The golden thread: why one endpoint change should rewrite five documents — automatically

A structured argument for treating protocol documents as a derived view of a canonical model, not as the model itself.

Every clinical trial produces roughly the same shape of document lattice: protocol, SAP, ICF, screening forms, IRB narratives, investigator brochures. In most tooling today, each of these is a separate Word document maintained by hand.

This is the design mistake the entire industry inherited from the 1990s — documents as the canonical artefact, structure as an afterthought. The cost shows up as three-week sync cycles whenever a primary endpoint moves by two weeks.

Trelice’s answer is to invert that: the structured protocol is the source of truth, and every "document" is a view over it. Change the endpoint once, and the dependent fields regenerate. The Word output is a render target, not a master.

This post walks through how we model the thread, where the current boundaries are, and what we plan to extend it to next (screening, IRB Q&A, investigator brochures).

---