A New Kind of Scrutiny on an Old Part of the Dossier

In late June 2026, the House Select Committee on China opened inquiries into major drugmakers — including Merck, AbbVie, Eli Lilly, Pfizer, and Bristol Myers Squibb — over clinical-trial activity in China. The committee raised specific concerns about trial sites in the Xinjiang region and sites linked to Chinese military hospitals or institutions, and asked the companies for information by July 17, 2026. The framing spanned national-security, ethical, data-protection, and informed-consent concerns; the letters were careful to note there was no evidence of illegal activity.

This is usually read as a geopolitics story, and it is one. But there is a quieter reading that matters to a regulatory writing organisation: clinical-trial-site geography — always a factual element of a submission — is becoming a scrutinised feature of it. And when a feature of the dossier comes under scrutiny after the fact, the operational question is the same one a retraction poses: can the writing organisation account for it precisely — where every trial site was, what data came from where, and what the submission relies on from each — quickly and completely.

The political scrutiny is new, but the regulatory question is not. FDA has already challenged submissions when pivotal evidence was generated largely in China and reviewers questioned its applicability to the U.S. population. The House inquiry adds a national-security dimension on top of a generalizability question the agency was already asking.

Trial-Site Geography Was Always in the Dossier. Now It Is Load-Bearing.

Every submission already contains its trial-site information. The list of investigator sites, their locations, the enrollment by site and region, the multi-regional trial structure — all of it is in the clinical study reports and the Module 5 detail. It has always been there. What changes when site geography comes under scrutiny is that this information stops being background and becomes load-bearing: a reviewer, or a congressional committee, or an agency responding to pressure, may now ask a sponsor to explain and defend the geographic composition of its evidence.

That shifts a writing burden that most dossiers were not built to carry. A submission assembled to demonstrate efficacy and safety was not necessarily written to make its site-provenance story easy to interrogate. The site data is present, but it may be scattered across study reports, not synthesised into a clear account of what share of the pivotal evidence came from which regions, which sites contributed which patients, and how the multi-regional structure supports applicability to the population the sponsor is seeking to treat. When the question comes, the sponsor that can answer it cleanly is in a very different position from the one that has to reconstruct the answer from a dozen study reports.

The Two Questions the Sponsor Has to Be Able to Answer

Scrutiny of trial-site geography reduces, for the writing organisation, to two questions it has to be able to answer on demand.

Where did the evidence come from? Not "which sites were in the trial" as a buried appendix, but a clear, current, queryable account: the geographic composition of the pivotal data, the proportion of key endpoints driven by each region, the specific sites and what they contributed. This is the provenance question applied to sites rather than citations — and it has the same failure mode. If the answer lives only as scattered site lists across multiple documents, assembling it under time pressure is slow and error-prone, and the error that matters is the one you miss.

What happens to the benefit-risk case if a site, region, or data subset is challenged? This is the harder question, and it is the one the House inquiry ultimately points toward. If the data from a scrutinised site or region were discounted, challenged, or excluded, would the submission's conclusions survive? A sponsor that has written its benefit-risk argument with a clear understanding of which conclusions depend on which regions' data can run that stress test. A sponsor that has not is exposed to a scenario where a geographic challenge propagates unpredictably through a dossier nobody mapped for it.

Neither question is answerable by a submission that treated site geography as a formality. Both are answerable by one that treated it as provenance.

Multi-Regional Trials Make This a Design Decision, Not Just a Reporting One

The modern pivotal trial is frequently multi-regional by design — run across the US, Europe, and Asia, including China, to enroll faster and to support global filings. That structure has real scientific and operational advantages, and it is not going away. What the House inquiry signals is that the writing around that structure now has to do more work.

A multi-regional trial's submission has to make an applicability argument: that the pooled, multi-region result applies to the specific population and regulatory context the sponsor is filing in. ICH E5 and E17 frameworks exist precisely for this — consistency of effect across regions, the influence of intrinsic and extrinsic ethnic factors, the justification for pooling. That argument has always been part of good multi-regional submission writing. Under geographic scrutiny, it becomes central: a submission that has written a rigorous region-consistency and applicability argument is far more defensible than one that pooled the data and left the regional structure implicit.

The writing decision, then, starts upstream — at the point where the trial's regional composition and the submission's applicability narrative are designed together, rather than the narrative being retrofitted after the data are in and, now, after the geography has become contentious.

What This Asks of the Writing Organisation

The defense against site-geography scrutiny, like the defense against a retraction, is built before the scrutiny arrives.

A site-provenance layer that spans the submission — a current, queryable account of where the evidence came from, synthesised rather than scattered, so the "where did this data come from" question has a fast, complete answer.

A benefit-risk and efficacy argument that knows its own regional dependencies — an explicit understanding of which conclusions rest on which regions' data, so the "does it still stand if a subset is questioned" scenario can be assessed rather than feared.

A region-consistency and applicability narrative written as a genuine argument — grounded in the multi-regional frameworks, not a boilerplate paragraph — so that the geographic structure of the evidence is defended, not merely disclosed.

This is not an argument against China trial sites, or against any geography — those are strategic and operational decisions well outside the writing organisation's remit. It is an argument for being able to account for trial sites with the same precision as any other load-bearing source of evidence: so that whatever the geographic composition turns out to be, the submission can account for it, defend it, and withstand a challenge to it — because the site information was written as provenance, not as a formality.

The Wider Pattern

The House inquiry is one instance of a broader shift: features of a dossier that were once treated as settled background — the sources it cites, the sites that generated its data, the regions its evidence came from — are increasingly subject to after-the-fact scrutiny, sometimes scientific, sometimes political. The through-line for a writing organisation is provenance in the widest sense: the ability to say, quickly and completely, where every load-bearing element of a submission came from, and what the submission would look like without it.

The Probe Is the News. The Writing Lesson Is Provenance.

The probe is the news. The writing lesson is provenance.

Trial-site geography was always present in the dossier: site lists, enrollment tables, regional subgroup analyses, CSR appendices, and Module 5 detail.

What changes now is that geography may become a challenged feature of the evidence.

The sponsor that can answer quickly — where the data came from, what each region contributed, how the pooled result holds, and what happens if one subset is questioned — is in a very different position from the sponsor that has to reconstruct the answer under pressure.

This is not about avoiding any geography.

It is about making trial-site provenance visible before someone asks for it.