Nutrient data

National food tables vs brand-level nutrient data: Implications for dietary assessment

When conducting dietary assessment, whether for population nutrition research or clinical support, it’s important that nutrient estimates are as accurate as possible. Accuracy is especially vital when supporting vulnerable patients in clinic (such as diabetic and renal patients), where small differences in nutrient values could make big differences to their health.

When it comes to accuracy, the choice of dietary assessment tool, and the nutrient composition tables which sit behind it, have an important role to play. This article takes a deep dive into food and nutrient databases and their influence on the accuracy of dietary estimates.

Where does nutrient information come from?

There are two main sources of nutrient information:

National food composition databases

Many countries compile their own databases detailing the nutrient composition of typically-consumed dishes within their population. For example, the UK Government maintains the McCance and Widdowson’s Composition of Foods Integrated Dataset (CoFID). National food tables are usually free to access and are commonly used in dietary research.

Branded product nutrient databases

These databases tend to be maintained by a third-party company who ensures they provide an up-to-date snapshot of products available on the market. This means there is typically a cost associated with access. In some cases, data may also be shared directly by food manufacturing or retail companies.

Both types of food composition databases have their pros and cons for the accuracy of dietary estimation which we explore below.

While accuracy and precision errors are inherent in observational studies, we can take steps to minimise them – a key factor in influencing the accuracy of dietary estimates is the food composition database used. In an ideal world, we want our dietary estimates to be as accurate and precise as possible, meaning our measurements are consistently close to the truth. But often we experience a trade-off between accuracy and precision.

The nutrient estimate

Once food items have been reported, they must be coded to a database containing food composition information to determine the corresponding nutrient composition. Accuracy of the nutrient estimate therefore relies upon:

  • The ability to match entered foods to an appropriate product in the composition database
  • The accuracy of nutrient information for products in the food composition database
  • The coverage of nutrients in the food composition database

Now, let’s take a look at the pros and cons associated with national food tables, and brand-level nutritional information, and see how they can influence the accuracy of nutrition estimates.

National Food Composition Databases (FCDBs)

Pro: Good nutrient coverage

National food tables collect data for a large range of nutrients (the UK’s CoFID tables covers over 100 nutrients/food components), however nutrient values are often missing. The core UK database for myfood24 combines data from McCance and Widdowson’s Composition of Foods Integrated Dataset with other national food tables from across the globe, to fill in these gaps. By doing this we’ve reduced nutrient gaps by 20% to improve the nutritional coverage of food and drink items, providing complete data on at least 60 nutrients, for every product.

Pro: Robust methods

Values in national nutrient databases are derived from lab-testing on multiple samples, to account for variety and seasonal variation. The method and sample size are also stated for transparency.

Pro: Captures all food types

Nutrient data is captured for all types of foods, regardless of whether they are required by law to display on-pack nutrition labels. This includes foods which are commonly consumed out of home (e.g. fish and chips), or home-prepared, and alcoholic beverages.

Con: Limited number of foods

National food tables are limited to only the most commonly consumed foods which represent the diet of the nation. The latest CoFID database captures around 3,000 foods and beverages. This sounds like a lot, but with tens of thousands of food and drink products on the supermarket shelves, it doesn’t cover everything on the market. Food composition tables also tend to under-represent ethnic foods, which may have implications for the accuracy of dietary estimates for minority ethnic groups.

Con: Limited range of variants

The food industry is constantly innovating to offer new products in new flavour variants and formats, contributing to the huge range of products on the market. National composition databases only represent a limited number of common variants. For example, in the UK’s CoFID database, the entry for a muffin is only captured by three sub-variants (fruit muffins, chocolate chip, or bran muffins). But what about triple chocolate, and lemon and poppy seed? We know the offering in reality is much broader. Selecting a standard database entry is likely to under-estimate the sugar and calories.

Con: Infrequent updates

Laboratory tests take time, so dietary composition tables aren’t updated very frequently. This means values can become outdated and may not reflect current market availability, farming/production practices, reformulation, and trends, such as the recent rise in availability of plant-based meat and dairy alternatives.

Branded product data

Pro: Large number of foods

There are many thousands of foods on the market, which each have their own nutrient composition. Research across Europe has found stark differences in the nutrient compositions of similar foods sold by different brands. Evidence suggests that lower-cost supermarket own-brand products often have a more favourable nutrient profile than their branded equivalents. In a study of the UK grocery market, lower cost ready meals were found to have around 2g less fat per 100g of product compared with higher cost versions.

Branded composition databases therefore reflect the granularity of these product-level differences, highlighting where swaps may be helpful to improve a patient’s diet.

Pro: Regular updates

Brands are constantly updating their products in response to consumer trends and the changing policy landscape. Analysis of supermarket websites showed 11% of pizzas changed their nutritional composition over a 6-month period. Furthermore, in response to the Soft Drinks Industry Levy, brands have introduced low/zero-sugar options, reformulated their products, and reduced the pack sizes they offer, resulting in a reduction in purchased sugar from soft drinks.

Using brand-level nutrient composition data means dietary estimates keep up with product-level changes.

Pro: Portion quantification

Research has shown that portions sizes for commonly consumed foods have crept up over the last few decades. Between 1993 and 2003, the average size of a bag of crisps increased by 50%. Without regular updates, ‘typical’ portions captured by food composition tables may now be out of date.

The inclusion of on-pack portion size information on branded products may help users to more accurately quantify consumed portions. That said, as we discussed in a previous blog, serving sizes are often missing from packaging, and don’t necessarily reflect the amount we eat (particularly for sharing bags of crisps/popcorn where we tend to consume more than the recommended serving).

Con: Limited coverage of food types

Product nutrition information captures only data which is reported on the product label. This means, for products which aren’t legally required to display a nutrition label (such as alcoholic drinks or those sold by small out of home retailers), no nutrient data will be available which could have a significant impact on the accuracy of the nutritional assessment.

Con: Limited nutrient coverage

Labelling requirements have a bearing on which nutrients are reported in product-level data too. In the UK, declared nutrients are limited to energy, fat, saturated fat, carbohydrates, sugar, protein, and sodium, which are mandatory, and fibre which is optional unless a fibre claim is made. This makes branded nutrient data unsuitable for assessing micronutrient intakes. What’s more, different labelling regulations in different countries may hinder between country comparisons.

Con: Lack of transparency

By law, nutrient estimates on the product label may be derived from laboratory testing, or recipe analysis based on typical CoFID values. The choice of method may have implications for the accuracy of information. However, the method is not required to be declared on pack, meaning we must take labels at face value.

How does myfood24 improve the accuracy of dietary intake estimation?

At myfood24 we pride ourselves on our novel and extensive food composition dataset, which is designed to maximise accuracy and coverage. So, what makes our nutrient database unique?

The myfood24 food and nutrient dataset includes both generic items from food composition tables as well as branded product data. Going a step further than other nutrient analysis software, we map our branded products to similar items in FCDBs to provide the missing micronutrient data which is so desperately needed. This unique data enhancing process provides a more complete nutrient profile for branded food items, and blends the pro’s of using both FCDBs and product-level data. This combination creates a unique and robust dataset containing 92,960* food and drink items, 96% of which are branded (representing 7,288 different brands), plus a wide range of nutrient coverage.

The unique myfood24 database makes it easier for users to select exactly what they ate, providing a more accurate representation of their dietary intake.

What might the future hold for dietary monitoring?

With such an active policy space around food retail in the UK currently, we can expect to see branded nutrient information being increasingly important for public health monitoring. With the ability to identify an exact product and to integrate nutritional data with information on sustainability metrics, purchases, and more, it is likely that barcode scanning technology will play a bigger part in the future of dietary assessment.

*figures relate to the core UK food and nutrient dataset and are correct as of February 2022

If you’d like to find out more about the myfood24 food and nutrient dataset you can get in touch using our contact form or try a free demo of the myfood24 food diary available in multiple languages with international food composition tables.

Interested In Using myfood24?
Why Not Get In Touch or Try Our Free Demo