Case Study 1 — Walt's Two Interventions

A clinical case. Walt Prosser is an illustrative composite; the comparison is not.

Setup

Walt Prosser, 68, retired postal carrier, widower, lives alone. Type 2 diabetes with an A1c of 7.4%, on metformin, atorvastatin and lisinopril.

He gets his health information from a wellness podcast he listens to on his morning walk, and he has acted on it comprehensively. Nine supplements. $187 a month. $2,244 a year, on a pension.

Among them is berberine, which he takes specifically for blood glucose, at 1,500 mg a day, because the podcast described it as "nature's metformin."

His fiber intake is 14 grams a day against a target of 30.

Two interventions are available to him for the same outcome. This case study is the comparison, done properly, because it is the clearest illustration in this book of how badly the attention economy allocates effort.

Intervention A: berberine

What it is. A plant alkaloid, available over the counter, widely marketed for glucose control.

Does it work? This is the part people expect me to skip, and I won't: yes, to a degree. Berberine has genuine glucose-lowering activity, with a plausible mechanism (AMPK activation among others — Chapter 6 §6.9), and there are randomized trials showing reductions in fasting glucose and A1c. It is not a placebo and it is not homeopathy. Some of the trials are small and many are of limited quality, but there is something there.

So what's the problem? Four things, and none of them is "it doesn't work."

1. Dose is unregulated. Under DSHEA 1994 (Chapter 16), what's on the label may not be what's in the bottle, and content varies between brands and between batches. Walt is taking an unverified dose of a compound with a real pharmacological effect.

2. Nobody is monitoring the interaction. Berberine lowers glucose. So does metformin. So does his diet. Stacking an unmonitored glucose-lowering agent on a prescribed one is exactly how hypoglycaemia happens, and neither his GP nor his pharmacist knew he was taking it — because he didn't think of a supplement as a drug.

3. It has genuine drug interactions. Berberine affects hepatic enzyme systems involved in metabolizing a range of medications. Walt takes three prescriptions. This has never been reviewed.

4. The cost. Roughly $35 a month, $420 a year, of his $2,244.

Intervention B: fiber, 14 g → 30 g

What it is. Beans, lentils, oats, and a pear.

Does it work? For glycemic control: yes, with a considerably larger and better body of evidence than berberine has. Viscous fiber slows gastric emptying and glucose absorption, flattening post-meal excursions (§11.3, mechanism 2). Higher fiber intake is associated with substantially lower type 2 diabetes incidence, with clear dose-response, across cohorts and multiple countries (§11.4). Trials support improvements in glycemic markers.

And it does six other things at the same time — LDL, satiety, energy density, bowel function, microbiome, colonocyte fuel supply — every one of which is relevant to a 68-year-old with type 2 diabetes.

The cost: dried lentils at about $0.14 per 10 g of fiber. Closing his 16 g gap costs approximately $0.22 a day. $80 a year.

The comparison

Five times the cost, weaker evidence, narrower benefit, real interaction risk, no monitoring — and it is the one he's doing.

⚠️ When to see a professional. Two things needed to happen here, and neither was optional. First, Walt needed to tell his GP and pharmacist about all nine supplements — the single most useful thing any patient in his position can do, and the thing almost nobody does, because supplements are mentally filed as food rather than as drugs. Second, any change to his fiber or his supplements needed to be made with glucose monitoring, because improving glycemic control while on metformin can eventually mean his medication needs adjusting — which is a good problem and still a problem to manage rather than discover.

What actually happened in the conversation

I did not tell him berberine doesn't work. That would have been false, and he'd have known it was false, and I'd have lost him.

I drew the table.

He looked at it for a while and then asked the question I was hoping for: "So why has nobody told me about the beans?"

Which is the entire book in one sentence.

The honest answer — §11.11 — is that nobody makes money from beans, that the advice hasn't changed in fifty years so it isn't news, that it offers him no identity as someone taking charge of his health, and that his podcast needs a new topic every week and "eat more fiber" has been true since before it launched.

He then said something that stayed with me: "The bottles feel like I'm doing something."

That's not irrationality. That's an accurate description of what the bottles are for, and it's why "just eat beans" fails as advice while nine supplements succeed as a habit. The supplements provide a daily ritual, a sense of agency, and visible evidence of effort. A tin of chickpeas provides none of those.

The plan we built

We kept the ritual and changed its contents, because removing the ritual would have removed the adherence.

Annual supplement spend: $2,244 → $310. Annual fiber cost: +$80.

Net: about $1,850 a year back, on a pension, plus an intervention with better evidence than the thing it replaced.

At five months: A1c 7.4% → 6.9%. LDL down modestly. Reported "regular for the first time in years," which he raised himself and which mattered more to him than the A1c did.

Analysis

1. "It doesn't work" was the wrong argument and would have failed. Berberine does something. The winning argument was comparative: same outcome, five times the cost, weaker evidence, narrower benefit, unmonitored interaction risk. Comparison beats debunking, and it beats it specifically because it doesn't require the patient to have been wrong.

2. The ritual was the adherence mechanism. Removing nine bottles and offering lentils would have removed the daily act that made him feel he was managing his health. Keeping two bottles preserved it. This is Chapter 10 §10.10's identity problem, solved rather than lectured about.

3. The interaction risk was the genuinely urgent part and it was invisible to everyone. Unmonitored berberine on metformin, biotin interfering with cardiac assays, zinc depleting copper — none of it was on any medical record, because supplements are filed as food.

4. The fiber intervention did seven things while the supplement did one. That breadth is the real argument and it's the one that never gets made, because breadth is boring and specificity sells.

5. "So why has nobody told me about the beans?" Because it scores zero for seven (§11.11). Walt is not the failure in this story. The information environment is, and he navigated it exactly as a reasonable person would.

Discussion Questions

1. I declined to say berberine doesn't work. Was that the right call? What would have been lost by leading with "that supplement is nonsense" — and what would have been gained?

2. "The bottles feel like I'm doing something." Take that seriously as a design requirement. What would a fiber intervention look like if it were engineered to deliver ritual, agency and visible effort?

3. Walt's supplements were on no medical record. Whose responsibility is that — his, his GP's, the pharmacist's, or the regulatory framework's? What single change would fix the most of it?

4. The plan kept two supplements that are, at best, marginal. Defend that clinically. Then argue it's an unnecessary concession that undermines the evidence-based position.

5. Compare the effect size of Walt's fiber change (A1c 7.4 → 6.9) to what a medication adjustment might have achieved. Does the comparison strengthen or weaken the case for the dietary intervention?

Your Turn

If you take any supplements, build Walt's table for one of them.

Then the question that decides it: if I could only do one, which has more evidence per dollar?

And separately, whether or not you take anything: write down every supplement, vitamin, herbal product and "functional" food you take, and take that list to your next medical appointment. It takes four minutes. Walt's list contained an interaction that nearly cost him a correct emergency-room diagnosis, and he had no idea, and neither did anyone treating him.