Methodology & coverage
A defined research sample
Initial coverage focuses on industry-sponsored Phase 2/3 drug trials associated with US-listed companies, across indications. Coverage expands as sources and mappings are verified. Trial and sponsor counts describe records present in this database, not every trial or every company program.
Historical coverage
The research archive contains 1,642 trials and 14,436 selected field versions. These are selected versions and fields, not complete registry snapshots. Some endpoint text was truncated to 300 characters by the research collector. Individual trial pages show the available coverage. Historical backfill is an offline operation; the application never depends on undocumented history pages at request time.
What a change means
A registry edit is evidence that the submitted text changed. It does not establish the reason, scientific significance or commercial impact. A hand check of 30 flagged endpoint events found 19 substantive changes and 11 rewordings. Unreviewed text edits remain labeled as edits.
Dates and attribution
Registry submission and posting dates differ. Timelines display the stored posting date when available. Completion dates may be estimates. Sponsor-level observations are limited to the included sample and are not complete track records.
Regulatory coverage
The regulatory calendar covers source-backed events linked to tracked companies and known drugs. SEC discovery scans recent filings; FDA sources include original CDER approvals, annual CBER approval tables, complete response letters and advisory committee notices. Unmatched and ambiguous records remain unpublished. The calendar shows source scan status; a completed scan is not a claim of complete market coverage.
Automated evidence review
AI-assisted review checks candidate regulatory events against cited primary-source evidence and known entity relationships. Ambiguous or conflicting candidates remain unpublished. Automated verification can be wrong; sources remain available for independent checking.