Skip to content
Heavy Metal Index

WHO GEMS/Food contaminants database (Global Environment Monitoring System — Food Contamination Monitoring and Assessment Programme)

Source

The Global Environment Monitoring System — Food Contamination Monitoring and Assessment Programme (GEMS/Food) is the WHO’s global repository of individual-measurement contaminant occurrence data: one record per sample, per analyte, per food, with the food (Codex food code), region, year, de...

Page snapshot
Cited by10 pages
Metals measured9
Evidence tierA
Year2026

Overview

The Global Environment Monitoring System — Food Contamination Monitoring and Assessment Programme (GEMS/Food) is the WHO’s global repository of individual-measurement contaminant occurrence data: one record per sample, per analyte, per food, with the food (Codex food code), region, year, detection limit (LOD/LOQ), sampling representativeness, food origin, and analytical basis. It is the authoritative, agency-curated occurrence dataset that national exposure assessments and the JECFA evaluations draw on, which is why it is graded A-tier here.

This source is not a paper and is not ingested through the PDF/Cochrane pipeline. It enters the wiki as structured occurrence data through tools/evidence/ingest-gems.mjs, which streams the raw CSVs (held in raw/, never committed — Cd alone is ~190 MB), normalises units to ppb (µg/kg), separates censored (below-detection) from detected measurements, keeps wet (“as is”) and dry bases apart, and emits per-(analyte × food × basis) distribution aggregates to the structured evidence register. Every value derived from GEMS carries source_id: who-gems-food-database for provenance, so any cell can be re-derived from, or contested against, the GEMS records.

Coverage in this export

The June 2026 export covers Pb, Cd, tHg, iAs/tAs, Al, Cr, Ni, and Sn across the full Codex food taxonomy, with dedicated infant-and-children subsets for the priority analytes. Record counts run to hundreds of thousands per analyte (Lead ~588k rows; Cadmium larger).

How GEMS feeds contamination profiles

GEMS distributions are pooled, per the synthesis methodology, into the contamination_profile blocks on ingredient and product pages: the detected-value percentiles supply typical_ppb and p95_ppb, the record count supplies sample weight, and the censoring fraction and per-region/per-year spread supply the geographic and temporal variance the defensibility argument relies on. The GEMS food taxonomy maps to wiki ingredient/product slugs through the food crosswalk before any value lands on a page.

Update history

This page has no substantive edit history yet. Future revisions will be recorded here once the change lands in git.