DEV Community

dicaria
dicaria

Posted on Edited on Originally published at dicaria.com

What gets a drug a Class I recall? Mostly particles

I counted the Class I entries in the 600 most recent FDA drug recall records. There are 33 of them, and 15 — nearly half — are the same story: something visible in the product that has no business being there.

The class field reads like a severity ranking, and it isn't one. Taken alone, it invites the wrong conclusion — "these are the dangerous drugs." Read against the reason field and the dosage form, a different picture appears: 25 of the 33 are contamination events in otherwise approved products, and the escalation logic runs through the delivery route, not the molecule.

The short version

In a sample of the 600 most recent FDA drug recall entries, 33 (5.5%) are Class I, the tier the FDA defines as a reasonable probability of serious adverse health consequences or death. Of those 33:

  • Foreign particles or visible contaminants: 15 (45.5%)
  • Sterility or microbial contamination: 10 (30.3%)
  • Unapproved or undeclared ingredients: 4 (12.1%)

Particles and sterility together are 25 of 33. The sample is recent, not the full history.

Does Class I mean the most dangerous drug? No

The FDA defines Class I as a situation where "there is a reasonable probability that the use of or exposure to a violative product will cause serious adverse health consequences or death." That is a statement about the recall action, not about the molecule.

A drug can be perfectly safe at its approved dose and still produce a Class I recall, because a single lot was contaminated. Conversely, a drug with a narrow therapeutic window can get a Class II recall for a labelling problem. The class describes what the violation could do to a patient, not how toxic the drug is in general.

This is why the class field and the reason field have to be read together. In the sample, the reason group with the highest Class I count is particles, and the group with the highest Class I share is unapproved ingredients. Neither is "the most dangerous drug" category.

What is the largest Class I group? Particles, 15 of 33

Thirty-three Class I drug recalls by reason: particles 15, sterility 10, unapproved ingredients 4, container or labeling 2, potency 1, impurities 1

Fifteen Class I entries name a particle or foreign object. The materials named in the reason text include glass, stainless steel, iron oxide, fiberglass, cellulose, polyethylene and polystyrene. A few entries name more than one.

Fourteen of the fifteen are injectable or infusion products: saline bags, dextrose, lactated Ringer's, imaging contrast, chemotherapy preparations, and an anticoagulant. One is an oral tablet. This is the same structural point that explains the sterility group. A particle in a tablet is a quality defect. A particle in a solution that enters a vein is a risk of embolism or granuloma, and the FDA treats it as Class I.

One detail worth noting for anyone building a filter: the entries do not say where the particle came from. The reason text names the material, not the source. Equipment wear, packaging, and the fill line all appear in individual case reports, but the field does not carry that. So the sample can tell you the material and the dosage form, and it cannot tell you the cause.

Why is sterility only second, with 10 of 33?

Ten Class I entries are sterility or microbial contamination. That is fewer than the particle group, and it is a smaller share of its own category: 10 of 159 sterility entries reached Class I (6.3%), while 15 of 94 particle entries did (16.0%).

The reason is the same dosage-form logic. Sterility problems are common in the whole sample because sterile injectables are a large part of the recent recall mix. But "lack of assurance of sterility" is a documentation failure: the firm cannot prove sterility, which is different from finding a microbe. Only one of the ten Class I entries is a bare assurance failure with nothing else named. The others name what was found: an identified organism, an out-of-specification endotoxin result, or a confirmed non-sterile product. Three of the ten are microbial contamination of products that were never meant to be sterile, such as oral liquids, where a confirmed organism is a direct hazard rather than a process gap.

That distinction is the most useful one in this article. "Assurance" failures are a process and paperwork problem at scale. Confirmed contamination is rarer and more serious.

What are the 4 entries about undeclared ingredients?

Four Class I entries fall under unapproved or undeclared ingredients. These are products marketed without an approved application, where FDA analysis found undeclared active ingredients. Three of the four name sildenafil, tadalafil or meloxicam in a product that did not declare them.

This group is small, 13 entries in the whole sample, but its Class I share is the highest at 30.8%. It is also the one group where the risk does not come from a manufacturing defect at all. The product was never evaluated for the use it was sold for. Thirteen entries is too few to build a rule on, but it is worth knowing that the class is not only a manufacturing story.

What are the last 3 entries? Label mix-up and potency

Two Class I entries are labelling problems. One is a mix-up where a product was labelled as one drug and contained another. A label mix-up is the clearest case of a recall that can put the wrong medicine in a patient's hand, which is why it reaches Class I with no contamination at all.

One Class I entry is potency. A superpotent product, meaning more active ingredient than labelled, is the potency failure that escalates. Subpotent products are far more common in the sample, 41 entries share that exact text, but they mostly land in Class II or III because the patient receives less drug rather than more.

How common is Class I? 5.5% of this sample, not the population

Thirty-three of 600 is 5.5%. It would be a mistake to read that as "5.5% of US drug recalls are Class I", for three reasons.

First, the sample is the most recent 600 entries by report date, not a random draw from the full table of roughly 18,000. Second, recent entries are disproportionately still ongoing: 597 of the 600 have status Ongoing, and a recall can be reclassified while it is open. Third, the sample is heavily weighted to 2026 (592 entries), so it reflects one period's mix.

What the number does support is a comparison inside the sample. Particles produce the most Class I entries, sterility produces more recalls overall, and unapproved ingredients have the highest escalation rate on a small base. Those three statements hold within the 600 rows.

How the sample was pulled

The recall data is openFDA's drug enforcement endpoint. No API key is needed:

# the 600 most recent entries by report date
curl "https://api.fda.gov/drug/enforcement.json?limit=600&sort=report_date:desc"

# or only Class I, newest first
curl "https://api.fda.gov/drug/enforcement.json?search=classification:%22Class%20I%22&limit=100&sort=report_date:desc"
Enter fullscreen mode Exit fullscreen mode

Two caveats if you rerun this. The sample in this article is a snapshot from 5 November 2026, and the window moves as new entries are posted — rerunning the same query today returns a different, more recent 600. And the reason groups in this article (particles, sterility, unapproved ingredients, and so on) are a keyword classification of the free-text reason_for_recall field; they are ours, not an FDA category.

What this data cannot answer

  • How many patients were harmed. The entries have no patient or injury field. Class I is a probability statement, not an outcome count.
  • Whether Class I recalls are increasing. The sample covers about twelve months, so no trend is claimed.
  • Why a lot was contaminated. The reason names the material found, not the failure that let it in.
  • Whether a specific product is safe now. The record describes a recall action and its status, not the current state of the market.
  • The full Class I population. The base is 600 recent entries, not the complete FDA file.

Frequently asked questions

Q. What does Class I recall mean?

A. The FDA defines it as a reasonable probability that use of or exposure to the product will cause serious adverse health consequences or death. It is the most serious of the three recall classes.

Q. What is the most common reason for a Class I drug recall?

A. In this 600-entry sample, foreign particles: 15 of the 33 Class I entries. Sterility is second with 10, and unapproved or undeclared ingredients third with 4.

Q. How common are Class I recalls?

A. In this sample, 33 of 600 entries, or 5.5%. That is a share of recent entries, not of all FDA drug recalls, because the sample is the most recent 600 by report date.

Q. Does a Class I recall mean the drug itself is dangerous?

A. No. It means the violation in that lot could cause serious harm. Most Class I entries in the sample are contamination events in an otherwise approved product.

Data notes

  • Source: FDA Drug Recalls endpoint, accessed through the agentdata SDK on 5 November 2026. Fields used: recalled_on, classification, status, reason_for_recall, product_description, distribution_pattern.
  • Sample: the 600 most recent entries by report date, not a random sample. recalled_on runs from 8 October 2025 to 13 October 2026; 592 entries are dated 2026.
  • Class: taken directly from the classification field; 33 Class I, 520 Class II, 46 Class III, 1 not yet classified.
  • Reason groups: keyword classification of the free-text reason, first match wins. The grouping is ours, not an FDA category.
  • Class definitions: quoted from FDA's recall background and definitions page, read on 4 November 2026.
  • Derived figures: all shares, the 25-of-33 combined figure, and the per-group escalation rates are computed from the 600-row sample.
  • Excluded: firm names are not used anywhere in this article. The firm field exists in the data and was not used for any statement here.

What we're building on this endpoint. I'm with dicaria — a data API for AI agents, currently in beta. The FDA recall endpoint above is one of the sources we serve: same records, but with normalized fields (class, reason group, status, distribution scope), filters on any of them, and every row carrying its official recall number and source link, so an agent can cite what it returns. If that's your stack, the quickstart is two steps — register for a key, then run one MCP command.

No API key is needed to reproduce anything in this post: the raw endpoint is free and public, and the queries above are complete.

Originally published on dicaria.com.

Top comments (0)