About Our Data

Where we get our data

Nutriofia draws on two authoritative, peer-reviewed nutrition databases to cover the whole foods, dairy and eggs in our database:

🇬🇧 CoFID 2021 — McCance & Widdowson's Composition of Foods

Published by Public Health England (now UKHSA), this is the UK's primary food composition reference and has been the gold standard for UK nutritional research since the 1940s. It uses UK measurement methods that align directly with UK food labelling regulations. CoFID itself covers the full range of UK foods, including meat and fish — but Nutriofia draws only on its vegetables, fruits, grains, pulses, nuts, seeds, dairy and egg entries. We do not track meat or fish.

🇺🇸 USDA FoodData Central 2025 — adjusted to UK method

For foods not covered by CoFID — primarily tropical fruits, some US-origin grains (quinoa, teff, farro), fermented foods (tempeh), and other foods common in a modern plant-based diet — we draw on USDA FoodData Central Foundation Foods (2025 release). Before import, these values are adjusted to match UK conventions. The adjustments are documented below. You will see a 🇺🇸 badge on pages that use USDA data.

Every food page shows which database its values came from. We do not mix sources within a single food — every nutrient value for any given food comes from the same dataset.

One food, one source

We have a strict rule: every food entry uses exactly one primary source for all its nutrient values. We never fill a gap in a CoFID food with a USDA value, or vice versa.

The reason is straightforward. Different databases use different measuring methods. CoFID measures fibre one way; USDA measures it another. CoFID calculates carbohydrate by direct measurement; USDA calculates it by subtracting everything else from 100g (the "by-difference" method). If you mix values from both in a single food entry, the numbers will never add up correctly — and the apparent precision will be false.

Where a food has gaps in our primary source, the gap stays empty rather than being filled with a different source's value. This is why you may occasionally see a nutrient listed as "no data" even for a well-studied food.

Fibre: NSP vs AOAC

Fibre is the nutrient most affected by measurement conventions, and it's worth explaining because the numbers you may have seen elsewhere could differ from ours.

How CoFID measures fibre (NSP — Englyst method)

CoFID uses Non-Starch Polysaccharides (NSP), the method developed by Dr Hans Englyst. NSP measures structural plant cell wall polysaccharides only. It does not include lignin or resistant starch. NSP values are the standard used in UK food labelling and form the basis of UK dietary guidelines. Typical NSP values are lower than AOAC values for the same food.

How USDA measures fibre (AOAC method)

The USDA uses AOAC total dietary fibre, the method adopted by the US FDA. AOAC includes lignin, resistant starch, and non-digestible oligosaccharides that NSP does not count, so AOAC values are higher for the same food.

How much higher varies enormously. Public Health England uses a population conversion factor of 1.33 — that is, AOAC ≈ NSP × 1.33. We tested that against the 791 foods in CoFID 2021 that happen to report both methods, and found a median of 1.30 and a mean of 1.43, which is close agreement.

But the spread between individual foods is wide. At the 10th percentile the ratio is 0.91; at the 90th it is 2.00. It depends on what the food actually contains — resistant starch in cooled potatoes and legumes, lignin in seed coats and berry skins, fructans in onions and wheat. Walnuts, for instance, run at roughly 3.5g NSP against 6.5g AOAC, a ratio of 1.86, which is near the top of the range rather than typical of it.

This is why we convert at the population level only, and never food by food. Applying a single multiplier to individual foods would put a false precision on values that genuinely differ.

What Nutriofia shows

All fibre values in our database are in NSP (Englyst method) — consistent with UK labelling and UK dietary guidelines. When we import USDA fibre values, we multiply them by 0.7 before storing them, to convert from AOAC to approximate NSP. This is a population-average approximation; the actual conversion factor varies by food. We flag this conversion in the source tag on each value.

The practical consequence: our fibre numbers will often look lower than values you find on US nutrition sites or packaging. They are not wrong — they are simply measured differently, in accordance with UK standards.

Why we show a fibre range, not a single target

There is a genuine mismatch here, and rather than hide it we would rather explain it.

Our food values are NSP. But the UK's headline recommendation — the 30g a day you will see on the NHS website — is an AOAC figure, set by SACN in 2015. Comparing NSP food values against an AOAC target understates how much fibre someone has actually eaten, by roughly a third.

The history explains where the number came from. The previous UK recommendation was 18g a day of NSP, set by COMA in 1991. Converting that to AOAC gives about 24g — and SACN then rounded it up to 30g as a public health goal. So the current target is part conversion and part deliberate ambition, which is also why so few people reach it.

Working back the other way, 30g AOAC is equivalent to roughly 22.5g NSP. Both numbers are defensible: 22.5g is the current recommendation expressed in the units our data uses, and 30g is what the government actually publishes.

So we show both. Where Nutriofia gives you a daily fibre goal, you will see a range of roughly 22–30g rather than a single figure. The lower end is where you have met the recommendation on our measuring system; the upper is the published target. The honest position is that nobody knows the correct figure to within 8g — which is precisely why 24 was rounded to 30 in the first place.

We do not adjust the fibre value of any food to achieve this. The food values stay exactly as measured. Only the target is expressed as a range.

Vitamin A: RE, not RAE

Vitamin A comes in two forms. Retinol, found in dairy and eggs, is ready to use. Carotenoids — the orange and dark green pigments in carrots, sweet potato, spinach and peppers — have to be converted by the body first, and the conversion is inefficient. Any vitamin A figure is therefore a sum: the retinol, plus a fraction of the carotene.

Which fraction is where the two conventions part company.

In 2001 the US Institute of Medicine reviewed the evidence and concluded that the vitamin A a body actually obtains from dietary carotenoids is about half what the older RE method assumed. The ratio was revised from 6:1 to 12:1 for beta-carotene, and from 12:1 to 24:1 for the other provitamin A carotenoids. Most of the world adopted the revision. The UK did not — the COMA figures were set in 1991 and SACN has not revisited vitamin A since.

What Nutriofia shows

All vitamin A values are RE, taken directly from CoFID's own retinol equivalent column. Not converted, not estimated — the figure the UK dataset publishes. The daily target of 700µg is also RE, so the number and the target are measured on the same basis.

For a raw carrot this is 1,961µg per 100g: CoFID records 11,764µg of carotene and no retinol, and 11,764 ÷ 6 is 1,961. On the RAE basis the same carrot would read 980µg.

The honest caveat, which matters on a plant-based site

RE is internally consistent but it is generous. The 2001 revision exists because the 6:1 ratio overstates how much vitamin A the body genuinely gets from plant carotenoids — and some researchers argue even 12:1 is optimistic.

That matters more here than it would elsewhere. Nutriofia is a wholefood database, so most of the vitamin A in it is carotene rather than retinol — precisely the part the older method flatters. If you are relying on plant sources for vitamin A, treat the figures on this site, and the UK's 700µg target, as the optimistic end of the range rather than the middle of it.

We show RE because mixing two conventions in one column is the only option that is definitely wrong, and because the UK target is expressed in RE. Showing RAE values against an RE target would understate plant foods by up to half.

Daily reference values: which ones we use, and where

Percentages need something to be a percentage of, and there is more than one official answer. Nutriofia uses two different sets, deliberately, in two different places.

On food pages — EU Nutrient Reference Values (NRVs)

The % Daily Value column on an individual food page uses EU NRVs: the labelling standard, and the same basis used for the "%RI" figures printed on food packaging in the UK. We use these here so that what you read on a Nutriofia food page is directly comparable with what you read on a packet in a shop.

NRVs are a single adult figure. They do not vary by age or sex, and they are not always the same as UK dietary reference values — for example the NRV for vitamin C is 80mg where the UK RNI is 40mg, and the NRV for iron is 14mg where the UK RNI is 8.7mg for men.

In reports and trackers — UK Dietary Reference Values

When Nutriofia assesses what you personally have eaten, it uses UK Reference Nutrient Intakes from the Department of Health's Dietary Reference Values, updated by SACN where newer advice exists — vitamin D (2016) and fibre (2015). These vary by age and sex, which matters: the iron RNI for a teenage girl is 14.8mg against 11.3mg for a boy of the same age.

Why the two differ, and why that is not a mistake

They answer different questions. A food page asks "what does this food contain, compared with the label standard?" A report asks "is this person getting what they need?" The first wants a fixed public benchmark; the second wants something matched to the individual. Using one set for both would make one of those answers worse.

Units matter too

Some reference values are expressed in equivalents rather than raw amounts, and we record which: vitamin A as retinol equivalents (RE), folate as dietary folate equivalents (DFE), vitamin E as alpha-tocopherol equivalents. These are noted against each target in our database so the comparison is like for like.

Carbohydrate: available carbs, not by-difference

CoFID reports available carbohydrate — the directly measured starch and sugars that are actually absorbed and raise blood glucose. The USDA "by-difference" method computes carbohydrate as whatever is left over after subtracting protein, fat, water, and ash from 100g. Because the USDA carbs figure includes fibre, it will be higher than CoFID's available carbs figure by the fibre amount.

When we import USDA carbohydrate values, we subtract the (already-converted) fibre value to arrive at available carbohydrate, which is then stored and displayed. All carbohydrate values on Nutriofia are therefore available (net) carbohydrate, consistent with CoFID convention and with UK food labelling.

Source tags on every value

Every nutrient value stored in our database carries a source tag recording exactly where it came from. You won't normally see these on the food page, but they're there for full audit traceability. The tags look like:

Adjustments applied to USDA values at import

Where a straightforward conversion factor is inadequate (for example, folate forms with different bioavailability), we store the value with a note rather than applying an unreliable formula.

Who these values are reliable for

Nutriofia is built for a UK audience but the data is broadly reliable for users across Western and Northern Europe for the types of foods we cover. For raw vegetables, fruits, nuts, seeds, and legumes — our largest categories — typical country-to-country variation is around ±10%, which is within the natural variation of the foods themselves.

Where CoFID figures may diverge from other European databases is in fortified products. UK flour is mandatorily fortified with iron, calcium, and B-vitamins to levels above those common in mainland Europe. If you are using Nutriofia to analyse intakes from fortified breakfast cereals or bread products, the values will reflect UK fortification levels and may overstate the nutrient content of equivalent products bought in Germany, France, Italy or the Netherlands.

For US users: our fibre values will appear lower than US labelling because we use NSP, not AOAC. Our carbohydrate values will also differ because we show available (net) carbs rather than by-difference carbs. The values are not wrong — they reflect a different, equally valid measurement standard.

What we don't do

No automatic on-page conversions. Normalisation to UK conventions happens at import time and is already built into the stored values. When you read a fibre number on a food page, you are reading a stored NSP value — there is no further calculation happening in the browser.

No overwriting of source data. Once a value is imported from a primary source, it is never silently replaced by a value from a different source. Any correction to a stored value is documented with an updated source tag.

No invented values. If a nutrient is not in our database for a particular food, the page will show no data rather than a zero or an estimate. A zero value on a food page means the source database recorded zero; an absent value means the source database did not report it.

Coverage is visible, not hidden. Because we never gap-fill, the completeness of our data varies by nutrient — some compounds are measured for almost every food; others are measured for only a handful. Each nutrient page shows a coverage percentage telling you what proportion of our foods have a recorded value for that nutrient. This turns what could look like a weakness into something you can actually see and judge for yourself.

Questions about the data