Read the document, not the listing: regulatory intelligence needs a primary-source rule

The label and the document are not the same thing

Start with the plainest example in this week's records. DG SANTE has published MDCG 2026-6, guidance on clinical evidence for orphan in vitro diagnostics. The accompanying entry description, however, refers to a different IVD classification guidance. Either the title or the description is wrong. The records do not say which, and I won't guess. The practical point is that anyone who filed this item from the description alone has logged the wrong guidance, or has missed the right one.

The MHRA's herbal list shows a different failure. The regulator refreshed its list of Traditional Herbal Registrations on 2 November 2026, but the page does not say which products changed. A notice that something changed, with no indication of what, is not intelligence. It is a prompt to go and do the comparison yourself.

Then there is the FDA's final potency assurance guidance for cell and gene therapies. The Federal Register notice describes it as replacing the December 2023 draft. CBER's own pages describe it as superseding the January 2011 potency tests guidance. These descriptions are not necessarily in conflict, but they frame the document differently. A team reading only one would form a different view of what has been displaced and what still applies.

Second-hand regulatory facts carry a second-hand risk

Much of the rest of this week's regulatory news reached us through sponsors. SeaStar says CBER determined that its QUELIMMUNE paediatric registry fulfils the post-approval study requirements attached to the device's Humanitarian Device Exemption approval. NRx says the FDA cleared its preservative-free ketamine ANDA ahead of the GDUFA goal date. Inovio's Voceryq is described as accelerated approval, with continued approval potentially contingent on confirmatory study results. Compass says it will complete its rolling NDA within two weeks and plans a first-half 2027 launch if approved.

None of these is implausible, and a company statement about its own regulatory status is a legitimate source. But it is a statement of the company's reading of a regulator's decision. The wording of a post-approval study determination, the scope of an accelerated approval, or the conditions around a launch plan all sit in the underlying letter or label, not the press release. In the Compass case the launch is conditional on approval, and the approval has not yet happened.

The strongest counter-argument

The obvious objection is proportionality. Regulatory teams are stretched, and most listing errors are cosmetic. A mismatched description on a guidance page is an embarrassment for the Commission, not a compliance event for you. Requiring every item to be traced to a primary document would slow the flow of information to the people who need it quickly. And company filings are legally accountable documents.

That is fair, and I don't propose reading everything at full depth. But the cost asymmetry runs the other way. Checking that a guidance title matches its content takes minutes. Discovering after a submission, an inspection or a diligence exercise that the team relied on the wrong document, or on a sponsor's gloss of an FDA determination, costs far more. The checking effort should scale with what you will do with the item, not with how official it looks.

A rule, not a heroic effort

What I am arguing for is a standing rule: no regulatory item enters a tracker, briefing or decision without a link to the primary document, and a named person who has opened it. Where the regulator's notice is silent on what changed, as with the MHRA herbal list, the owner records the comparison or records that none was made. Where a regulator's action is reported by a company, the tracker marks it as company-reported until the agency's own document is seen.