When "I Want to Help Patients" Becomes Confirmation Bias

"I want to help patients get a good medicine" is a legitimate starting point for pharmaceutical information provision. But the stronger that conviction grows, the more the certainty that "this drug works" runs ahead of the data — and interpretation follows behind it. Framing a pharmacological mechanism as "safety evidence." Citing another institution's reputation as "evidence." Walking through only the favorable subgroup in detail. None of this starts from malice. Each deviation begins the moment mission becomes confirmation bias.

So What / So Why — The Core of This Issue

So what (what actually happens) — belief justifies "just a little." When strong product confidence is the starting point, guideline constraints look like barriers blocking information that should reach patients. What follows is local deviation: "just this one slide," "I didn't say it because no one asked." Each step feels small, within tolerable limits. But when these accumulate, the informational foundation on which physicians base prescribing decisions erodes, increment by increment.

So why (why this is serious) — when confirmation bias is active, disconfirming data becomes harder to see. The hypothesis that "mechanistically, this drug should be safer than alternatives" can look like something permissible to say even without clinical trial evidence. Physicians incorporate the information they receive into prescribing decisions as truth. When a hypothesis becomes fact upstream of the prescription, even an excellent physician cannot protect the patient.

Across all seven annual reports of the Ministry of Health, Labour and Welfare's pharmaceutical information provision oversight and monitoring programs, "explanations without evidence" has ranked as the single largest or top category of suspected violations in each fiscal year on record (per the suspected-violation item tallies in individual reports). What most of those incidents share is not fraudulent calculation, but a structure in which people who believed in their product read data through the lens of that belief.

The Architecture of Pressure — Who, by What Deadline, for What Number

The external pressures that turn conviction into confirmation bias arrive from multiple directions at once.

Sales and share targets, quarterly quotas. A product just launched moves on expectations. If prescription volume doesn't grow in the first quarter after a new indication, the activity plan for the next quarter shrinks. Field MRs see those numbers every week. The moment a trade-off appears between securing a prescription and providing accurate information, the phrase "there are no comparative data" stalls in the throat.

Managerial approval and incentive design. Slides prepared by MSLs and MRs pass through managerial review before reaching the field. When the cost of sending something back — "this phrasing is too strong" — is weighed against the cost of silently accepting modest overstatement, the latter is lower in the short term. And where compensation is tied to prescription volume, accuracy of information competes directly with individual performance scores.

The KOL relationship. Words spoken by a prominent physician at a symposium or in-hospital meeting carry more weight than anything an MR says. "Professor X said the same thing" moves prescriptions. The problem is that the KOL's statement may itself be a personal opinion without supporting evidence — and when the company transfers that statement into a slide, the professor's authority substitutes for a rationale.

The squeeze from competitors, the gap with regulation. When a competing product has already told physicians "that drug class has more side effects," responding with only "we have no head-to-head data" sounds like tacit agreement. The desire for "at least comparable, if not better" builds up in the field as an undeclared gap with regulation.

The Inner Reconstruction — Creed, Sentiment, Deep Psychology

The creed — "The mechanism of action of this drug is scientifically sound. I understand that science. I am convinced this drug helps patients."

This is a legitimate starting point. Deeply understanding one's own product and taking pride in the pharmacological mechanism is, in fact, the genuine strength of a good MSL or MR. The problem is what happens when that creed becomes an unquestionable premise: the eye that selects data changes.

The sentiment — imagine being asked by a physician: "What makes this different from the alternatives?" You have no head-to-head trial. If you answer "there are no comparative data," the temperature in the room drops. You may lose the call. And that means the patient won't receive this drug — or so it feels. "But the mechanism makes logical sense. I want the physician to understand."

That sentiment is sincere. But sincerity, here, is precisely what justifies the logical leap.

The deep psychology — how four drivers operate

① Motivated reasoning (conclusion first, data second). When you hold "this drug has a superior safety profile" as a prior conclusion, the fact that it is not metabolized by CYP enzymes looks like evidence of safety. Whether that claim has been clinically demonstrated becomes a secondary question. The internal script runs: "The mechanism is different, so the clinical profile must be different too. Helping the physician understand that is my job."

② Local rationalization ("just this one slide"). Even if a single slide creates the impression of superiority over alternatives, the overall presentation also contains accurate information — so you tell yourself. "No one is going to take issue with one slide. I can supplement verbally." The existence of a broader context makes individual deviations invisible.

③ The sin of omission (not volunteering, waiting to be asked). The fact that "no head-to-head trial has been conducted" does not transmit unless actively stated. Having decided "I'll say it if asked," the absence of a question becomes absolution. But physicians bear no obligation to verify whether comparative data exist. The provider's silence creates a de facto falsehood. The internal script: "Since I wasn't asked, I have no obligation to say it. That is proper information provision practice."

④ Responsibility externalization (KOL, manager, "it wasn't requested"). "A distinguished professor recommended it" shifts the burden of proof from oneself to the KOL. Or: "My manager reviewed the slides, so there's no problem" — suspending individual judgment. "It wasn't requested" signals an unawareness of the duty to provide information proactively. The internal script: "What I conveyed was factual. I am not the one who chose it."

The Real Incidents Behind the Analysis

The three cases below show that the psychological analysis above is not theoretical. Each is a documented incident from the Ministry of Health, Labour and Welfare's FY2019 "Advertising Activity Monitoring Project for Ethical Pharmaceuticals" report.

① A mechanism of action recast as proof of safety — analgesic (FY2019)

Analysis vol.03 — "Explanations Without Evidence / Information Lacking Credibility" compiles all incidents of this pattern by fiscal year.

② One word — "STRONG" — standing in for a comparative trial — antiallergic drug (FY2019)

Analysis vol.05 — "Exaggerated Expressions" covers all incidents of catchphrase overstatement, authority transfer, and logical leaps.

③ "Physicians want Japanese data" as the stated rationale for data selection — bronchial asthma drug (FY2019)

Analysis vol.02 — "Arbitrary Extraction, Manipulation, and Framing of Data and Graphs" examines the structure of cherry-picking and endpoint substitution in detail.