01The foreword is the constitution of the whole guide
The body of the creation guide does not begin with Part I and the product information summary. The opening foreword, "Introduction," is the higher-order norm that runs through every provision that follows. Skip it, and you lose sight of why each clause reads the way it does. Hold onto its spirit, and you keep a compass even where the guide says nothing explicit.
The foreword states several things plainly. Companies have a duty to convey accurate information to healthcare professionals. Conveying it without misleading the recipient matters as much as the accuracy itself. The basis for proper-use information is the electronic package insert; the product information summary and similar materials merely supplement it. Media have shifted from print to digital and the Web, yet the principles to uphold have not changed. The guide sets out basic matters rather than covering every territory, and anything outside its provisions still falls under the Pharmaceuticals and Medical Devices Act, the Standards for Fair Advertising, the sales information provision guidelines, and the JPMA Code. Materials for additional risk minimization under an RMP are handled separately, by a voluntary agreement.
Read the foreword as a mere preface and you will misread the structure of the guide. Each sentence here is the basis for the detailed prohibitions that come later. The single word "supplement," for instance, is the starting point for the strict rule that a summary may not exceed the approved scope by even one character.
02Five implications the foreword carries
The foreword does more than state; it carries operational implications between the lines. Putting them into words sharpens how you read the individual pages. Here they are, broken into five.
① "Supplement" builds the wall of the approved scope
The product information summary sits below the electronic package insert. The insert is the approved content itself; the summary is a document that complements it. For exactly that reason, the summary cannot exceed the approved scope by even one character. Indications and dosage may be written only within the approved frame. This single point is the source of the many "must nots" that follow.
② "Accurate and not misleading" is two separate duties
The duty not to lie and the duty not to mislead are not the same. Every word may be factually true, yet if the presentation leads the reader to form a false picture, that can be a violation. How an axis is drawn, where emphasis is placed, the order in which things are arranged. The demand is to manage not only the truth of the facts but the impression they raise.
"We wrote nothing false" is no defense. Closing off expressions that are factual yet misleading is one of the foreword's heaviest assignments.
③ "Not exhaustive; the unwritten still applies" forbids treating it as an answer key
Not written in the guide does not mean permitted. The guide sets out only basic matters, and unwritten territory is governed by the higher norms. This is also a device that keeps the guide from shrinking into a checklist of banned acts. It forecloses, from the start, the notion that dodging the letter of a clause gets you through.
④ The evolution of media is an assignment to re-translate print practice
Media have moved from print to digital and the Web. But because the principles are the same, the craft built up in print must be translated into the new media. How do print assumptions — character size, placement, ease of reference — get implemented in a world of screens and links? The guide devotes later chapters to each medium precisely to make this re-translation concrete.
⑤ RMP materials are tools for safe use before they are tools for selling
That materials for additional risk minimization under an RMP are handled in a separate frame reflects the true character of materials. A material is a tool for getting a product used safely before it is a tool for selling that product. This order of priority flows into the whole guide through the way safety information is handled.
03Understanding the hierarchy of scientific evidence
When the guide says "accurately," that accuracy rests on the quality of the evidence. The same result — "it worked" — carries entirely different weight depending on the study design behind it. Generally, meta-analyses and systematic reviews sit at the top, followed by randomized controlled trials, then observational studies (cohort and case-control), then case reports, and finally expert opinion and review articles.
The watershed of quality is peer review. Whether a claim has passed verification by independent experts changes its credibility greatly. And results from in vitro experiments or animal studies must not be tied directly to the human clinical setting, because of differences in species, dose, and experimental system. The guide repeatedly requires the notation "(animal species)" or "(in vitro)" to fix this hierarchy awareness into the writing itself.
04Trial design determines the weight of a conclusion
Before the numbers, ask what design produced them. This is the basis of reading a clinical trial.
Devices that even out bias
Randomization evens out both known and unknown bias across the two arms at once. Double blinding, by hiding the allocation from both patients and assessors, keeps expectation and subjectivity from seeping into the result. The choice of comparator carries meaning too: a placebo control shows the absolute effect of the drug, an active control shows its standing among existing drugs. What you can read off the trial changes with which one was placed.
The weight gap between confirmatory and exploratory
A trial may be confirmatory — going to confirm one hypothesis fixed in advance — or exploratory — going to search for the hypothesis itself. The two bear entirely different weight as conclusions. A relationship seen in an exploratory trial is a "candidate to be verified next," not an established fact.
05Discipline of analysis — what it means to decide in advance
Even with the same data, a conclusion bends if the craft of analysis breaks down. The guide is strict on analysis because the room for arbitrariness here is large.
One primary endpoint, in principle
The primary endpoint to be confirmed is, in principle, narrowed to one in advance. Test many endpoints at once and the probability that one of them turns out "significant" by chance swells. To avoid this multiplicity, what counts as the primary endpoint is decided before the trial begins.
Pre-specified, or post hoc
An analysis written into the plan in advance (pre-specified) and one done after seeing the data (post hoc) are treated differently. A post hoc analysis should state, at its outset, that it is post hoc and why, and be presented modestly. A re-analysis that differs from the original plan is treated as post hoc even if it has been published.
Subgroups and ITT
Subgroup analyses are treated as exploratory, limited to those planned in advance, and presented alongside the whole-population result. To avoid the trick of dropping discontinued cases to improve appearances, the ITT principle (analyzing all randomized cases) is valued. A meta-analysis must follow the procedure of a systematic review.
"Searching afterward for a convenient angle" is what the scientific procedure guards against most. The seemingly plain discipline of deciding in advance is what supports the trust in a conclusion.
06Do not mistake what the statistics mean
Each statistic answers a different question. Mistake the question, and a false conclusion is drawn from a correct number. The table below sorts the main measures by "what they say and what they do not."
| Measure | What it says | What it does not say |
|---|---|---|
| p-value | Probability of a difference this large or larger by chance, assuming no difference | The size of the effect or its clinical value |
| 95% confidence interval | The likely range of the effect (precision of the estimate) | Clinical significance asserted on its own |
| Hazard ratio | The ratio of the relative pace of events | The real picture without the CI and absolute difference read together |
| Nominal p-value | Any p-value from analyses other than the pre-specified confirmatory analysis (reference only) | A basis for a confirmatory conclusion |
The relationship between statistical significance and clinical significance is especially easy to misread. In a large trial, even a clinically tiny difference can reach statistical significance. Conversely, the absence of a significant difference does not mean "equivalent" — it may only mean the power to detect a difference fell short. A p-value is only a comparison against chance; it says nothing about effect size or clinical value itself. A hazard ratio, too, cannot be grasped without reading the confidence interval and the absolute difference together. A nominal p-value is any p-value from analyses other than the pre-specified confirmatory analysis, and cannot serve as the basis for a confirmatory conclusion.
07Three pillars running underneath — read them as design philosophy
Gather the foreword and the discipline of science together, and three design ideas run underneath the whole guide.
First pillar — closing off "factual yet misleading"
Beyond forbidding falsehood, close off misdirection through how facts are shown. This is implication ② carried straight into implementation. The fine restrictions on graph emphasis and selective extraction all derive from this pillar.
Second pillar — implementing balance as structure
The balance of efficacy and safety is secured by structure, not by good intentions. Make the safety text size equal to or larger than the efficacy text, show the page that serves as basis, hold reference information to a set share of the page. Such concrete specifications drop balance into "implementation." Safety in particular is imposed as an asymmetric duty: it must be disclosed even when unfavorable to the company. The very safety information that is inconvenient is what must not be hidden.
Third pillar — securing verifiability through structure
Sources, statistical methods, the date of creation or revision. By always requiring these to remain, the guide keeps a state in which a third party can later retrace a claim. Building verifiability in as structure is itself the guarantee of trust.
The order in which to unpack the creation guide starts from this foundation. The foreword, as a constitution, sets out "what must be protected"; the discipline of evidence and statistics gives "how to verify it." The fine clauses from Part I onward are merely the result of translating these two into individual media and items.
So the place to return to when in doubt is not a clause number but the spirit of the foreword. Do not exceed the approved scope. Do not mislead even with facts. Govern the unwritten by higher norms. Disclose safety even when unfavorable. Leave it in a verifiable form. Carry these five, and you are unlikely to misjudge even in situations the guide never spells out.