Fact-Checking Policy

Last updated: June 2026

Every number on trialfinderdata.org is meant to be traceable to a primary public record. This policy describes where our facts come from and how we check them before they are published.

Primary sources only

TrialFinderData builds its pages from the ClinicalTrials.gov v2 API. We do not republish second-hand summaries, scraped aggregations, or unattributed figures. Each page cites the dataset it draws from and links back to the source so any reader can verify a figure independently.

No invented numbers, no synthetic data

If a value is not present in the underlying public data, it does not appear on trialfinderdata.org. We never fabricate statistics to fill a gap, and we never use model-generated estimates in place of reported figures. Where the data is genuinely missing, we say so rather than guess.

How figures are verified

• Faithful processing. Data is fetched directly from the source, then transformed with documented, repeatable code — not hand-transcribed, where transcription errors would creep in.
• Source-of-truth checks. Derived values (rankings, grades, composite scores) are computed from the source fields using stated formulas, so the inputs to any number remain auditable. We index every U.S.-recruiting study from the ClinicalTrials.gov v2 API, parse inclusion and exclusion criteria into a searchable eligibility profile (condition, age, sex, phase, location), and link each result to the original registry entry for full protocol detail.
• Sanity bounds. Cross-cutting totals and outliers are checked against known ranges to catch unit errors, duplicates, and mis-scaled values before they reach a page.
• Dating. Every dataset page carries a “Last updated” date so readers know how current a figure is.

The limits of what we check

TrialFinderData verifies that our pages faithfully represent the ClinicalTrials.gov v2 API. We do not independently audit the issuing agency’s collection methods, and we cannot guarantee the source itself is free of error. Source records may be incomplete, reported late, or restated after publication. We treat the primary record as authoritative and follow its corrections — always verify a consequential figure against the original source before acting on it.

Found an error?

Tell us and we will fix it. See our Corrections Policy for how to report an issue and what happens next, and our Editorial Policy for how our content is produced.