Data overview

Datasets

Dataset	What it is	Size (September 2026)
Programs	Drug programs from company pipelines, filings and trial registrations: drug, indication, phase, modality, target, mechanism, status, linked trials	14,000+ active programs (28,000+ in all)
Drugs	One drug product across every company that develops it, under all its names	14,000+ drugs
Companies	Biopharma companies with owners, tickers and identifiers, and their programs, trials, deals, patents and changes	3,300+ with active programs
Clinical trials	ClinicalTrials.gov studies, linked to Pav programs and sponsor companies	ClinicalTrials.gov registry
Deals	M&A, licensing, collaborations, options, joint ventures; terms, lifecycle, drugs covered, source documents	3,300+ deals
Patents	US patent families with owners, members, ownership history, FDA-approved drugs and statutory term	123,000+ families
FDA records	Orange Book, Purple Book, orphan designations, drug applications, complete response letters, accelerated approvals, advisory committee meetings, warning letters and recalls	115,000+ records
Changes	Detected movement: programs added, removed or changing phase; trials changing status, dates, phase or results	Continuous feeds

Current active program and company counts are available at any time from GET /v1/stats.

Every dataset joins through Pav ids.

A company (company_id) owns programs (id), and can own other companies (parent_company_id).
A drug (drug_id) groups every company’s programs of the same product.
A program lists its linked clinical trials (nct_id). A trial lists the Pav programs and companies linked to it.
A deal lists its party companies and the drugs it covers; filter deals by company_id or drug_id.
A patent family lists its owner companies and the FDA-approved drugs listing its patents; filter patents by company_id.
An FDA record carries its linked company and FDA application number; filter FDA records by company_id or application_key.
A change records a program’s movement with its program_id and company_id, or a trial’s with its nct_id.

Look up a company’s id once, with GET /v1/companies?q=..., then reuse the id across every dataset.