Zero as a Blueprint── What to Think About Before Fearing a Missed Finding

01The Day the Comment Column Passed Forty

Looking at a review sheet completed by a newly assigned reviewer, my hand stopped. Forty-three comments were lined up. The first was serious: dosage and administration written differently from the electronic package insert (= the official prescribing document for a drug). A finding that could cause real harm if it reached a patient. But below it stretched an endless run of particle fixes, comma placement, and okurigana usage that the company had never standardized in the first place. The one heavy item lay buried among forty-two light ones.

I cannot blame him. I used to do the same thing. Zero missed findings. That was once my goal. I had nightmares about the monitoring program's report (= a program in which the Ministry of Health, Labour and Welfare covertly inspects sales information provision activities) carrying a case that looked like one of our own materials. From the next morning I made the mesh of my net finer. The finer I made it, the more the comments grew. Cases that used to close after two rounds of return (= a material being sent back by review) now routinely took a third. Marketing, unable to move the print deadline, kept sending revised proofs to the production agency, and the printing and delivery schedule slipped in a chain reaction. When lead time to a finished material stretches, everything planned beyond it is delayed too. Did patients become safer in proportion to their exhaustion and the lost days? Back then, I never thought about that.

02What Happens When You Aim for Zero

Signal detection theory (= a theory in psychology dealing with the relationship between misses and false alarms, systematized by Green and Swets) teaches a cold fact. If you loosen the decision criterion so that everything suspicious gets flagged, misses go down, but false alarms invariably go up. Moving the criterion alone can never reduce both at once. If you want to reduce both, you have to improve the performance of the net itself. This is not a matter of willpower. It is a matter of structure.

In a review office that seriously commits to zero missed findings, the criterion tips endlessly toward "flag everything." Then, on the side that makes the materials, the alarm never stops ringing. The world of medical safety has a term, alarm fatigue (= the phenomenon in which people stop responding to alarms altogether when too many keep sounding). A person who receives forty comments every month eventually stops reading all forty-three with equal weight. They become unable to. The one heavy item gets processed on the assembly line. The net that aimed for zero lets slip the very fish it most wanted to catch.

Then why are we drawn to zero? Behavioral economics knows this as zero-risk bias (= the tendency to place irrationally high value on eliminating a risk entirely rather than reducing it). "Cut misses from three a year to one" feels less satisfying, as an explanation and as a goal, than "make it zero." But that satisfaction comes at a price: the exhaustion of the departments around you, and serious risks buried out of sight.

Let me say this to be safe. I am not arguing that missed findings do not matter. If a material with missing safety information reaches a physician's hands, a patient stands beyond that. A serious miss is not permissible. Which is exactly why, instead of "watch everything with equal intensity," we should re-aim the goal. Not zero missed findings, but zero misses of serious risk. That one word of difference determines every part of the design that follows.

03Three Paths

I believe there are three paths.

The first is to reduce risk through mechanisms. Grade findings by severity. At the top, anything that could cause real harm to patients — that is, deviation from laws and guidelines (including missing safety information). Below that, deviation from internal rules, then appropriateness of data citation, and last, wording and formatting. If the grades are separated on the sheet itself, the one heavy item will not get buried among forty-three. Take formatting out of the conditions for review approval, separate mandatory fixes from optional ones on the face of the sheet, and return the latter in a different column as reference comments. Trim the checklist rather than expanding it. And if two people review, do not have them look through the same lens. If their viewpoints overlap, the second person misses what the first person missed. Changing the net is what matters. This is also the practical form of what signal detection theory calls "improving the performance of the net."

The second is to work on people's state of mind. The hospital study by Edmondson that gave rise to the concept of psychological safety (= a workplace condition in which people can say difficult things without being punished) contains a counterintuitive finding. The better teams reported more errors. Not that they made more. They hid fewer. A room where a reviewer can say "I may have missed something" the next day, and a room where they cannot: in which one is a serious risk found sooner? The answer is obvious. If misses are treated as defects in a reviewer's ability and fed into performance appraisals, people will hide their misses and sit on borderline cases. If misses are treated as material for learning, the design of the net keeps improving. Do you read failure as evidence of ability, or as material for growth? The stance of the one doing the evaluating determines the volume of reports.

The third is the path of management. Neither severity grading nor error reporting survives on frontline ingenuity alone. The month after a circular about another company's enforcement case makes the rounds, the whole review office tilts at once toward "flag everything." If, in a month when returns went down, someone upstairs says "has the review gotten soft?", the same thing happens. Conversely, if a leader says "cut the formatting comments and spend the time debating the serious items — and if a miss occurs, that is not a mark against you, it is homework for the system," the air changes. What an old boss of mine said was shorter still. "I trust the month you can explain what you didn't catch more than the month you caught everything."

04The Duty of Senior Management, and Where Psychology Belongs

The Sales Information Provision Activity Guidelines set out, in Chapter 2, Section 1, the "duties of senior management." Senior management bears responsibility for the workplace conduct of every officer and employee engaged in sales information provision activities. Building the internal system, engaging with evaluation and education, responding to improper activities. The responsibility is not lightened by "I delegated it" or "I distributed the authority." The structure allows no "the frontline did it on their own." And the same Chapter 2 also requires the establishment of an independent review and oversight department. How the review office designs its net is not an internal matter for the frontline; it is part of the system senior management is expected to put in place (for details, see this site's commentary at /compliance/04-msa-guidelines/msa2-1.html).

This duty sometimes gets translated into pressure on each individual reviewer: "don't miss anything." I was once on the side doing that translation. Apply the pressure, and reviewers start flagging everything suspicious; false alarms multiply, the recipients' responses dull, and misses get hidden. The result plays out exactly as the theory in 02 and 03 predicts.

Fulfilling the duty, I think, means management doing the opposite. Approve the severity design, offload formatting checks to automation or a separate lane, and shift reviewers' time toward the serious items. Draw an institutional line: voluntary error reports are, as a rule, not subject to discipline — concealment and intent excepted. Since the guidelines say to engage with evaluation and education, writing that line into the personnel evaluation documents is management's job. And share the reviews that went well — cases where a serious risk was caught early and quietly fixed — with the same enthusiasm as the failure cases. That is the practice of what Hollnagel calls Safety-II (= a view of safety that does not only stamp out failures but studies and supports the reasons things usually go right).

Zero missed findings is too crude to serve as a goal. Define what counts as serious risk. Decide how many false alarms you will tolerate. Keep the air in which reports come up. These three decisions belong not to the reviewers but to management, and so does the responsibility for having made them. That the guidelines demand this in senior management's name is probably no coincidence. What the old me who wrote forty-three comments needed was not a finer net, but someone who would sit down and decide with me what the net was for.