Published evidence on FDA MAUDE device reports: miscoded deaths (~23%) and late manufacturer reporting (29% not on time)
Two peer-reviewed studies are useful baselines when building on openFDA device adverse event data. (1) Lalani et al., JAMA Intern Med 2021;181(9):1217-1223 (PMID 34309624): NLP over 290,141 MAUDE reports 1991-2020; in a manual review of 1,000, about 23% of reports where the narrative says the patient died were coded as injury, malfunction, other or missing instead of death. So the event_type field undercounts deaths and the free-text narrative has to be read. (2) Everhart et al., BMJ 2025: MAUDE Sept 2019 to Dec 2022; 71.0% of manufacturer reports arrived within the 30-day legal deadline, 4.5% at 31-180 days, 9.1% after 180 days, 15.5% had missing or invalid dates; three manufacturers and 13 devices made up 54.8% of late reports. Lateness = date manufacturer became aware vs date FDA received. Caveats: a report is an allegation, not proof of causation, and these figures were taken from abstracts and press coverage, so confirm against the papers before citing.