Skip to content
ppharmaranks
Menu

How long a drug recall stays invisible

pharmaranks measured the gap between a manufacturer starting a US drug recall and that recall appearing in the FDA's public Enforcement database, across 1,687 recalls from Jan 2, 2024 to Jun 29, 2026. The median gap is 28 days; 43% take longer than a month.

What this measures — and what it doesn’t.The FDA’s Enforcement database is the systematic public record of every drug recall — it is what recall trackers (including ours) and most news coverage rely on. The gap here is the time between the manufacturer’s recall start date and the recall’s appearance in that database, both dates the FDA’s own. It is not the time the public had no warning at all: for the most dangerous recalls, the firm or the FDA usually issues a separate press announcement much sooner. During the gap, the recall exists — but not yet in the database most tools watch.

28
days — median gap
43%
take longer than 30 days
37
days median for Class I — the longest

The most serious recalls take the longest

Median days from recall start to the public Enforcement report, by FDA class. Class I is the FDA’s label for a defect that could cause serious harm or death — and it has the longest median gap. A likely reason is that classification itself takes time (the FDA assigns the class before the entry appears), but the effect for a reader is the same: the worst recalls surface in the database last.

Class I37median days · 125 recalls
Class II28median days · 1,401 recalls
Class III28median days · 160 recalls

Is the gap improving?

Median gap in days, by the year the recall reached the database.

202428median days · 502 recalls
202530median days · 765 recalls
202626median days · 420 recalls

The longest gaps on record here

The recalls that took the longest to reach the public database — several are Class I.

Methodology & sources

Built from the openFDA drug Enforcement (recall) database. For each recall we take the FDA’s own two dates — recall_initiation_date(“the date that the firm first began notifying the public or their consignees of the recall”) and report_date(the date the recall entered the FDA’s weekly Enforcement report) — and compute the difference in days. Recalls missing either date are excluded (1,687 of the recalls we track qualify). Medians are used because a few extreme gaps would distort averages. Counts are recall events, not units. The full underlying data is downloadable as CSV. This is public regulatory data, not medical advice.

See also the FDA Drug Recall Report, the contamination report and the latest recalls tracker.