How the data is built
Methodology
Where every figure on our data registers comes from, how often it is refreshed, how records are matched to companies and medicines, and what the data cannot tell you.
01
The rules every register follows
Copied from the source, never estimated
Every record comes from an official register or a company's own regulatory filing. Figures are copied as published. Where we calculate something, such as a 12-month total, a share or a year-on-year change, the page says so. No figure is estimated, extrapolated or generated by AI.
Dated and sourced
Each register shows when it last synced with its source and lists its sources with their licences at the foot of the page. Most registers refresh nightly (from about 03:15 UTC); slower-moving data such as trade statistics and company filings refresh weekly.
A failed update keeps the last good data
If a source is unreachable or returns an error, the previous data stays up with its original sync date rather than an empty or partial register. The regulatory, company-financial and medicine-spending updates also refuse a refresh that would suddenly shrink the data, treating it as a possible error until it is checked.
Scope stated on the page
Registers say what they include and leave out, for example which trial statuses or which approval route. Counts describe the register as stated, not a whole market.
02
Data families
sources, method and limitsClinical trials
SourcesClinicalTrials.gov API v2 (US National Library of Medicine) for the country registers and the Irish dashboard; the EU Clinical Trials Information System (CTIS) for the EU register.
MethodA country register lists studies with at least one site in that country, selected on the registry's location-country field, not on free text. Each register states on the page which statuses and sponsors it includes. EU trials are shown with the CTIS status they carry; trials that were not authorised are left out.
LimitsRegistry records are entered by sponsors and can be late or incomplete. The same study can appear on ClinicalTrials.gov and CTIS under different identifiers, so the two are never added together.
Approvals, safety and patents
SourcesopenFDA (Drugs@FDA, device 510(k) clearances, enforcement reports), FDA Orange Book and Purple Book downloads, European Medicines Agency data files (medicines, orphan designations, safety communications), Health Canada Notice of Compliance data (Open Government Licence, Canada).
MethodRecords are copied field by field from the regulator's file. "First approved" dates are found by matching the active substance across regulators with a normalised substance name; biosimilars and generics are excluded from first-approval dates.
LimitsUS approvals from Drugs@FDA exclude most biologics regulated by CBER (vaccines, gene therapies). EU approvals cover the centralised procedure only, not national authorisations. Patent and exclusivity dates are the regulator's listed dates, not legal advice on when a product can be copied.
Medicine supply
SourcesThe FDA drug shortage database, EMA medicine shortage communications, the German BfArM supply-shortage register, and the HPRA list of medicines authorised in Ireland.
MethodEach shortage keeps the status and dates the regulator gives it. A company is linked to a shortage only when it is named on the regulator's record.
LimitsShortage lists change daily and are only as current as the last refresh shown on each page. A shortage can involve several manufacturers.
Drug targets and mechanisms
SourcesThe Open Targets Platform (CC0 1.0), whose drug mechanism, target and drug-disease records come from ChEMBL (EMBL-EBI, CC BY-SA 3.0); checked against the IUPHAR/BPS Guide to PHARMACOLOGY (database ODbL 1.0, contents CC BY-SA 4.0).
MethodEach compound and drug page name is looked up in Open Targets. A match is accepted only when ChEMBL's own name for the medicine, or a synonym ChEMBL itself curates, is the same molecule; close spellings, automatically mapped trial-registry names and combination products are never matched, and a short list of names checked by hand and found wrong is refused. Target pages are made only for mechanisms that name one protein. Every medicine-target link is then checked against the Guide to Pharmacology, matched by ChEMBL ID or exact name, with each target resolved to its human gene whatever species the study used; pages say whether a link is in both sources or one. The Guide to Pharmacology adds links of its own only for a medicine's primary target.
LimitsOnly the trial stages of drug-disease records are shown. Open Targets also labels some records "approval", but it splits combined label wording into separate diseases, so approved uses are taken only from the regulators' own records. The data covers medicines ChEMBL has curated; many early-stage compounds have no record yet.
Reference identifiers
SourcesWikidata (CC0 1.0): ATC codes, DrugBank, PubChem, CAS and FDA UNII identifiers and English Wikipedia articles for medicines; Wikidata items, official websites, founding years and headquarters for company groups.
MethodMedicines are matched to Wikidata only through their ChEMBL ID, and company groups only through their SEC filer number (CIK) or Legal Entity Identifier, never by name. An identifier carried by two Wikidata items is dropped. The data is re-read every 30 days.
LimitsWikidata is edited by volunteers and can be incomplete or out of date; each value links to its source so it can be checked. Company facts describe the group, not its Irish subsidiaries. Logos are not taken from Wikidata because most are trademarks.
NICE appraisals
SourcesNICE technology appraisal guidance pages and NICE's technology appraisal recommendations data, shown with NICE's written permission (October 2026): appraisal number, title, recommendation category and dates, with a link to the guidance and NICE's attribution.
MethodA nightly job reads NICE's recommendations data, reads each new appraisal page once for its title, dates and status, and re-checks every page every 60 days so replaced and withdrawn guidance is marked. An appraisal is linked to a medicine page only when the medicine NICE appraised is exactly the same molecule; medicines given alongside it are not counted. Company pages list appraisals of medicines whose first EU or US authorisation was held by the company.
LimitsNICE decisions apply to the NHS in England only. No guidance text is reproduced: for who a medicine is recommended for, and on what terms, follow the link to NICE. Recommendation types are summarised in our own words.
Medicine spending
SourcesNHSBSA Prescription Cost Analysis, monthly (Open Government Licence v3.0); CMS State Drug Utilization Data (US public domain); FDA National Drug Code Directory, current and delisted products.
MethodNHS England figures are the net ingredient cost and items of every medicine and appliance dispensed in the community, summed by BNF chemical substance over the latest 12 published months and the 12 before. Medicaid figures are the national rows of the State Drug Utilization Data for the latest complete year and the year before, summed by NDC product and grouped by active ingredient using the FDA NDC Directory. Where the BNF moves products to a new code during the year, no year-on-year change is shown for the codes involved.
LimitsBoth are gross amounts: NHS net ingredient cost is list price before discounts, and Medicaid amounts are before manufacturer rebates, which neither system publishes per drug. NHS figures exclude hospital medicines. Medicaid rows covering fewer than 11 prescriptions are suppressed by CMS and left out.
Irish economy and companies
SourcesCSO Ireland PxStat tables TSM09, TSM10 and MIM05 (CC BY 4.0); SEC EDGAR XBRL financial data from 10-K and 20-F annual reports; GLEIF legal entity data (CC0); CRO Open Data (CC BY 4.0).
MethodTrade values are the CSO's published figures; 12-month totals, shares and changes are calculated from them and labelled. Company financials are the values each group tagged in its annual report, in the currency it reports in, and each year comes from the latest report that includes it. Irish companies are linked to a group through GLEIF parent records or by name and registered address, checked by hand.
LimitsThe CSO revises recent months and suppresses some series for confidentiality. Groups that do not file with the SEC have no financial figures here. Register records are linked to company pages by the organisation names they carry, which can miss or mismatch a name; each company page lists the names it matched.
03
Corrections and questions
When we get it wrong
Errors are fixed and recorded in the public corrections log, with what was wrong and what changed. Our wider policy is in the editorial standards.
Report an error or reuse the data
Email info@priyalifescience.com or use the contact form, naming the page and the figure. Most registers can be downloaded as CSV from their own page; please credit the original source and Priya Life Science.