Chapter 40 — Further Reading

This chapter is not about peptides, so neither is this list. Almost nothing below mentions a peptide. Everything below is about the thing Chapter 40 actually taught: how to hold a claim, how to check whether you are holding all of them the same way, and how to notice when your own file has started drifting.

The internal references come first, because for this chapter they are the primary sources.

Tier 1 — Start here

Appendix C — Your Peptide Evidence Dossier: Blank Workbook. The template, the field notes, and a completed entry filled to the standard the rest should aim at. If you read one thing alongside this chapter, read C.4 (what actually goes in each field) with your own document open.

Appendix A — the book's ratings, and Chapter 37 — the master table. Not to copy. To compare against, so §40.4's column exercise has something to be a column of.

Appendix F — red flags and Appendix H — worked evaluations. F is the fast heuristic layer; H is the twelve fields run at speed on compounds the chapters did not cover. Use H as calibration when your own entries feel either too generous or too severe.

Chapters 5, 6, 38, and 39. The four chapters this one assembles. Chapter 5 built Field 5, Chapter 6 built Field 12 and the testimonial dynamic behind §40.6, Chapter 38 built Field 7, and Chapter 39 built the verdict-to-question conversion in §40.7. Re-reading Chapter 6 immediately after §40.6 is unusually productive.

Evans, Thornton, Chalmers, and Glasziou, Testing Treatments: Better Research for Better Healthcare, 2nd edition (Pinter & Martin, 2011). The best plain-language book on why fair comparisons matter and what goes wrong without them. Written for non-specialists, freely available online, and the closest thing in print to a general-audience version of Field 5. If you buy nothing else on this list, this is the one that would have taught you the most before Chapter 5.

Woloshin, Schwartz, and Welch, Know Your Chances (University of California Press, 2008). Short, practical, and aimed squarely at the numeric confusions this book keeps returning to — particularly relative versus absolute risk, which is the most common way a trial result is oversold and which Appendix C's completed entry deliberately separates.

Cochrane's free "Evidence Essentials" modules (cochrane.org). Structured, short, and free. Covers randomization, bias, and systematic reviews at roughly the level Chapter 5 assumed.

Goldacre, Bad Science (Fourth Estate, 2008). The most entertaining route into publication bias, regression to the mean, and the anatomy of a bad health claim. Read it for the pattern recognition rather than the specific cases, several of which have moved on.

Tier 2 — Going deeper

Straus, Glasziou, Richardson, and Haynes, Evidence-Based Medicine: How to Practice and Teach It (Elsevier). The standard practical text. Its structure — ask an answerable question, find the evidence, appraise it, apply it — is recognizably the skeleton of the twelve fields, and its chapter on framing answerable questions is the formal version of §40.7's verdict-to-question conversion.

Guyatt, Rennie, Meade, and Cook, eds., Users' Guides to the Medical Literature (JAMA/McGraw Hill). A reference rather than a read-through. When you need to appraise a specific study design — a diagnostic test study, a harm study, a non-inferiority trial — this is where you look up what to check.

The CONSORT statement (consort-statement.org) and the PRISMA statement (prisma-statement.org). Reporting guidelines for randomized trials and for systematic reviews respectively. Read them backwards: the checklist of things a paper is supposed to report is also a checklist of things whose absence should worry you. Field 5's sub-lines are a compressed version of the same idea.

The GRADE working group (gradeworkinggroup.org). How professional evidence bodies grade certainty in a body of evidence, and how they separate certainty from the strength of a recommendation. That separation is the formal counterpart of §40.9's claim that a rating is an input to a decision rather than the decision.

ClinicalTrials.gov and the WHO International Clinical Trials Registry Platform. These belong in Tier 2 not as reading but as tools: they are how you answer Field 12's last line, is such a study underway or planned? Learn to search by intervention and by condition, and learn to read a registration record for its population, primary endpoint, comparator, and completion date. A registration is also how you notice that a trial reported something other than what it registered.

The FDA 510(k) database, the FDA and EMA product-label databases, and equivalent national registers. Field 7 work. Case Study 2 turns on the difference between clearance and approval, and these are where you check which one a given product actually has, and for what indication.

The WADA Prohibited List (wada-ama.org). Field 10 work, and the only correct source for it. This book states repeatedly that sport status must be verified against the current List rather than against a book, and this is why: it is revised annually.

Prasad and Cifu, Ending Medical Reversal (Johns Hopkins University Press, 2015). On established practices that were later shown not to work, and why they became established. The best available argument for why mechanism plausibility is not evidence — rating rule 3, told through cases where medicine itself broke the rule.

Tier 3 — Specialist and primary

Popper, Conjectures and Refutations (Routledge, 1963). The source of this chapter's epigraph and the philosophical spine of Field 12. Read the first essay. The claim that a theory's scientific status lives in what would refute it is, restated, the whole of §40.3's fourth check.

Turner, Matthews, Linardatos, Tell, and Rosenthal, "Selective Publication of Antidepressant Trials and Its Influence on Apparent Efficacy," New England Journal of Medicine (2008). The mechanism in §40.6, demonstrated. The authors compared trials registered with a regulator against what reached the published literature. It is the clearest single demonstration that what you can read is not what was found, and it is worth reading precisely because it examines the literature of an approved, well-studied drug class rather than a fringe one.

Ioannidis, "Why Most Published Research Findings Are False," PLoS Medicine (2005). The most-cited statement of why a single positive study is a weaker object than it appears. Field 5's replication line exists because of this argument.

Chalmers and Glasziou, "Avoidable Waste in the Production and Reporting of Research Evidence," The Lancet (2009). On how much research fails to answer a question anyone asked, or fails to be reported in a usable form. This is the formal version of Field 5's "what is conspicuously absent" line — the shape of a missing literature is informative.

Tetlock and Gardner, Superforecasting (Crown, 2015), and Tetlock, Expert Political Judgment (Princeton University Press, 2005). Not medical, and the closest thing on this list to a manual for §40.4 and §40.5. The central practices are dating your predictions, specifying in advance what would count as being wrong, and scoring yourself over time to find the direction of your error. That is the version history, Field 12, and the bias column, developed in a domain with no trials at all — which makes these the most useful preparation available for Part VIII.

Primary literature in your own entries. The real Tier 3 for this chapter is the two or three papers behind whichever Field 5 you are least sure of. Read the methods section and the participant table before the abstract. If you cannot state the population, the endpoint, the comparator, and the duration after ten minutes, the paper has told you something about itself.

If you only do one thing

Take the entry you care about most, read the study you specified in Field 12, and go look for it in a trial registry.

Search ClinicalTrials.gov or the WHO ICTRP for the intervention and the condition. Then answer one question: is anyone running the study that would settle this?

Three outcomes, all of them worth the twenty minutes:

• A registered trial matches what you specified. Record its identifier and its expected completion date in Field 12, and set a reminder for that date. You have converted an open question into an appointment.
• Trials exist but none of them tests what you specified — wrong population, surrogate endpoint, no comparator, too short. That is a finding about the field, and it belongs in Field 5's conspicuously-absent line.
• You cannot search for it, because your Field 12 was not specific enough to turn into search terms. This is the most common outcome and the most useful one. A Field 12 that cannot be searched cannot be checked, which means it cannot do the job §40.6 assigned it. Go back and rewrite it until it can.

Whichever you get, date the entry and save the version. That is the whole discipline of this chapter, executed once, on the entry that matters to you.