Sign the Whole — What Happens When "This Slide Is Fine" Compounds

"This graph alone isn't a problem." "I'm only adding one slide." Assembled slide by slide under those quiet reassurances, the finished document ends up as something no one can put their name on. Stacking individually accurate slides does not produce an accurate overall impression. The question this conviction asks is not "is each part correct?" but "can I sign the whole thing?" The opposite of piecemeal rationalization — that is the heart of this entry.

So What / So Why — Why This Conviction Matters

The core of the problem is straightforward: stacking individually accurate slides does not make the overall impression accurate. You included a slide for the primary endpoint. You included a slide for secondary endpoints. But the secondary endpoint graph was printed twice as large while the primary endpoint was tucked into a corner. Every slide is technically accurate, yet the impression a physician walks away with no longer matches what the trial actually showed.

Most of the deviations recorded across seven years in the Ministry of Health, Labour and Welfare's surveillance reports on promotional activities were not cases of outright lying — they were cases of selecting. Removing the control arm from an efficacy graph. Raising the threshold on an adverse-event list so that serious items disappear. Presenting only secondary endpoint figures in detail when the primary endpoint failed to reach significance (Series 1 vol.02: "Selective Extraction, Manipulation, and Presentation of Data and Graphs"). Each individual decision, examined in isolation, ends with "the data are real" and "we answered if asked." But look at the full bundle of information a physician receives and the bias is unmistakable.

Why is this so consequential? When a physician decides on a prescription, the information at hand is their starting point. The person who designed that bundle is the material author. Every choice about what to show — and what to withhold — accumulates into the evidence base a physician uses to judge. Few physicians go back to the regulatory review report themselves. In that sense, the overall impression a document conveys is a product of the author's choices, and that is precisely why the author must own the whole.

The Conviction — How I Want to Work

When I finish a document, I want to ask myself what impression a physician will be left with after the final slide. Not "is this graph accurate?" but "can I put my name on what this document as a whole communicates?"

Going through slides one by one and watching the complete document from start to finish are two different acts. Even after each slide has passed its own review, I want to look at the finished material as a single, connected piece of information. The sense that surfaces in that read-through — "has the primary endpoint been pushed to the margins?" "does the ratio of favorable to unfavorable data reflect what the trial actually found?" — that is something I want to take seriously.

Material production is divided work. Different people draw the graphs, build the slides, write the copy, and run the review. Inside that division, unless someone takes explicit responsibility for "what does this, as a whole, deliver to a physician?", the individual steps can each be accurate while the whole tilts. I want to be the person who holds the whole — not as a persuader, but as a steward of the evidence.

I will be honest about my own uncertainty: when a deadline is pressing, or a manager has already approved, the relief of "this is fine" is genuine. I want to make a habit of pausing that relief and asking myself, "can I sign the whole?" — before moving on.

Daily Practice — Concrete Actions and Checks

Below are the actions and questions I use to anchor "signing the whole" in actual work.

• Always save a full read-through for the end. Separate from slide-by-slide review, carve out fifteen minutes after completion to watch the document from the first slide to the last. Write one sentence describing the impression a physician would have after finishing it.
• Compare how primary and secondary endpoints are treated. Is the primary endpoint graph at least as large as the secondary endpoint graph? Does the oral-delivery script give the primary endpoint equal or greater weight?
• List what was not shown. Write down data that appeared in the original paper or regulatory report but did not make it into the material. Ask whether "it didn't fit the theme" is sufficient justification.
• Cross-check the adverse-event list against the source paper. Confirm that no threshold was raised and no serious items were rolled up or combined. Check whether anything removed happens to be the item most unfavorable to this drug.
• Ask: can I put my name on the whole? If this material were published and I were introduced as the author, would I be comfortable? If something feels off, trace that feeling to a specific location in the document.

When It Gets Hard — Working Under Pressure

This conviction is most at risk when deadline, division of labor, and managerial approval all land at once.

Imagine building slides immediately after a supplemental approval, in an atmosphere that says "the next six months are everything." Speaker program dates are already on the calendar. The option in front of you is to "add just one slide" to an already-reviewed deck. That one slide carries efficacy data from the original indication, even though the new indication reached only non-inferiority. "It's just one slide" does the work of minimization. The MHLW surveillance report has on record the phrase: "it is only one slide, but it conveys superiority over existing therapies." That phrase — "only one slide" — is a verbatim document of the moment responsibility for the whole disappeared.

In divided work, graph sizing becomes a design team decision; copy phrasing becomes a brand team decision. Each person finishes their own section and no one is watching the full picture. The fact that a manager approved generates a sense that "the organization signed off." This structure is examined in detail in Series 2 vol.04: "'Just This One Slide' — Piecemeal Rationalization".

The fork in the road is here. Refuse to use "just one slide" as an excuse. Even inside divided labor, keep for yourself the role of asking "what does this deliver to a physician, in full?" That is where the line gets drawn. Even if a manager approved, do not finalize a document you cannot put your own name on. That decision is quiet and hard to articulate — but it is how work stays defensible over time.

Something to Return To — When the Line Is About to Break

The Guidelines on Promotional Activities for Prescription Drugs (September 2019) set out as a basic principle that "information shall be provided accurately on the basis of scientific and objective evidence" and require that providers "take care not to cause misunderstanding by providing only specific information." The weight of that sentence is that "accuracy" extends beyond the truth or falsity of individual data points — it reaches the selection of information and the manner of its presentation.

When I feel the line bending, the question I return to is simple: "Could a physician who receives this document come away with an impression that does not match what the trial actually found?" Efficacy and safety both, primary endpoint before secondary endpoint, control-arm data alongside the study drug's data — when that balance is off, I look for where exactly it broke.

The deviations the MHLW surveillance reports have documented across seven years did not all start from bad intent. The majority were cases where a starting point of "this drug helps patients" drifted, through deadlines and division of labor, into "select the data that helps the argument." Noticing that drift is easier when I frame myself not as "someone presenting data" but as "someone managing evidence." Rereading the record of how selective extraction, manipulation, and presentation of graphs distorted physicians' prescribing decisions (Series 1 vol.02: "Selective Extraction, Manipulation, and Presentation of Data and Graphs") sharpens the outline of what I am about to do.

Finally: this conviction also protects me. If I hold myself responsible for the whole document, I never have to fall back on "that slide was fine" as a repeated defense after the fact. Work I can sign in full is work I can explain in my own words, later, to anyone who asks.