Ch.1 Basic considerations | JPMA Guide Commentary | Advertising Regulation in Japan (Promotional Material Review) and Artificial Intelligence

Chapter 1, "Basic Considerations," is a single chapter within Part I, yet its character differs from the rest. Where Chapter 2 handles the sixteen items of the Comprehensive Product Information Summary and Chapter 3 the narrowed-down version, this chapter sets out the premises that apply equally to every material. Comprehensive or specific-item, journal advertising or any other material — unless the promises made here are met, nothing else can proceed. In the "foundation → I → II → III" descent, this is the ground beneath your feet.

The chapter has two sections. Section 1 covers considerations for the text and wording itself (twenty items); Section 2 covers the handling of data, figures, and tables (five items). The former asks "what to say, and how to word it"; the latter, "how to present numbers and graphs." Both translate the spirit of the foreword — the package insert is the original and the summary merely supplements it; do not just avoid falsehood but avoid misleading; govern even unwritten areas by higher norms — into concrete working rules. Many provisions carry a [detailed rule] that pushes an abstract principle down to the line of "how far you may go." This chapter reads at the level of those detailed rules.

01Seven questions that run through the twenty items

The twenty items of Section 1 are not a scattered list of prohibitions. Pressed to their core, they converge on a few questions: Is it scientifically correct? Are efficacy and safety balanced? Does it stay within the scope of approval? Is reference material kept out of the main line? Is the presentation free of exaggeration? Does it avoid disparaging competitors? Is it aligned with the latest facts? Below, the twenty items are regrouped along these seven axes, descending into the specifics of the detailed rules attached to each.

The reason for going down to the detailed rules is simple. No one objects to "write accurately, fairly, and objectively" as a generality. The problem is that even a well-meaning author has a boundary somewhere that gets crossed unconsciously. The Guide names that boundary at a physical level — the way you quote, the size of the type, the axes of a graph. So this chapter, too, does not stop at generalities; it traces those lines one by one.

02The discipline of quotation — eight lines that keep a guideline undistorted

Item 1 requires content to be "based on scientific grounds, accurate, fair, and objective." It looks like a generality, but the attached detailed rule trains on quotation — above all the quotation of academic diagnostic and treatment guidelines — and draws eight concrete lines. Quotation is, by nature, the act of backing one's own claim with third-party authority, and for that very reason a single choice of excerpt can distort the reader's picture. That is why this item carries the densest detailed rule.

Detailed rule: what to observe when quoting

Quote verbatim. Quotation from commentary that includes figures and tables must faithfully reflect the guideline's intent.
Do not quote only the passages favorable to your own drug. When quoting from several places, make the quoted locations explicit so that this is apparent.
Do not emphasize the parts relevant to your own drug with color or bold type.
Do not include a guideline that contains off-label statements about your own drug. Where a competitor's off-label statement is included, reproduce it unaltered, annotate that it is off-label, and do not word it so as to recommend off-label content.
Do not let it become disparagement of competitors or their products.
Where comparable guidelines exist in Japan and abroad, quote the domestic one in principle. An overseas guideline may be translated, but accurately.

Spelling out "do not emphasize with color or bold type" matters because a technique can leave a false impression without containing a single lie. Without changing a single character of the original, bolding just the one sentence favorable to you skews where the reader's eye and memory settle. The detailed rule on quotation is the most elementary implementation of the foreword's idea of closing off the "true but misleading."

03Balance and the scope of approval — guarded by the physical quantity of type size

Item 2 addresses the balance of efficacy and safety; items 3 and 9, the scope of approval. Both are places where the central idea of the foreword is made concrete — on the page and in approval information respectively.

Detailed rule: balance of efficacy and safety

If you state efficacy in clinical results, you must also state safety. And the safety description must be in type the same size as, or larger than, the body text presenting efficacy. Reaching down to the physical quantity of type size is not a matter of formatting. It is a device to keep the reader's first gaze from being drawn away from safety. The asymmetric obligation — safety information is disclosed even when unfavorable to the company; item 11 goes so far as to require that "important safety information be recorded even if unpublished" — is guaranteed through the design of the page itself.

The scope of approval — not a step beyond, in either efficacy or dosage

Neither efficacy nor dosage may be written outside the approved scope. Conditions that limit the scope (so-called qualifying expressions) must be copied accurately so that the approved indication, including those conditions, is conveyed correctly. Even where dosage says "adjust as appropriate," stay within the explicitly stated range. Electronic media must make content clearly understandable, just as print does. The root of this restraint is one word from the foreword: "supplement." The summary is a document subordinate to the package insert, and to exceed the scope of approval by even a single character is no longer supplementing but departing.

04Quarantining reference information — separating secondary findings, QOL, and pharmacology from the main line

Results obtained secondarily in treatment within the approved scope, and pharmacological actions whose relation to the indication is not sufficiently clear, are kept clearly distinct as "reference information" and not allowed to mislead about the indication. And item 8 draws the line: "reference information" must not be written as a feature/characteristic. The aim is to prevent a secondary finding from looking, in the reader's mind, continuous with the principal indication. The idea is to make the grading of information visible on the page.

Detailed rule: handling daily activity and QOL

Daily activity and QOL are, in principle, treated as reference information. However, where the definition of the evaluation index or score is clear and generalized, it does not fall under reference information — yet even then, no expression may mislead about the efficacy. "Reference information or not" decides not only where the finding sits within the material but also the ceiling on how strongly that finding may be stated.

05The ban on exaggeration and disparagement — closing off the step beyond falsehood

Items 10 through 16 name and foreclose a family of expressions that are not false yet still mislead. The ban on exaggeration is a demand one step beyond the ban on falsehood — it targets techniques that leave a false impression without containing a single lie.

Prohibited expression	Concrete example (from the detailed rules)
Emphasizing/guaranteeing safety	Do not emphasize or guarantee "it is safe." Do not write anything inconsistent with the warnings, contraindications, and other precautionary information
Appeals to authority	Do not create material whose layout is built mainly around portrait photographs of healthcare professionals and the like
Linking nonclinical directly to clinical	Do not phrase animal or in vitro results as directly connected to clinical efficacy or safety
Generalization	Do not present exceptional, limited data as though it were a general fact
Appeals to dignity/emotion	Do not use expressions that provoke anxiety, fear, or discomfort, that damage the credibility of the drug, or catchphrases, photos, or illustrations that impair dignity

Detailed rule: statements that lead to disparagement

No statement may lead to disparagement of competitors or their products. In clinical comparative trials the line is concrete:

The control drug's trial results may be presented, but their evaluation or interpretive commentary must not be written.
Do not write anything emphasizing that a competitor's product was ineffective, insufficiently effective, or not tolerated.
Do not provide commentary on prior therapy (a competitor's product).
In the trial outline, record as accurately as possible the drug names used (competitors by generic name), the dosing period, dose, and number of cases.

"The control drug's results may be shown, but not commented on" — this is the crux. Comparison as a fact belongs to science; the judgment of superiority belongs to advertising. Separating the two keeps comparative data from becoming a tool of self-praise or disparagement. Share the fact; leave the evaluation to the reader.

06Alignment and updating — binding a material that ages the moment it is printed

Items 17 through 20 address the axis of time. For a new drug, write with full regard to the deliberations of the Pharmaceutical Affairs and Food Sanitation Council; where conditions or instructions were attached at approval, keep consistency with the related items. Align the material with the latest package insert, review report, and re-examination/reevaluation results, and where particularly important matters among efficacy, dosage, and the warnings/contraindications precautionary information are revised, promptly revise the summary. And comply, as a matter of course, with related laws such as the Pharmaceutical and Medical Device Act and the Standards for Fair Advertising of Drugs, and with self-imposed norms such as the JPMA Code and JPMA notices.

A material begins to age the moment it is printed. Mandating synchronization with the latest version supports, on the operational side, the foreword's idea of verifiability — keeping the point in time of the information traceable after the fact.

07Section 2, data (figures and tables) — which results may be presented

Section 1 handles words; Section 2 handles numbers and graphs. This is a domain where the discipline of science becomes almost directly the rule, divided into five: reliability, accuracy, statistics, charting, and quotation. First the entrance — reliability, which decides "the range of data you may present."

Detailed rule: securing reliability

Statistical results may be presented only for a pre-specified analysis with scientific validity. To present primary/secondary endpoints or subgroup analyses, the statistical method must have been stated in the analysis plan.
For a post-hoc analysis (= an extra analysis thought up only after the trial has ended) where the information is nonetheless important (a result suggesting a risk factor for a serious adverse reaction; a tabulation used as evidence of efficacy in, say, an orphan drug (= a drug for a rare disease with very few patients) where statistical significance is hard to show), state at the head of the relevant results that it is post-hoc and why it is shown, and use restrained wording.
A meta-analysis (= a method that pools the results of several trials into a single evaluation) must be a systematic review (= a method that gathers the literature comprehensively by a pre-set procedure and evaluates it). Record the search source, search keywords, the number of records identified, the total number of records assessed for eligibility, and the number excluded with reasons.
When a registration-stage clinical trial is published with a re-analysis different from what the original analysis plan specified, that result is post-hoc even though published, and must not be used.

The assumption that "since it is published, it ranks with a pre-specified analysis" is the pitfall. Whether it passed peer review and whether the analysis was planned in advance are separate matters. An analysis cut out conveniently after the fact remains post-hoc even in print — mistake this single point and you speak of the result's certainty beyond what it is. The statistical discipline of strictly separating confirmatory analysis, which checks one pre-specified hypothesis, from exploratory analysis, which searches for hypotheses, here becomes the very condition for inclusion.

Detailed rule: accuracy (do not distort)

Do not intentionally distort the underlying information. Do not run various analyses and present only the result convenient to your interpretation. Do not crop the figures and tables of an original paper so as to show only the part favorable to you. It sounds obvious, yet a single choice of graph axis or excerpted interval moves the impression greatly. The principle of closing off the "true but misleading" appears here as the discipline of charting and presentation.

08Section 2, statistics and charting — binding method, limits, and scale

From item 2 onward, the writing of statistics and the making of graphs are specified concretely. To present a statistical result, accompany it with the statistical method and its result (confidence interval, p-value, etc.), and where a significance level other than two-sided 5% is used, state that level too.

Detailed rule: when statistics are recorded

If you adjusted for covariates (= patient-side conditions that can affect the result, such as age, severity, or medical history; background factors, prognostic factors, etc.) or stratification factors (= factors set in advance as the basis for dividing patients into groups), state them in the results.
If you used a statistical model without a common name for the test or estimation, state the model formula (= the calculation formula expressing the relationships in the data) in the results.
If you imputed missing values (= data that could not be measured or are absent) — that is, filled them in by estimation — for tabulation or analysis, record the imputation method in the results.
Subgroup analyses, being largely exploratory, are limited to those written into the trial plan from the outset and scientifically valid. Where a whole-population analysis exists, present the subgroup result together with the whole-population result.

Mistake what a statistic can and cannot say, and the same number turns into an overreaching conclusion. The Guide requires method and limits side by side precisely to prevent that confusion on the page.

Statistic	What it can say	What it cannot say (typical misreading)
p-value	The probability that, assuming no difference, a difference this large or larger arises by chance	The size of the effect or its clinical value. A small p does not mean a large effect
95% confidence interval	The range the effect is likely to lie in (precision of the estimate)	It does not guarantee the effect at the single point estimate
Statistical significance	That a difference is hard to explain by chance alone	Clinical meaning. In large trials a tiny gap can be significant, and "no significance" is not proof of equivalence
Nominal p-value (= a p-value from an analysis outside the pre-specified plan, treated as reference only)	A p-value obtained from any analysis other than the pre-specified confirmatory analysis	A confirmatory conclusion. It cannot, on its own, support "it was proven"

Detailed rule: showing results in graphs and tables

Charting carries concrete lines that close off manipulation of impression.

When showing numbers, state in a clear place whether they are means, medians, geometric means (= an average taken by multiplication, used for data that move multiplicatively, such as ratios or concentrations), or least-squares estimates (= an estimate computed by a statistical model so as to minimize the scatter).
When showing the presence or absence of significance in a graph or table, state the statistical method in a clear place.
Do not merge separately obtained data from differing trial conditions into the same graph or table.
Do not chart so as to emphasize a difference, such as by altering the vertical or horizontal scale more than necessary.
In figures comparing against a control (including placebo) or showing before/after, do not use arrows or the like to emphasize the difference. Do not emphasize differences with type size or color either.
Do not embellish the size of a difference with adjectives that have no basis.

One line weighs especially heavily: where no significant difference was found, or where no statistical analysis was performed, present only the numerical results. Add an arrow or an adjective where there is no difference, and a picture of "a difference exists" rises up. So the very ornaments that speak of a difference are banned, leaving only the numbers. With the same data, blowing up the relative risk reduction alone looks dramatic, while placing the absolute difference alongside conveys its realistic size. The fine requirements of Section 2 are a bundle of brakes against the author drifting (even unintentionally) toward the former.

Detailed rule: quotation from original papers

When quoting data from an original paper, record it so the content is conveyed accurately, do not excerpt only the part where the conclusion favors your product, take care not to harm the original's true intent, and state the source. The same thinking as the Section 1 quotation rule (section 02) is repeated once more in the world of data — cropping distorts the picture without changing a single character of the original.

In closing

Chapter 1 is not a parade of dramatic prohibitions. Be scientifically correct, balance efficacy and safety, stay within the scope of approval, quarantine reference material, do not exaggerate, do not disparage competitors, stay aligned with the latest — and attach method and limits to your numbers. The twenty items and their detailed rules, and Section 2's reliability, statistics, and charting, are all nothing but concretizations of the foreword's three pillars: closing off the "true but misleading," implementing balance on the page, and guaranteeing verifiability through structure.

This foundation unfolds in Chapter 2 (Comprehensive Product Information Summary) into the detailed rules for each item, and in Chapter 3 (specific-item) into the discipline of the narrowed-down version. The related ideas also run through the Standards for Fair Advertising (/en/compliance/03-ad-standards.html) and the JPMA Code (/en/compliance/06-jpma-code.html); Chapter 1 is the first gate that brings those principles down into "how to make a material."

Source & DisclaimerThis page is a study note explaining the intent of the Japan Pharmaceutical Manufacturers Association (JPMA), "Guidelines for the Preparation of Product Information Summaries for Prescription Drugs (with commentary)," October 2023 edition (original PDF). The individual provisions of the Guidelines are summarized and paraphrased based on that document, the copyright of which belongs to the JPMA. This page is not an official position of the JPMA and does not guarantee the accuracy or currency of its content. Always consult the original document before applying it in practice.

Pages in this level

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

Chapter 1's Basic Precautions translate the principles of citation, balance, approval scope, segregation of reference information, and prohibition of exaggeration into concrete work rules. The cases below trace the reasoning of those who believed those rules could be evaded by a careful reading — and explain why each reading fails.

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

The scheme: "I quoted one sentence from the guideline verbatim — no bold, no colour, source noted. The rule says 'don't quote only favourable passages,' but that must mean: if you quote from multiple places, flag each. A single self-contained sentence drawn from one place is fine."

Verdict: The rule against quoting "only favourable passages" does not carve out an exemption for single-sentence citations. When a guideline sets conditions or warnings in other sections and only the favourable efficacy line is lifted, readers assume the guideline endorses the product without reservation. The supplementary rule about flagging multiple-source citations does not imply that single-source quotes need no disclosure. The overarching requirement to "faithfully reflect the intent of the guideline" bars selective extraction as a matter of spirit.

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

The scheme: "Nothing says the y-axis must start at zero. The prohibition is on changing the scale 'more than necessary.' If I can argue the scale is objectively necessary, starting the axis at 50% is fine even if the gap looks huge."

Verdict: The provision bans "graphs that emphasise differences by, for example, altering axis scales more than necessary." The test is the impression left on the reader, not the producer's characterisation. A non-zero baseline that exaggerates a difference visually violates the rule regardless of a self-serving "necessary range" label. The companion rule against using arrows to highlight differences in comparison graphs reflects the same principle: layout choices that amplify perceived effect size are prohibited even when the underlying numbers are accurate.

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: "I matched the safety font size to the efficacy font size exactly. The Guidelines say nothing about font weight or colour. So using bold coloured text for efficacy and light grey text at the same point size for safety is compliant."

Verdict: The Fair Advertising Standards prohibit making safety information "difficult to understand," addressing legibility as a whole rather than point size alone. Manipulating weight and colour to suppress safety information's visual prominence renders it practically invisible, regardless of matching point size. The JPMA Code's requirement for "balanced information provision" likewise covers visual asymmetry. Achieving the same effect through means other than font size while satisfying the letter of the point-size rule defeats the rule's purpose.

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

The scheme: "The post-hoc subgroup result was strong, so I included the figure without the note that it is post-hoc. A reader can tell from the subgroup label, and adding an ugly disclaimer at the top hurts the layout."

Verdict: Chapter 1's supplementary rules require that post-hoc analyses included in materials must carry, at the head of that section, an explicit statement that it is post-hoc and a reason for its inclusion, and must be presented in restrained language. Omitting the header statement fails all three conditions. "Readable if you look closely" is not an acceptable substitute for "stated at the head." The material must be revised.

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

Chapter 1 imposes its rules on every section of every material — quotation, balance, approval scope, isolation of reference information, and prohibition of exaggeration — at the level of working procedures rather than mere principles. The three patterns below show how a writer who believes "the regulation cannot reach what it does not name" ends up implying what the rules forbid. No individual sentence contains a prohibited word; no single passage looks clearly wrong. The improper intent surfaces only when the document is read as a whole.

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

How it is built: "A favourable passage from the guideline is quoted verbatim; the sentence is closed with the connective 'Notably,' followed by one sentence adding context. Slightly wider spacing separates this passage from the surrounding text."

Intent that can be read: The word "Notably" signals supplementary comment rather than a selective excerpt. The connective creates the appearance that the quoted passage is simply a note clarifying the main guideline position, concealing the fact that only the favourable clause was extracted.

Verdict: Chapter 1 requires that quotations "faithfully reflect the intent of the guideline." The choice of connective does not change what was omitted. Extracting one favourable clause while a separate section of the guideline sets out conditions and cautions is not remedied by the word "Notably." The correct approach is to treat favourable and unfavourable passages equally, quoting in a way that conveys the overall intent of the guideline.

Implication 2 — Adjacent: read across neighboring clauses

How it is built: "The incidence of adverse events was comparable to that of the comparator (p=0.18). Patients tolerated the treatment well, and the trial completion rate was 98.3%." — these two sentences appear in the same paragraph without a line break.

Intent that can be read: Following "no significant difference" with "tolerated well" and "high completion rate" implies that the adverse-event data have put safety concerns to rest. "Tolerated well" is a claim about the mildness of adverse events, not a neutral relay of facts.

Verdict: Chapter 1, item 2, prohibits emphasis on or assurance of safety. Placing "tolerated well" immediately after an adverse-event rate turns a statistical fact into a safety assurance through adjacency. The completion rate and the absence of a significant difference are facts that can each be stated, but chaining them into a sequence that leads to the conclusion "adverse events are not a problem" is not permitted. The correct approach is to record safety and efficacy as separate facts, without evaluative language.

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

How it is built: "Efficacy data occupy the first three pages with detailed tables and figures. Safety is addressed only at the end: 'For the full safety profile, please refer to the electronic package insert.' The product name and approved indication are repeated in the header of each page."

Intent that can be read: The document's centre of gravity is placed on efficacy, and safety is treated as information that needs only a pointer to another document. The cross-reference to the package insert is used to discharge the obligation in Chapter 1, item 2 — that safety must be shown in type no smaller than that used for efficacy — without actually doing so. The reader's overall impression is of a drug that works.

Verdict: Chapter 1 requires that when efficacy is presented, safety information must appear in type at least as large. A single sentence directing readers elsewhere is not a substitute for that obligation. A design in which efficacy spans multiple pages while safety fits into one line fails the balance requirement of Chapter 1, regardless of whether each page's text is individually correct. The correct approach is to give safety and efficacy equal weight within the document and to include important safety information in the body text.