Dietary statistics

Dietary Statistics for the Norwegian Population

Section for Health, Care, and Social Statistics

Figures are published for the entire country.

Annually

Not relevant

In addition to data on sold individual items throughout the year from the selected grocery stores, and data on product declarations from Tradesolution and kasal.app, results that have undergone processing programs are stored at the individual item level.

Data are anonymized.

Custom tables can be built and API extracts can be made from the Statistics Bank.

The statistics are published on https://www.ssb.no/en/helse/helseforhold-og-levevaner/statistikk/kosthald

Other dietary statistics can be found here:

• Utvikling i norsk kosthald (Helsedirektoratet): https://www.helsedirektoratet.no/rapporter/utviklingen-i-norsk-kosthold
• Norkost 3 (Helsedirektoratet): her
• Nasjonale kosthaldsundersøkingar (FHI): https://www.fhi.no/studier/nasjonale-kostholdsundersokelser/

Publikasjonar og andre lenkjer

• Survey of consumer expenditure: https://www.ssb.no/en/inntekt-og-forbruk/forbruk/statistikk/forbruksundersokelsen
• Food Table (Matvaretabellen): https://www.matvaretabellen.no/

The Norwegian Directorate of Health has previously published dietary statistics based on data from consumption surveys from 1974 to 2012. The statistics were published in various publications.

Since 2018, the data sources have been aggregated data on sold food items over a calendar year from a selection of stores operated by the four largest players in the grocery industry, as well as detailed data on product declarations from the EPD database of Tradesolution AS, www.kassal.app, and matvaretabellen.no.

The main purpose of the statistics is to provide a detailed overview of the diet of the Norwegian population over a calendar year and per day per person across various food groups and nutritional content. As the statistics are published annually, they provide the opportunity to monitor the development of the diet of the Norwegian population.

The statistics are intended to meet knowledge needs as expressed in the action plans of central health authorities and in an agreement of intent concluded between the food industry and health authorities.

The main areas of use at Statistics Norway (SSB) include National Accounts, economic statistics, and statistics on health and lifestyle.

The statistics are important for central health authorities, who require updated and relevant information on the diet of the Norwegian population to meet knowledge needs as expressed in the action plans of central health authorities and in an agreement of intent concluded between the food industry and health authorities.

Key users include central health authorities, researchers, the food industry, the media, and the general public.

No external users have access to the statistics before they are simultaneously published for everyone at 8:00 AM on ssb.no, following advance notification at least three months prior in the Statistical Calendar. For more information, see the Principle of Equal Treatment.

Dietary statistics are closely linked to the Consumption Survey. Previous dietary statistics have been published based on data from the Consumption Survey. While dietary statistics use their own code system for food grouping, the Consumption Survey is based on COICOP-HBS (Classification of Individual Consumption by Purpose - Household Budget Survey). Consumer groupings in official statistics.

Statistics Norway collects data in accordance with the Duty to Provide Information under Section 10 of the Statistics Act (https://lovdata.no/dokument/NL/lov/2019-06-21-32?q=statistikkloven). At the same time, Statistics Norway uses this information for the preparation of official statistics in accordance with Section 10 of the Statistics Act (Act of 21 June 2019 No. 32).

Not relevant

The first publication includes data for the 2018 dataset. Subsequently, additional datasets for subsequent years will be published over time.

The statistics include figures on sold food items from a selection of grocery stores belonging to the chains Bunnpris, COOP, Norgesgruppen, and Rema 1000 over a calendar year (the same selection used for the Consumer Price Index (CPI).

The figures are estimated to be representative for the country using weights based on sales figures for the four different chains and independent stores as explanatory factors. The sales figures used for the relevant statistical year are within the NACE 47.111 sector "Retail sale in non-specialized stores with food, beverages, and tobacco predominating."

The statistics do not include food purchased from kiosks, gas stations, cafes, restaurants, and canteens. They also do not include alcoholic beverages purchased at Vinmonopolet or cross-border/tax-free shopping. Furthermore, they do not account for food waste.

Data sources:

The statistics gather data from the following sources:

• Aggregated data on sold food items from a selection of stores operated by the chains Bunnpris, Coop, Norgesgruppen, and Rema 1000.
• Detailed product declarations from the EPD database of Tradesolution AS.
• Detailed product declarations from www.kassal.app.
• Information on nutritional content and edible portion from Matvaretabellen (www.matvaretabellen.no).
• Machine learning for predicting nutritional content for items lacking this data.
• Enterprise and Establishment Register (VoF) for total store population/sales in the relevant year.
• Population statistics from Statistics Norway (SSB).

Sample:

The statistics include data on the quantity of sold food items (at the individual item level) from a sample of just under 200 grocery stores operated by Coop, Norgesgruppen, Rema 1000, and Bunnpris for a calendar year. The sample is the same as that used for the Consumer Price Index: https://www.ssb.no/en/prices-and-prices-statistics/consumer-price-index.

The quantity of sold food items is further linked to detailed product declarations from the EPD database of Tradesolution, kassal.app, and matvaretabellen.no. These declarations include information on nutritional content, edible portion, and details on weight and volume for the sold food items.

Data collection:

Aggregated data on sold items at the individual item level from a selection of stores in the four largest grocery chains (Bunnpris, Coop, Norgesgruppen, and Rema 1000) are collected through extracts from the respective grocery chains' systems on a monthly basis.

Product declarations from Tradesolution AS, which include nutritional content and information on quantity/volume for sold food items, are collected through extracts from the EPD (Environmental Product Declaration) database.

Product declarations from www.kasal.app are collected through extracts from the database.

Quality control/revision:

Various logical and manual checks have been performed on the data collected for the entire year. These checks primarily concern nutritional content, quantity/volume, and grouping of food items into food categories related to the sold products.

Calculations: The statistics are presented as sold quantities of food and beverages per person per year (kg/liter) and per day (grams).

Calculation for per year: Total edible amount of sold food and beverages (kg/liter) / total population = Edible amount of sold food and beverages per person per year

Calculation for per day: (Total edible amount of sold food and beverages (grams) / total population) / 365 days = Edible amount of sold food and beverages per person per day

Calculation for energy percentage (E%): Example calculation for E%: 80 grams of fat provides approximately 3,000 kJ (80g x 37kJ/g = 2,960 kJ). For a person with a total daily energy intake of 10,000 kJ, such a fat intake contributes 30 E% (3,000 kJ / 10,000 kJ) x 100 = 30%

For calculations of uncertainty, refer to the section on "Sources of Error".

Not relevant

Not relevant

Not relevant

Not relevant

In any data source, there can be inaccuracies that arise both during data collection and processing.

The data collection in the dietary statistics is based on the sale of food items from a selection of stores reported by the grocery store chains COOP, Norgesgruppen, Rema 1000, and Bunnpris over the course of a calendar year. The sold food and beverage items from the store selection are then compiled with product information, including nutritional content, from the EPD database of Tradesolution, www.kassal.app, and the Food Composition Table.

The EPD database of Tradesolution AS contains updated information on the nutritional content of various food and beverage items in real-time. This means that changes in nutritional content in already existing food and beverage items will be updated in the database when they receive new information from the supplier of the specific food item. In connection with the preparation of statistics for the 2018 dataset, the nutritional content is based on an extraction from the EPD database as of January 2021. This implies that the nutritional content underlying certain food and beverage items in the dataset may not necessarily be the same as it was in 2018.

The nutritional content

A large portion of the sold food and drink items lack information on their nutritional content in the dataset. This is particularly true for fresh foods, fruits, and vegetables. To assign nutritional values to these items, a combination of different approaches and methods has been used.

Firstly, manual reviews have been conducted on the most sold items that lack nutritional information. Standard values from similar items sourced from the Food Composition Table or nutritional information about a product sourced directly from the manufacturer's website have been applied.

Secondly, machine learning has been used to predict and provide nutritional content for items lacking this information. The methods used are based on examining similar foods or items within the same food group to provide likely nutritional content values.

COICOP classification

To assign the correct COICOP classification to the sold food and beverage items, goods are classified into COICOP using both manual and machine learning processes. The UN 2018 standard for COICOP is used for classification. Item names along with food chain groupings and prices contribute to the training of the algorithms. A Support Vector Machine Classifier model is used for prediction.

For the most sold food and beverage items, manual reviews have been conducted to ensure that these items have been correctly classified. However, it is possible that some foods have been placed in the wrong food group.

For the dietary statistics, data are not published according to the COICOP classification, but rather based on a separate national food grouping system. However, the COICOP classification is important in the work related to the data foundation for dietary statistics.

Weight and volume (quantity)

Calculation of weight and volume for a significant portion of the food and drink sold is prepared based on a combination of different methods:

Method 1: Retrieves weight and volume from product declarations from the EPD database of Tradesolution AS.

Data from Tradesolution gives figures on quantity and what type of unit is referred to. All products that have specified quantity units of the type weight or volume can be retrieved from the data set and standardized to litres/kg or grams. The best-selling items with quantity types "portion", "piece" etc. have been assigned an average weight per unit and converted to kilograms.

Method 2: Retrieves weight and volume based on information about this from the product name text field in data from the store selection of the grocery chains.

Method 3: Goods with a non-zero decimal number on the quantity sold are counted as bulk goods and receive one kilogram.

Method 4: Estimation of weight and volume based on COICOP groups.

The goods that cannot be given a weight/volume value from text or from Tradesolution AS data can be estimated from average values ​​from products in the same COICOP group.

Method 5: Conversion from volume to grams of edible part for certain foods

For certain foodstuffs, separate conversions from volume to grams have been made to obtain the most correct amount possible. This applies, among other things, to ice cream, oil, jam, where you have used your own conversion factors based on the food table to convert from liters to grams.

Method 6: The few items that have not been weighted according to these methods have been assigned the average weight of all food items of around 0.5 kg.

All quantities are normalized to kilograms, so that energy and nutrient content per 100g can be multiplied by 10 and used to calculate total quantities.

By combining the use of these methods, there will be individual items sold that may have the wrong weight and volume (quantity) stated.

Through various checks, an attempt has been made to find obvious errors and rectify them.

When errors have been corrected, as far as possible, the experience is that the statistical results are in most cases relatively little affected by both collection errors and processing errors. The effects of errors can still be significant in some cases, e.g. for certain food groups and nutritional content, and it is clear that not all errors are detected.

SSB stores collected and revised data in a secure manner, in accordance with current legislation on data processing. SSB can provide access to the underlying data (de-identified or anonymized microdata) on which the statistics are based to researchers and public authorities for the purpose of producing statistical results and analyses. Access may be granted upon application and under certain conditions. For more information on this, please refer to the Access to Data from SSB page.