A Map of Violations Drawn from 7 Years and 186 Cases — Why Ordinary Practitioners Err

In the seven years since Japan's Sales Information Provision Activity Guidelines took effect (2019), healthcare-institution monitors captured a cumulative total of 186 suspected violations. Three categories — claims without evidence, selective data manipulation, and omission of safety information — together account for roughly 60% of all cases. This overview maps the full picture through a year-by-category table, introduces the nine-category framework used throughout this series, and examines four psychological drivers that explain why ordinary practitioners commit violations.

So What / So Why — What This Series Is Really About

Japan's Sales Information Provision Activity Guidelines, enacted in September 2019, govern the content and methods pharmaceutical companies use when providing information to healthcare professionals. Since implementation, the Ministry of Health, Labour and Welfare (MHLW) has collected suspected violation cases each fiscal year through monitors placed at medical institutions nationwide, publishing annual reports under the "Prescription Drug Sales Information Provision Activity Monitoring Project." The cumulative total from FY2019 (Heisei 31) through FY2025 (Reiwa 7) is 186 cases (FY2019: 51; FY2020: 40; FY2021: 18; FY2022: 20; FY2023: 19; FY2024: 23; FY2025: 15).

These 186 cases do not represent a handful of bad actors. The majority arose during ordinary product briefings, promotional materials, and routine MR (medical representative) visits. Practitioners who genuinely believed they were helping physicians prescribe better drugs participated in violations without realizing it — that observation is the central concern running through this entire series.

Physicians rely on information from pharmaceutical companies when making prescribing decisions. When distorted information enters that channel, evidence-based treatments go unselected and adverse-effect risks go uncommunicated. Individual cases may be hard to see as harmful, but systematic distortion of information can cause harm at a population level. That is precisely why the guidelines and the monitoring project exist, and why cases are made public.

Year × Category Case Count Matrix

The annual reports classify each suspected violation into up to eight categories. Because category label wording was adjusted slightly across years, the table below uses standardized terminology based on content. In FY2023, three cases were counted under both "No Evidence" and "Disparagement of Competitors," creating a gap between the total-count and case-count figures (19 instances across 16 cases).

※ In FY2023, three cases were double-counted across categories ③ and ⑤. The cumulative total of 186 reflects instance counts, not unique cases.

By cumulative category count, ③ No Evidence (60 cases, 32%) ranks highest, followed by ⑥ Safety Downplayed (27 cases, 15%), ① Unapproved/Off-label (26 cases, 14%), and ② Data Manipulation (23 cases, 12%). These top four categories alone account for 70% of all instances. ⑦ COI Not Disclosed has been near zero since FY2021, likely reflecting the industry-wide adoption of COI disclosure as standard practice. ⑤ Competitor Disparagement shows an upward trend, with eight cases combined in FY2024 and FY2025.

A Map of the Nine Categories

Each category is examined in depth in the chapters that follow. Here is a consolidated overview of all nine.

Why Ordinary Practitioners Violate — Four Psychological Drivers

Reading through the cases, outright intentional deception is the minority. Most representatives acted under the belief that "showing this data helps the physician" or "I was honestly explaining the difference from a competitor." Understanding the four mechanisms that produce this distorted perception is the starting point for prevention.

① Motivated Reasoning

A representative holds the conclusion — "this drug is good for patients" — first, then selects only the data that supports it. Contradictory information is interpreted as an exception or an unusual patient subgroup and filtered out. Most of the "selective data omission" (category ②) and "claims without evidence" (category ③) cases arise through this mechanism. It is not malice but conviction that produces the violation.

② Local Justification

Decisions are justified at the level of "just this one slide" or "I was only answering the question," without recognizing the cumulative distortion in how a physician perceives the overall presentation. Prefacing an off-label topic with "as a general reference" before proceeding (category ①) is a classic instance. Each individual step appears legitimate; the aggregate result is a violation.

③ Omission Bias

Safety information goes unmentioned, with the rationale "I wasn't asked, so I didn't say" (category ⑥). Unlike active fabrication, simply not saying something feels less culpable. But the guidelines require balanced provision of both efficacy and safety information. Omission is also a violation.

④ Diffusion of Responsibility

"The physician makes the final call." "My manager approved these slides." "Other companies do the same." Shifting responsibility to others creates psychological distance from one's own decisions. In company-sponsored symposia where a speaker (physician) over-promoted the sponsor's product (category ⑧), the pharmaceutical side often also engaged in responsibility avoidance under the premise that "the speaker said it."

These four drivers combine. "For the patient (motivation)" → "just this data (local justification)" → "I don't need to mention adverse effects (omission)" → "the physician decides (diffusion)" — this chain completes a single violation. The chapters that follow trace how this chain appears in each of the nine categories through specific documented cases.

Cases — Representative Patterns by Category

This overview chapter does not document every individual case. Representative patterns for each category are sketched here; full detail is in the corresponding chapter. Case counts are based on the matrix above.