Skip to content
Pharma Observatory a Scientist.com platform

Data

What is in Pharma Observatory

Counts of what we hold, rounded down, each with a plain definition of what one record is and where it comes from, so you can set them beside any other provider's figures.

Individual plan, $249/seat/mo. Card required.

Clinical trials 600,000+ Investigators 380,000+ Organizations 350,000+ Publications 790,000+

The counts

A + means at least that many: each is rounded down to two significant figures.

  • Data sources One dataset or registry we load on its own schedule, such as ClinicalTrials.gov, PubMed or a type of SEC filing, counted once however many files it comes in. Sources no commercial licence covers, and data we use only internally, are not counted.Sources: ClinicalTrials.gov, PubMed, SEC EDGAR, FDA, CMS, NIH, and other public sources. 94+ data sources
  • Source records Records loaded from the public sources we draw on, before matching: registry entries, filings, publications, payments and similar. One real-world thing can appear in many source records.Sources: ClinicalTrials.gov, PubMed, SEC EDGAR, ORCID, PatentsView, and other public sources. 930 million+ source records
  • Clinical trials One study registered on ClinicalTrials.gov, interventional or observational, at any status.Sources: ClinicalTrials.gov. 600,000+ trials
  • Investigators One person named as an investigator on at least one registered trial, with name variants and affiliations matched into a single record where we can.Sources: ClinicalTrials.gov, the NPI registry (NPPES), and ORCID. 380,000+ investigators
  • Research sites One physical place where a trial is run. Listings that share a facility name, city and country are counted once.Sources: ClinicalTrials.gov and the NPI registry (NPPES). 1 million+ research sites
  • Organizations One organization after its name variants and registry identifiers are matched: trial sponsors, manufacturers, institutions, filers and grant recipients.Sources: ClinicalTrials.gov, SEC EDGAR, ROR, GLEIF, FDA registrations, NIH RePORTER, and other public registries. 350,000+ organizations
  • Molecules One drug molecule, with its salts, formulations and trial-arm names folded into a single record. Combinations are counted under their components, and an investigational compound we could not link to a named, branded or approved molecule may not be counted.Sources: ClinicalTrials.gov, FDA Drugs@FDA and NDC, RxNorm, ChEMBL, EMA, and Health Canada. 4,800+ molecules
  • Drug substances One substance with its own registry identifier (a UNII), counted once across every name it was registered under. A salt or other form that has its own identifier is counted separately. A substance that never matched a registry identifier, whether a sponsor's code or a name the registry does not recognise, is not counted.Sources: FDA Drugs@FDA and NDC, RxNorm, ChEMBL, and ClinicalTrials.gov. 8,600+ drug substances
  • Brands One approved brand name (trade name), linked to the drug it sells. Several brands can point to one drug.Sources: FDA Drugs@FDA and RxNorm. 6,500+ brands
  • Development programs One drug name, as registered, studied in one condition, counted when at least two registered trials cover the pair.Sources: ClinicalTrials.gov. 330,000+ development programs
  • Conditions One disease or condition in the MeSH vocabulary, including the broader terms trials roll up to.Sources: NLM MeSH and ClinicalTrials.gov. 5,100+ conditions
  • Rare diseases One entry in the Orphanet rare disease classification: a group of disorders, a disorder or a subtype.Sources: Orphanet. 11,000+ rare diseases
  • Drug targets One gene and its protein in the Open Targets dictionary, with the evidence recorded against it.Sources: Open Targets. 78,000+ drug targets
  • Biomarkers One gene, variant or other marker with at least one of: an FDA listing, a cancer-actionable assertion, a mention in a trial, or a link to a disease. Many carry an association only, so this counts biomarkers with evidence on record, not validated tests. Markers with no evidence recorded are left out.Sources: FDA, the Cancer Genome Interpreter, ClinicalTrials.gov, and Open Targets. 25,000+ biomarkers
  • Biomedical patents One granted US patent classified in a medical or biological field: medical, dental and veterinary science, including devices and drug or cosmetic preparations; peptides; genetic engineering and enzymes; or tests on biological material. The classification comes from the patent's official patent-office codes. A patent appears on an organization's page when its assignee is matched to that organization.Sources: USPTO PatentsView. 1.1 million+ biomedical patents
  • Publications One PubMed article cited by at least one registered trial.Sources: PubMed and ClinicalTrials.gov. 790,000+ publications
  • Funding events One financing or grant event: an NIH or SBIR/STTR award, an SEC Form D offering, or a financing reported in an SEC filing.Sources: NIH RePORTER, SEC Form D, and SEC filings. 950,000+ funding events
  • Investors and funders One investment adviser or public funding agency, matched into a single record.Sources: SEC Form ADV and NIH RePORTER. 23,000+ investors and funders
  • FDA enforcement events One FDA enforcement record: an inspection, a refused import line, a warning letter, an untitled letter, a debarment or a recall.Sources: FDA inspections and Form 483s, FDA import refusals, FDA warning and untitled letters, FDA debarments, and openFDA recalls. 640,000+ FDA enforcement events
  • Hospitals One Medicare-certified hospital, one per CMS certification number.Sources: CMS hospital directory. 5,400+ hospitals
  • Research suppliers One supplier published on the Scientist.com marketplace with at least one located site, including contract research organizations. Duplicate and retired records, Scientist.com's own service accounts, and suppliers with no located site are left out.Sources: the Scientist.com supplier directory. 16,000+ research suppliers

How we count

Counted when the data is rebuilt

Each figure is rounded down to two significant figures and shown with a "+", so rounding never overstates a figure. Most are exact counts of the records behind them, taken when the data was last rebuilt. Data sources is the number of sources we run, and source records is an estimate from the database's own statistics.

One record per real-world thing

Name variants of one organization, one investigator or one molecule are folded into a single record, so a count is of things, not of source lines. A thing a source repeats under different spellings is counted once. Source records are the exception: they count what the sources contain before any matching, so one real-world thing can appear in many.

What is counted

Only data a subscriber on a paid plan can see. Data no commercial licence covers, and data we use only internally, is not counted. A provider's own figures may be defined differently, so compare the definitions as well as the numbers.

Source licences: ChEMBL is CC BY-SA 3.0. The Cancer Genome Interpreter is CC BY 4.0. EMA content is reused with EMA acknowledged as the source. Health Canada data is used under the Open Government Licence - Canada. Rare disease classification data is from the Orphanet Scientific Knowledge Base, Orphadata Science (INSERM), CC BY 4.0. Clinical trial records come from ClinicalTrials.gov through the AACT database of CTTI (CC BY).

Look inside

Condition pages Trial activity for each disease, free to browse. Biomarker pages Gene and variant evidence by tier, free to browse. Pricing What full access costs, plan by plan.