Chapter 16 — Supplements: The $50 Billion Industry Built on Hope — What Works, What Doesn't, and What's Dangerous

The Hook: The night Walt's blood test lied

Walt Prosser woke at about half past four with a pressure in his chest.

Not dramatic. Not the clutching-the-shirt version from films. A heaviness, and an ache down the inside of his left arm, and a feeling he later described as "like something was sitting on me." He is sixty-eight, has type 2 diabetes, takes a statin and a blood pressure tablet, and has all the risk factors you would list if you were making a list.

He woke his neighbour. The neighbour drove him to the emergency department, which was the right decision.

They did what you do. History, examination, ECG, and troponin — the blood test that detects cardiac muscle damage and which, more than anything else, determines whether a person with chest pain is admitted or sent home.

His troponin came back low.

Not equivocal. Low. Combined with a non-diagnostic ECG and a settling symptom, that is a reassuring picture, and the working impression moved toward something non-cardiac.

In the bag of medications his neighbour had grabbed off the kitchen counter were nine supplement bottles.

One of them was an "immune support" blend. Among its ingredients, in a quantity nobody had any reason to look at, was 3,000 micrograms of biotin — vitamin B7, roughly a hundred times the adequate intake, included because biotin is marketed for hair, skin and nails and gets added to a great many things.

High-dose biotin interferes with laboratory immunoassays.

Many immunoassays — including several troponin assays — use a biotin–streptavidin binding system as part of how they capture and detect the target molecule. Flood the sample with free biotin from a supplement and it competes with that system. Depending on the assay's design, the result can be falsely low or falsely high.

In sandwich-format assays, which many troponin tests are, the interference produces a falsely LOW result.

The FDA issued a safety communication about this in 2017, updated it in 2019, and reported that it had received a report of a death associated with a falsely low troponin result in a patient taking high-dose biotin.

What happened to Walt is that a junior doctor, going through the bag properly because someone had taught her to, read the label on the immune blend, saw the biotin, and said something.

They repeated the troponin on a different platform. It was elevated. He was admitted, investigated properly, and treated.

He is fine. He was, in the most literal sense, lucky in his junior doctor.

This is the chapter's thesis, and Walt is the argument for it: supplements are not "can't hurt."

They are unregulated, frequently unnecessary, occasionally dangerous, and — in ways almost nobody anticipates — capable of making your medical care worse without doing anything to your body at all.

And yet. This chapter also contains the strongest supplement recommendation in the entire book, for a compound that is cheap, extremely well studied, and genuinely works.

Both halves. That's Part III.

🏃 Fast Track: §16.3 (what works), §16.5 (what's dangerous), and §16.9 (how to evaluate any claim). Twenty-five minutes.

🔬 Deep Dive: §16.1 (DSHEA — the regulatory fact that explains everything else), §16.6 (drug interactions), and §16.7 (assay interference) are where students and clinicians should spend real time.

16.1 DSHEA 1994: the law that built the industry

Almost every strange thing about supplements traces to one piece of legislation.

The Dietary Supplement Health and Education Act of 1994 (DSHEA) established that, in the United States, dietary supplements are regulated more like food than like drugs.

The practical consequences:

Read the disclaimer again. It appears on essentially every supplement sold in the US, in small print, and it is a complete and accurate description of the regulatory situation. It has been hiding in plain sight for thirty years.

📊 Diagram (described). Picture two pipelines running left to right, side by side, from an idea to a product on a shelf.

The drug pipeline is long and has gates. Discovery. Preclinical work. Phase I — is it safe in a small number of healthy volunteers? Phase II — does it do anything, at what dose? Phase III — does it work, against a comparator, in a large randomized trial with a pre-specified endpoint? Then regulatory review, then approval, then post-marketing surveillance. Every one of those is a gate: fail it and the product stops. The pipeline takes years and most candidates die in it.

The supplement pipeline has one meaningful gate, and it is at the end. Idea → formulation → manufacture → sell → and then, if enough harm accumulates and is reported and investigated, the FDA may act to remove it.

Now draw the arrow of proof. In the drug pipeline it points from the manufacturer toward the regulator: you must demonstrate this works and is safe. In the supplement pipeline it points from the regulator toward the product, after the fact: the FDA must demonstrate it is unsafe, using evidence generated by people already taking it.

That reversal is the entire chapter. It is why proprietary blends are legal, why "supports immune health" is the ceiling of permitted language, why adulterated products reach shelves, why third-party certification exists as a voluntary market response to a regulatory gap — and why the assumption that "if it were dangerous it wouldn't be sold" is precisely backwards.

The safety data for supplements is generated by the consumers, and Walt was one of them.

What "structure/function claim" actually permits

This is the mechanism that generates the language you see on every label.

A disease claim — "prevents osteoporosis," "treats depression" — requires drug-level evidence.

A structure/function claim — "supports bone health," "promotes a positive mood" — requires substantially less, needs no pre-approval, and is what the industry uses.

So the vocabulary is not vague by accident. Supports. Promotes. Helps maintain. Contributes to. Every one of those is a legally-selected word doing legally-specific work.

💡 Aha moment. The phrase "supports immune health" is not marketing fluff obscuring a claim. It IS the claim — and it is precisely as strong as the law requires, which is barely at all.

Once you can read that vocabulary as a regulatory signal rather than as advertising, the entire shelf becomes legible. A product that could say "reduces the risk of X" would say it. When it says "supports," that is the strongest thing it is permitted to say.

(Other jurisdictions differ. The EU operates a permitted-health-claims register that is considerably more restrictive; Australia and Canada have their own frameworks. But the global market is shaped by the US one, and much of what's sold internationally is formulated for it.)

16.2 What's actually in the bottle

Since nobody checks before sale, the obvious question is whether the contents match the label.

Frequently, they don't. This has been examined repeatedly, and the findings are consistent:

Under- and over-dosing. Independent testing has repeatedly found products containing substantially more or less of the labelled ingredient than declared.

Substitution and absent ingredients. Analyses of herbal products in particular have found labelled botanicals absent, substituted with cheaper species, or bulked with fillers.

And the serious one — adulteration with actual drugs. The FDA maintains a tainted products database, and the recurring finding is undeclared pharmaceutical ingredients in supplements marketed for three categories in particular: weight loss (sibutramine, a withdrawn appetite suppressant), sexual enhancement (sildenafil and analogues), and bodybuilding (anabolic steroids and steroid-like compounds).

⚠️ This is the genuinely dangerous end. A person taking nitrates for angina who unknowingly consumes a sildenafil analogue in a "natural" product is at risk of a severe hypotensive event. And athletes have failed drug tests from contaminated supplements they had no reason to suspect.

Third-party certification

Which is where the useful practical answer lives.

None of these tests whether the supplement works. They test whether the bottle contains what it says and nothing it shouldn't.

That's a lower bar than people assume and a higher one than the baseline — and given §16.1, it is the only quality signal available to a consumer.

⚠️ For any competitive athlete subject to drug testing, third-party certification is not optional. Strict liability applies: "it was in my supplement" is not a defence.

16.3 The short list that actually works

Here it is. This is the constructive half of the chapter, and it is shorter than the aisle and longer than a sceptic would guess.

The ones already established in this book

Notice the pattern across all of them: every ✅ is either correcting a measured deficiency or meeting a genuinely elevated requirement in a defined group. Not one improves a replete person.

And the two that don't fit that pattern

Because they're the exceptions, and they're the strongest evidence in the chapter.

🔬 Claim → Evidence → Verdict

The claim: "Creatine monohydrate improves strength, power, and lean mass."

Where it comes from: Creatine phosphate is the immediate energy buffer for very short, very intense effort (Chapter 6 §6.2) — the first few seconds of a sprint or a heavy lift. Supplementation raises muscle creatine stores, increasing the capacity of that system.

What the evidence actually shows: Creatine is among the most-studied supplements in existence — hundreds of trials over three decades — and the findings are unusually consistent. It reliably increases muscle creatine stores, improves performance in repeated high-intensity efforts, and increases strength and lean mass gains when combined with resistance training. The International Society of Sports Nutrition position stand describes it as the most effective ergogenic nutritional supplement available for increasing high-intensity exercise capacity and lean mass.

On safety: long-term studies in healthy people have not found the kidney damage that was widely feared. The elevation it causes in serum creatinine — a kidney marker — is a laboratory artefact of creatine metabolism, not evidence of kidney injury, though it will confuse a clinician who doesn't know you take it. (Which is §16.7's theme again.)

There is also a growing and genuinely interesting literature on cognitive effects, particularly under sleep deprivation and in older adults, and on creatine in vegetarians — who have lower baseline stores and may respond more. That literature is younger and less settled.

📉 Evidence quality: ✅ Rung 7. Extensive RCTs, meta-analyses, and position stands.

Verdict: ✅ Well supported for strength, power and lean mass with resistance training. Creatine monohydrate, roughly 3–5 g/day, no loading phase required, taken consistently. ⚠️ Buy a third-party certified product; skip every "advanced" or "buffered" form, which cost more and have not outperformed plain monohydrate.

🧾 Cost: roughly $0.10–$0.20 a day. It is the cheapest thing on the shelf with the strongest evidence, and it is stocked next to products costing thirty times as much with none.

🔍 Why this works. Creatine is worth understanding mechanistically, because it is the clearest example in the book of a supplement that works for exactly the reason its advocates say, and because the mechanism explains its limits as precisely as its benefits.

Chapter 6 §6.1 established that ATP is the only currency cells spend, and that you carry almost none — a few seconds' worth. So how do you sprint for ten seconds?

Creatine phosphate. Muscle stores creatine bound to a phosphate group, and when ATP is spent (becoming ADP), creatine phosphate donates its phosphate back — regenerating ATP almost instantaneously, without oxygen and without waiting for glycolysis to spin up. It is the fastest energy system you have, and it is a buffer rather than a fuel: a rechargeable battery that covers the gap between the first heartbeat of effort and the slower systems arriving.

Supplementation raises the size of that battery — muscle creatine stores rise by something like 20–40% in people who start with lower baseline levels.

Two things follow, and they define the whole verdict.

First: it works where the battery is rate-limiting. Repeated short maximal efforts — sets of heavy lifting, sprint intervals, the fifth rep rather than the first. A bigger buffer means you recover it faster between efforts, which means more total work, which — over months — means more adaptation. The strength gain is downstream of being able to train slightly harder, not of the creatine itself.

Second: it does essentially nothing for a marathon, because at marathon pace the creatine phosphate system is irrelevant and the limitation is elsewhere entirely (Chapter 6 §6.2, Chapter 14's Devi). A supplement that works for a specific reason works only in the specific situation where that reason applies — which is the shape of every honest ✅ in this book, and the shape almost no marketing respects.

🔬 Claim → Evidence → Verdict

The claim: "Caffeine improves exercise performance."

Where it comes from: Adenosine receptor antagonism reduces perceived effort; there are also effects on central drive and on calcium handling in muscle.

What the evidence actually shows: One of the best-supported ergogenic aids there is. Consistent improvements across endurance performance, high-intensity work, and some strength and power measures, in a large body of trials. Typical effective doses are around 3–6 mg/kg body weight, taken roughly an hour before exercise.

The caveats are real but modest: individual response varies (partly genetic); tolerance develops to some effects; higher doses bring anxiety, tremor, GI upset and disturbed sleep without further benefit; and ⚠️ caffeine taken later in the day measurably degrades sleep, which will cost you more than the session gained.

📉 Evidence quality: Extensive RCTs and meta-analyses.

Verdict: 🟢 Probably true — genuinely effective, and available as coffee for a fraction of the price of a "pre-workout." (Chapter 5 §5.8 gave caffeine's metabolism-boosting claim a much cooler reception. Different claim, different verdict — which is the point of doing this claim by claim.)

16.4 The long list that doesn't

Not exhaustive — there are tens of thousands of products — but these cover most of the aisle.

🔬 Claim → Evidence → Verdict

The claim: "Our proprietary blend delivers a synergistic complex of clinically-studied ingredients."

Where it comes from: A regulatory allowance — manufacturers may declare a proprietary blend as a total weight without disclosing individual ingredient amounts, to protect formulations.

What the evidence actually shows: The practical effect is that you cannot know whether any ingredient is present at a dose resembling the one in the studies cited. A blend listing twelve ingredients totalling 500 mg may contain 480 mg of the cheapest and trace amounts of the rest — a practice known as fairy dusting, where an ingredient is present in sufficient quantity to appear on the label and insufficient quantity to do anything.

⚠️ And "clinically studied" is doing specific work. It means a study exists involving the ingredient. It does not mean this product was studied, at this dose, in this combination, for this outcome, with a positive result.

📉 Evidence quality: Unassessable by design.

Verdict: ❌ Not supported. If a product won't tell you the dose, you cannot evaluate it, and you should treat unassessable as unsupported. This is the single most useful heuristic on the shelf.

16.5 The dangerous list

The part that matters most, because "can't hurt" is the assumption this chapter exists to dismantle.

🔬 Claim → Evidence → Verdict

The claim: "It's natural, so it's safe."

Where it comes from: An intuition with a genuine kernel — foods humans have eaten for millennia are, on the whole, tolerated well, and that is real information.

What the evidence actually shows: The inference fails immediately on examples. Digitalis, atropine, ricin, amatoxins, aconitine and nicotine are all natural. Many pharmaceuticals are plant-derived — aspirin, digoxin, morphine, paclitaxel — which demonstrates that natural compounds are pharmacologically active, and pharmacological activity is exactly what carries risk.

And "natural" supplements are frequently more concentrated than any food. A green tea extract capsule may deliver catechins equivalent to many cups of tea. Concentration is the mechanism by which a traditionally-safe plant becomes a hepatotoxin, and it's why the LiverTox database contains entries for products people think of as tea.

📉 Evidence quality: Refuted by counterexample and by the case literature.

Verdict: 🟠 Probably false. "Natural" describes an origin, not a safety profile. The relevant question is dose and evidence, and it always was.

16.6 Drug interactions: what your doctor doesn't know

The most consequential clinical content in this chapter, and it turns on a single behavioural fact: most people do not tell their clinicians about supplements, because supplements are mentally filed as food.

⚠️ When to see a professional — and what to actually do.

Tell your clinicians and your pharmacist about everything you take. Not just prescriptions. Vitamins, minerals, herbals, protein powders, "functional" foods, anything from a health shop.

The practical version, which takes four minutes: photograph every label, or write a list with doses, and keep it in your phone. Show it at every appointment, and especially in an emergency department.

Walt's neighbour grabbing the bottles off the counter is why this chapter has a happy ending. The pharmacist is often the best person to review it, and in most systems will do so for free.

16.7 The interference problem

Walt's section, generalized — because the biotin case is not unique and the category is barely known.

Supplements can make your test results wrong without doing anything to your body.

Why biotin is the serious one: the interference is common (biotin is in a great many hair, skin and nail products at doses far above requirement), the affected assays include one that determines whether someone is admitted for a heart attack, and the direction of error can be toward false reassurance.

The FDA safety communication (2017, updated 2019) advises patients to inform clinicians about biotin and advises clinicians to consider interference when results don't fit the clinical picture. It noted a reported death associated with a falsely low troponin.

🍽️ On your plate. ⚠️ Stop biotin supplements several days before planned blood tests, and tell the clinician if you take them. Specifically check "hair, skin and nails" products, "immune" blends, and B-complexes — biotin appears in all three at doses hundreds of times the adequate intake, and almost nobody thinks of it as a drug.

And if you take creatine, say so before a kidney function test, or you will be investigated for something you don't have.

16.8 Reading a supplement label

Six things, in order.

1. The Supplement Facts panel. Ingredients, amounts, and % Daily Value where one exists. Compare amounts to the UL (Chapter 13 §13.2), not to the RDA.

2. "Proprietary blend." ❌ If you see it, the product is unassessable (§16.4).

3. The other ingredients line. Fillers, binders, allergens.

4. Third-party certification marks. NSF, Informed Sport, USP. ⚠️ Non-negotiable for tested athletes.

5. The claims. Read the verb. "Supports," "promotes," "helps maintain" = structure/function claim, minimal evidence required (§16.1). A disease claim would need drug-level evidence — and if the product had it, it would say so.

6. The disclaimer. "This statement has not been evaluated by the FDA." Believe it.

And three phrases worth learning to read as signals rather than as information:

🔄 Check your understanding. A bottle reads: "Advanced Immune Complex — proprietary blend 800 mg: elderberry, echinacea, zinc, vitamin C, biotin, astragalus, quercetin. Supports healthy immune function.† †This statement has not been evaluated by the FDA." It costs $34/month.

Identify everything wrong.

Answer

1. Proprietary blend — you cannot know the dose of anything. 800 mg across seven ingredients could be 780 mg of the cheapest and traces of the rest (§16.4's fairy dusting). Unassessable = unsupported.

2. ⚠️ Biotin, at an unknown dose, in a product taken daily. This is Walt's problem exactly. Nobody buying an immune product is thinking about troponin assays.

3. ⚠️ Zinc at an unknown dose — and if it's toward the high end, sustained daily use approaches or exceeds the UL and risks copper deficiency (Ch 14).

4. "Supports healthy immune function" is a structure/function claim — the strongest thing it is legally permitted to say without evidence. It is not vagueness; it's the regulatory ceiling.

5. The disclaimer tells you plainly that nobody assessed it.

6. No third-party certification mark.

7. 🧾 $34/month = $408/year for a product whose contents are undisclosed and whose evidence is absent — against creatine at $0.15/day, which has hundreds of trials.

And the meta-point: every single one of these problems is visible on the label, in under thirty seconds, without any nutritional knowledge at all.

16.9 How to evaluate any supplement claim

The framework, which generalizes past the specific products in this chapter — because the products turn over every eighteen months and the framework doesn't.

Six questions:

1. What specifically does it claim, and is that a disease claim or a structure/function claim? If "supports," expect nothing.

2. What's the evidence, and at what rung? (Chapter 2 §2.1.) Mechanism? Animals? A twelve-person crossover? Meta-analysis of RCTs?

3. Is the dose in the product the dose in the studies? ⚠️ Frequently not — and with a proprietary blend, unknowable.

4. Was it tested in people like me? Deficient or replete? Trained or untrained? Sick or healthy? This distinguishes almost every ✅ in this chapter from almost every ❌.

5. What's the harm profile? Interactions, UL, hepatotoxicity, assay interference. "Can't hurt" is a hypothesis, not a default.

6. What does it cost per year, and what's the alternative? 🧾 Multiply monthly by twelve, and compare against the food or behaviour targeting the same outcome (Chapter 11's Case Study 1).

And a seventh, from Chapter 1: what happens to the seller's income if you decide against?

🧩 Productive struggle. Six minutes before reading on. This is the chapter's synthesis exercise and it's worth doing properly.

Design a supplement that would earn a ✅ from this book.

Not a real one — invent it. Specify the compound, the claim, the population, the dose, the evidence you'd need, and the trial that would generate it. Then check it against §16.9's six questions.

What a ✅-earning supplement looks like

Work backwards from the six ✅ verdicts in Part III and the pattern is unmistakable. A supplement earns one when it has all of the following:

1. A specific, named deficiency or elevated requirement — not "adults who want more energy." Folic acid has people who could become pregnant. B12 has vegans and people with malabsorption. Iron has people with a ferritin below threshold. The population is defined by a measurable state, not by an aspiration.

2. A mechanism that specifies why this group and not everyone. Vegans need B12 because it isn't in plants. Metformin users need it because absorption is blocked. The mechanism names the gap.

3. A measurable status marker — so you can tell whether the person is in the group and whether the intervention worked. Ferritin. 25(OH)D. Serum B12. This is what magnesium lacks (Ch 14 §14.4) and why its verdict is 🟡 despite a plausible case.

4. A hard outcome, not a marker — neural tube defects, not homocysteine. Chapter 13 §13.7's graveyard is full of supplements that moved the marker beautifully.

5. A dose that matches the tested dose, disclosed on the label.

6. A trial in the deficient population, not in replete volunteers — because the antioxidant trials mostly weren't.

Now the uncomfortable part of the exercise. Take any product currently in your cupboard and score it against those six. Most score zero or one, and the ones that score highly are the boring cheap ones.

And notice what creatine and caffeine did instead: neither corrects a deficiency. They earned their verdicts by having hundreds of trials, in the relevant population, on outcomes people care about, at disclosed doses. There are two routes to a ✅ — fix a real gap, or accumulate overwhelming evidence for an ergogenic effect. Almost nothing on the shelf attempts either.

16.10 Walt's cabinet, in full

The audit, completed — begun in Chapter 11, finished here.

Annual spend: $2,244 → $310.

Added instead: targeted B12 (~$18/year) · a **fiber ramp** from 14 g to 30 g (~$80/year of lentils and oats) · and a written list of everything he takes, in his phone and on his record.

At five months: A1c 7.4% → 6.9%. LDL down modestly. And — the thing he raised himself — "regular for the first time in years."

💡 Aha moment. Look at what was actually wrong with Walt's cabinet, because it isn't what people expect.

The problem was not that his supplements did nothing. Two were justified. One (berberine) worked and was therefore more dangerous, not less, because it was unmonitored and stacked on metformin.

The problem was that a $2,244-a-year regime built to take charge of his health had, in aggregate: masked a real deficiency, exceeded a UL, created an assay interference that nearly cost him a cardiac diagnosis, and left the two highest-yield interventions available to him — B12 and fiber, costing under $100 a year combined — entirely unaddressed.

He was not under-treating himself. He was mis-treating himself, expensively, in a direction nobody was checking.

16.10b What happened to Walt's neighbour

A short postscript, because the most useful thing in this chapter came out of the ER visit rather than the clinic.

Walt's neighbour — the man who drove him at half past four in the morning and grabbed the bottles off the counter — asked me afterwards what he should have done differently.

Nothing. He did the single most useful thing available, and he did it by accident.

Grabbing the bottles is the intervention. Not a list, not a memory, not "he takes some vitamins" — the physical containers, with the labels on, in the room where the decisions are being made. A junior doctor read one and changed the outcome.

What would have been better still is a list, because bottles get left behind and because a person who arrives by ambulance alone has nothing. Which is why this chapter has asked for one three times.

⚠️ The four-minute version, and then I'll stop asking.

Open your phone. Photograph every supplement label you own, or type a list with doses. Put it somewhere you'd find it under pressure — a note titled Medications, or in your phone's medical ID, or on a card in your wallet.

Include: prescriptions · vitamins and minerals · herbals · protein powders · "functional" foods · anything from a health shop.

Then tell someone it exists. A partner, a neighbour, an adult child. Because ⚠️ the situation where this matters most is precisely the one where you may not be able to tell anyone anything — which is what makes Walt's story a piece of luck rather than a system working.

Cost: four minutes. Expected value: occasionally enormous.

16.11 Who genuinely benefits

To be fair to the category, because a chapter this sceptical owes a clear statement of where supplements earn their place.

Notice how specific that list is. Every row names a group with a reason. None of them is "adults who want to be healthier," which is who the aisle is aimed at.

16.12 The economics

🧾 Cost check — the whole book's spending, in one table.

Everything with strong evidence is in the top half. Everything in the bottom half costs ten to fifty times more.

The US supplement market is estimated at well over $50 billion annually. Given §16.3's list — most of which costs under twenty dollars a year and applies to defined groups rather than everyone — that figure is not a measure of nutritional need. It's a measure of hope, sold at retail.

What we don't know

We don't know how large the adulteration problem actually is. The FDA's tainted products database records what has been found, which is a function of what was looked for — and enforcement capacity is limited relative to a market of tens of thousands of products. The detected cases are a lower bound of unknown tightness.

We don't know whether most supplements do anything at population scale, because the trials that would tell us are expensive, unglamorous, and — under DSHEA — entirely unnecessary for the manufacturer. The absence of evidence here is structural rather than accidental, which is a different situation from a question that has been asked and not answered.

And the emerging creatine literature is genuinely open. The strength and power evidence is settled; the cognitive findings — under sleep deprivation, in older adults, in vegetarians with lower baseline stores — are interesting, plausible, and much less established. I've given creatine ✅ for what it has been shown to do, and I'd want considerably more before extending it.

🪞 Learning Check-In

Fifth of these, and Part III is complete. Three minutes.

• Did Part III change what's in your cupboard? Not what you believe — what you own. If you've read three chapters on micronutrients and thrown nothing away and bought nothing, either you were already well-calibrated or the reading didn't reach the shelf. Both happen; only one of them is fine.
• Which affirmative verdict surprised you most? For most readers it's creatine, or ORS. Notice that surprise — it means you'd absorbed the sceptical frame more thoroughly than the evidence warranted, which is exactly the failure Chapter 1 §1.8 warned about.
• Have you written the list? Every supplement, with doses, in your phone. It's been asked for in Chapters 13, 14 and 16. It takes four minutes and it is the single highest-value action in Part III.

One thing to carry into Part IV: you now have a framework (§16.9) that works on claims that don't exist yet. Part IV is six chapters of live arguments — detoxes, sugar, seed oils, organic, fasting, ultra-processed food. Try running §16.9's questions and Chapter 2's Claim Filter before reading each verdict. Being wrong is informative; the prediction is what does the work.

Spaced Review

Answer before reading on.

1. (Chapter 13) What do all six ✅ supplement verdicts in Part III have in common?

Every one corrects a measured deficiency or meets a genuinely elevated requirement in a defined group — folic acid preconception, B12 in vegans and malabsorption, vitamin D in the deficient, iron in the iron-deficient, iodine in pregnancy, ORS in illness. None improves a replete person. (Creatine and caffeine are the deliberate exceptions, and they're ergogenic rather than nutritional.)

2. (Chapter 14) Walt's "immune blend" contained two separate hazards. Name both.

Biotin 3,000 µg — immunoassay interference, which nearly cost him a cardiac diagnosis. And zinc 50 mg — above the ~40 mg UL, sustained, risking copper deficiency with anemia, neutropenia and potentially irreversible neurological damage. One hazard was diagnostic, one nutritional, and neither was on any medical record.

3. (Chapter 2) Why is "clinically studied ingredient" a weaker claim than it sounds?

Because it means a study exists involving the ingredient — not that this product, at this dose, in this combination, for this outcome, produced a positive result. It's Chapter 2's abstract-reading problem compressed into two words, and with a proprietary blend the dose is unknowable by design.

Project Checkpoint: Your Supplement Audit

Component sixteen, and the last of Phase 3 — Debiasing. This is the checkpoint that saves people the most money.

Step 1 — Empty the cupboard. Literally. Put every bottle on the table: vitamins, minerals, herbals, protein powders, "functional" products, anything bought for a health claim.

Step 2 — Build the table. One row per product.

Step 3 — Run §16.9's six questions on anything you're unsure about. Claim type · evidence rung · dose vs. studied dose · tested in people like me · harm profile · annual cost and alternative.

Step 4 — ⚠️ Check three things specifically:

• Biotin — in anything? "Hair, skin and nails," "immune," B-complex. If yes, note it and tell your clinician before blood tests.
• Zinc — add up the total across all products. Over ~40 mg/day?
• Proprietary blends — any? Those are automatically unassessable.

Step 5 — Total the annual spend.

My current annual supplement spend: $__

Step 6 — Decide, and write the reason. For each: keep, cut, or ask a clinician (anything glucose-lowering, anticoagulant-interacting, or that you're unsure about — don't stop it unilaterally either).

Step 7 — ⚠️ Write the list. Everything you're keeping, with doses, in your phone. Show it at every appointment. Four minutes.

Step 8 — Redirect one thing. Take the money from one cut product and put it toward whatever Chapters 8, 11, 13 or 14 identified as your actual gap. Walt's went to lentils and B12, and moved his A1c half a point.

Non-tracking alternative. Skip the costing. Do Steps 1, 4, 6 and 7 — the cupboard, the three safety checks, the decision, and the list. The list is the one that matters most.

Next checkpoint (Chapter 17): your myth audit — revisiting the Belief Inventory you sealed in Chapter 1 and grading it. Phase 3 ends where it began.**

Chapter Summary

DSHEA 1994 regulates supplements like food, not drugs: no pre-market approval, burden of proof on the FDA after marketing, and structure/function claims — supports, promotes, helps maintain — which require minimal evidence.

"Supports immune health" is not marketing fluff obscuring a claim. It IS the claim, and it's the strongest thing the product is permitted to say.

What's in the bottle: under- and over-dosing, substitution, and ⚠️ adulteration with undeclared pharmaceuticals — recurring in weight-loss, sexual-enhancement, and bodybuilding products. Third-party certification (NSF Certified for Sport, Informed Sport, USP) verifies contents, not efficacy — and is non-negotiable for tested athletes.

The short list that works: folic acid preconception ✅ · B12 for vegans and malabsorption ✅ · vitamin D in the deficient ✅ · iron in the iron-deficient ✅ · iodine in pregnancy ✅ · ORS in illness ✅ · creatine ✅ · caffeine 🟢 · psyllium 🟢 · electrolytes in prolonged exercise 🟢 · omega-3 without fish 🟡 · protein powder 🟡.

Every ✅ corrects a deficiency or meets an elevated requirement in a defined group — except creatine and caffeine, which are ergogenic rather than nutritional.

This chapter's verdicts:

⚠️ The dangerous list: biotin (assay interference) · zinc >40 mg (copper) · preformed vitamin A in pregnancy · B6 (neuropathy) · iron without testing · green tea extract, garcinia, kava (hepatotoxicity) · ephedra/synephrine · kelp · yohimbine · adulterated products.

⚠️ Interactions: St John's Wort is the most dangerous common one — a potent CYP3A4 inducer affecting contraceptives, warfarin, immunosuppressants, antiretrovirals. Plus bleeding risk (fish oil, ginkgo, garlic, vitamin E), berberine on diabetes drugs, mineral–thyroid timing, potassium.

⚠️ Interference: biotin (troponin, thyroid) · creatine (raises creatinine — artefact, not injury) · vitamin C (some glucose meters).

Walt: $2,244 → $310, plus B12 and lentils; A1c 7.4% → 6.9%.

And the two routes to a ✅, from §16.10's exercise: fix a real, measurable gap in a defined group — which is what every nutritional ✅ in Part III does — or accumulate overwhelming evidence for an ergogenic effect, which is what creatine and caffeine did. Almost nothing on the shelf attempts either.

The one thing to remember: Walt's troponin came back low, and it was wrong. Supplements are not "can't hurt" — they can make your medical care worse without doing anything to your body at all.

What's Next

Part III is complete, and the pattern across it is worth naming: the nutrients where deficiency is real have overwhelming evidence and cost almost nothing; the products sold to people who aren't deficient have neither.

Part IV is where you use everything. Six chapters of live arguments — detoxes and superfoods · sugar · seed oils · organic and GMO · intermittent fasting · ultra-processed food.

Chapter 17 is the densest 🔬 chapter in the book. But the verdicts are the least interesting part of it — what it's really doing is dissecting why these particular ideas are sticky, so you can recognize the shape of the next one before it has a name.

And your Chapter 1 Belief Inventory has been sealed for sixteen chapters. You open it in Chapter 17.