Food composition databases compared: USDA, CoFID, Ciqual and Frida

USDA FoodData Central

FoodData Central is maintained by USDA and includes multiple data types such as Foundation Foods, FNDDS, Branded Foods and SR Legacy. These types differ in source, update frequency and intended use.

USDA states that FoodData Central data are in the public domain and published under CC0 1.0. Foundation and Experimental data are updated periodically, while branded-food data are updated monthly.

• Strong breadth and API access
• Clear separation of data types
• Useful branded-food coverage
• Source type should be considered when matching an ingredient

UK CoFID

McCance and Widdowson’s Composition of Foods Integrated Dataset consolidates nutrient data for foods in the UK food supply. The current GOV.UK publication is the 2021 dataset and user guide.

The dataset is published as an Excel file and the documentation states that the information may be reused under the Open Government Licence, subject to the stated terms and exclusions.

• Useful UK-oriented generic-food data
• Downloadable spreadsheet with documentation
• Current published release is older than some other national tables, so release date should remain visible

France Ciqual

ANSES published Ciqual 2025 as the current French national food-composition table. The 2025 release covers 3,484 foods and 74 constituents and added more than 300 foods compared with the 2020 version.

Ciqual documentation also exposes source information and methodological detail, which is valuable when a reviewer needs to understand how an average value was produced.

Denmark Frida

Frida is the Danish Food Composition Database. Version 5.4, released in May 2025, updated multiple food groups and continues to provide detailed food-level nutrient and source information.

Frida can be especially useful for Nordic foods and for parameters not consistently available in older datasets, but as with any national table, the food description still needs to match the ingredient actually used.

How to choose between databases

Choose the food record that best represents the ingredient rather than selecting a database solely because it is familiar. Geography can matter, but food form, analytical basis, brand formulation, fortification and data completeness may matter more for a particular ingredient.

• Prefer the closest food description and preparation state.
• Keep the database, release/version and food identifier with the ingredient record.
• Do not merge apparently identical nutrient fields without checking definitions and units.
• Treat missing values as missing unless the source explicitly defines them as zero.

Why a multi-source strategy can be useful

A professional recipe system may need more than one trusted source because no national database covers every ingredient equally well. A multi-source approach can improve matching, provided the application preserves provenance and uses consistent nutrient definitions.

The goal is not to maximise the number of databases. It is to use the most appropriate defensible record and retain enough metadata to review that choice later.