Case Study 2 — Margarine: Sixty Years of Being Confidently Wrong, Twice

A historical and policy case. The sequence of events, the regulatory actions, and the evidence described are real; the reconstruction of what a person would have been told in each era is my own.

Setup

Case Study 1 was one patient's panel. This one is a single product across sixty years, because margarine is the best available demonstration of something this book keeps asserting: that confident dietary advice can be wrong, can be corrected, and that both the wrongness and the correction are informative.

Imagine one family. Four generations. Same kitchen, same question: butter or margarine?

1955 — "Butter, obviously"

What she'd be told: nothing, because nobody was asking. Butter is food. Margarine exists, is cheaper, and is regarded as an inferior substitute — in some jurisdictions it had been subject to restrictions and taxes at the behest of dairy interests, and in several it was legally required to be sold uncoloured, so it arrived white with a sachet of dye.

The state of the evidence: cardiovascular disease is rising sharply in industrialized countries and nobody knows why. The diet-heart hypothesis is being formulated.

Verdict of the era: butter, and no one thought about it.

1985 — "Margarine, and it's not close"

What he'd be told: switch to margarine. This is not fringe advice; it is mainstream medical guidance, delivered by physicians, reinforced by heart-health charities, and printed on the tub.

The reasoning: saturated fat raises blood cholesterol; blood cholesterol drives heart disease; butter is 63% saturated fat; margarine is made from vegetable oil. Therefore margarine.

Every step of that chain was defensible on the evidence available. The first two steps remain defensible now (§9.6, §9.10). The failure was in the third and fourth: partial hydrogenation, which turned liquid vegetable oil into a spreadable solid, produced industrial trans fats — and nobody was measuring them, because nobody suspected them.

Verdict of the era: margarine, confidently, on medical advice.

Verdict in retrospect: the margarine of 1985 was worse than the butter it replaced. Trans fats raise LDL and lower HDL — the only fat that does both.

2005 — "Actually, some margarine"

What she'd be told: it depends on the margarine. By now trans fats are recognized. Labelling requirements are arriving. Manufacturers are reformulating, replacing partial hydrogenation with interesterification and blends of oils that achieve spreadability without producing trans fats.

The situation is genuinely confusing for a shopper, because "margarine" now describes two quite different products — the old hydrogenated kind and the reformulated kind — sold side by side under similar branding.

Verdict of the era: read the label. Which almost nobody did, and which the label at that point often didn't make easy.

2025 — "Neither, mostly, and here's why"

What he'd be told now:

Industrial trans fats are largely gone. The FDA's 2015 GRAS determination and equivalent action elsewhere removed partially hydrogenated oils from most of the food supply. Modern margarine and spreads are not the 1985 product, and the trans fat objection to them no longer applies.

But the question has moved. The best-supported recommendation from the entire diet-heart era is §9.5's: replace saturated fat with unsaturated fat. A modern soft spread based on rapeseed, olive, or sunflower oil does exactly that, and does it better than butter.

And the honest answer for most cooking isn't either. Use olive oil, which is what the population with the best trial evidence uses (§9.9, PREDIMED), which has good oxidative stability (§9.11), and which requires no reformulation debate at all.

Butter is fine as butter — for flavour, for baking, on toast, in the quantities a person actually uses. The saturated fat objection to it is real, modest, and comparator-dependent, and if the alternative is a low-fat spread eaten alongside more refined carbohydrate, it isn't an improvement.

Verdict of the era: olive oil for cooking; either spread or butter in the quantities you actually use; and the whole question is smaller than any of the four generations was told.

What each generation got right and wrong

Look at the second row again. The 1985 advice was wrong not because the underlying science was wrong, but because the implementation introduced a harm nobody had thought to look for. The premises survived. The recommendation didn't.

That's a specific and important failure mode, and it isn't fixed by being more skeptical of the premises.

Analysis

1. This is the single strongest argument for humility — in every direction

Anyone confident about dietary fat should sit with 1985. Serious, well-intentioned scientists and clinicians, reasoning correctly from the best available evidence, told a generation to eat something that was worse than what they were eating.

And anyone using that to argue that nutrition science is worthless should sit with 2015. The field detected the error, quantified it, published it, and drove the substance out of the food supply. That is not what a worthless field looks like.

2. The error was in the substitution, and nobody specified it

Trace it back to §9.5. The 1985 advice was "reduce saturated fat." It did not say what to replace it with, so the market decided — and the market chose the cheapest spreadable solid, which happened to be a trans fat delivery system.

The advice that would have been right in 1985 is the advice that's right now: replace saturated fat with unsaturated fat. That instruction, given clearly, would have selected against partial hydrogenation, because hydrogenation exists precisely to make unsaturated fat behave like saturated fat.

The failure was a specification failure. It's the same failure as the low-fat era (Chapter 7 §7.8), and it's why this book keeps insisting on the comparator.

3. Advice outlives its evidence

Ruth Kaminsky (Chapter 8) was still avoiding eggs and cheese eight years after the guidance changed.

Somewhere right now there is an eighty-year-old still eating hydrogenated margarine on medical advice from 1985, and a fifty-year-old avoiding all margarine because of a trans fat problem that was solved a decade ago.

Guidance changes quietly and advice is given once. Anyone working with older patients should routinely ask: what were you told to avoid, and when?

4. The current answer is the least dramatic one available

Not "butter is back." Not "seed oils are poison." Not "margarine is heart-healthy." Cook in olive oil, use whichever spread you prefer in the amounts you actually use, and spend your attention elsewhere.

That's an unsatisfying resolution to a sixty-year argument and it is very probably the right one.

Discussion Questions

1. The 1985 advice was correct in its premises and wrong in its recommendation. What kind of error is that, and how would you guard against it? Notice that being more skeptical of the premises wouldn't have helped.

2. Suppose you're a public health official in 1985 with the evidence available then. Would you have recommended margarine? What would it have taken not to — and would that caution have cost lives elsewhere?

3. The chapter says the field "detected its own error, published it, and drove regulation." It took roughly two decades from serious evidence to regulatory action. Is that a success or a failure? What would a faster process get wrong?

4. Identify a present-day dietary recommendation that could be making the 1985 error — correct premises, unspecified substitution, and a market that will fill the gap with something unmeasured. Defend your candidate.

5. If you were writing food labelling policy, what would you have required in 2005 that would have made "read the label" a workable instruction?

Your Turn

Ask someone over sixty-five what they were told about butter and margarine, and when.

Then ask what they eat now, and why.

Most people discover one of three things: they're following advice from an era whose evidence has moved; they abandoned all of it and eat whatever, having concluded that nobody knows anything (Chapter 1 §1.7's most damaging conclusion); or — occasionally — they updated, and can tell you exactly when and what convinced them.

That third group is small, and worth studying. Whatever made them update is more interesting than anything in this case study.

Then check your own kitchen. What's in it, and what year is the reasoning from?