Pharmaverse

A clinical trial does not hand you a tidy data frame. It hands you data shaped by two CDISC standards: SDTM, the tabulation model that organises what was collected, and ADaM, the analysis model that organises what gets tested. Both are required for submission to the FDA and to the PMDA in Japan, and both are defined down to the variable names, so a demographics table built in one company reads the same as one built in another.

The pharmaverse is where the R tooling for that pipeline lives. It is a network of companies and individuals that curates open-source R packages for clinical reporting, work that used to happen behind each company’s firewall as closed, duplicated code. A council drawn from Roche, GSK, Sanofi, Novo Nordisk, Novartis, Atorus, Appsilon and Johnson & Johnson decides what is in the set and what is recommended. This section works through the pipeline with the packages that Roche and GSK built or co-built: sdtm.oak and the admiral family for the data standards, rtables, tern and tfrmt for the reporting end.

Every number and every table on these pages comes from a real run of the code you can read on the page, in one pinned container, on packaged example data.

The path

Page	What it does
CDISC and SDTM data	what an SDTM study looks like in R, and how raw data becomes a domain
ADaM with admiral	deriving the analysis datasets ADSL and ADLB from SDTM domains
Tables for clinical reports	a demographics table and an adverse event table from ADaM data

The data

Two packages exist only to hold example data, and the whole section computes on them. pharmaversesdtm carries SDTM domains, several of them sourced from the CDISC pilot project, a real study run through the standards and published for exactly this purpose. pharmaverseadam carries the ADaM datasets that the admiral template programs produce from those domains, regenerated on each admiral release, so the target of a derivation on the ADaM page can be checked against a reference the admiral team shipped.