The food pyramid: who decided the base was bread?

Draw a triangle. Bread, cereal, rice and pasta across the wide bottom. Fat and sugar in the point at the top. You do not need a caption to know what it is asking: plenty of the bottom, very little of the top.

That is the 1992 Food Guide Pyramid: six to eleven servings of grains a day, with butter moved into the naughty corner.

So where did those numbers come from?

Not from a trial. Nobody put one group on the pyramid and another on a different diet and counted what happened. The shape and the numbers came from work on nutrient adequacy: getting enough of everything into a population. Then the drawing went on school walls and cereal boxes and was read for thirty years as an answer to a different question: whether that diet beats another one.

None of that means the pyramid poisoned anyone. It means a document built to stop deficiency got read as a claim about heart attacks, with the serving counts doing the arguing.

What the leaflet actually said

The pyramid went out as Home and Garden Bulletin 252, with three worked daily plans: 1,600 calories with 6 grain servings, 2,200 with 9, 2,800 with 11, against 3 to 5 vegetables, 2 to 4 fruits, 2 to 3 dairy, and 2 to 3 of meat, fish, eggs and nuts, 140 to 200g in total [1].

A serving turns out smaller than people remember. One slice of bread. 30g of ready-to-eat cereal. Half a cup of cooked rice or pasta, roughly 100g [1].

Do the sum in rice or pasta. Nine servings at roughly 100g each is close to a kilogram of cooked grain a day, before a single vegetable is counted.

Wholegrains got a mention. They were not a condition of hitting the number, and the number is what ended up on the wall. White bread met it neatly.

Britain drew the same shape and called it a plate

The UK's first national food model was the Balance of Good Health in 1994, the eatwell plate in 2007, the Eatwell Guide in 2016, after a refresh that used linear programming to set the segment sizes [5].

The proportions survived. The NHS page gives fruit and vegetables "just over a third" of the diet, and starchy carbohydrates "just over a third" too [6]. The model report puts the split at 37 per cent starchy against 39 per cent fruit and vegetables [5].

America threw its pyramid out. Britain kept the base at close to two-fifths of the plate and renamed the furniture.

The nearest thing anyone ran to a test

The Women's Health Initiative Dietary Modification Trial randomised 48,835 postmenopausal women. One arm aimed at 20 per cent of calories from fat, five servings of vegetables and fruit a day, at least six servings of grains. The other got leaflets and no support. Follow-up averaged 8.1 years [2].

They missed their target. By year 6 the intervention group was eating 8.2 percentage points less fat than the comparison group, saturated fat down 2.9 points, plus 1.1 more servings of vegetables and fruit and half a serving of grains.

The biochemical changes were real and small. LDL cholesterol fell 0.09 mmol/L, diastolic blood pressure 0.31 mmHg, factor VIIc 4.29 per cent. HDL, triglycerides, glucose and insulin did not differ [2].

Then the part that decides it. Coronary heart disease ran at 0.63 per cent a year against 0.65, a hazard ratio of 0.97 with a confidence interval of 0.90 to 1.06. Stroke was 1.02, the combined cardiovascular outcome 0.98 [2].

A doctor looking at LDL down 0.09 mmol/L and blood pressure down a third of a millimetre of mercury would say what they say about most printouts: nothing much has moved. Over eight years, nothing much happened to the hearts either.

The trial has a fair defence: they aimed for a 20 point gap in fat and got 8, so the groups ended up closer together than designed, and both ate a Western diet. An experiment nobody delivered is not proof the diet does nothing. It is proof the claim was never properly tested.

The question the pyramid answered for you

The pyramid treats the proportion of energy from fat as the lever: move fat to the top, cut it, and the rest sorts itself out. DIETFITS tested that lever directly, building both diets from whole foods so the comparison was the fat-to-carbohydrate ratio rather than crisps against salad. It randomised 609 adults with a body mass index of 28 to 40, ran 22 group sessions over 12 months, and told each arm to reach the lowest fat or carbohydrate intake it could sustain [3].

Achieved splits were 48 per cent carbohydrate and 29 per cent fat on the low-fat diet, against 30 and 45 on the low-carbohydrate one. Twelve-month weight change was minus 5.3 kg against minus 6.0 kg, a difference of 0.7 kg with a confidence interval of minus 0.2 to 1.6 [3].

Neither genotype nor baseline insulin secretion predicted who did better, the question the diet books sold hardest. When both diets are whole food, the ratio stops deciding much.

Then America changed its mind and Britain did not

The United States capped total fat at under 30 per cent of calories from 1980, moved the range to 20 to 35 per cent in 2005, and dropped the numeric limit in 2015, leaving saturated fat as the constraint [9]. Britain kept its ceiling: total fat still no more than 35 per cent of food energy, set in 1991, with saturated fat at about 10 to 11 per cent [10]. The country that invented the pyramid dropped the fat limit. The country that copied it did not.

What the country actually ate

The National Diet and Nutrition Survey for 2019 to 2023 reads badly against every version of this advice. Free sugars supplied 10 per cent of adult energy against a 5 per cent target, and 19 per cent of adults met it. Saturated fat supplied 12.6 per cent against 10, with 82 per cent of adults over. Ninety-six per cent of adults and of 11 to 18 year olds missed the 30g fibre target. Adults averaged 3.3 to 3.7 portions of vegetables and fruit a day, with 17 per cent reaching five against 9 per cent of 11 to 18 year olds [4].

Adult obesity in England was 26.5 per cent in 2023 to 2024, up from 22.6 per cent in 2015 to 2016 on the same measure [8].

That is a correlation across a decade in which the guidance did not change, and on its own it convicts nobody. What it rules out is the confident claim that the advice was working.

What survives, and what does not

Advice that names a food has held up better than advice that counts a percentage. 30g of fibre a day and free sugars below 5 per cent of energy came from SACN's 2015 review, and both are still the UK position [7]. Two portions of fish a week, one of them oily, is a food-level instruction with measurements behind it [6]. The percentage-of-energy framing has the weakest support, which is why the two countries diverged on it.

The starch base is weaker still: an adequacy calculation from 1992, never tested against a replacement, and it did not deliver the fibre it was meant to carry. What would settle it is a randomised trial of the pyramid against a low-carbohydrate diet, run long enough to count heart attacks rather than risk factors. Nobody has run one, so the honest position on either side is that the largest relevant trial came back null and the rest is inference, this page included.

The six to eleven servings are gone from American advice. In Britain the same share of the plate is still drawn, still printed on the leaflets, and still missed by most of the people it is drawn for.