(14) Key references | JPMA Guide Commentary | Advertising Regulation in Japan (Promotional Material Review) and Artificial Intelligence

The sixteen items of the comprehensive product information summary open with the development story, pass through the clinical results that form the core of efficacy, and close with the location of the evidence and the marking of the version. One of the two closing pillars is this item ⑭, Main References. There is no dramatic prohibition here. The task is simply to list, in a form anyone can later trace, the references that support the text. Yet this unassuming list is the keystone that turns the document's claims into claims that can be verified.

In one phrase, the references section is the convergence point for sources. Which statement in the text rests on which trial, which paper? Creating a state in which the reader can confirm that correspondence unaided is the section's sole, and greatest, job.

01Why "just listing references" earns its own item

The foreword places on companies a duty to convey accurate information, and at the same time positions the product information summary as a "supplement" to the electronic package insert. Being a supplement, its contents may not step beyond the scope of approval, and must also be in a state where a third party can confirm whether they are correct. The means of confirmation is precisely the source. A claim whose source is not shown offers the reader no entry point for verification, even if its content is true.

The third of the design principles that run through the Creation Guide — securing verifiability through structure — takes concrete form here. Citing sources, disclosing statistical methods, marking the date of preparation or revision: these look like separate rules, but their aim is one. To place into the reader's hands the answer to "what is this statement based on?" The references section is where that handover is performed all at once, in the shape of a list.

02The weight of an obvious act: listing the supporting references

What goes into the references is the literature that supports the statements in the text. Not academically impressive papers added for decoration, nor loosely related reviews used to borrow authority. Only references with which the body's statements on efficacy, safety, pharmacokinetics, and pharmacology can each be matched to their foundation earn a place here.

This condition — that the correspondence holds — is continuous with the discipline of the evidence hierarchy. Meta-analyses and randomized controlled trials, observational studies, case reports, and expert opinion carry vastly different weight of conclusion. Whether peer review was undergone is the watershed of quality. What is listed in the references implicitly discloses on which tier of evidence the text's claims rest. By looking at the nature of the sources, the reader can judge how strongly to take each statement.

What the nature of a source tells the reader about how strong the evidence is

For the very same statement that something "works," the certainty a reader should assign differs by orders of magnitude depending on whether the support is an original randomized controlled trial, a case report of a handful of patients, or a review that never left the company. The reference list is not a bare roll call of sources; it is itself the information that hands the reader the confidence level of each claim.

Peer-reviewed original papers — published and subjected to critical examination by third parties. The easiest to trace, with the clearest character.
Approval-time evaluation materials — primary material the regulatory authority used in its approval decision, publicly scrutinized through a route separate from publication.
Case reports and expert opinion — strongly individual, calling for caution before generalization. Listing them is not barred, but they must not be made to look like strong evidence.

03When the clinical results are approval-time evaluation materials

When the basis of the clinical results is material that was evaluated during the domestic approval review process, this fact is stated explicitly in the references. Approval-time evaluation material is the primary material the regulatory authority used in its decision to approve, and ranks among the heaviest of sources. That is exactly why writing "this rests on material evaluated at approval" carries meaning. The reader learns that the statement underwent public scrutiny through a route distinct from whether it was ever published as a paper.

One must not write only "approval-time evaluation material" and leave it vague which trial the body's clinical results correspond to. Noting that something is approval-time evaluation material is not a license to sever the correspondence. The notation gains meaning only once it is traceable which statement it supports.

04Bibliographic details on the clinical results page too — why place them twice

References relating to clinical results are not only gathered in the references section at the end; their bibliographic details are also recorded on the very page of the relevant clinical results. At first glance this is duplication, but it is intended doubling, not redundancy.

The reason lies in the reader's path. The moment a reader looking at a clinical-results figure wonders "what is this based on?", if the source cannot be confirmed on the spot, the flow of verification breaks. Going back to the end, then re-matching the body's claim against the list, is effort that pushes confirmation away. With the bibliographic details right beside the data, claim and evidence sit within the same field of view. Verifiability is not enough as "confirmable in principle"; it works only when designed to the point of being "easy to confirm in reality."

Location	Role
Each clinical-results page	Shows the source right next to the figures and data, enabling immediate, on-the-spot confirmation
Closing references section	Lists the document's supporting literature in full, consolidating the whole picture and location of the evidence

05When the source is internal company material

Sometimes the basis is neither a peer-reviewed original paper nor approval-time evaluation material, but internal company material. What the Guide demands then is not to stop at writing "internal material" as the source name. The specific content of that material must be recorded so that it can be understood.

Behind this lies the fact that there is a step in the verifiability of sources. Published papers can be traced to the original by anyone. Approval-time material has passed public evaluation. Internal material has neither form of openness. Because it is hard for the reader to obtain the original directly and confirm it, at the very least one adds enough specificity to trace "what was measured, in what subjects, and how." This is not a rule that excludes internal material; it is a brake that keeps weakly verifiable material from passing through opaquely while still weak.

The specificity to secure when citing internal material

The subjects and conditions the material addresses must be readable. The four characters "internal material" alone support nothing.
It must be traceable which statement in the text the internal material corresponds to. A source whose correspondence cannot be traced is a blank dressed as a source.
It must not be presented as if it were "strong evidence" on par with published literature. Not misrepresenting the nature of the source is part of the duty not to mislead.

06What it secures paired with ⑯ the date of preparation or revision

Main references (⑭) show "what it rests on"; the date of preparation or revision (⑯) shows "as of when the information stands." It is no accident that these two stand at the end of the sixteen items. The location of the evidence, and the point in time of the version. Only when these two are together can the reader grasp, in complete form, "as of when, and on what basis, this material stands."

The electronic package insert is revised. The evaluation of clinical results can be updated through re-examination and re-evaluation. That is exactly why the material's ability to state for itself "as of when, and on what basis it was made" enables later verification and recall. The references section bears that "on what basis" as the closing item.

In closing

Item ⑭, Main References, is not a section full of prohibitions. What is asked is to keep the correspondence between the body's claims and their evidence unbroken, and to list sources without misrepresenting their nature — that is all. If approval-time evaluation material, say so; the clinical-results literature also on the data page; if internal material, with specificity that lets the content be traced. Each converges on a single point: leaving things in a state that can be confirmed later.

The foreword's "accurate, without misleading" is not satisfied by avoiding exaggeration alone. Accuracy becomes verifiable accuracy only when the reader can confirm unaided what a statement is supported by. The quiet accumulation of an unassuming reference list is what sustains the trust of the whole material.

Case studies — the "it isn't written, so…" mindset, and how each lands

The reference list is the entry point that converts claims into verifiable assertions. Blocking or narrowing that entry point undermines the credibility of the whole body text. Four levels.

GRAY 1 — Straying from the spirit (untouched by the wording, but against the foreword's spirit)

The scheme: "For a body-text statement supported by an internal study report, write only 'data on file' in the reference column. The guidelines don't require disclosing that the source is inaccessible externally or that it is an in-house document."

Verdict: Spirit violation. "Data on file" does not function as a traceable reference. Readers have no path to verify the claim, which contradicts the foreword's requirement that the summary enable readers to return to source documentation. When citing internal data, disclose it as "Internal study report (study number, completion year); enquiries: MSL/MR contact point."

GRAY 2 — Exploiting the gap (deliberately working a loophole to invite misreading)

The scheme: "Include author, journal, volume, issue, pages, and year, but reduce the author field to 'Smith et al.' with only the first-author surname. Identification is 'not impossible' but requires significant effort."

Verdict: Gap exploit. PubMed and similar databases require a combination of author name, journal, and year to isolate a specific paper; a first-author surname alone may match multiple articles by that author. The guidelines require traceability, and "not impossible" is not the standard. Record all authors (or first three + "et al."), journal name, volume, issue, pages, and year in full.

GRAY 3 — Violating higher norms (arguably "not in the text," yet almost certainly out under the PMD Act, Fair Advertising Standards, and the JPMA Code)

The scheme: "Deliberately omit the Phase III pivotal trial paper from the reference list and cite only Phase II supporting data — a paper that presents the product more favourably. 'References are included' remains technically true."

Verdict: Higher-norm violation. The Phase III trial is the direct basis of approval; omitting it hides the approval rationale from healthcare professionals. The Fair Advertising Standards explicitly prohibit omitting key trials relevant to approved indications. List references for every pivotal trial that underpins the approval.

BLACK — Breaching the explicit text (a plain violation of the provision itself)

The scheme: "Fabricate a citation that appears to support a body-text claim but that was never actually referenced."

Verdict: Explicit breach. The guidelines require that reference entries correspond to sources that actually underpin the body-text statements. Citing a source that was never reviewed constitutes false representation, qualifying as false advertising under the PMD Act with exposure to criminal penalties and administrative sanctions. Verify that every listed reference has a corresponding body-text passage.

Improper intent by implication — three scopes of "insinuating" without stating it

Item ⑭ (key references) exists solely to let any reader trace every claim back to its source. Yet through disguised attribution, mixing internal reports with peer-reviewed journals in an identical format, and strategically weighting the list toward efficacy, the list can misrepresent the depth and independence of the evidence without stating anything false.

Implication 1 — Local: carried in a single word or sentence

How it is built: "XX Pharmaceutical Research Institute (2022). Phase IIb Study Report. Internal report." → changed to "XX Pharmaceutical Research Institute (2022). Phase IIb Study Report." (the words "Internal report" omitted)

Intent that can be read: Removing the "internal report" designation makes the entry indistinguishable from an independently conducted and published study. A single omission disguises the source type.

Verdict: An in-house study report has not undergone independent peer review. The "internal report" identifier is mandatory information for verifiability. Its omission invites readers to assume external validation that does not exist — an implied credential borrowed through a single deletion.

Implication 2 — Adjacent: read across neighboring clauses

How it is built: "1. [Authors]. Randomised trial. New Engl J Med. 20XX. 2. [Authors]. Long-term safety. J Clin Oncol. 20XX. 3. XX Pharmaceutical Research Dept (20XX). Internal Study Report No. A-025."

Intent that can be read: Listing the internal report in the same numbered format immediately after two prestigious peer-reviewed articles implies that all three carry equivalent independence and methodological rigour.

Verdict: Uniform formatting serves readability, not equivalence. The substantive difference in source type must remain visible. Internal reports should be listed in a clearly distinct sub-section or accompanied by a notation identifying their status. Adjacent formatting that implies equivalence is an implied credibility transfer.

Implication 3 — Whole-context: arising from the document as a whole

How it is built: Of 20 references listed, 14 relate to efficacy and are placed first; 6 addressing safety or adverse events appear in the latter portion of the list, with less descriptive context.

Intent that can be read: The numerical preponderance of efficacy references, combined with their lead position, constructs an overall impression that "efficacy is extensively substantiated, safety concerns are thin" — amplifying the efficacy narrative of the body text through volume and order.

Verdict: Key references must comprehensively correspond to body text claims, and the composition ratio must not be strategically manipulated. Placing safety literature at the rear turns the reference list into an implied efficacy advertisement. References must be distributed to match all body text claims with equal traceability.