Case Study 39.1 — The Same Appointment, Twice
What this is. One patient, one clinician, one presenting problem — followed twice. In the first version the patient conceals what they have been taking. In the second, they disclose. Everything else is identical: the same symptom, the same laboratory result, the same procedure scheduled six weeks out, the same competent clinician working in the same short appointment.
The point is not that concealment is wicked. It is that concealment removes inputs from a reasoning process, and you can watch what the reasoning does without them.
No one in this case study has a name. The people in it are the patient, the clinician, the gastroenterologist, the anesthetist, and the pharmacist — roles, not characters.
The setup, common to both versions
A patient in their late forties has been taking two things their clinician does not currently know about. One is a GLP-1 receptor agonist obtained through an online telehealth service that their regular practice has no record of. The other is a compound sold as a peptide for joint recovery, purchased from an overseas supplier, of unverified composition.
They present for an appointment about something unrelated to either: intermittent upper abdominal discomfort and a period of unusual fatigue over roughly six weeks. Routine bloods drawn a week earlier show a mild elevation in liver enzymes that was not present at their last check eighteen months ago.
Separately, they have a colonoscopy scheduled in six weeks — an ordinary interval screening, booked months ago, involving sedation.
The clinician has fourteen minutes.
Version A — the patient conceals
The appointment
The clinician asks about alcohol (unchanged, modest), about new medications (the patient says none), about supplements (the patient says a multivitamin), and about recent illness (none). They examine the abdomen, find nothing alarming, and look at the enzyme pattern.
Working from what they have, they build a differential: fatty liver disease, given the patient's history; a viral cause; an unreported medication or supplement; less likely, something biliary. They order a repeat panel in six weeks, a hepatitis screen, and an abdominal ultrasound. They advise reducing alcohol further and mention weight as a contributor to fatty liver — noting, mildly, that the patient looks like they have lost some weight, which the patient attributes to "being more careful."
The clinician makes a note of the weight loss but does not pursue it. Unexplained weight loss with abnormal liver enzymes is a combination worth watching; in the absence of any other information, they file it as probably favorable.
The appointment ends. Nothing has gone wrong.
What happened next
The ultrasound shows hepatic steatosis, which is consistent with the working hypothesis and also consistent with several other things. The hepatitis screen is negative. The repeat panel shows the enzymes have drifted slightly higher.
The clinician, working conscientiously, refers to hepatology. That referral has a nine-week wait. In the meantime the patient is advised to stop anything hepatotoxic — advice they cannot act on usefully, because they are not going to volunteer the two things they might need to stop, and they do not know whether either is relevant.
Meanwhile, the joint compound continues. The patient has begun to wonder whether it is connected, and has quietly stopped it — but has not told anyone, so the timing of the stop is not recorded anywhere, and its relationship to the next set of results cannot be interpreted.
The colonoscopy
Six weeks later the patient attends for the scheduled procedure, having followed the fasting instructions exactly.
The pre-procedure questionnaire asks about medications. The patient lists the ones the practice knows about. It does not occur to them that a weight-management injection is relevant to a bowel examination, and nobody asks a question specific enough to prompt it.
Under sedation, the gastroenterologist finds a stomach that is not empty.
What happens from there depends on when it is noticed and on the anesthetist's judgment. In the better case, the procedure is abandoned, the patient is recovered, and it is rescheduled — a wasted preparation, a wasted day, an anxious explanation. In the worse case, gastric contents reach the airway. Aspiration under sedation is not a small event. It can mean an unplanned admission, respiratory complications, and in a minority of cases something considerably worse.
Nobody involved has been careless. The anesthetist assessed the patient against the information available. The information available was wrong.
The accounting for Version A
- A differential built without a relevant input, producing a workup aimed at the wrong problem
- An ultrasound, a hepatitis screen, and a specialist referral, at least partly avoidable
- A nine-week wait carrying real anxiety
- An uninterpretable stop, because it was not documented
- A weight loss recorded as favorable that was, in fact, an unexplained finding with a known cause
- A peri-operative event that the entire apparatus of pre-procedure assessment exists to prevent, and could not
Version B — the patient discloses
The appointment
Same room. Same fourteen minutes. Same enzyme result.
When the clinician asks about new medications, the patient says: "There are two things I should tell you about. I've been getting a weight-loss injection through an online service since March — I don't think you have that on file. And I've been using something sold as a peptide for my shoulder that I bought online. I don't actually know what's in that one. I should have said earlier."
Three sentences. Roughly eleven seconds.
Note what the clinician does not do. They do not lecture. They ask what the injection is, how long, what dose interval, and whether the patient has the label for the second item. They ask about nausea, vomiting, early fullness, and reflux — the symptoms that matter for a class that slows gastric emptying — and record the answers. They ask when the second compound was started relative to the fatigue.
They do say, briefly, that they would rather know about things like this and that the patient should tell them going forward. It takes about six seconds and it is not a lecture, because it is framed as a request for information rather than a judgment about character.
What the disclosure changes
The differential reorders. Fatty liver is still on it. But so now are: a reaction to a compound of unverified composition; a contaminant; and — importantly — the possibility that the enzyme elevation is unrelated to either and the compound is a coincidence that must nonetheless be accounted for before anything else is concluded.
The weight loss stops being a mystery and stops being reassuring. It has an explanation. It is no longer a finding requiring investigation, and it is also no longer a favorable sign. The clinician removes it from both lists, correctly.
The sequence becomes interpretable. The clinician asks the patient to stop the unverified compound — the one with no evidence, unknown composition, and a plausible temporal relationship — and to continue the GLP-1 receptor agonist for now, since it is a characterized drug with a known hepatic profile and stopping both simultaneously would destroy the information. They repeat the panel in six weeks.
This is the part patients most often miss. Stopping one thing at a time is a design decision, and it is only available to a clinician who knows what the things are. In Version A the patient stopped one compound silently at an undocumented moment, and the next result meant nothing.
The pharmacist gets involved. The clinician suggests the patient take the label photograph to their pharmacist, who spends ten minutes on it and identifies that the product listing names two ingredients, one of which is not a peptide at all. That is not a diagnosis. It is a fact that was available for the cost of asking.
Nothing is ordered that does not need to be. The hepatitis screen still happens, because it is cheap and it belongs on the list. The ultrasound is deferred pending the repeat panel. The hepatology referral is not made yet.
The colonoscopy
Six weeks later, the pre-procedure assessment includes a current medication list that contains the GLP-1 receptor agonist, because it is now in the record.
The anesthetist sees it. What follows is a normal clinical decision made with normal inputs: a question about whether the patient has symptoms suggesting delayed emptying, a decision about the timing of the last dose relative to the procedure, possibly an extended clear-liquid period, possibly a pre-procedure gastric ultrasound depending on local practice, and possibly no change at all.
The specific decision is not the point of this case study, and it will differ between institutions and will keep changing as guidance evolves. The point is that a decision was made. In Version A no decision was made, because the question never arose.
The accounting for Version B
- A differential containing the actual candidates
- One test deferred, one referral not yet made
- A stop sequence that will produce interpretable information
- A pharmacist consultation that cost ten minutes and identified something real
- A weight change correctly attributed
- A peri-operative assessment performed with accurate inputs
- One mildly uncomfortable eleven-second disclosure
The comparison, stated plainly
VERSION A VERSION B
(concealed) (disclosed)
differential wrong candidates right candidates
tests ordered more fewer
specialist referral yes, 9-week wait deferred
stop sequence silent, undocumented staged, recorded
weight loss read as favorable explained
vial identity unknown partially resolved
peri-operative decision none made made, with inputs
cost to patient money, time, worry, ~11 seconds of
and an avoidable discomfort
airway risk
The two columns differ by one input. Every downstream difference follows from it mechanically. Neither clinician was better than the other. They were the same clinician, given different information.
Discussion questions
-
In Version A, the clinician records the patient's weight loss as probably favorable. Explain precisely why that inference was reasonable given their information and wrong given the facts. What does this illustrate about how a missing input distorts a differential rather than merely shortening it?
-
Version B's clinician stops one compound and continues the other. Reconstruct the reasoning. Why does stopping both at once destroy information, and why was this option unavailable in Version A even though the patient did stop something?
-
The disclosure in Version B takes about eleven seconds. Write your own version of those three sentences for a real or hypothetical exposure. Then identify anything in your draft that is a justification rather than a fact, and remove it. Why does the chapter want justifications removed?
-
In Version A, the pre-procedure questionnaire asked about medications and the patient answered it without lying by their own understanding — they did not think a weight-management injection was relevant to a bowel examination. Whose failure is this? What would have to change, on the clinician's side and the patient's side, to prevent it? Consider §39.2's argument about who is structurally able to judge relevance.
-
Version B's clinician spends about six seconds noting that they would rather know about things like this. Compare that with the category dismissal described in §39.7. What makes the difference between a brief request for future information and a lecture, and why does §39.7 argue that the difference matters more than the verdict?
-
Suppose Version A's peri-operative outcome had been the better one — the procedure abandoned, no aspiration, no harm. Would that change the argument of this case study? Answer with reference to the distinction between an outcome and a decision process, and explain what it means to say that a process failed even when nobody was hurt.