Claims Matrix: One Row Per Claim, One Source Per Row

A claims matrix is a controlled list of the claims a company has agreed it can make about a product, each paired with the evidence supporting it and the conditions under which it can be used.

It exists to stop the same argument happening on every piece of promotional material.

The problem it solves

Without one, a hundred-page deck arrives for MLR review and the reviewer evaluates every claim on it from scratch: is this supported, by what, and was it agreed before?

With one, the reviewer is checking a different and much faster question: is each claim on this piece already in the matrix, used within its stated conditions? Anything that matches passes. Anything that does not is the small set that needs real discussion.

That difference is the whole value. It moves review from re-adjudication to matching.

What a row contains

• The claim, in the exact wording approved.
• The evidence: the specific study, endpoint, table or label section, cited precisely enough to find in seconds.
• The label basis, meaning where in the approved labelling this claim sits. A claim with no label basis is off-label promotion regardless of how good the data is.
• Audience: healthcare professional, patient, payer. A claim acceptable to one may not be to another.
• Required context: the fair balance, qualifier or footnote that must appear with it.
• Geography, where the label differs by market.
• Status and expiry, with the date agreed and by whom.

The rules that keep it honest

Approved wording, not paraphrase. A matrix listing concepts rather than sentences invites drift, and the drift is where violations live.

Comparative claims need head-to-head data. The standard is substantial evidence, and "no head-to-head study exists" means the claim does not go in the matrix.

A claim is scoped to its market. Local labels differ. A claim supported by the EU SmPC is not automatically supported by the US label, and the core data sheet is where you check.

Expire claims when the evidence changes. A label update or a new study can invalidate a row. A matrix nobody prunes gradually becomes a list of claims that used to be true.

Where it breaks down

The matrix lives in a spreadsheet, the deck lives in PowerPoint, and matching one against the other is manual. So it gets skipped under deadline, which is exactly when the mistakes happen.

The second failure is subtler: a claim in a deck is a paraphrase of a matrix row rather than the row itself. It reads as the same claim, and it is not the approved wording.