Chapter 39: Talking to Your Doctor About Peptides

"Listen to your patient; he is telling you the diagnosis." — attributed to William Osler

Overview

Thirty-eight chapters of this book have been about evidence. This one is about a room.

It is a small room, usually. There is a person in it who has read something, tried something, or is considering something, and who has spent real effort trying to work out whether it is true. And there is a person in it who has roughly eight minutes, a screen full of other people's results, and a professional obligation to be careful. Those two people want, in almost every case, the same thing. And the conversation between them fails constantly.

It fails in predictable ways, which is the good news, because predictable failures can be prepared for. It fails because the patient arrives braced for judgment and tells a partial story. It fails because the clinician arrives braced for a request they cannot responsibly grant and answers a question that was not asked. It fails because "I don't know" costs a professional something to say, and because "what did you read that on?" costs a patient something to hear. Nobody in that room is behaving badly. The room is badly shaped.

This chapter is about reshaping it, and the goal is easy to get wrong. The goal is a better conversation, not a won one. Nothing here is a technique for getting a prescription. If a passage ever reads to you like a negotiation tactic — a way of maneuvering a clinician toward a particular yes — I have failed, and you should discard it. A conversation won by maneuvering produces a decision made on bad information, which is the outcome this entire book exists to help you avoid.

What you are after is narrower and more useful: a clinician who knows what is true about you, and a set of questions that get real information out of the encounter regardless of which way the answer goes. That is achievable in eight minutes, and it does not require your clinician to be unusually good, unusually interested in peptides, or familiar with the compound you are asking about.

One section matters more than the others for your physical safety. If you read nothing else, read §39.2.

In this chapter, you will learn to:

• Recognize the structural pressures that shape the appointment before either person speaks
• Disclose what you are taking, and understand the clinical — not moral — reasons this is not optional
• Assemble the five pieces of information your clinician actually needs from you
• Ask the five questions that work on any medical claim, in any specialty
• Bring one study, in the form that opens a conversation
• Tell "we don't know" apart from "it doesn't work" and from "I don't know"
• Respond usefully to dismissal, and skeptically to enthusiasm
• Name what a clinician genuinely cannot do for you
• Turn a dossier verdict into a question, and use the conversion to test the verdict

Learning Paths

All five paths read §39.2 in full. It is the only section in this book with a direct, immediate, physical safety consequence.

💊 GLP-1 — §39.2 is not optional for you. The peri-operative gastric-emptying issue is live for anyone on a GLP-1 receptor agonist facing any procedure with sedation, including ones you would not file as surgery. 🏋️ Performance — §39.2 and §39.7 are written for you: the population most likely to be taking something undisclosed and most likely to have had a bad experience disclosing. §39.4's fifth question is the one that protects you. 🔬 Science — §39.6 is the most interesting section; the three-sentence distinction is the epistemics of the whole book compressed into one diagnostic. 💄 Cosmetic — lighter stakes, same structure. §39.8 is yours: cosmetic dermatology has the tightest coupling between recommending and selling, and the reasoning transfers directly. 🏥 Clinical — read this from the other chair. §39.1, §39.7, and §39.9 describe what your patients experience and are not saying.

39.1 Why this conversation goes badly

Two people walk into an appointment. Before either of them says anything, both are already braced.

The patient expects to be judged. Often this is not a projection — it is a memory. Many people interested in compounds outside the ordinary prescribing repertoire have had at least one conversation that went badly: an eyebrow, a sigh, a change in tone, a question that was really an accusation. Once that has happened, the next disclosure gets shorter, and the one after may not happen at all. What the clinician sees as an unremarkable question — are you taking anything else? — lands on a patient who has already decided how it ends.

The clinician expects to be asked for something they cannot responsibly provide. Also not a projection. A primary care clinician in 2026 has had some version of this conversation several times a month for years, and a meaningful fraction ended with a request for a prescription the evidence did not support, a referral to a clinic they had reservations about, or a laboratory panel with no clear purpose. They are not braced because they think you specifically are unreasonable. They are braced because the base rate is what it is.

So both people enter with a script, and the scripts do not match. The patient's is how do I say this without being dismissed. The clinician's is how do I decline this without damaging the relationship. The result is a negotiation in which neither says what they actually want, because both decided in advance that saying it plainly would not work.

WHAT EACH PERSON IS ACTUALLY BRACED FOR

  THE PATIENT                              THE CLINICIAN
  ───────────────────────────              ─────────────────────────────
  fears: being judged                      fears: being asked for
         being lectured                           something unsupportable
         it going in the record                    being blamed if it goes wrong
         being labeled "difficult"                 losing eight minutes they
                                                   do not have

  so they:  understate                     so they:  answer the question they
            omit                                     expect rather than the
            ask hypothetically                       one asked
            ("a friend of mine…")                    close the topic early
                                                     reach for the category
                                                     ("that's all unregulated")

  ─────────────────────────────────────────────────────────────────────
  RESULT: a conversation about a compound neither person has described
          accurately, ending in a decision made on incomplete information.
          Both people leave frustrated. Neither behaved badly.

That diagram is the whole diagnosis. The failure is not caused by a bad clinician or a difficult patient but by two reasonable people optimizing against each other's expected behavior. Everything else in this chapter breaks one of those loops from your side, which is the only side you control.

Three structural pressures deserve naming, because misreading them as personality is the most common mistake patients make.

Appointments are short. A substantial clinical conversation has to happen inside a slot designed for a simpler one. A clinician who cuts a topic short is frequently not uninterested; they are triaging against your blood pressure, your last results, and four other items on the list. §39.5 develops the implication: ask for the time explicitly rather than smuggling a real evidence discussion into a slot that cannot hold it.

No clinician can know every compound. Thousands circulate in the supplement, research-chemical, and gray-market spaces, most with no literature attached. A physician who has never heard of a particular pentadecapeptide is not out of date. They are ordinary. §39.6 turns this into a rating, because the inference people draw — they'd never heard of it, so it must be very new — is almost always backwards.

"I don't know" is professionally expensive. Medicine, like most expert fields, rewards confidence. A clinician who says it risks being read as incompetent by a patient who came for an answer, and the incentive to produce a confident-sounding response instead is constant. When you get a fast, certain answer about an obscure compound, it is worth gently checking whether it is a considered answer or a comfortable one — and §39.6 gives you the question that checks.

None of this is anyone's fault. It is the shape of the room, and recognizing the shape is the first move.

💊 In the Clinic — the other side of the desk

Patients almost never see the encounter from the chair opposite theirs, and seeing it changes how the responses read. The clinician has, at minimum: a limited slot; a chart that may not include the medications you were prescribed elsewhere; a screen demanding structured entries while they are also listening to you; documentation and quality requirements; and a legal environment in which this conversation's record may be read later by someone looking for a mistake. If they prescribe something unapproved for an unapproved purpose and it goes badly, the exposure is theirs.

They also have a real epistemic problem. You are describing a compound they may not recognize, in terms that came from marketing, and asking them to evaluate it live. Most clinicians assess a class, a mechanism, and a study design quickly; very few can assess an unfamiliar molecule's evidence base from memory, because that requires a literature search they cannot run while you sit there.

The practical translation: the response you get is shaped as much by these constraints as by what the clinician thinks. "I'm not comfortable with that" may mean I think the evidence is bad, or I do not know the evidence and will not pretend, or my institution will not let me, or I would need forty minutes to answer that properly. Four different statements wearing the same words. Asking which one it is costs you nothing and is not confrontational.

39.2 The disclosure problem, and why it is a clinical question rather than a moral one

Some readers of this book are currently taking something they have not told a clinician about. I am not going to pretend otherwise, and I am not going to spend this section on disapproval. The position here has no conditions attached:

Tell them. Tell them the truth. Tell them all of it.

Not because it is the honest thing to do — although it is — but because the alternative makes you less safe in four mechanical ways. The reasons here are clinical rather than moral, because the moral framing is precisely what makes people lie. A person who experiences disclosure as a confession will minimize, and minimizing produces the failures below. A person who experiences it as handing over a missing instrument reading hands it over completely.

1. Differential diagnosis

When you present with a new symptom, your clinician builds a list of things that could be causing it, ordered roughly by likelihood, and works down it. That list is constructed from what they know about you.

If what they know about you is wrong, the list is wrong, and the entire workup that follows solves a different problem than the one you have. A liver enzyme abnormality in a person taking nothing has a certain differential. The same abnormality in a person who has been injecting an unregulated compound of uncertain composition for four months has a substantially different one, containing items — contamination, an unexpected constituent, a reaction to something that is not what the label said — that the first does not even list. The clinician cannot put those items on the list if they do not know to.

The failure mode is expensive as well as dangerous: undisclosed exposure routinely produces workups that chase the wrong thing — extra imaging, extra testing, a referral for a problem one more piece of history would have explained.

2. Interactions

Interaction checking is a real, systematic, largely automated process. Your clinician and your pharmacist both run it, and it works well — for the things it is given.

It cannot be run against a compound nobody knows about. That is not a limitation of the software; it is a definitional impossibility. And peptides are not exempt from interaction concerns simply because they are peptides. Compounds that affect gastric emptying change the absorption timing of oral medications taken with them. Compounds that affect blood glucose interact with everything else that does. Anything affecting fluid balance, blood pressure, clotting, or thyroid function sits next to whole classes of prescription drugs. And a compound of uncertain composition, per Chapter 34, may contain things nobody has accounted for at all.

3. Surgery and emergencies

This is the most concrete safety reason in this chapter, and it deserves its weight.

GLP-1 receptor agonists slow gastric emptying. That is not an incidental side effect — it is part of how the class produces satiety, and Chapters 7 and 8 covered the mechanism. It has a consequence that took the surgical and anesthetic world by surprise: patients who have fasted correctly before a procedure may nonetheless have retained solid or liquid contents in the stomach. Retained contents raise the risk of aspiration — material entering the airway — during sedation or general anesthesia, when the reflexes that normally protect the airway are suppressed. Aspiration is not a minor complication.

This became an active area of professional guidance beginning in 2023 and has been revised since. Early advice leaned toward holding the medication before elective procedures; more recent multi-society guidance has moved toward individualized assessment considering the specific agent, the timing, whether the patient has symptoms suggesting delayed emptying, and in some settings a pre-procedure gastric ultrasound. The details will keep moving, and you should not try to memorize them.

You should not need to. That is the point. What matters is not that you know the current rule — it is that your anesthesiologist has to know you are taking it. A patient who does not disclose cannot be protected by any version of the guidance, current or future. The apparatus of pre-operative assessment is built to be given accurate inputs and fails silently when it is not.

This applies more broadly than "surgery" suggests. Procedures with sedation include endoscopy, colonoscopy, many dental procedures, some imaging, cardiac catheterization, and a long list of things people do not mentally file as operations. And it applies in emergencies, where the person treating you has only the record and whatever you or a companion can say.

🧬 The Molecule — why a satiety mechanism became an anesthetic problem

The peri-operative story illustrates something this book keeps returning to: a molecule's effects do not sort themselves into "the intended one" and "the side effects." There is only physiology, and we apply the labels afterward.

GLP-1 is a gut hormone released in response to food. Among its actions is a slowing of gastric emptying, which contributes to fullness and to flattening the post-meal glucose rise. The native hormone does this in a pulse and is gone in minutes. A long-acting receptor agonist does it continuously, at a receptor occupancy the body never produces on its own. Chapter 3 made this the central distinction between a hormone and a drug, and here it produces a consequence nobody was looking for.

One wrinkle matters for interpretation: the degree of slowing is not uniform across the class or across time. Shorter-acting agents produce more pronounced slowing than long-acting ones, there is evidence of partial adaptation with continued use of some agents, and individual variation is substantial. That is precisely why guidance moved from a blanket rule toward individualized assessment — the population is not homogeneous, which is a Chapter 5 point wearing surgical scrubs.

What it means for you: a real, mechanistically expected, clinically recognized consideration whose management is still being worked out. That is a normal state for a young drug class. It is not a scandal and not a reason for alarm. It is a reason to say the words out loud to the person about to sedate you.

🩺 Safety and Risk — the undisclosed compound in an emergency

Consider what a treating clinician works with when someone arrives acutely unwell and cannot give a history. They have the medication list in the record, whatever a family member knows, examination findings, and laboratory results. From those they build a working explanation and act on it, often quickly.

An undisclosed injectable is a hole in that reasoning in a particularly unhelpful place. It does not merely leave an item off the list — it makes the remaining items look more likely than they are, because probability belonging to the missing explanation gets redistributed across the visible ones. An unexplained finding gets attributed to the most plausible listed cause, and the treatment follows the attribution.

There is an easy fix people do not think of: what is written down where someone can find it. A current, accurate medication list — including things you did not get from a pharmacy — is the input to a decision that may be made by someone who has never met you, at a time when you cannot speak for yourself. If you are unwilling to have something in the record, you have decided the social cost is higher than the clinical one. That is your decision, and it should be made with the trade in view rather than by default.

4. Monitoring

Chapter 19 established what supervision actually adds: baseline and interval measurements, someone watching for a set of predictable problems, and a defined pathway if something goes wrong. That is genuinely valuable, and §39.9 is careful about what it is not.

None of it is available for something undisclosed. A clinician cannot monitor for a harm they do not know to look for. They order the tests that make sense given what they believe about you, and if that belief is missing a component, the relevant test is the one that does not get ordered. This is the quietest of the four failures and in some ways the most complete, because nothing visibly goes wrong. The monitoring simply is not happening, and nobody notices until there is something to notice.

The fear is real, and here is how to weigh it

Everything above assumes the disclosure happens. Let me deal directly with why it often does not, because the reasons are not stupid ones.

People worry about being lectured. About a note in the record that follows them, changing how the next clinician reads them before the appointment starts. About being labeled: drug-seeking, non-compliant, difficult. In some cases about professional or legal consequences. And some of these worries are grounded in an experience that already happened.

I am not going to tell you those fears are baseless. Some clinicians do respond badly. The literature on patients underreporting supplement and non-prescribed substance use is extensive and consistent across many settings, and the reasons patients give in it are mostly the ones above — expectation of disapproval, belief that it was not relevant, and never being asked in a way that invited an answer. This is a well-documented pattern, not a rare one.

Here is the weighing, and it is not close:

The risk of an uncomfortable conversation is smaller than the risk of being treated by someone working from an incomplete picture. The first has a bad afternoon as its worst case. The second has a wrong diagnosis, a missed interaction, or an aspiration event under sedation. Those are not comparable magnitudes and it helps no one to pretend they are.

There is a further point, which I want to put carefully because it is easy to hear as adversarial. A clinician you cannot tell things to is one worth reconsidering — not because they are a bad doctor, but because the entire value of the relationship is the accuracy of the information moving through it. A relationship in which you filter produces decisions from filtered data. That is not a moral judgment about either of you; it is a description of what the relationship can and cannot do.

Most clinicians, told plainly and without drama, respond fine. The most common actual response to "I've been taking something I bought online and I want you to know" is a couple of clarifying questions and a note in the chart. The catastrophe people brace for is rarer than the bracing suggests.

📊 Evidence Rating

Claim form — "disclosing all substance use to a treating clinician improves care." Rating: ✅ Strong clinical evidence (as of this writing, 2026) Why: the reasoning is structural rather than statistical — differential diagnosis, interaction checking, monitoring, and peri-operative assessment each take a medication history as an input and produce a wrong output when that input is incomplete; the GLP-1 gastric-emptying issue is a concrete, mechanistically understood case where non-disclosure removes a specific protection. What would change it: essentially nothing available. It would require showing clinical reasoning is insensitive to the accuracy of the history, which would invalidate most of clinical medicine. A narrower question — whether a particular way of asking elicits more accurate disclosure — is genuinely open and actively studied.

39.3 What your clinician actually needs from you

If §39.2 persuaded you, the next question is what to actually say. It is a shorter list than people expect, and knowing it in advance removes most of the awkwardness, which comes from not knowing where to start.

What you are taking. The name as you know it, and — if you have it — what it is supposed to be. "A compound sold as BPC-157" is more accurate and more useful than "BPC-157," and Chapter 34 explains why that is not pedantic. If you have the label, the vial, or the order confirmation, bring a photograph; a clinician extracts more from a label than you can.

How long. Start date if you have it, approximate duration if not. Continuous or intermittent. Whether anything changed recently — a new source, a different presentation, a gap and a restart. The temporal relationship between an exposure and a symptom is among the most informative things in a history, and it is unrecoverable if nobody wrote down when things started.

Why you started. Not as a justification — as a clinical fact. "For a shoulder that has hurt for two years" is a different presentation than "because I read about longevity," and both point the conversation different directions. The first contains a symptom your clinician may want to look at independently. Quite often the underlying problem that motivated the compound is itself treatable and has never been examined.

What you have noticed. Including nothing. Especially including nothing. "No change at all" is genuinely useful information that almost never gets reported, because people volunteer changes and stay quiet about their absence. Report the good, the bad, the ambiguous, and the injection-site or application-site observations that feel too minor to mention. Minor local findings are exactly what a clinician wants to hear about early.

What you want to happen next. This one gets skipped and changes the conversation most. Continue and be monitored? Stop, and unsure how? An opinion on the evidence? An interaction check? A referral? These lead to completely different appointments, and if you do not say which you are asking for, your clinician will guess — and their guess, per §39.1, will be shaped by the last several people who sat in that chair.

FIVE THINGS TO HAVE READY   (write them on your phone before you go in)

  1  WHAT      name as sold, and what it is claimed to be
               "a compound sold as ___"        + photo of label if you have it
  2  HOW LONG  start date · continuous or on and off · anything changed recently
  3  WHY       the problem you were trying to solve — this is clinical data,
               not a justification
  4  NOTICED   changes, non-changes, and anything local at the site.
               "Nothing at all" is a real and useful answer.
  5  WANT      continue and monitor · stop · opinion on evidence ·
               interaction check · referral        ← say which one

Two minutes of preparation, and it converts a defensive exchange into a handover. The list is short on purpose: this is not a dossier presentation, it is the minimum a clinician needs to think clearly.

The asymmetry that makes this hard

There is a structural reason these conversations feel like two people talking past each other, and naming it helps.

You usually know the marketing claim and not the compound. Your clinician usually knows the drug class and not the marketing.

You know it is supposed to help with tendon recovery, that people online report results in a few weeks, that it comes as a lyophilized powder, and that the community around it has strong opinions. You may not know its molecular class, its half-life, whether it has ever been through a human trial, or what it is likely to do to your other systems.

Your clinician knows what a peptide is, what a receptor agonist does, how to read a trial, and what categories of harm to watch for. They may have no idea what is being claimed about this compound, what it is being sold as, or how it is presented — and therefore no idea what you are actually asking.

Each of you holds half the picture. The conversation works when both halves get onto the table and fails when each assumes the other has the whole thing. Saying this out loud is allowed: "You'll know things about this class that I don't, and I probably know what's being claimed about it that you haven't seen. Can I tell you the second part?" That sentence does a remarkable amount of work, and it is not a tactic — it is an accurate description of the situation.

🔍 Check Your Understanding

1. Why is "I haven't noticed anything" worth reporting rather than skipping?
2. Explain the asymmetry in §39.3 in your own words, and give an example of a question a clinician might misanswer because they were missing your half.
3. Of the five items, which would you be most likely to leave out, and what would that cost?

39.4 The five questions worth asking

This is the practical core of the chapter and the most portable thing in the book.

These five questions work on any medical claim — a peptide, a supplement, a procedure, a screening test, a psychiatric medication, a specialist's recommendation. They are not peptide questions. They separate a claim with evidence behind it from a claim with a story behind it, and memorizing them will serve you for life in settings with nothing to do with this subject.

None is adversarial. All are questions a good clinician is glad to be asked, because they are the ones the clinician is already asking internally.

1. "What's the evidence for this in someone like me?"

The population question, built in Chapter 5.

The three most important words are like me. Evidence is generated in specific populations and does not automatically travel. Age, sex, body composition, kidney and liver function, other conditions, other medications, and — often the decisive one — whether you have diabetes all change what a trial result means for you.

Chapter 8 gave the cleanest illustration in the book. STEP 1 enrolled adults with overweight or obesity without diabetes; STEP 2 enrolled adults with overweight or obesity with type 2 diabetes. Same drug, different populations, different average weight change, with the group with diabetes losing less. Read only STEP 1 while having type 2 diabetes and you carry an expectation the evidence does not support for you.

The question is useful even when the answer disappoints. "The evidence is in a population quite different from you" is real and actionable. It tells you where you are standing.

2. "What's the endpoint — what would actually get better?"

The surrogate-versus-outcome question, built in Chapter 16.

A surrogate endpoint stands in for something you care about: a laboratory value, an imaging finding, a score on a scale. An outcome is the thing itself: living longer, fewer heart attacks, walking without pain, sleeping through the night, not being hospitalized.

The two come apart more often than anyone would like. Chapter 16 cataloged treatments that moved a number impressively and did nothing for patients — in a few notorious instances making them worse. A number on a report is not the same as a life that goes better.

So, plainly: if this works, what would actually be different? If the answer is a laboratory value, follow up with is that value linked to something I would notice? Sometimes that is a well-validated yes. Sometimes the honest answer is that the link is assumed. Either way you now know which, and can hold your expectations at the right altitude.

3. "What are the alternatives, including doing nothing?"

The comparator question, also Chapter 5's.

Every claim of benefit is implicitly a comparison, and the comparison is where the meaning lives. Better than nothing? Than the standard treatment? Than the cheaper one? Than what you would have done anyway?

Doing nothing belongs in the list. It is not a rhetorical addition — it is a real arm of real trials, and in a great many conditions it performs better than people expect, because bodies improve on their own, symptoms fluctuate, and much of what looks like treatment effect in an uncontrolled setting is regression toward the mean and natural history. Chapter 6 is about why this makes personal testimony such an unreliable instrument.

Including it also changes the emotional shape of the decision, converting a binary — do I take this or do I fail to address my problem — into a comparison among options, one of which is watchful waiting with a plan to reassess. That is often the right answer and almost never the one offered, because it does not feel like doing anything.

4. "What would you expect to see, and by when?"

This question is the whole book in a sentence. It converts a vague hope into a falsifiable prediction. Before it, you have "I'm hoping this helps my shoulder." After it, you have "if this is working, the morning stiffness should be noticeably better within about six weeks." The second one can be wrong. That is its entire value.

An unfalsifiable expectation cannot fail, which sounds comfortable and is a trap. It is the mechanism behind the escalating-dose spiral of Chapter 2 §2.8: with no prediction, there is no point at which the honest conclusion is this is not doing what I hoped — there is only maybe it needs longer, then maybe it needs more, then maybe it needs a partner compound. Each step is locally reasonable and the sequence has no exit. A prediction with a date on it installs one.

Write the answer down, in your dossier, with the date. Chapter 40 will ask you to look back at it, and the looking back is where the learning is.

5. "What would make you stop this?"

This is the stopping rule, and it is the question people skip.

Ask it before starting. It is nearly impossible to ask honestly once you are invested — once you have spent the money, told people, felt something, or simply become attached to the hope. Chapter 6's testimonial dynamic describes how the reinterpretation happens, and it happens to intelligent, self-aware people who can watch it happening and cannot stop it.

A stopping rule set in advance is a commitment made by the version of you that is still neutral. It should have content: a symptom that means stop immediately, a laboratory finding that means stop and reassess, and a date at which no benefit means stop. Chapter 10 covered the formal version — trials pre-specify interim stopping rules for the same reason, because people watching accumulating data without a rule reliably find reasons to continue.

If the answer is "nothing, really," that is itself a significant finding. An intervention with no stopping condition cannot be evaluated.

THE FIVE QUESTIONS — what each one is actually testing

  1  "evidence in someone like me?"   →  POPULATION      (Ch 5; STEP 1 vs STEP 2)
  2  "what would get better?"         →  ENDPOINT        (Ch 16; surrogate vs outcome)
  3  "alternatives, incl. nothing?"   →  COMPARATOR      (Ch 5; the missing arm)
  4  "what, and by when?"             →  FALSIFIABILITY  (Ch 2 §2.8; the exit)
  5  "what would make you stop?"      →  STOPPING RULE   (Ch 10; ask while neutral)

  Not peptide questions. Medical-claim questions. They work on a statin, a
  supplement, a knee operation, and an antidepressant, and they work the same way.

Two closing observations. A good clinician will often ask you some of them — let's see where you are in six weeks is question 4 in ordinary clothes, and if they get there first, that is a good sign.

And a clinician who welcomes these questions is showing you something. Not because welcoming them proves competence, but because they are uncomfortable for anyone whose recommendation rests on enthusiasm rather than evidence — question 5 especially. A practitioner with a crisp answer to "what would make you stop this" has thought about the failure case, and thinking about the failure case is most of what separates careful practice from confident practice.

🔬 Read the Study — why "like me" is not a technicality

Take the two trials named above and look at what changes.

Population. STEP 1: adults with a body mass index at or above 30, or at or above 27 with a weight-related condition, and without diabetes. STEP 2: adults with a body mass index at or above 27 with type 2 diabetes. That is the deliberate design difference; the rest of the programs is deliberately similar.

Endpoint. Percent change in body weight over 68 weeks, with a co-primary of achieving at least five percent loss — a surrogate for the things people actually want (cardiovascular events, mobility, function, mortality), which were addressed later in different trials. That sequence is normal and correct.

Comparator. Placebo, with lifestyle intervention in both arms. Everyone got the counseling. The placebo arm lost weight too, which is exactly why the comparator exists.

Size. Both large multi-center randomized trials — the kind of size that makes an average effect trustworthy while telling you nothing about any one person.

The finding that matters here: average weight loss in the population with type 2 diabetes was meaningfully smaller than in the population without. The drug did not change. The population did. A trial result is a statement about a group that met a list of criteria, and the first question about any such statement is whether you would have met the list.

39.5 Bringing evidence to an appointment

You have built a dossier. You have read papers. You have a folder.

Do not bring the folder. I say this as someone who has spent thirty-eight chapters teaching you to read papers. The folder does not work, and understanding why is more useful than any technique for making it work.

Bring one study. Not a stack, not a forum thread's citation list, not a review with sixty references. One — the one that most influenced your thinking, or the one you found hardest to interpret.

Bring the actual paper, or its abstract — not an article about it. The gap between a study and the coverage of a study is where most public misunderstanding of medicine is manufactured; a press release describing a mouse experiment and a news article describing the press release are two degrees of separation, and by the second one the mice have usually disappeared. Your clinician can read an abstract in ninety seconds and will immediately see things you may not have: whether the comparator was appropriate, whether the endpoint was a surrogate, whether the population resembles you, whether the effect size is clinically meaningful or merely statistically detectable.

Know its four basics before you walk in: population, endpoint, comparator, size. If you cannot state those four from memory, you have not read the study closely enough to discuss it, and reading it once more with those four questions in hand takes about ten minutes and will change what you think of it surprisingly often.

Ask about it. Do not present it.

The difference between a productive appointment and an unproductive one is entirely in the framing.

"I read this and I'm not sure how to weigh it — what am I missing?" opens a conversation. It gives the clinician a defined task they are good at, signals that you are after information rather than agreement, and makes it easy for them to tell you something uncomfortable.

"This study proves it works" closes one. It converts the clinician from interpreter to opponent and invites a defensive dismissal of the whole topic. And if the study genuinely did prove it, you would not need to say so — the clinician would see it in the abstract.

I want to be careful here, because this is where a chapter like this can slide into being a manipulation manual. The reason to ask rather than present is not that asking works better. It is that asking accurately describes your epistemic position: you genuinely do not know how to weigh the study, which is why you brought it. The framing that gets better results is the framing that is true — and used as a technique while actually wanting agreement, it will not work anyway, because it will be obvious.

Respect the clock, or ask for more of it

An eight-minute appointment cannot absorb a literature review. That is arithmetic, not a failure of will. If you want a serious evidence discussion, ask for the time explicitly — many practices will book a longer or separate slot if you say what it is for. "I'd like to discuss something I'm taking and go through some evidence — can I have a longer appointment?" is an ordinary request schedulers handle daily, and it tells your clinician what the visit is so they are not discovering it at minute six.

If you cannot get more time, prioritize ruthlessly. §39.2's disclosure and §39.4's question 5 are worth more than any study you could bring.

💊 In the Clinic — the most underused person in the building

Pharmacists are the most underused clinical resource available to most patients, and for the specific questions this book raises they are frequently better positioned than a physician.

Consider what a pharmacist has. Training concentrated on exactly what you are asking about: what a compound is, what class it belongs to, how it is absorbed and cleared, what it interacts with, what the formulation implies. Interaction databases they use daily rather than occasionally. Familiarity with compounded preparations and with what a legitimate label should contain. And — the part that matters most — often more available minutes than a physician.

A community pharmacist can usually answer an interaction question at the counter, and many will spend ten minutes on a question a physician does not have ten minutes for. If your question is "does this interact with what I'm already on," "what does this label actually say," or "what class is this," a pharmacist may be the fastest and most precise source you have.

What a pharmacist cannot do: diagnose you, interpret your symptoms against your history, order or interpret your laboratory results, or take responsibility for your ongoing care. They supplement the relationship in §39.10 rather than replacing it, and what you learn from them should still make its way into your record.

39.6 When the honest answer is "we don't know"

You ask the five questions. Your clinician looks at you and says: we don't know.

Most people hear that as evasion, or a polite refusal, or an admission that the clinician will not be helpful. It is none of those. It is information, and often the most valuable sentence in the appointment.

Think about what has to be true for a clinician to say it. They are in a system that rewards confidence (§39.1). They know it risks disappointing you. They could produce a plausible-sounding answer instead and nobody would check. A clinician who says the evidence is not there is telling you something true and unpopular at a small professional cost — a considerably better signal about how they practice than a confident answer built on nothing.

The absence of evidence is a real state of the world and a common one. Chapters 17 through 27 are substantially a catalog of compounds where the honest verdict is that human evidence does not exist. A clinician who reproduces that verdict is agreeing with the best reading of the literature. They are not failing to know something; they are correctly reporting that nobody knows it.

Three sentences that sound alike and mean different things

Here is the distinction that is worth the whole section.

THREE SENTENCES, THREE DIFFERENT WORLDS

  "WE DON'T KNOW"        →  THE EVIDENCE IS ABSENT
                            Nobody has run the study. The field does not know.
                            Your move: is anything under way? What would you
                            want to see? (This is a ❌ or 🔬 in dossier terms.)

  "IT DOESN'T WORK"      →  THE EVIDENCE EXISTS AND IS NEGATIVE
                            Someone ran the study and it failed. This is a
                            STRONGER statement than the first, and rarer.
                            Your move: which trial? (Ch 28's nesiritide is the
                            canonical example — a compound that looked excellent
                            until an adequately powered trial said otherwise.)

  "I DON'T KNOW"         →  THIS CLINICIAN HAS NOT LOOKED INTO IT
                            Says nothing about the state of the field.
                            Entirely reasonable — nobody knows every compound.
                            Your move: is there someone who would? Can we find
                            out? Would you look at one abstract with me?

  ─────────────────────────────────────────────────────────────────────────
  ASK WHICH ONE YOU ARE HEARING. It is a friendly question and the answer
  reshapes everything you do next.

These three get conflated constantly, in both directions, by patients and by clinicians. A patient hears "I don't know" and concludes the field is uncertain — when the evidence may be clear and this clinician simply has not encountered it. A patient hears "we don't know" and concludes the compound was tested and found wanting, a strictly stronger claim than what was said. A clinician says "it doesn't work" as shorthand for "there is no evidence that it works," and the patient who later finds a positive study reasonably concludes the clinician was wrong.

The distinction between the first two is the one Chapter 5 built into the rating system: a ❌ is a statement about the evidence, not a verdict on the molecule. Some ❌ compounds have never been tested. Some have been tested and failed. Those are different situations with different futures, and a rating system that collapses them throws away the information you most need.

So ask. The question is easy and non-confrontational: "Is that 'nobody knows,' or 'it's been studied and didn't pan out,' or 'this isn't something you've looked into'? Any of those is fine — I just want to know which." Adding any of those is fine is not a softener. It is true, and it makes the honest answer available.

⚠️ Hype Check — "my doctor had never heard of it, so it must be cutting edge"

The claim, in its usual form:

"I asked my doctor about it and she'd never even heard of it. That's how far ahead of the mainstream this stuff is — the research is years ahead of what they teach in medical school."

What's true in it. Clinical practice does lag research. Physicians are not systematically taught about compounds outside the prescribing repertoire. Genuinely new agents exist that a general practitioner would not recognize. The premise is usually accurate.

Where it fails. The inference. There are two explanations for unfamiliarity, and the claim picks the far less likely one.

A: it is very new and the literature has not reached practice. This happens, and it is checkable — a genuinely new agent in serious development has a trail: a generic name from an international naming authority, registered trials, published phase results, a sponsor. Chapter 1 §1.8 gave you the naming test in a sentence.

B: there is no literature to reach practice. A compound with no completed human trials generates nothing for a clinician to have encountered — nothing in the journals they read, the guidelines they follow, or the references they consult, because it does not appear in the evidence base at all.

B is far more common and trivially distinguishable: look for the trail. A compound decades old, known only by a laboratory code, with no registered active trials, is not ahead of the mainstream. It is outside it, which is a different geometry entirely.

And the scale problem. No clinician can know every compound — there are thousands, most with no literature attached. Unfamiliarity with any one carries essentially no information about it. Treating it as a signal is treating the size of a haystack as evidence about one straw.

Verdict: the observation is usually true and the inference is usually backwards.

📊 Evidence Rating

Claim form — "my doctor had never heard of it, so it must be cutting edge." Rating: ❌ Hype outpaces evidence (as of this writing, 2026) Why: unfamiliarity is far more often explained by the absence of a literature than by novelty, and no clinician can know every compound in a space containing thousands with no published human data; the inference treats a base-rate fact about medical knowledge as a signal about one molecule. What would change it: for a specific compound, a checkable development trail — an assigned generic name, registered and recruiting trials, published human results — would make "genuinely new" the better explanation for that compound. This rates a claim form, and the fix is not to upgrade it but to replace it with the specific check.

39.7 When your clinician is dismissive

Sometimes the response will be flat, fast, and dismissive. What that means depends, and there is a reliable way to tell.

Sometimes dismissal is correct. If you ask about a compound with no human evidence and a great deal of online enthusiasm, "there is no good evidence for that" is not dismissiveness. It is accuracy, and it is the verdict this book issues in several chapters. A clinician who declines to be enthusiastic about an unevidenced compound is doing their job. Do not confuse I did not get the answer I wanted with I was dismissed.

Sometimes dismissal is reflexive. Per §39.1, your clinician has had this conversation many times and the pattern-match is fast. Occasionally it fires on something that deserved an individual look.

Here is the diagnostic, and it is clean:

Does the response engage with the claim, or with the category?

"There's no good human evidence for that. There are animal studies, which is where most of what you read online comes from, and those have been poor predictors in this area. Here's what I'd want to see." — This engages the claim. It may be entirely negative and it is still a real answer, checkable, leaving you better informed. A good clinical response even though it is a no.

"That's all nonsense. Where did you read that?" — This dismisses the category. Notice that the underlying verdict may be completely correct; the compound may well be nonsense. But the response tells you nothing you can check, attributes the problem to your judgment rather than to the evidence, and carries a real cost.

The cost is this: it teaches you not to disclose next time. That is the outcome §39.2 spent its whole length trying to prevent, and a category dismissal produces it efficiently. A patient made to feel foolish for asking about one compound will not mention the next one, and the next one might be the one that matters when they are being prepped for a procedure. The dismissal is a failure of the conversation even when the verdict is right, because it degrades the channel all future information has to travel through.

The repair move

If you find yourself in a category dismissal and want to recover the conversation, one question reliably works and is genuinely not adversarial:

"What would change your mind?"

That is it — the same question your dossier's Field 12 asks about every entry, and the question this book has been asking for thirty-eight chapters. It is not a challenge: it does not assert they are wrong, does not defend the compound, and does not require you to know anything. And the answer is informative either way.

"A randomized trial in humans with a clinical endpoint. There isn't one." — Excellent. Specific, falsifiable, exactly right. You now know what the gap is and what would close it.

"Nothing would change my mind." — Informative in a different direction. A position immune to evidence is not an evidence-based position, and you have learned something about how this clinician reasons that will matter in every future conversation.

"Honestly, I'd have to look into it." — The best of the three: honest, and it opens a door. This is §39.6's third sentence, offered voluntarily.

A note on tone, which matters more than the words. Ask this the way you would ask a colleague, not an opponent. Delivered as a gotcha it becomes a challenge and you get a defensive answer that tells you nothing; delivered as curiosity it usually gets a real one. That is not manipulation, just the ordinary requirement that a question sound like a question.

When to move on

Rarely, the door stays closed across more than one appointment and more than one topic. If a clinician cannot be told things, the relationship cannot do what relationships are for (§39.2), and considering a change is legitimate. Two guardrails, because this is where readers go wrong.

One bad conversation is not a pattern. Clinicians have bad days, run late, and get short. Judge across encounters, not one.

Do not shop until you find agreement. Changing clinicians until one says yes does not improve your information — it selects for agreement, the same error as selecting the testimonials that confirm what you hoped (Chapter 6). The correct reason to change is that the channel is broken, not that the answer was no. Those feel identical from the inside and are completely different; the test is whether you would have stayed had a different clinician given the same answer more carefully.

39.8 When your clinician is enthusiastic

Almost everything written about this topic assumes your clinician will be skeptical and that your problem is being taken seriously. There is a second situation that gets almost no attention and that many readers will actually encounter: the clinician who is enthusiastic. Who recommends the compound. Who has a protocol. Who tells you they have seen remarkable results. Who is warm, engaged, unhurried, and interested in exactly what your last clinician brushed past.

That feels like validation, and after §39.7 it can feel like relief. Slow down there, because of an uncomfortable structural fact:

A clinician recommending a compound is not automatically a stronger signal than a clinician refusing one.

Both are clinical judgments made by trained people. Neither is evidence about a molecule. And enthusiasm is easier to produce than restraint, because yes costs a clinician less social friction than no — the patient is happy, the appointment ends well, and nobody feels dismissed.

Ask the five questions regardless of direction

This is the operational rule and it is simple: §39.4's five questions get asked no matter which way the recommendation points.

Most people do the opposite without noticing. When a clinician says no, they interrogate. When a clinician says yes, they relax. That asymmetry is not skepticism — it is motivated reasoning wearing skepticism's clothes, and it is the bias Chapter 1's dossier asked you to write down and date.

Run all five. Evidence in someone like me? — an enthusiastic recommendation should answer this as specifically as a cautious one; if the answer is clinical experience rather than trial data, hear it as what it is, Chapter 6's territory, valuable as a hypothesis and weak as evidence. What's the endpoint? — if the expected benefit is a number on a panel, ask what that number is linked to; practices built around optimization frequently work in surrogates, because surrogates are measurable at every visit and outcomes are not. Alternatives, including nothing? — the most diagnostic of the five here, since a recommendation that never mentions an alternative or waiting was not weighed against anything. What would you expect, and by when? — a prediction, with a date. What would make you stop this? — ask it and listen carefully; it is the question that most reliably distinguishes a considered protocol from an enthusiastic one.

And then ask the structural question

There is one more question, and it applies only here: does this practice sell what it is recommending?

You can usually establish this without asking anyone. Does the clinic dispense the compounds it prescribes from its own inventory? Does it run a membership model in which continued treatment is continued revenue? Does it sell an in-house line? Is the follow-up interval tied to a purchase?

None of those facts is an accusation. They are structural features, and structural features are knowable in a way that motives are not.

Chapter 42 rated exactly this claim form, and the conclusion there is the one to carry: a conflict of interest is a reason to check the evidence, not evidence of dishonesty. The cynical reading is easy and wrong. Physicians who dispense are not thereby corrupt, and many practices dispense for sound reasons — supply reliability, convenience, quality control over compounded preparations, and the fact that in some areas the alternative is a pharmacy an hour away. A financial relationship raises the value of checking. It settles nothing.

What it does mean is that you cannot use the recommendation itself as your evidence. This is the part that generalizes:

"A doctor recommended it" is not evidence about a molecule.

Chapter 38 §38.10 established the parallel. A prescription is a clinical judgment about one patient at one moment, made under uncertainty with incomplete information by a person exercising professional discretion. It is a legitimate and often excellent thing. It is not a demonstration that a compound works, and it was never intended to be.

A reader who has learned to discount a testimonial from a stranger but not a recommendation from an enthusiastic clinician has learned the easy half of the lesson. Both are individual judgments about individual cases. One comes with a white coat, which changes how it feels and not what it is.

📊 Evidence Rating

Claim form — "my doctor prescribed it, so it must be well supported." Rating: ❌ Hype outpaces evidence (as of this writing, 2026) Why: a prescription is a clinical judgment about one patient, not a statement about an evidence base; it may be legitimate and off-label (Chapter 38 §38.5), may rest on clinical experience rather than trial data, and may reflect a practice's structure — the inference from "prescribed" to "well supported" skips every step that would establish the latter. What would change it: nothing about the prescription itself. Repair the claim by replacing it — ask what the prescription rests on (§39.4, question 1) and rate that. A prescription backed by an approved indication with adequately powered trials in your population is well supported, and the support comes from the trials, not from the prescribing.

39.9 What a clinician cannot do for you

Unrealistic expectations cause a large share of the frustration in these conversations. Here are five limits, directly. None is a failure of any individual clinician.

They cannot verify what is in an unregulated vial. Chapter 34 explained why: no chain of custody, no trustworthy analytical certificate, no assay. They cannot tell you whether the contents match the label, whether the concentration is right, whether it is sterile, or whether the peptide inside is the one claimed. Bringing the vial is still useful — a clinician reads a label better than you do — but that is information about what someone wrote, not what is inside. Closing that gap requires a laboratory, a method, and a reference standard.

They cannot make an unapproved compound safe by supervising its use. This is worth stating carefully in both directions, because both the credulous and the cynical version are wrong.

Supervision adds real things: baseline measurements, so a later abnormality has something to be compared against; interval monitoring aimed at foreseeable harms; interaction checking against your full medication list; someone who knows your history and can interpret a finding in context; and a response pathway — a person to call, a plan for what happens if something goes wrong. Chapter 19 §19.7 laid this out and none of it is trivial. It is the difference between an unknown risk you are carrying alone and an unknown risk someone is watching with you.

And it adds no evidence. Supervision does not tell you whether the compound works. It does not convert an untested molecule into a tested one, and it cannot reduce risks nobody knows to look for — which, for a compound with no human trial data, is by definition most of them. Monitoring is a search, and a search finds what it is looking for. Supervision changes the risk profile without changing the unknowns.

They cannot generate the missing trial. If there is no adequately powered human study for your indication, your clinician cannot produce one in an appointment and their opinion does not substitute for one. A thoughtful clinician's judgment about an unstudied compound is worth having and is not worth mistaking for data, and the good ones will say so unprompted.

They cannot tell you what will happen to you individually. This is Chapter 5's central distinction and the one patients most want overridden. Evidence describes what happened, on average, to a group who met a list of criteria. A clinician can tell you that most people in a trial like you lost a certain amount of weight, or that a certain fraction had a certain side effect. They cannot tell you which group you will be in, because nobody can, and a clinician who confidently claims otherwise is telling you about their calibration rather than about your future.

This limit is hard to accept, and the desire for a personal answer is not irrational — it is your life, and the average is not what you experience. But the honest response to "what will happen to me" is a range and a probability, and a clinician who gives you that is more useful than one who gives you a number.

They may have institutional or licensing constraints you will experience as unwillingness. I want to name this one specifically, because the reader will otherwise read a constraint as a personality.

Clinicians work inside systems. Employers and health systems set formularies and prescribing policies; malpractice insurers attach conditions; licensing boards in some jurisdictions have taken explicit positions on categories of prescribing; institutional protocols govern what can be ordered and documented.

So "I can't do that" sometimes means I am not permitted to, and it sounds exactly like I do not want to. Those are different situations with different responses, and you can tell them apart by asking — not as a challenge, just as a question. "Is that a policy thing or a clinical judgment thing?" is ordinary, and most clinicians answer it directly. If it is a constraint, you have learned something about the system rather than about your clinician, and you have not damaged the relationship by treating a rule as a refusal.

📊 Evidence Rating

Claim form — "medical supervision reduces the risk of using an unapproved compound." Rating: ⚠️ Promising but preliminary (as of this writing, 2026) Why: supervision genuinely adds baseline and interval monitoring, interaction checking against a full medication list, and a defined response pathway (Chapter 19 §19.7), which plausibly reduce the severity and duration of foreseeable, detectable harms; it adds no evidence about whether the compound works and cannot detect harms nobody knows to monitor for, which for an untested compound is most of them. ⚠️ rather than ✅ because the direction is sound and the magnitude is unquantified — there is no trial of supervised versus unsupervised use of unapproved peptides, and unlikely ever to be one. What would change it: comparative harm-rate data under supervision versus without, for this class of use. Note that ⚠️ here is not an endorsement of the underlying use — it rates what supervision adds to it.

39.10 The conversation as a repeated game

One frame remains, and it reorganizes everything above.

Most people approach an appointment as a single encounter with a defined outcome. There is a thing they want, a person who can grant or refuse it, and the appointment succeeds or fails. That framing makes people adversarial, and it is wrong about what is actually happening.

A clinical relationship is a repeated interaction, and its value compounds.

Consider what a clinician who has known you for five years has that a new one does not. They know what your laboratory values normally look like, so an abnormal result can be compared against your own baseline rather than a population range — and the difference between "high" and "high for you" is frequently the whole diagnosis. They know how you describe symptoms, whether you understate or overstate, and how you respond to treatment. They know what you have already tried and what happened. They have seen you well, which is what lets them recognize you unwell.

That accumulated context is not a nice-to-have. It is diagnostically load-bearing, and it is built one appointment at a time out of accurate information. Which reframes everything in this chapter as an investment rather than a tactic:

Disclosing (§39.2) builds a record that is true, the substrate everything else runs on. Asking answerable questions (§39.4) establishes what kind of patient you are — one who wants information rather than agreement; clinicians calibrate to this quickly and it changes how they talk to you. Accepting "we don't know" (§39.6) makes it cheaper for your clinician to be honest next time, since uncertainty punished with disappointment gets replaced by confident-sounding filler. Not treating a constraint as a refusal (§39.9) keeps you from burning a relationship over something they did not control.

And there is one more, which is the reciprocal move and the note this chapter should end on.

Bring back the outcome

Come back and report what actually happened. Including when nothing happened. Including when you stopped.

Almost nobody does. Think about the selection: a patient who feels better mentions it, sometimes glowingly. A patient who notices nothing quietly stops and never brings it up, because there is nothing to say and it feels faintly embarrassing. A patient who stops because of a side effect often just stops.

So most clinicians never learn how these stories end. The information reaching them is filtered by exactly the mechanism that makes online testimony unreliable — Chapter 6's mechanism, running inside a clinic. Your clinician's impression of a compound is assembled from a non-random sample biased toward success, in precisely the way that produces overconfidence.

You can fix a small piece of that, and it costs one sentence. "That thing I mentioned — I stopped it in March. Nothing happened, good or bad." Or: "I stopped because of the nausea, about three weeks in." Or: "It seemed to help for two months and then it didn't." Every one is data your clinician does not otherwise have, and they will carry it forward to other patients.

This connects to Chapter 19 §19.8, and it is the right place to close. The gray market cannot produce evidence: no trials, no registries, no adverse-event reporting, no follow-up, no mechanism by which a bad outcome in one person becomes information available to the next. That is a consequence of how the market is organized rather than a failure anyone could fix from inside it.

Which means the observations reaching clinicians, one patient at a time, are the only systematic information ever generated about these compounds. It is weak evidence — uncontrolled, unblinded, unrandomized, subject to every bias in Chapter 6 — and for a great many compounds in Part III it is the only evidence there will ever be. A patient who reports back is contributing to it. That is a small thing, it is not nothing, and it is the one contribution available to a person who is not running a trial.

🔍 Check Your Understanding

1. Give two specific things a clinician who has known you for years can do that a new one cannot, and explain what each depends on.
2. Why is a clinician's impression of what a compound does likely to be biased toward success, and which chapter's mechanism is producing that bias?
3. A reader says: "There's no point telling my doctor it did nothing — that's not information." Answer them.

📋 Your Evidence Dossier

Field 11 (Verdict) — as a conversation.

Your dossier now has verdicts. Field 11 holds a sentence you wrote for yourself: a rating and a short statement of where the evidence stands. Appendix C has the full field definitions; Chapter 40 assembles everything.

This chapter asks one new thing of Field 11: convert it into a question.

Why a verdict is the wrong thing to bring into an appointment

A Field 11 verdict is written for your own use, and its natural grammar is that of a conclusion:

"Not enough evidence. No completed human trials I could find. ❌ as of March."

That sentence does its job perfectly in your dossier and is exactly the wrong form to bring into an appointment. The reason is structural rather than stylistic.

A conclusion invites agreement or disagreement. A question invites information.

Say it out loud to a clinician and you have handed them a binary: agree, in which case you learn nothing you did not already know, or disagree, in which case you are in a debate about your conclusion rather than a conversation about the evidence. You did not go in to have your verdict scored. You went in to find out what you were missing.

The conversion

Each rating converts into a characteristic question:

FIELD 11 VERDICT  →  QUESTION FOR AN APPOINTMENT

  ❌  "no good human evidence"
      →  "What would you want to see before you'd consider something like this?"
         Gets you: their evidentiary standard, stated explicitly. Useful on every
         future topic, not just this one. (This is §39.7's repair move, asked
         before anything has gone wrong.)

  ⚠️  "some human data, doesn't settle it"
      →  "Is there a population where this is better established than it is for me?"
         Gets you: the boundary of the evidence. ⚠️ almost always means the data
         exist somewhere and the question is whether the somewhere includes you.
         (§39.4, question 1.)

  ✅  "well supported — for a different indication"
      →  "That's established for ___. Does any of it apply to my situation?"
         Gets you: an honest assessment of transfer. The most common error with a
         ✅ is assuming the rating travels with the molecule rather than staying
         with the claim.

  🔬  "frontier — too early to rate"
      →  "Is there anything worth watching here, or is it too early to have an
         opinion?" Gets you: whether trials are running and when they read out.

  ─────────────────────────────────────────────────────────────────────────────
  Note what all four have in common: none of them asks the clinician to agree
  with you, and none of them is a request for anything.

Notice that last line. None of these questions asks for a prescription. That is deliberate and it is not squeamishness. A question that is secretly a request gets answered as a request, and you will get the clinician's answer to the request rather than to the question. If you want to ask for something, ask directly and separately — that is a legitimate conversation deserving its own sentence rather than being smuggled inside an inquiry.

The conversion is also a test

Here is what makes this worth doing even if you never have the appointment.

A verdict that cannot be turned into an answerable question was probably a feeling.

Try it on your entries. Most convert cleanly, almost mechanically. Some will not — you will find yourself writing something vague, unanswerable, or on inspection a statement with a question mark attached.

When that happens, go back to Field 11. You will usually find one of three things: the verdict rested on a general impression rather than anything specific; or it rested on how the sellers sounded rather than what the evidence showed, which is rule 4, never downgrade with distaste, catching you in the act; or it was about a molecule rather than a claim with a population and an endpoint, and molecules do not have verdicts (rule 1). All three are fixable and worth finding.

Do this now

1. Pick your top dossier entry — the one you would actually raise in an appointment.
2. Write out Field 11 verbatim, without improving it.
3. Convert it to a question using the pattern above.
4. Read it aloud. If you would not be comfortable asking it, or it sounds like an argument, rewrite it. The test: could a clinician who disagreed answer it without either of you becoming defensive?
5. Note whether the conversion was easy. If not, write one line on why — often more valuable than the verdict was.
6. Repeat for every entry, then add the questions beside the verdicts with today's date.

Chapter 40 will ask you to look at both. The gap between what you concluded and what you could actually ask is one of the more revealing things in the project.

Conclusion

The conversation goes badly for structural rather than personal reasons. Both people arrive braced — the patient for judgment, the clinician for a request they cannot responsibly grant — and neither says what they want. Appointments are short, no clinician can know every compound, and "I don't know" costs something in a system that rewards confidence. None of that is anyone's fault.

Tell your clinician what you are taking. The reasons are clinical, not moral: differential diagnosis fails on an incomplete history, interactions cannot be checked against a compound nobody knows about, peri-operative protection cannot be applied to an undeclared exposure, and monitoring cannot look for a harm it does not know about. The GLP-1 gastric-emptying issue is the concrete case. The risk of an uncomfortable conversation is smaller than the risk of being treated by someone working from an incomplete picture, and it is not close.

Five questions work on any medical claim. Evidence in someone like me? What is the endpoint? What are the alternatives, including nothing? What would you expect, and by when? What would make you stop? The fourth converts hope into a falsifiable prediction; the fifth has to be asked before you are invested. Ask all five regardless of which direction the recommendation points — interrogating a no and relaxing at a yes is motivated reasoning, not skepticism.

Learn to hear the three sentences. "We don't know" means the evidence is absent. "It doesn't work" means it exists and is negative. "I don't know" means this clinician has not looked into it.

And know what a clinician cannot do. Verify a vial. Make an unapproved compound safe by watching you use it — supervision adds monitoring and a response pathway, and adds no evidence. Generate the missing trial. Tell you what will happen to you individually. And some of what looks like unwillingness is a constraint they did not choose, distinguishable by asking.

The frame holding all of it together is that this is a repeated game, compounding on accurate information. Which is why the last move belongs to you: come back and say what happened. Including when nothing did. Including when you stopped.

Chapter 40 closes the dossier and asks what you now believe — and, more usefully, what you believed in Chapter 1 and were wrong about.

Key Terms

Disclosure — telling a treating clinician what you are actually taking, including unprescribed and unregulated substances. A clinical necessity rather than a moral obligation: differential diagnosis, interaction checking, peri-operative assessment, and monitoring all take the medication history as an input and fail silently when it is incomplete.

Differential diagnosis — the ordered list of possible causes a clinician builds and works through, constructed from what they know about you. An incomplete history produces a workup that solves the wrong problem.

Peri-operative — the period surrounding a procedure: preparation, procedure, recovery. Where undisclosed exposures cause some of their most concrete harm.

Gastric emptying — the rate at which stomach contents pass into the small intestine. Slowed by GLP-1 receptor agonists as part of their mechanism, so a correctly fasted patient may still have retained contents.

Aspiration risk — the risk of stomach contents entering the airway, elevated when protective reflexes are suppressed by sedation or general anesthesia.

Stopping rule — a condition, defined in advance, under which an intervention will be discontinued; the clinical analogue of a trial's pre-specified interim stopping criteria.

Falsifiable prediction — an expectation specific enough to be wrong, with a magnitude and a date. An expectation that cannot fail provides no point at which to stop.

Comparator — what an intervention is measured against. Doing nothing is a legitimate comparator and a real trial arm.

Surrogate endpoint — a measurement standing in for an outcome people care about. Frequently what an optimization-oriented practice is actually treating.

Off-label — prescribing an approved drug outside its approved indication, population, or route. Legal, common, often appropriate, and not a statement about evidence strength.

Conflict of interest — a structural situation in which a financial or professional interest could influence a judgment made impartially. A reason to check the evidence, not evidence of dishonesty.

Dispensing practice — a practice that supplies the products it recommends from its own inventory. A knowable structural fact, not an accusation.

Category dismissal — rejecting a claim by rejecting its class without engaging the claim. Often reaches the correct verdict and still damages the conversation.

Monitoring — baseline and interval measurement aimed at foreseeable harms. Unavailable for an undisclosed exposure, and not a substitute for evidence.

Chain of custody — a documented, unbroken record of a substance's handling from manufacture to administration. Absent for gray-market compounds.

Clinical judgment — a trained decision about a particular patient under uncertainty. Legitimate, often excellent, and not a demonstration that a compound works.

Shared decision-making — clinician and patient combining evidence with the patient's values and circumstances. The five questions of §39.4 are its patient-side toolkit.

Repeated interaction — a relationship in which the same parties engage many times, so each encounter's value depends on accumulated history.

Therapeutic relationship — the ongoing clinical relationship, whose diagnostic value comes substantially from accumulated context: your baselines, your patterns, what has been tried.

Spaced Review

1. (Ch 39) A clinician tells you a compound "doesn't work." Using §39.6, write the one-sentence question you would ask next, and explain what each of the three possible answers would mean for what you do afterward.

2. (Ch 5 + Ch 39) You find a positive trial of a compound for a condition you have. Which two of the four basics from §39.5 most directly test whether the result applies to you, and why does Chapter 5's population argument make one of them decisive?

3. (Ch 19 + Ch 39) A clinic offers "medically supervised" use of an unapproved compound. State what the supervision adds, what it does not add, and why the rating in §39.9 is ⚠️ rather than ✅ or ❌.

4. (Ch 38 + Ch 39) Someone says a compound must be well supported because a physician prescribed it. Name the two distinct errors in that inference, referring to Chapter 38's treatment of off-label prescribing, and explain why the same reasoning applies to an enthusiastic recommendation from a practice that dispenses.

5. (Ch 39) Take one ⚠️ entry from your dossier, write its Field 11 verdict verbatim, and convert it to a question. Was the conversion easy? If not, what does that tell you about the verdict?