FAERS Adverse Event Data: A Complete Guide for Pharmacovigilance Teams

The FDA Adverse Event Reporting System, commonly called FAERS, is one of the most important public resources for post-market safety surveillance. It is also one of the most time-consuming to use directly. Each quarterly release contains linked tables, duplicated case versions, inconsistent drug names, MedDRA terms, outcomes, indications, therapies, and reporting source metadata.

What FAERS contains

FAERS is organized into seven core tables: DEMO for case demographics, DRUG for suspect and concomitant products, REAC for reactions, OUTC for patient outcomes, INDI for indications, THER for therapy dates, and RPSR for report sources. A real safety question usually requires several of these tables at once.

Why the raw data is hard to analyze

Drug names may arrive as brand names, generics, misspellings, abbreviations, or free-text variations. Case identifiers need version handling. Serious outcomes must be interpreted across multiple rows. Reported indications and reactions are not a clean substitute for labeled indications and warnings. These issues are manageable, but they should not be solved repeatedly by every safety analytics team.

What PRR means

The Proportional Reporting Ratio compares how often a reaction is reported for a drug versus how often that reaction is reported for all other drugs. It is a disproportionality signal, not a causal finding. A typical PRR workflow filters to primary suspect drugs, counts drug-event pairs, computes background event rates, and applies minimum count thresholds.

SELECT
  canonical_drug_id,
  reaction_pt,
  case_count,
  expected_count,
  prr_score,
  serious_outcome_rate
FROM `tdd.safety.faers_prr_signals`
WHERE prr_score >= 2
  AND case_count >= 3
ORDER BY prr_score DESC;

Label coverage is the missing context

A high PRR score is more useful when analysts know whether the event is already described in current labeling. Label coverage requires a join between FAERS reactions and structured label sections such as boxed warnings, warnings and precautions, contraindications, and adverse reactions. Historical labels add another dimension: did the language change after the signal started appearing?

Common data quality issues

How PV teams should use FAERS

FAERS should be used as a triage and hypothesis-generation system. The best workflows combine PRR, seriousness outcomes, report geography, reporter type, label coverage, REMS status, and lifecycle context. The analyst's time should be spent evaluating biological plausibility and clinical context, not cleaning quarterly text files.

Warning: FAERS reports are spontaneous adverse event reports. They cannot estimate incidence and should not be interpreted as proof of causality.

Related data products

• FDA Adverse Event Signal Intelligence (FAERS)
• FAERS Drug-Reaction Signal Intelligence
• FDA Structured Drug Label Intelligence