Chapter 10 — Exercises
Items marked † have worked solutions in Appendix M.
A. Recall
10.1 State the four candidate explanations for why one drug appears to work on several diseases.
10.2 † What did FLOW study, in whom, and what happened to the trial?
10.3 What do MASLD and MASH stand for, and what were they previously called?
10.4 What is HFpEF, and roughly what share of heart failure does it account for?
10.5 † What is the apnea-hypopnea index, and why is sleep apnea the chapter's clearest case of a weight-mediated benefit?
10.6 State the ratings issued in this chapter and the population attached to each.
B. Early stopping
10.7 † Explain why "stopped early for efficacy" is simultaneously good news and a statistical complication.
10.8 Why are trials that cross an interim boundary enriched for chance-favorable results? Relate this to Chapter 9's Phase 2 selection argument.
10.9 Given an early-stopped trial, which should you trust more — the direction of the effect or its magnitude? Explain.
10.10 A data monitoring committee recommends stopping. Describe the ethical argument for stopping and the scientific argument for continuing. Is there a resolution?
C. Endpoints
10.11 † List four distinct problems with a liver biopsy endpoint, and state which one makes it a surrogate rather than merely a difficult measurement.
10.12 Design an alternative endpoint for a MASH trial that avoids biopsy. What do you lose?
10.13 HFpEF trials in this chapter used symptom and function endpoints. State what a hard-endpoint trial would use instead, and why the softer endpoints were used.
10.14 The apnea-hypopnea index is an objective measure. Is it a surrogate? For what?
D. Mechanism
10.15 † For each indication in §10.10's table, state whether the benefit is more plausibly weight-mediated or more plausibly direct, and give your reason.
10.16 §10.9 argues "anti-inflammatory" is the vaguest mechanism claim in medicine. Explain, and state what would have to be specified to make it a strong claim.
10.17 Weight loss itself reduces inflammation. Explain why this makes explanation ③ hard to distinguish from explanation ①, and propose a study design that would help.
10.18 † Design a study that would determine whether the cardiovascular benefit is mediated by weight loss. Is your design feasible? What are its limitations?
E. Why 🔬 and not ⚠️
10.19 State the difference between ⚠️ and 🔬 in this book's rating system, in one sentence each.
10.20 † Explain precisely why the Alzheimer's program is 🔬 despite having a mechanistic rationale, animal data, and supportive epidemiology.
10.21 What is confounding by indication, and why does it specifically undermine the observational dementia findings?
10.22 The addiction hypothesis came from patients rather than from a pharmaceutical target hypothesis. Does that make it more or less credible? Argue carefully.
F. Compression errors
10.23 † Define upward compression and downward compression in your own words, give an example of each, and state what they have in common.
10.24 A friend says "they're claiming it treats everything, so I don't believe any of it." Identify which compression error this is and respond.
10.25 A different friend says "it's proven for heart disease, so the Alzheimer's research is probably right." Identify the error and respond.
G. "Explain this to a friend"
10.26 † Explain why a drug being proven for one disease tells you nothing about another — without implying the other research is illegitimate.
10.27 Explain what "the trial was stopped early because it was working" means, and why that is slightly less straightforwardly good than it sounds.
10.28 Your friend read that these drugs help with alcohol use. Respond accurately, taking the observation seriously without endorsing the claim.
H. Judgment
10.29 §10.1's explanation ④ is "some of these will not replicate." Which indication in this chapter would you bet against, and why? Write it down and date it.
10.30 † Sleep apnea benefit is probably mechanical. §10.6 says this is "a complete explanation rather than a criticism." Do you agree? When would a mechanical explanation be a criticism?
10.31 This chapter rates one drug class ten different ways. Is that a useful level of granularity or an unusable one for a patient? What would you give a patient instead?
I. Evidence Dossier extension
10.32 † For every peptide in your dossier, list every distinct claim you have encountered and rate each separately, with population, endpoint, rating, and falsifier.
10.33 Check your entries for upward compression: did you write favorably about a second indication on the strength of a first? Note any instance.
10.34 For any compound where all your rows have the same rating, ask whether that is because the compound genuinely has one claim or because you have not looked hard enough. Record which.