default output.
city / state / countrystring | nullApplicant location. For 510(k), country is a country code (for example IL) taken from the source country_code field, not a full country name.
decisionDatestring | nullFDA decision date (YYYY-MM-DD)
decisionDescriptionstring | nullDecision description (e.g. Substantially Equivalent)
dateReceivedstring | nullSubmission receipt date (YYYY-MM-DD)
productCodestring | null3-letter FDA product code
advisoryCommitteeDescriptionstring | nullMedical specialty advisory committee
regulationNumberstring | null21 CFR regulation number
clearanceTypestring | nullOne of: Traditional, Special, Abbreviated, Direct, Post-NSE, Dual Track
statementOrSummarystring | nullSummary or Statement. Null when the source record is blank, which is common (about a third of 510(k) records).
fdaUrlstring | nullDirect CDRH 510(k) link

Sample output row (Enforcement)

{
"dataset": "device_enforcement",
"recallNumber": "Z-3183-2026",
"eventId": "99634",
"productType": "Devices",
"classification": "Class II",
"status": "Ongoing",
"recallingFirm": "Stryker Sustainability Solutions",
"city": "Tempe",
"state": "AZ",
"country": "United States",
"reportDate": "2026-09-23",
"recallInitiationDate": "2026-08-19",
"centerClassificationDate": "2026-09-16",
"terminationDate": null,
"voluntaryMandated": "Voluntary: Firm initiated",
"initialFirmNotification": "N/A",
"distributionPattern": "US Nationwide distribution...",
"productDescription": "REF:2B5ST Ethicon Endo-Surgery, Inc. XCEL BLADELESS TROCAR W/OPTIVIEW TECHNOLOGY and SLEEVE 5MM X 75MM",
"productQuantity": "2838 units",
"reasonForRecall": "Reusable access port system may crack or seperate during use.",
"codeInfo": "UDI: (01)00885825017231/ Lots: 16188647 16210212...",
"openfda": null
}

How pricing works

Pay per result. You only pay for rows returned.

Data source and terms

Data is pulled from the U.S. Food and Drug Administration's openFDA API (https://api.fda.gov), updated weekly from FDA Recall Enterprise System (RES) Enforcement Reports and regularly from the CDRH 510(k) database. openFDA data is provided under the openFDA license and Terms of Service.

openFDA disclaimer: Do not rely on openFDA to make decisions regarding medical care. While we make every effort to ensure that data is accurate, you should assume all results are unvalidated. We may limit or otherwise restrict your access to the API in line with our Terms of Service.

Data provided by the U.S. Food and Drug Administration (https://open.fda.gov).

This Actor is an independent tool. It is not affiliated with, endorsed by or an official product of the FDA or openFDA. It reshapes openFDA records (selected fields, renamed keys, blank values returned as null) without changing their content; check the official FDA record before acting on it.

Troubleshooting

  • The run ends PARTIAL: with a stale source reason. The Actor compares the newest source record for the dataset as a whole against a 28 day threshold. If the newest record is older, the status names the newest date and the threshold. Enforcement data updates weekly and 510(k) data less often, so a stale result usually means the source itself has stalled.
  • The run ends Complete: 0 rows matched your filters. That is a healthy zero match, not an error: the source is up and has records, but none match your filters. Widen the date range, add Class III or Not Yet Classified, or clear statuses, keywords, firmName, and states. A Pending-only run is expected to return zero today, because Pending currently has zero live records.
  • The run ends PARTIAL: and mentions maxItems or the max charge limit. You hit a limit you set (or the run charge cap). Raise maxItems or the charge limit to get more rows. These user-limit endings carry no Issues CTA on purpose.
  • The run ends PARTIAL: and mentions skipped rows, unparsed dates, duplicates, or unknown values, with an Issues line. The source data drifted from its documented shape. The counts in the status say how much was affected.
  • The run fails with a missing core column message. A source field the Actor depends on was renamed or removed. That failure is deliberate: the Actor refuses to ship silent nulls.
  • The run fails with a source error after retries. openFDA was unreachable or kept returning server errors through 6 tries. Wait and run again.
  • No API key errors. No key is required. The optional key only raises your own daily quota.

If this keeps happening, please open an issue on this actor's Issues tab and we'll fix it fast.

Limits and FAQ

  • Why must enforcement dateFrom be on or after 2012-06-20? openFDA's enforcement datasets begin coverage on June 20, 2012; earlier dates are rejected up front with a clear validation message (device_510k covers clearances back to 1976).
  • How are large result sets (over 25,000 rows) paginated? openFDA caps skip at 25,000, so this Actor uses limit=1000 pages and follows the Link: <...>; rel="next" (search_after) cursor header, pushing rows in chunks of 100 as they arrive.

Changelog

  • v0.2 (audit fix build, audit tag only): core-column sentinel per dataset; per-row schema validation with skip counting and a 20 percent loud-failure rule; Complete/PARTIAL status contract with scanned, matched, and emitted counts and the Issues CTA only on source-caused partials; healthy zero matches now succeed as Complete: 0 while an empty source fails loudly; retry standard raised to 6 tries with jittered 2 to 32 second backoff, a 120 second budget, and Retry-After honored; 28 day staleness check against the newest source record for the dataset as a whole; strict date parsing with unparsed dates counted, never fabricated; numeric ID coercion and non-object row skipping; hash fallback dedupe so distinct records never collapse, with duplicates counted and disclosed; canonical casing for classification, status, and state, with Not Yet Classified added; openfda subobject restricted to an allowlist of known keys; push chunk reduced to 100 rows; README canary, Troubleshooting, and source disclosures added; TEST_NOTES.md added; categories set to BUSINESS only in source.
  • v0.1.2: Added the openFDA disclaimer and a non-affiliation notice.
  • v0.1.1: Results are sorted newest first (report date for recalls, decision date for 510(k)). Blank text fields are returned as null.
  • v0.1: Initial release supporting device_enforcement, drug_enforcement, food_enforcement, and device_510k with search_after cursor pagination.