MAUDE is public, searchable, and routinely over-trusted. A report in it is what someone said happened, not a finding about what happened.
Who files, and on what clock
The Manufacturer and User Facility Device Experience database — MAUDE — is FDA’s public window into 21 CFR Part 803, the medical device reporting rule. Two different obligations feed it. Manufacturers generally have 30 calendar days from becoming aware of a reportable death, serious injury, or malfunction to file, under 803.53. Device user facilities — hospitals and similar sites — work on a tighter clock under 803.30: generally 10 work days to report a death to both FDA and the manufacturer, or a serious injury to the manufacturer. Knowing which kind of reporter filed a given entry tells you something before you’ve read a word of the narrative — a user-facility report is a clinician’s account of what they saw; a manufacturer report is usually a complaint investigation the company is required to disclose whether or not it agrees with the complainant’s account.
The coded fields are an index, not a diagnosis
Every report carries standardized codes for the device problem and the patient problem, picked from a controlled list by whoever completed the form — often not the clinician who witnessed the event, and rarely someone with the device’s engineering in front of them. Treat those codes as a rough index for searching, not as an adjudicated cause. The narrative field, usually labeled something like Event Description or Additional Manufacturer Narrative, is where the specific, useful detail tends to live: what the user was doing, what the device did or didn’t do, and what the manufacturer’s investigation found when it looked at the returned unit. Some entries are supplemental reports that amend or update an earlier one — check the report number and any reference to a prior submission before assuming two entries describe two separate events.
- What was reported to have happened, in the reporter’s own account.
- Who reported it — manufacturer or user facility — and on which clock.
- Whether it’s an original report or a supplement to an earlier one.
- What it doesn’t tell you: confirmed cause, or how many devices didn’t have a problem.
A count is not a rate
MAUDE has no reliable denominator. There’s no field for how many units of a device are actually in use, so a search that returns a rising number of reports can’t distinguish between a genuinely worsening problem, a device that simply sold more units this year, and one hospital that filed a batch of reports after a single training gap. Underreporting cuts the other way for the same reason — a quiet search result is evidence of a quiet database, not proof of a clean safety record. Treat a report count as a lead worth investigating further, the same way a regulatory intelligence habit treats any single signal: something to track over time against other information, not a number that resolves the question on its own.
Two habits that don’t hold up
Citing a report as a confirmed cause
An MDR records an allegation the rule required someone to disclose. Say what the report claims, not that the device caused it, unless the narrative itself documents a confirmed investigation finding.
Reading report volume as incidence
Without a denominator, a count of reports can’t be converted into a rate. Compare device volume and reporting source before treating a spike as a trend.
Sources & further reading
- eCFR — 21 CFR 803.53, Manufacturer Reports ecfr.gov
- eCFR — 21 CFR 803.30, Individual Adverse Event Reports (Device User Facilities) ecfr.gov
- Regulatory Academy — How to Read an FDA Recall regulatoryacademy.com
- Regulatory Academy — How to Build a Regulatory Intelligence Habit regulatoryacademy.com
This essay is provided for general educational purposes and reflects the regulatory landscape as of its publication date. It is not legal, regulatory, or career advice.
