Where the data come from, how they were checked, and how to report an error.

Cancer Evidence Explorer

• IARC Monographs, table “Classifications by cancer site”, Volumes 1–141 (last update 18 May 2026): the 185 agents with at least limited evidence in people for a named cancer, and their 469 links to 68 types of cancer. Adapted from Schubauer-Berigan et al., JNCI Monographs 2026, and augmented for later evaluations and for the single HPV types named in Volume 100B.
• IARC Monographs, List of Classifications, Volumes 1–142 (last updated 23 July 2026): the group of each agent, and the 874 agents IARC evaluated without naming a cancer site.
• The evaluation in each agent’s Monograph volume: what IARC found in studies of people, in studies of experimental animals and on how the agent acts in the body.

About this tool →

Chronic Disease Evidence Explorer

• Systematic reviews with meta-analysis, pooled cohort consortia and expert syntheses, such as the GBD Burden of Proof analyses and the Lancet Commissions. A single study is used only as randomised or genetic evidence that supports or contradicts a finding from observational studies.
• The reference index lists the 641 published sources behind the links, each with a link to the paper.

Reference index

Cancer Evidence Explorer

• Every link was compared with IARC’s cancer-site table. Where the table lists a cancer that the evaluation text of the volume does not name, the entry says so.
• The volume and year behind each classification were checked against IARC’s List of Classifications and the Monograph volumes, in two independent passes (October 2026). The evidence in people, in animals and on how each agent acts was taken from IARC’s own evaluations, checked in a second pass, and every quote compared with its source.
• IARC’s List of Classifications was last compared with the tool on 28 October 2026. It had not changed since IARC’s update of 23 July 2026.

Chronic Disease Evidence Explorer

• Every effect estimate was compared with the paper it cites, for the exact comparison stated (August 2026). In September 2026 a second pass checked 100 flagged links against their papers again. Each grade follows the written decision table of the framework.
• The change log records every change to every link since the first published version (24 August 2026). In it, 343 of the 431 links now in the tool have at least one dated check or correction from September or October 2026. The log lists changes only, so a link that was checked and found correct leaves no row: more links were checked than this count shows.

Change log Framework: what has been checked and what has not