Search FAERS for individual case safety reports (ICSRs).

Returns compact triage cards by default: report id, dates, seriousness flags, suspect drugs, reactions and outcomes. Pass full=true for complete records.

Total, serious and fatal case counts for a drug, optionally by event.

All counts are issued concurrently as a single round trip.

Build the 2x2 contingency table and compute ROR, PRR and chi-square.

Any date window or raw_filter is applied to all four marginals including the grand total N, so the table stays internally consistent. Screening criteria are reported individually by name rather than collapsed into a verdict.

Aggregate FAERS reports for a drug by any field (SQL GROUP BY equivalent).

Top MedDRA preferred terms reported with a drug.

Convenience form of faers_count_by_field with the reaction field preset.

Retrieve one complete ICSR by Safety Report ID.

The only tool that returns an unprojected record; a single ICSR averages ~72 KB.

Sex, age, reporter qualification and country breakdown for a drug.

Age is reported two ways: the coded patientagegroup (populated on ~18% of reports) and bands derived from patientonsetage in years (~55%), the same bands the stratifiers use. Each carries its own coverage against the total.

Reaction outcomes and seriousness criteria for a drug, in one concurrent round trip.

Yearly reporting trend for a drug, aggregated from receivedate.

count=receivedate returns a complete date histogram irrespective of limit, so the yearly rollup is exact. Date buckets arrive keyed "time", not "term".

Substances most often appearing on the same reports as an index drug.

Co-REPORTED at any role, not co-suspect: openFDA cannot restrict a count to substances carried as suspect, so the list is dominated by concomitant medication.

Screen a drug against all its reported events, or an event against all its drugs.

ROR, PRR and chi-square for every term with named criteria. The database-wide marginals come from a single cached count call rather than one call per term, so a 200-term screen costs ~13 calls cold and ~10 warm. Any date window or raw_filter is applied to every marginal.

Escape hatch: run an arbitrary Lucene query against the FAERS endpoint.

The query is checked for balanced quotes and parentheses and for the "+AND+" mistake before being sent; the same compact projection and ceilings apply.

Searchable FAERS field paths with coded values and traps, plus the available stratifiers.

Consult before writing a raw_filter or count_field rather than guessing a path.

Empirical Bayes signal scores (EBGM with EB05/EB95) via the Gamma-Poisson Shrinker.

EBGM shrinks the observed-to-expected ratio toward 1 in proportion to how little evidence supports it. EB05 is the conventional screening statistic; EB05 > 2 is the usual threshold. The prior is fitted across a drug x event background table, so the FIRST call for a FAERS release builds and caches that table (~100 s, 3 + background_drugs calls). Call faers_warm_cache first, or raise the client timeout. Later calls take a few seconds.

Build the EBGM background table and fit the prior ahead of time.

The first faers_ebgm call for a FAERS release takes ~100 s and 100+ API calls, which exceeds many MCP client timeouts. Calling this tool first moves that cost to a turn that expects it; subsequent faers_ebgm calls take a few seconds. Reports whether each cache was already warm.