Editorial methodology

How an entry is written.

Every entry in The Nutrition Dex is produced against the same five rules. We publish them so readers can audit our copy against the standard.

1. Term selection

Candidate terms are drawn from three places: the USDA FoodData Central documentation, the FDA's nutrition-label regulations (21 CFR 101), and the peer-reviewed dietary-assessment literature — particularly the Journal of Nutrition, American Journal of Clinical Nutrition, Public Health Nutrition, and JAMA. We prioritise terms that carry a measurable quantity (an error figure, a tolerance, a laboratory method) and that are either inadequately defined elsewhere or scattered across sources of uneven quality.

2. Source tiers

Entries cite sources in this order of preference, and we state the tier in the References block:

• Tier 1 (primary regulation or dataset): the USDA FoodData Central release (SR Legacy, Foundation Foods, FNDDS, Branded Foods), the relevant section of 21 CFR 101, the ICH or AOAC method document.
• Tier 2 (peer-reviewed methodology): the journal paper that supplies the validation protocol or the laboratory-reference meal set. We prefer studies with a reported n, a stated reference method, and the statistical measure (MAE, MAPE, RMSE, or a paired t-test) named explicitly.
• Tier 3 (authoritative synthesis): NIH Office of Dietary Supplements fact sheets, Harvard T.H. Chan School of Public Health reference pages, Academy of Nutrition and Dietetics position papers.

We do not cite consumer blog posts, app marketing pages, or aggregator sites as authority.

3. Statistical reporting standard

Where an entry makes an accuracy claim, the claim carries:

• The specific error statistic (MAE in kcal, MAPE in per cent, RMSE, or percentage within the FDA 20% tolerance).
• The reference meal set against which it was measured, including the n.
• The year the measurement was taken — because database freshness matters.

"Accurate" is not a claim. "±1.2% MAPE against an n=500 Bitebench reference meal set in 2026" is a claim.

4. Commercial disclosure

When a consumer app, scale, or device is named in an entry — and this happens only where editorially relevant — we follow three rules:

• It is named alongside at least two competitors covering the same use-case.
• Each named product carries an accuracy figure drawn from a source we would otherwise cite (a peer-reviewed validation, a public benchmark report, or a regulatory filing).
• The publication's relationship with the vendor, if any, is disclosed in plain English at the bottom of the page. We currently hold no advertising contracts, no affiliate relationships, and no equity in any nutrition-technology company.

5. Correction policy

Entries are reviewed on a rolling basis and substantively re-reviewed whenever a cited source is superseded (a new USDA FoodData Central release, a revised FDA rule, a methodology paper that replaces the validation protocol we were leaning on). Reader-submitted corrections we can verify are honoured in place, with a dated corrigendum line at the foot of the entry. We do not silently edit published entries.

Funding

The Nutrition Dex is funded by reader subscriptions and, in due course, by an institutional-licensing tier for dietetics programmes and clinical teams. We do not accept advertising, sponsored content, paid placement, or affiliate commissions. We decline press-trip invitations from nutrition-technology companies. If a vendor offers data for a piece, we use it only when the underlying method is documented to the standard above, and we disclose the arrangement on the page.