Methodology
The sourcing standard every figure has to clear before it reaches you — and what we do when one cannot clear it.
Every AXLRx brief is a synthesis of published evidence, assembled and interpreted by analysts. This page sets out where the evidence comes from, what each figure has to survive before it is used, and what happens when a number cannot be verified. It is the same standard we hold ourselves to on commissioned work and on everything published here.
A figure ships only if it can be traced to the document it came from. If it cannot, it is removed rather than estimated.
That sounds obvious. In practice it is the rule that gets broken most often in commercial research, because an approximate number is more useful to a deck than a missing one. We take the opposite view: a gap you know about is safer than a figure you cannot defend in front of a payer, a regulator, or your own review committee.
Primary sources only. In practice that means:
What we do not use: syndicated market-research summaries, aggregator sites that restate a figure without its origin, or anything recalled from memory. A number that appears in three secondary sources and no primary one is not corroborated. It is repeated.
A live citation. Each claim is tied to a specific identifier — a PubMed ID, a trial registry number, a regulatory document reference or a URL — captured at the point of writing, not reconstructed afterwards. Regulatory identifiers in particular are looked up live every time. They are easy to recall incorrectly and expensive to get wrong.
A numeric cross-check. The figure must actually appear in the source cited, not merely be consistent with it. This is a stricter test than it sounds, and it is where most inherited errors surface: a percentage that drifted between a press release and a publication, a patient count that belongs to a different subgroup, an appraisal number attached to the wrong medicine.
An independent audit. Before delivery, a separate pass re-checks the links, the claims and the numbers against their sources. The purpose of a second pass is to be unconvinced by the first.
It is dropped, and where its absence matters we say so on the page rather than quietly closing the gap.
Some categories of number are simply not public. Negotiated prices under confidential commercial arrangements, patient access scheme discounts and net-of-rebate figures are, by construction, not disclosed. Where a brief needs to reason about them, it reasons about the published list price and the mechanism, and states plainly that the net figure is confidential. It does not print an estimate formatted to look like a fact.
Where two credible sources genuinely disagree — a registry count against a modelled prevalence estimate, for instance — we present both, explain what each is measuring, and let the difference stand as the finding. Reconciling them into a single tidy number would destroy the most useful thing on the page.
Every published report carries a named reviewer, shown on the page. The reviewer is responsible for the sourcing standard on that report, not merely for reading it. Analysts work to defined areas — competitive intelligence and payer access, disease landscape and patient flow, pricing and market sizing — so that the person checking a figure understands the context it sits in.
A source ledger listing every claim against the source it came from, so the evidence base can be audited rather than taken on trust. An appendix with the underlying detail. The exhibits in editable form, with their assumptions visible rather than buried. And an analyst readout, because the questions that matter usually arrive after the document has been read.
We would rather you know the limits than discover them.
A page reflects what was true when it was published or last reviewed, and each carries its date. Regulatory positions, health technology assessments and competitive positions move, sometimes within weeks — an appraisal is updated, a licence is withdrawn, a confirmatory trial reads out. We revisit published pages when a material change lands, but we do not warrant that every page is current at the moment you read it. Check the date, and check anything time-sensitive against the source.
Tell us. If a figure we published cannot be traced to the source we cited, we want to know, and we correct it. For commissioned work, a brief that fails the standard set out on this page is corrected at no charge, and refunded in full if we cannot correct it — the terms are in section 7 of our Terms of Use.
Corrections and questions about method go to hello@axlrx.ai, or through our contact page.