Instructor Notes — Chapter 14

Opens Part III. Whatever else this session accomplishes, it should install four frames that the rest of Part III leans on by name: mechanism versus evidence, surrogate versus outcome, replacement versus override, and how to reason from a natural experiment. Growth hormone was chosen to open the part precisely because all four are load-bearing here and none of them can be skipped.

The one thing to protect. This chapter must not become a debunking. A substantial fraction of any real audience is currently using growth hormone or is seriously considering it, and a session that plays the anti-aging claim for laughs loses exactly the students the chapter was written for — and, as a bonus, teaches them that the ❌ tier is a sneer rather than an evidentiary judgment. If you find yourself getting laughs at the clinic email in §14.6, redirect: ask the room which clause in it is true.

Common misconceptions

1. "❌ means it doesn't work." The single most common misreading in the entire book, and growth hormone is where it bites hardest, because the molecule demonstrably does something. Correct it early and correct it out loud. The ❌ says the confident claim is unsupported by outcome data in this population. Students who leave with the wrong version will lose arguments to anyone who has read Rudman.

2. "A rating belongs to a molecule." Students arrive expecting a verdict. Growth hormone is the cleanest available demonstration that a verdict is the wrong shape of answer. Make them write both ratings out with populations and endpoints before you discuss either.

3. "Growth hormone only matters for growth." The name misleads. Get the direct/IGF-1 split on the board early; it resolves most downstream confusion, including why the same hormone can raise glucose while its mediator lowers it.

4. "Low growth hormone on a blood test means deficiency." Almost nobody arrives knowing about pulsatility, and it is genuinely counterintuitive that a hormone can be undetectable in a healthy person. This is the fastest, most durable practical takeaway in the chapter and it should be non-negotiable by the end of the session.

5. "A normal IGF-1 rules out adult deficiency." It does not, and students who have just learned to love IGF-1 will over-apply it. Worth one sentence, because it is the kind of nuance that separates someone who has understood the diagnostic logic from someone who has memorized a shortcut.

6. "Acromegaly is what will happen to you." The alarmist misreading. Correct it as firmly as you correct the dismissive one — a student who overclaims the acromegaly analogy has committed the same category of error as the clinic, in the opposite direction.

7. "The Rudman study was bad science." It was not. It was a competent small study asked, for thirty-five years, to answer questions it was never built to address. Students find "the paper is fine and the citation is not" genuinely surprising, and that surprise is the lesson.

8. "Suppression of endogenous production means the drug damaged something." Expected physiology. Worth flagging because it is one of the few places students can be given a correct, reassuring answer, which builds credibility for the places where the answer is not reassuring.

The hardest point to teach

That the effect is real and the claim is unsupported, simultaneously.

Students have two available templates for a claim: "true, and the evidence supports it" and "false, and the evidence refutes it." Growth hormone fits neither. The body composition change is real, replicated, mechanistically explained, and visible to the user in a mirror — while the claim built on it has been tested and did not hold up.

Most students will collapse this into one of the two familiar templates within about ninety seconds of leaving the room. The collapse in the debunking direction is more common in a classroom and the collapse in the credulous direction is more common everywhere else.

What works: make them state the surrogate and the outcome as separate objects, on the board, in their own words, before you ever discuss whether growth hormone "works." If lean mass and strength are two distinct things in their heads before Rudman appears, the whole chapter lands. If they are the same thing, nothing you say will fix it.

What does not work: repeating "surrogate endpoint" as a phrase. It becomes a password students produce on demand without the underlying distinction. Demand the concrete version every time — what did they measure, and what does anyone actually want?

Demonstrations

D1 — The empty search (10 min, live, high impact). Open ClinicalTrials.gov in front of the class. Search somatropin, filter by condition, and let them see the same molecule across GH deficiency, Turner syndrome, Prader-Willi, short bowel syndrome, HIV lipodystrophy. That is Rule 6 in one screen. Then search for a completed, multi-year, adequately powered anti-aging trial with functional endpoints in healthy adults. The absence is the demonstration. Nothing you say about the ❌ lands as hard as watching a search return nothing.

D2 — The label versus the landing page (15 min, pairs). Half the room reads the Indications and Warnings sections of a somatropin label on DailyMed; half reads a clinic's marketing page for the same molecule. Then they trade findings. Ask one question: what is on one document and not the other, in both directions? The marketing page has endpoints the label does not; the label has warnings and contraindications the marketing page does not. Students discover the chapter's thesis themselves, which is worth more than being told it.

D3 — Strike the sentence (5 min, opener). Put the Rudman finding on the board without its final sentence. Ask what it supports. Collect answers — they will overreach, reliably and enthusiastically. Then add: It did not measure strength, function, or long-term safety. Ask which of their answers survive. This takes five minutes and is the most efficient teaching device in the chapter; if you run only one demonstration, run this one.

D4 — Two natural experiments (10 min, board work). Draw an axis. Put acromegaly at one end and Laron syndrome at the other, with their outcomes underneath. Ask the room to sketch the curve between them and mark where they are confident and where they are guessing. Almost everyone draws a clean U-shape and almost nobody can defend the position of its minimum. That gap is the point.

D5 — Rewrite the email (15 min, small groups, best closer). Give groups the §14.6 clinic email. Task: rewrite it so every claim is defensible, keeping as much as can honestly be kept. They will discover that the honest version is much shorter, much less appealing, and — this is the part worth naming out loud — would not sell anything. That is a real finding about the economics of the space, not a moral lesson, and it should be presented as such.

D6 — The four sentences (5 min, drill). Have students recite the cancer construction from §14.7 in order. Then hand half the room "sentences 1–2 only" and half "sentences 3–4 only" and ask each half what a reader would conclude. Alarmism and false reassurance, generated from true statements by omission. Fast, memorable, and transferable to every risk question in the book.

Timing

If you have only 90 minutes: D3 opener → §14.5 Rudman → §14.6 industry + the ❌ → §14.9 two ratings → D5 closer. That path teaches the transferable skill. Everything else is support for it.

If you have only 50 minutes: D3, §14.5, §14.9. Assign §14.1–14.4 as reading.

Assessment notes

The quiz (quiz.md, 22 items) has four items under a dedicated "Rating skill" heading — 16, 17, 18, 19. Weight them. A student who scores well on physiology and poorly on those four has learned facts and not the method, which is the failure mode this book exists to prevent.

Items that discriminate well: - Q4 option (d) — students who think IGF-1 is unaffected by nutrition and age have memorized "IGF-1 is the stable one" without understanding why it needs age-referenced ranges. - Q8 — students who pick (a) "it's from 1990 and therefore outdated" have not understood scope; the date is not the problem. - Q14 — the only item where both wrong answers are overclaims in opposite directions. Getting this right is a good proxy for having internalized the four sentences. - Q16 — the answer is (c) even though (b) and (d) name real defects. Students who pick (b) or (d) are reasoning correctly but have not identified the structural error. Consider partial credit and discuss it; the discussion is more valuable than the mark.

Short answer 20 is the single best one-item check on the chapter. If the required final sentence is missing, the student has not internalized the chapter's central discipline regardless of what else they wrote. Grade it strictly and say in advance that you will.

Exercises. Sections D and E are where understanding actually shows. In Section D, grade the "what's true in it" half at least as heavily as the "what's wrong" half — students who can only find faults have learned skepticism, not evaluation, and skepticism is the cheaper skill. E4 (writing to someone already using growth hormone) is the best single assessment in the chapter: it tests evidence literacy, tone, and restraint simultaneously, and it has an automatic-fail condition (any protocol, any moralizing) that makes grading unusually clean.

Section F feeds the running dossier project. F2 in particular — finding an item you had miscategorized — is worth collecting and discussing anonymously; the misclassifications are consistent across cohorts and seeing that is reassuring to students.

A note on the daggers. exercises.md has 36 items, of which 12 are marked †; worked solutions for all 12 are in the answers fragment. Solutions for D1, D4, D6, E2, and E4 are long and are written to be read as models of the reasoning, not just as answer keys. Consider releasing E4's solution to the class after grading.