Own the Image That Lands — Accuracy on Paper Is Not Enough

Every number in the material was accurate. Every figure was taken straight from the published paper. And yet the image that remained in the physician's mind was not what we intended. Cherry-picking incidents repeat every year, and the reason may be that creators are protecting only the accuracy of what sits on paper. This piece goes one step further. Using correct data is the baseline — but integrity means owning the accuracy of the mental image the reader takes away. That is the core of this commitment.

So What / So Why — Why This Commitment Matters

"It was accurate on paper." That phrase comes up again and again in case reviews of cherry-picking incidents. The graph values matched the published paper. The cited trial was real. But the image left in the reader's mind was not accurate.

Consider one case documented in the March 2019 Advertising Activity Monitoring Project report. At an in-hospital product briefing for a bronchial asthma treatment, results from the full analysis set (approximately 250 patients per arm) were withheld. Only the Japanese subgroup (approximately 15 patients per arm) appeared in the slides and brochure. The data were genuine. But the image that stayed with the physician was "this drug works especially well in Japanese patients" — the full picture from the overall analysis never arrived. When the representative was asked to explain, the answer was "physicians ask for Japanese data." No lie was told. But the image was distorted. (See Analysis Series vol.02: Selective Extraction, Manipulation, and Presentation of Data and Graphs.)

Prescribing decisions are driven by impressions, not numbers. After reviewing a material, physicians rely on memory traces rather than data specifics — "that trial showed good efficacy," "the safety profile seemed manageable." Those traces shape the next prescription. Who controls their accuracy? Only the people who created the material. Physicians do not go back to the original paper or regulatory review report every time. That is precisely why creators must take responsibility for the accuracy of the image that is conveyed. Accuracy on paper is a necessary condition. It is not sufficient.

The Commitment — How I Want to Work

I want to treat data accuracy as the floor, not the ceiling — and not stop there.

Even when numbers are correct, a one-directional presentation leaves a skewed image in the reader's mind. A physician who sees a slide where secondary endpoints (predefined supplementary measures) are displayed more prominently than the primary endpoint (the central efficacy measure of the trial) may never learn that the primary endpoint showed no significant difference. That physician is not at fault. The time and attention available for prescribing decisions are finite; checking the regulatory review report every time is not realistic.

So I want to think of myself not as a persuader but as a custodian of evidence. A persuader selects information to make the audience believe. A custodian of evidence checks whether the full set of information the audience needs to make a judgment has been delivered accurately. The distinction looks subtle, but in practice it reverses the direction of decision-making.

Doubt arises in the middle of real work. A material cannot be infinitely long. Treating every piece of information with equal weight is sometimes impossible. Choices about emphasis are unavoidable. When I face those choices, I want to ask myself: "Is this emphasis here to help the reader make a prescribing decision, or to secure adoption?" The two often align, but they diverge. When they diverge, which do I choose? I want to stay with that question rather than sidestep it. I am not claiming perfect achievement. I am committing to keep asking.

Daily Practice — Concrete Actions and Checks

• The "read it backward" test: Once a material is finished, show it to a colleague unfamiliar with the product for thirty seconds and ask, "What impression did you get of this drug?" Any gap between that answer and your intent signals that the received image diverged from the design.
• Check the visual weight of the primary endpoint: Is the secondary endpoint graph larger or placed in a more prominent position than the primary endpoint? Placement and size directly shape the image.
• List what was cut: Write down information that did not make it into the final version. Examine whether each cut was made because the reader did not need it or because it was inconvenient. If the latter, stop.
• Articulate why the comparator was omitted: If data for a comparator drug were not included, practice explaining that reason to a manager or review officer. "It was easier to read this way" is not a reason.
• Measure the time balance between efficacy and safety: Track what proportion of a product briefing was spent on efficacy versus safety information. A large imbalance in the presentation produces the same imbalance in the image the reader retains.
• Ask what data were withheld: What is the next thing a physician who sees this material would want to know? How much of that is covered? If something is missing, what is the reason?

When It Gets Hard — Under Pressure

"The physician appeal is too weak" — that phrase erodes this commitment.

When a regional manager checks share progress at the end of a quarter, the impulse to make materials "hit harder" surfaces. The adjustments that follow tend to run in one direction: enlarge the favorable subgroup, quietly shrink the comparator arm, make the secondary endpoint more visually striking than the primary endpoint. Each decision is small. But they accumulate, and the result is a systematically skewed image in the reader's mind.

The motivation of "wanting to get a good medicine to patients" is legitimate in itself. But the stronger that motivation becomes, the more it pulls toward selecting only the data that support the prior conviction that "this drug works." A sense of mission slides into confirmation bias, and the transition is hard to notice — because there is no feeling of having lied. This psychological dynamic is examined in detail in Under Pressure Series vol.02: The Belief Trap — How 'I Want to Get a Good Drug to Patients' Becomes the Entry Point.

When pressure begins to wear down this commitment, I return to a single question. When an inner voice says "it is correct on paper" or "I have not lied," that voice may be the signal that a line is about to be crossed. Accuracy is a necessary condition, not sufficient. When someone notices that the image being conveyed is skewed but continues because "the numbers are real," they are moving away from integrity. That is where conscience and honesty diverge from technical correctness.

What I Return To — When Resolve Starts to Bend

"Provide accurate information based on scientific and objective evidence." That is the core principle of the Guidelines on Sales Information Provision Activities for Prescription Drugs (September 2018). Each time I read that sentence, I ask what "accurate information provision" actually means. If data are accurate but the image is not, the information provision cannot be called accurate.

The Guidelines go further: "provide information covering both efficacy and safety," and "take care not to cause misunderstanding by providing only specific information." "Not cause misunderstanding" is a demand for accuracy in the conveyed image, not merely in the numbers. Concealing the primary endpoint while amplifying a secondary endpoint causes misunderstanding, even if every individual figure is correct. Removing a comparator arm from a graph, starting a vertical axis from a non-zero baseline — both cause misunderstanding in the same sense.

The fact that cherry-picking patterns have appeared in monitoring reports every year from March 2019 through March 2025 (see Analysis Series vol.02) shows that "using correct data" alone is not enough to prevent the problem. What that approach fails to stop, I believe, is a lack of sensitivity to what forms in the reader's mind.

Returning to the principle in the Guidelines and to the weight of the documented record. Returning to the position of custodian of evidence. That is my anchor. Conscience and a sense of responsibility ask "is the image that was conveyed accurate?" before they ask "is it accurate on paper?" Keeping that order — that is what practicing this commitment means.