What 194,000+ adverse-event reports say about surgical staplers

HFE teams mostly look forward — formative studies, validation, design controls. But the FDA's MAUDE database is a backward-looking goldmine: real use errors, in real ORs, in the reporters' own words. This page mines the public openFDA device-event API for surgical staplers (product codes GAG & GDW) and shows how post-market signals can seed formative-study hypotheses.

193,938

Total reports in MAUDE

153,477

Malfunctions

38,283

Injuries

1,360

Deaths

Live from openFDA · dataset last updated 2026-06-02

Reports per year — and the hidden-database story

For years, stapler manufacturers could route adverse events through FDA's “Alternative Summary Reporting” program — tens of thousands of malfunction reports that never appeared in public MAUDE. After investigative reporting surfaced the practice, FDA ended ASR in mid-2019 and the hidden reports flooded into the public record. The lesson for anyone reading this data: the shape of a reporting curve reflects policy as much as risk.

A use-error lens on the narratives

Event narratives often encode perception, cognition, or action failures — the raw material of use-related risk analysis. These phrase counts are a deliberately simple heuristic for surfacing candidate reports to read, not a validated classifier:

Failed to fire

1,029

reports mentioning “failed to fire”

Often entangled with loading, positioning, or tissue-thickness selection

Difficult to remove

968

reports mentioning “difficult to remove”

Post-fire release problems often involve technique interaction

Misfire

836

reports mentioning “misfire”

Frequently involves firing sequence or reload handling

Inadvertent action

490

reports mentioning “inadvertently”

Marker for unintended activation or release — action-stage errors

Labeled 'user error'

178

reports mentioning “user error”

How reporters themselves attribute the event

Wrong size

reports mentioning “wrong size”

Cartridge/tissue mismatch is a classic perception-stage use error

The latest reports, in the reporters' own words

The most recent stapler narratives in MAUDE, with use-error phrases highlighted. Reading raw narratives is where the method earns its keep — counts point you somewhere; the words tell you why.

InjuryReceived 2026-05-29SUREFORM · INTUITIVE SURGICAL, INC

IT WAS REPORTED THAT DURING A DA VINCI-ASSISTED TRANSTHORACIC ESOPHAGECTOMY CHEST ANASTOMOSIS SURGICAL PROCEDURE, A SUREFORM 60 STAPLER WITH A BLUE RELOAD MISFIRED AND HAD A MISALIGNED STAPLE LINE. NO FRAGMENT FELL INTO THE PATIENT. IT IS UNKNOWN IF ANY INTERVENTION WAS REQUIRED TO ADDRESS THE MISALIGNED STAPLE LINE. THE PROCEDURE WAS COMPLETED WITH A DELAY OF LESS THAN 15 MINUTES. INTUITIVE SURGICAL, INC. (ISI) MADE MULTIPLE FOLLOW-UP ATTEMPTS TO OBTAIN ADDITIONAL INFORMATION. HOWEVER, NO FURTHER DETAILS HAVE BEEN RECEIVED AS OF THE DATE OF THIS REPORT.

InjuryReceived 2026-05-29PROXIMATE · ETHICON ENDO-SURGERY, LLC.

IT WAS REPORTED THAT FOLLOWING AN ILEOCECECTOMY PROCEDURE, THE SURGEON CLOSED THE COMMON ENTEROTOMY USING A TX STAPLER. ON POSTOPERATIVE DAY 1, THE PATIENT DEVELOPED A LEAK AND WAS TAKEN BACK TO SURGERY. UPON RE-EXPLORATION, IT WAS OBSERVED THAT THE STAPLED SITE APPEARED COMPLETELY OPEN, MEASURING APPROXIMATELY 3¿4 CM IN LENGTH. THE SURGEON SUSPECTED A POTENTIAL FAILURE OF THE TX STAPLER; HOWEVER, NO ISSUES WERE NOTED WITH THE DEVICE DURING THE INITIAL PROCEDURE.

Why this matters for HFE teams

1 · Signal

Trend breaks and phrase clusters point to where users struggle — before your own study budget is spent.

2 · Hypothesis

Each recurring narrative pattern (“wrong size,” “failed to fire”) becomes a candidate use error for task analysis and PCA classification.

3 · Study design

Formative scenarios and IFU probes get grounded in documented field failures instead of conference-room guesses.

Limitations — read before drawing conclusions

MAUDE has no denominator. Report counts can't be turned into rates — procedure volumes aren't in the data, so more reports ≠ more dangerous.
Reporting is biased and incomplete. Underreporting is well documented; media attention, litigation, and policy changes (see the ASR story above) all move the curve.
Keyword matching is a heuristic. Phrase counts surface candidates for human reading. A validated use-error classification needs trained reviewers and a coding scheme (e.g., PCA taxonomy) with reliability checks.
Duplicates and follow-ups exist. The same event can generate multiple MDRs; no deduplication is attempted here.
Narratives are secondhand. Most are written by manufacturers from user communications, with their own framing incentives.