FDA Recalls - Drug, Device and Food Enforcement

Every drug, medical device and food recall the FDA has published, as clean rows.

When a product is pulled from the US market the FDA records why, who made it, what the hazard class is and how far it was distributed. This Actor turns that enforcement database into a table you can filter, monitor and join against your own catalogue.

No API key needed. It works out of the box.

What you get

Field	Description
classification / classificationLevel	Class I, II or III, plus a number so you can sort by severity
recallingFirm	The company that issued the recall
productDescription	What was recalled, including size and packaging
reasonForRecall	Why, in the FDA's own words
status / isOngoing	Whether the recall is still open
recallInitiatedOn / terminatedOn / daysOpen	Timeline of the recall
distributionPattern	How widely it shipped, from one state to nationwide
voluntaryOrMandated	Whether the FDA had to force it
city / state / country / address	Where the firm is
codeInfo	Lot numbers and expiry codes for matching your own stock
brandName / genericName / manufacturerName / substanceName / productNdc	Drug identifiers, where the FDA has linked them

Hazard classes, in plain terms

Class	What it means
Class I	Reasonable probability of serious injury or death
Class II	Temporary or medically reversible harm
Class III	Unlikely to cause harm, but violates FDA rules

classificationLevel is 1, 2 or 3, so sorting ascending puts the dangerous recalls at the top.

Example input

Every ongoing Class I recall this year, across all three databases:

{
  "categories": ["drug", "device", "food"],
  "classifications": ["Class I"],
  "onlyOngoing": true,
  "startDate": "2026",
  "endDate": "2026"
}

Watch one company:

{
  "categories": ["device"],
  "firmName": "Abbott",
  "startDate": "2020"
}

Watch one hazard:

{
  "categories": ["food"],
  "searchTerm": "undeclared peanut",
  "startDate": "2024"
}

Example output

{
  "category": "device",
  "recallNumber": "Z-2372-2023",
  "classification": "Class I",
  "classificationLevel": 1,
  "status": "Ongoing",
  "isOngoing": true,
  "recallingFirm": "Medtronic Inc.",
  "productDescription": "MAHURKAR 12 Fr High Pressure Triple Lumen Acute Dialysis Catheter",
  "reasonForRecall": "Catheter center lumen was found to have an occlusion in the tip",
  "recallInitiatedOn": "2023-05-18",
  "distributionPattern": "Nationwide",
  "voluntaryOrMandated": "Voluntary: Firm initiated",
  "state": "MN",
  "country": "United States"
}

What people use it for

• Supplier monitoring - match recallingFirm and codeInfo against what you actually stock
• Competitive intelligence - recall history by manufacturer over time
• Insurance and legal - Class I history for a product line or a company
• Newsrooms - a filtered feed of serious recalls as they are classified
• Health apps - warn users about drugs and devices they have on file

Notes

• The FDA answers an empty result set with a 404. That is treated as "no matches", not as an error.
• openFDA will not page beyond 25,000 results in a single query, so wide pulls are split year by year automatically. If one year still exceeds that in a single category, the Actor says so rather than silently truncating.
• Drug identifiers such as brandName and productNdc come from the FDA's own linking, which is present on roughly a fifth of drug recalls and empty on the rest. They are left blank rather than guessed at.
• The optional API key is only worth setting if you are pulling many years at once. Without one, requests share a per-address daily allowance.
• Source: U.S. Food and Drug Administration, via openFDA. The FDA does not endorse any product or company, and this data must not be used to make clinical decisions.

Pricing

Pay per result. You are charged only for recalls actually returned.