Chapter 12 — Further Reading

How this list is organized. Three tiers, and the tier is part of the information.

• Tier 1 — Verified canonical. Primary sources whose existence and location this book can vouch for. These are institutional databases and official documents, not individual papers, because databases stay findable and paper citations rot. Go here first.
• Tier 2 — Attributed, specifics unverified. Real literatures and real bodies of work, described by topic and by search route rather than by exact citation. Where a figure appears it is given as a range and should be confirmed at the source.
• Tier 3 — Illustrative and constructed. Material created for this book. Useful for reasoning, not citable as evidence.

A warning specific to this chapter. Everything here about prices, coverage rules, shortage status, and what compounders may lawfully do changes on a timescale of months. As of this writing (2026) the compounding position in particular is unsettled and under litigation. Treat any secondary summary, including this book, as out of date until you have checked a Tier 1 source.

Tier 1 — Verified canonical

FDA compounding pages and compounding risk alerts — fda.gov, human drug compounding section. The authoritative statement of what compounders may and may not do, the difference between 503A pharmacies and 503B outsourcing facilities, the bulk drug substances lists, and the agency's specific alerts on compounded GLP-1 products including the salt-form and dosing-error concerns described in §12.5. If you read one thing from this list, read the current compounding alerts.

FDA Drug Shortages database — the searchable list that determines whether the compounding exception is open for a given drug. Also carries resolution notices. This is the single fact that drove the entire case study in case-study-01.md.

FDA registered outsourcing facility list — the roster of 503B facilities. A supplier claiming 503B status either appears here or does not, and checking takes under a minute.

DailyMed — dailymed.nlm.nih.gov. Full current labeling for every approved US product, including the exact indication wording, the eligibility criteria, and the "as an adjunct to a reduced-calorie diet and increased physical activity" clause discussed in §12.6. Read the label for the specific brand you care about; indications differ between products containing the same molecule.

Drugs@FDA — approval history, review documents, and supplements. Where to see what a regulator actually concluded, rather than what a press release said it concluded.

ClinicalTrials.gov — registration records for STEP and SELECT and their siblings. The registration record shows the pre-specified endpoints and the enrollment criteria, which is how you check whether a published headline matches the study that was designed.

National regulator equivalents outside the US — the European Medicines Agency, the UK MHRA, Health Canada, the TGA in Australia. Indications, thresholds, and availability differ by jurisdiction, and §12.6's refusal to state BMI numbers is the direct consequence. Use the regulator for the country you live in.

Health-technology assessment bodies — NICE in England, IQWiG in Germany, CADTH in Canada, PBAC in Australia. These publish their reasoning about whether a therapy is worth funding, which is the closest thing available to a public version of the decision described in case-study-02.md. Reading one of these appraisals end to end is the fastest way to understand what a payer is actually weighing.

Tier 2 — Attributed, specifics unverified

The STEP program (semaglutide weight management trials). Multiple randomized trials across populations with and without type 2 diabetes. Weight change at 68 weeks in the non-diabetic population is reported at approximately −15% of baseline against approximately −2.4% on placebo, with both arms receiving lifestyle intervention; the diabetic population is reported at approximately −10%. Search route: ClinicalTrials.gov for the registration records, then the primary publications.

SELECT (cardiovascular outcomes). Approximately 17,000 participants with established cardiovascular disease and overweight or obesity without diabetes, followed roughly three years. Approximately 20% relative reduction in major adverse cardiovascular events; absolute event rate approximately 8% → 6.5%, roughly 1.5 percentage points; number needed to treat roughly 65–70 over that period. Always report the relative and absolute figures together.

Randomized withdrawal and discontinuation literature. Trials in which participants are randomized to continue or stop after an initial treatment period consistently report substantial regain on discontinuation. This is the evidence base behind §12.9's claim that regain is expected physiology rather than treatment failure.

Real-world persistence and discontinuation studies. Claims-database analyses of how many people remain on therapy at 6 and 12 months. Reported persistence varies widely by population, coverage status, and data source; discontinuation within the first year is common rather than exceptional. Treat any single headline figure with suspicion and look at the denominator.

Set point, metabolic adaptation, and energy expenditure after weight loss. A long experimental literature showing that energy expenditure after weight loss falls by more than the loss of tissue mass predicts, and that hunger and satiety signalling shift in the direction of regain. Foundational to Position 1 in §12.9.

Heritability of body-mass index. Twin, adoption, and family studies. Reported heritability estimates vary substantially by design and population and are frequently over-interpreted; read at least two methodologically different studies before quoting a number, and remember that heritability within a population says nothing about what moved a population average.

Monogenic obesity and the leptin-melanocortin pathway. Rare single-gene disorders producing severe early-onset obesity. Small populations, enormous conceptual weight in the disease argument.

Weight stigma and health outcomes. A literature showing that experienced weight stigma independently predicts care avoidance and worse outcomes. Central to Position 4 in §12.9 and to §12.7's account of what telehealth actually removed.

GLP-1 agonists and eating disorders. Two distinct and under-studied literatures: concern about facilitation of restrictive disorders, and early clinical work on binge eating disorder. Both are small, short, and preliminary. Trials of weight management generally excluded people with active eating disorders, so the safety database does not cover them — absence of data, not evidence of safety.

Withdrawn weight-loss pharmacotherapy. Fenfluramine combinations (valvular heart disease), sibutramine (cardiovascular signal), rimonabant (psychiatric effects; withdrawn in Europe, never approved in the US). Read the regulatory withdrawal documents rather than retrospectives if you can find them; they show what was known when.

The inverse care law. Described in 1971 in The Lancet: the availability of good medical care tends to vary inversely with need in the population served. Short, famous, still accurate.

Pharmaceutical pricing structure. Literature on list-versus-net divergence, rebate flows, and the position of the uninsured as residual payer. Search terms: gross-to-net bubble, rebate walls, list-price inflation. The empirical work here is contested and the ideological loading is high; read at least one source you expect to disagree with.

International price comparison. Studies comparing prices for identical products across high-income countries. Directionally consistent — single-buyer systems pay less — with methodology disputes about the size of the gap. Do not quote a multiplier without reading how it was computed.

Aseptic processing and sterile fill-finish. Pharmaceutical engineering and regulatory guidance literature on why sterile injectable capacity is slow to build. Dry, technical, and the single best antidote to the assumption that manufacturing scales with money.

Tier 3 — Illustrative and constructed

These were built for this book. They are teaching devices, not evidence, and should not be cited.

• The four-price ladder (§12.1) — the ①②③④ diagram showing where a rebate acts and where it does not. Constructed to make the residual-payer structure visible in one image.
• The molecule-to-pen manufacturing chain (§12.3) — the six-stage diagram with the bottleneck marked at steps 4 and 5. Stage boundaries are simplified; the ordering and the location of the constraint are accurate.
• The 503A / 503B comparison table (§12.5) — accurate on the dimensions shown, not exhaustive.
• The off-label legitimacy spectrum (§12.6) — a five-row ladder from "published randomized evidence" to "the clinic sells it." A constructed heuristic, not a regulatory category.
• The five-feature rationing comparison (§12.8) — Chapter 11's framework applied to GLP-1 agonists. The framework is this book's; the underlying observations about insulin are not.
• The four positions table (§12.9 and key-takeaways.md) — a fair-presentation device. Real advocates hold mixtures rather than pure positions.
• The Field 10 worked demonstration (Dossier section) — compounded semaglutide, with the branded product as contrast. Populate it yourself and date it.
• The coverage-decision shape diagram (case-study-02.md) — orphan drug versus ordinary chronic drug versus this. Illustrative proportions, not measured ones.

If you only do one thing

Open the FDA compounding pages and the FDA Drug Shortages database, and look up the current status of one product yourself.

Not because the answer will be interesting — it may be entirely mundane — but because the entire case study in this chapter turned on a single database entry, and because the exercise permanently changes how you read a sentence like "there's a shortage, so it's legal." That sentence is either true or false on a given date, the answer is public, and it takes ninety seconds to find. Almost nobody arguing about it has looked.