Faers Adverse Event Quarterly Data: Dataset Structure and Field Coverage

Every figure below comes straight from the cataloged source records. Note what the refresh column implies: FAERS is the only one of the five that cannot be mirrored wholesale on a cron of hours - the other four archives together weigh less than 2 GB.

Report metadata opens the record: transmissiondate, receivedate, receiptdate, reporttype, and seriousness flags for death, hospitalization, life-threatening events, disability and congenital anomaly. primarysource.reportercountry records where the reporter sat - the bundled example reads CANADA - and companynumb preserves the manufacturer's own case number.

The patient block carries demographics and outcomes: patientonsetage interpreted against its unit code (801=years, 800=decades), patientsex (1=male, 2=female), death outcome fields, and a reaction array of reactionmeddrapt MedDRA preferred terms such as OVERDOSE. Every FAERS adverse event report therefore stores reactions as structured MedDRA-coded reaction values rather than free text, which is what makes cross-quarter counting possible without a terminology mapping project.

The drug array adds the analytical spine: medicinalproduct as entered by the reporter (example DURAGESIC-100), drugcharacterization (1=suspect, 2=concomitant, 3=interacting), drugindication, drugadministrationroute (041=transdermal) and drugauthorizationnumb. openFDA publishes the searchable-field list separately from the bulk schema, so confirm a field is queryable through the API before writing a filter against it. In the sample record 5801206-7 - a fatal serious case received 20080707 - those fields read DURAGESIC-100, drugcharacterization 1, route 041, indication DRUG ABUSE, with reactions coded DRUG ADMINISTRATION ERROR and OVERDOSE.