Much of the content here is summarised from the FAO e-learning course on Food Composition Data along with the videos on food composition, food matching and recipe calculations at Food Composition Explained)(Please note the views expressed on foodcomponline.info are those of the author(s) and do not necessarily reflect the views or policies of the Food and Agriculture Organization of the United Nations).
It is recommended that you complete the FAO e-learning course and watch the videos in advance of linking the food list to composition data. This will give you a better understanding of the issues and challenges when using/managing food composition data. There is also an FAO/INFOODS FOOD COMPOSITION STUDY GUIDE to go along with the FAO course and the book Food composition data: Production, Management and Use by Greenfield and Southgate, 2003 which can also be used to complement the information summarised here.
All of the foods in the food list need to be linked with appropriate food composition data. This should be carried out by a research nutritionist or dietitian who is experienced in collecting and coding dietary data from the population of interest.
Food composition overview
Food composition data detail the nutritional content (including energy, macronutrients and micronutrients) of individual foods or drinks (raw and or cooked) and mixed dishes such as pies, stews, soups and curries. They can be quite specific, e.g. pink lady apple, with skin and core, skin eaten, or generic e.g. Red apple, average. They allow the conversion of information on the foods and quantities an individual consumes into an estimate of energy and nutrient intakes.
Ideally each country would have its own comprehensive food composition table (FCT) containing the nutritional content of locally consumed foods, derived by chemical analysis. This is because significant variation in the composition of foods occurs due to differences in the cultivars of plants grown or the breed of animals used, differences in the soil, climate and farming practices, level of ripeness/ age of the animal, different policies around fortification, different brands of packaged foods and differences in processing and cooking practices.
In reality due to resource constraints the comprehensiveness and quality of FCTs differs between countries. They range from around 100 foods to over 10,000 foods and from around 20 nutrients to over 200 nutrients. The FAO website has a comprehensive list of food composition tables broken down by continent and country or region. These vary in format from searchable databases (FCDBs) and downloadable excel spreadsheets to Pdfs of books, and references to published journal articles. They vary greatly in their coverage of the local foods with some containing just a few local dishes and or only a few basic nutrients. As chemical analysis of the nutritional composition of foods is expensive many food composition tables will include analytical data which is compiled from published literature or unpublished laboratory reports, data which is borrowed from neighbouring food composition tables or calculated based on the nutrient composition of the ingredients. The quality of food composition tables varies, FAO have published the FAO/INFOODS Evaluation framework to assess the quality of published food composition tables and databases to help users select high quality FCT/FCDBs for their work, and also demonstrate areas of improvements to compilers for future editions.
Depending on the available food composition data and the resources available to develop the system, foods within the new Intake24 food list may be; matched to existing food composition data, matched to existing food composition data along with a small number of new composition codes added from recipe calculations and food packaging information or, a new country specific food composition database may be developed.
Matching to existing food composition data
Where comprehensive food composition data for the country or region exist the foods in the food list can be linked to one of more of the food composition databases within the system. For example, if the main source of imported food was the US then the USDA tables could be used along with any local tables from the target country or region. For instance, an Intake24 food list for Malaysia could use a combination of the Malaysian Food Composition database, Singapore’s Energy and Nutrient Composition of Food and the ASEAN food composition database.
If a food composition table is to be uploaded to Intake24 the Nutrient Mapping spreadsheet will need to be completed in order to map the nutrient columns in the FCT to the columns within Intake24.
Items within the food list should be matched with food composition data based on the:
Food description
Component definition
Analytical method
Unit and denominator
Further details are given below.
Food description
Ideally the food description given in the food composition table should give enough detail to identify the type of food and if appropriate the cooking method, for example not just “chicken” but “chicken breast, roasted with skin”.
In matching a food to a food code, you need to think about the qualities of that food which are likely to result in differences in nutritional content. For example, for fruit, the colour, ripeness and whether the inedible portion is included will be important. For meat, the type and part of the animal, the age of the animal, the cut of meat used and the cooking method will all need to be considered. For grains whether or not the husk is included in the analysis is important.
Use the description of the food to identify whether the food is a good match. For example, if matching an apple is it the same variety? The same colour? Does the value include the skin? Is it for the edible portion only or does it include the weight of the core?
If matching beef steak, is it the same cut of beef? Is the cooking method the same? If fried was the oil used the same? Was the visible fat cut off?
Food matching can take a long time and priority should be given to foods which are consumed frequently or in large amounts or that are significant contributors to a nutrient of interest (e.g. oily fish). For some foods you will find a high-quality match e.g. cow’s milk, semi-skimmed, winter, pasteurised or you may need to match to a generic code such as cow’s milk, average. If a food is consumed frequently and no exact match exists it may be desirable to take a selection of milk codes from several high-quality food composition databases and take an average. For infrequently consumed foods an approximate match e.g candlenuts matched to a food code for macadamia nuts may be sufficient. It is important to document in the food list the quality of the match so foods with a low-quality match can be re-visited, if necessary, when time and data availability allow.
High quality match
– The food description matches for all descriptors and so does the component definition
Medium quality match
– An average taken from several similar foods
– Derived by recipe calculation
– Matched to a similar food e.g boiled for steamed or peas for green beans
Low quality match
– Raw values for a cooked item
– Match at the food group level e.g Camel matched with Beef
Always look for a high quality match first, but if a low-quality match is the best available and the food is infrequently consumed that may be good enough.
When matching it is important to consider the components of interest. If fat content is of interest it will be important that details of cooking methods are given, for both the food to be matched and the food code along with whether the visible fat was removed from cuts of meat and that low-fat options are included for packaged food products. If salt is of interest it may be worth including options for cooked with and without salt. It is also important to be aware of differences in naming between different countries e.g. “Curry“ could be a cooked dish or the spice, “Jelly” could be a fruit preserve or a dessert. There may also be differences in the formulation of packaged food products between different countries and different policies on food fortification.
Further detailed information and examples for matching to existing food composition data are given in the FAO/INFOODS Guidelines for Food MatchingTable 2 and Annex 1.
If no food composition data exists for a commonly consumed dish and the composition of the individual ingredients is available the recipe information can be used to calculate the nutritional composition which can then be added to the appropriate food composition table for linking. This may also be necessary for frequently consumed items where only the raw composition data is available, e.g. calculating values for cooked pasta from dry pasta if only uncooked values are available (See Recipe calculation section for further details).
For packaged foods, data from published food composition tables should be used if these exist. Back of pack information can be used for the macronutrients, if no other data can be found. Ideally these would be used to guide matching with food composition data for a food with a similar profile. They could also be used along with the micronutrient composition of a similar item. The source and date of access of these data should be recorded in the food composition table to allow auditing.
Where a significant amount of new food composition data needs to be generated a new food composition table should be compiled. See Compiling a new food composition database for further details.
In addition to matching based on the food description, when matching across multiple food composition databases it is essential to make sure that the component definition and expression and the units, denominator and chemical analysis methods match.
Components
The components in an FCT are the values given for each food. These include the nutrients but also any other components of interest such as antioxidants and food additives (important for exposure assessment). The nutrients include the macronutrients Carbohydrate, Protein, Fat, Alcohol and Ash and sometimes also the macronutrient fractions such as Starch, Fibre, Sugars, Fatty acids and Amino Acids, along with the micronutrients (vitamins and minerals).
In addition to checking that the food description matches it is also important to make sure that the components are the same. For some nutrients the name will be enough, for example for the minerals, but for other nutrients there are several definitions, analytical methods and expressions. For example:
Definition – Carbohydrate can be given as total or available carbohydrate
Expression – Carbohydrate can be expressed by weight or in mono-saccharide equivalents.
Nutrients which need careful attention due to different analysis methods, definitions and methods of expression are:
Macronutrients
Micronutrients
Energy
Vitamin A
Protein
Vitamin B6
Fat
Niacin/ Niacin equivalents
Carbohydrate
Folate
Fibre
Vitamin C
Vitamin D
Vitamin E
Vitamin K
Some food composition tables will differ in the analytical methods used to derive the values for the component and therefore it is important to check that the analytical method used will result in equivalent values (See Chemical analysis methods).
Units and denominators
Units are the measurement in which the nutrient value is given. For example, gram, microgram, millilitre etc. The denominator is the amount of food containing that value. In food composition tables the denominator is usually per 100g of edible portion of food on a wet weight basis. However, it can sometimes be expressed as per 100ml for liquids, per 100g dry weight basis of the food or for amino acids as per g of protein and for fatty acids per 100g of fat in the food. It is important therefore when using multiple food composition tables, and especially data compiled from literature or unpublished laboratory reports, to check both the units and denominator for the value given (care should also be taken to determine whether data is expressed on a dry matter or wet weight basis, particularly for data sources such as lab reports, university thesis, etc) in case data conversion needs to be carried out.
It is possible to convert one denominator to another. For converting volume (ml) to weight (g) you need the density for the liquid which can be found in the FAO/INFOODS Density Database.
You can calculate the nutrient value per 100g of food using the value per 100ml of food by the following equation:
Non-numeric values in a food composition table include Tr which means just a trace of the nutrient is present in the food and N/ND/- or blanks to indicate missing data (e.g. the nutrient may be present but the analysis wasn’t performed). It is advisable to verify the definitions non-numeric values of the different FCT.
Nutrient values which are always calculated
Some nutrient values are always calculated, these include Energy, Protein, Carbohydrate and some vitamins.
Protein can be either calculated from measured nitrogen content using food/food group specific conversions factors (See Factors for the conversion of Nitrogen values to protein) or calculated from the individual amino acids, as measured by HPLC or Ion exchange chromatography.
Carbohydrate can be given as either available carbohydrate (that is the sum of sugars, oligosaccharides, starch and glycogen) or total carbohydrate which includes dietary fibre as well. It is recommended to use available carbohydrate. There are 3 methods for calculating available carbohydrate:
Available carbohydrate by weight, which is the sum of anhydrous sugars, starch and glycogen is the preferred method but is expensive and therefore rarely used.
Available carbohydrate by difference, uses the content of water, protein, fat, alcohol, ash and dietary fibre to work out the carbohydrate content, as the sum of the proximates should equal 100g. This is often used as it is a cheaper alternative but contains more error as the analytical error for each contributing nutrient is included.
Available carbohydrate by mono-saccharide equivalents which is the sum of the hydrous form of sugars, starch and glycogen. This is used only in the UK tables and results in higher values compared with the other 2 methods.
Metabolisable energy is calculated from the energy yielding components, fat, carbohydrate protein, fibre and alcohol (and sometimes also organic acids and polyols) using the sum of proximates methods.
The calculation uses Atwater factors which estimate the energy derived from 1g of each of the energy yielding components at 17 kJ/g (4.0 kcal/g) for protein, 37 kJ/g (9.0 kcal/g) for fat, 17 kJ/g (4.0 kcal/g) for available carbohydrates, 8KJ/g (2.0kcal/g) for fibre and 29 kJ/g (7.0 kcal/g) for alcohol. So:
Energy (in kJ) = 17*(XPROT) + 37*(XFAT) + 17*(XCHOAVL) + 8*(XFIBT) + 29*(XALC)
Vitamins are made up of several components each with vitamin like activity. For some vitamins the components have the same activity and so they can simply be summed to give the nutrient values for that vitamin of the given food. These include:
For other vitamins the components have different activity and are expressed as equivalents. These include Vitamin A, Vitamin B3 (Niacin), Vitamin B 9 (Folate), Vitamin D when cholecalciferol is included (This is high in pork and therefore tends to be included in countries where pork, ham and bacon are commonly consumed e.g. UK and Denmark) and Vitamin E. Unfortunately, there is variation between countries in how these equivalents are calculated, for example Vitamin A can be expressed as retinol equivalents or retinol activity equivalents which can be calculated in the following ways.
If combining food composition data from several tables which use different methods of expression or calculation the values will need to be converted. See the FAO/INFOODS document FAO/INFOODS Guidelines for Food Matching Annex 2 for further details on the different methods of calculation of these nutrients and their Tagnames. Details of how to convert the units are given here https://www.fao.org/3/i3089e/i3089e.pdf.
Tagnames
To correctly identify a food component, you need to know the component name, the definition, the expression in units, the denominator and also how the content was calculated or analysed.
In order to facilitate comparison across international datasets FAO/INFOODS have developed a series of component identifiers called Tagnames which enable the component definition (e.g. dietary fibre) and the analysis method (e.g. Kjeldahl method) to be easily identified. Tagnames are unique codes comprised of upper case letters, numbers and “_” or “-“. Tagnames can also be used to identify conversion factors, these will have an “X” at the beginning of the code. Where the analysis method is not specified in the FCT the Tagname will end with a hyphen. For example, FIB- represents a Fibre value where the analysis method is unknown.
If two FCTs contain a nutrient value with the same Tagname you can be confident that these values are equivalent and they can be combined. If the Tagnames are different they cannot be directly combined. More detail on Tagnames is given on the FAO website. These should be used in any FCT compiled in developing Intake24.
Calculated from the energy producing macronutrients
Calculated as the sum of the proximates
ENERC
Carbohydrate
Available carbohydrate by weight
Sum of measured sugars and starch (HPLC & enzymatic hydrolysis)
*ideal but expensive
CHOAVL
Available carbohydrate by difference
100 – sum of the remaining proximates
*recommended – not as accurate but cheaper
CHOAVLDF
Available carbohydrate as mono-saccharide equivalents
Sum of the hydrous form of sugars, starch and glycogen
*only used in UK gives higher values
CHOAVLDFM
Sugars
Total sugars – sum of free sugars
Specific enzyme method or HPLC
SUGAR
Starch
Total starch
Enzymatic hydrolysis and glucose-specific method
STARCH
Fibre
Total dietary Fibre
AOAC method (Prosky et al.)
*recommended
FIBTG
Non-Starch Polysaccharides
Englyst method
PSACNS/NSP
Fat
Total Fat
Mixed solvent extraction
FAT
Saturated fat
N/A
GLC
FASAT
Monounsaturated fatty acids
N/A
GLC
FAMS
Polyunsaturated fatty acids
N/A
GLC
FAPU
Trans fats
N/A
GLC
FATRN
n-3 polyunsaturated fatty acids
N/A
GLC
FAPUN3
n-6 polyunsaturated fatty acids
N/A
GLC
FAPUN6
Protein
Calculated from Nitrogen content
Kjeldahl or Dumas
PROTCNT
Calculated from Amino acid content
Ion-exchange chromatography
PROA
Alcohol
N/A
Distillation, GLC or Specific Enzyme method
ALC
Vitamin A
Retinol activity equivalents
Calculated from retinol and carotenoids measured by HPLC
(µg retinol + 1/12 µg β-carotene equivalents)
VITA_RAE
Retinol equivalents
Calculated from retinol and carotenoids measured by HPLC
(µg retinol + 1/ 6 µg β-carotene equivalents)
VITA
Vitamin B3 (Niacin)
Preformed
HPLC/Mircobiological
NIA
Niacin equivalents
Calculated from preformed niacin and niacin equivalents from tryptophan measured by HPLC/Mircobiological analysis
(Niacin equivalents from tryptophan: 1/60 x tryptophan)
NIAEQ
Vitamin B9 (Folate)
Total folate
*Recommended
Microbiological analysis
FOL
Dietary folate equivalents
*Used by Institute of Medicine
Calculated from food folate + 1.7 x synthetic folic acid.
Measured by Microbiological analysis
FOLDFE
Folate from food
Microbiological analysis
FOLFD
Folic acid (fortification)
Microbiological analysis
FOLAC
Vitamin C
HPLC
VITC
Vitamin D
Ergocalciferol and cholecalciferol
*Most common
Calculated as the sum of ergocalciferol and cholecalciferol (D2+D3) measured by HPLC
VITD
Ergocalciferol, cholecalciferol and 25-hydroxycholecalciferol
*Used in UK & Denmark
Calculated value - Vitamin D2 + vitamin D3 + 5 x 25-hydroxycholecalciferol measured by HPLC
VITDEQ
Vitamin E
α-tocopherol equivalents
Calculated value based on tocopherols and tocotrienols measured by HPLC
VITE
α-tocopherol
HPLC
TOCPHA
Calcium
N/A
Flame photometry
CA
Iron
N/A
AAS with electrothermal furnace
FA
Zinc
N/A
AAS with electrothermal furnace
ZN
Sodium
N/A
Flame photometry
NA
Missing Nutrient values
Many food composition tables will be incomplete for some of the less frequently analysed micronutrients e.g Vitamin K. Where this is the case missing nutrient values can be imputed, presumed or borrowed. There should be no missing values for the key nutrients in Intake24.
Missing nutrient values can be filled from other food composition tables, published literature, unpublished laboratory reports or from food labels. Focus attention on foods which are frequently consumed or that make a significant contribution to a nutrient of interest. For recipe calculations missing values should be added prioritising the major ingredients.
Imputed values are derived from analytical data for a similar food, for example the calcium value for green cabbage could be used for the calcium value for kale. Borrowed values are taken from data on the same food from a different data source. Presumed values are based on general knowledge for example the alcohol content of a biscuit can be presumed to be zero.
Plant foods can be presumed zero for the following nutrients; retinol, Vitamin B12 (except fermented foods and mushrooms), cholesterol, haem iron, alcohol (unless fermented), Vitamin D3, Vitamin K2 and folic acid.
Non- processed animal foods can be presumed zero in alcohol, dietary fibre (except for insects), carbohydrate (except for offal), starch, Vitamin D2, Vitamin K2 and folic acid.
Oils can be presumed zero for Vitamin C.
Eggs and milk can be presumed zero in dietary fibre.
Note - When using data from other FCTs it is important to check that the values are given per 100g of edible portion. If 100g of dry food is given we need the moisture content to be able to calculate the values in the prepared food, check that amino-acids aren’t expressed as per g protein or Nitrogen and that fatty acids aren’t given per 100g fat.
Assessing the quality of a food composition database
Before using data from a food composition database, it is important to assess the data to make sure that it is of sufficient quality. Food composition databases vary not only in their comprehensiveness but also in the quality of the value documentation and data. Values within food composition tables may be derived from; chemical analysis of the specific food in question sampled from a number of varieties available in the local population which have been selected based on the amount consumed (ideal), or, may be based on values for the same food from a neighbouring or close country-specific food composition table, imputed based on the values for a similar food, calculated (e.g. values for the food as consumed calculated from the value for the raw food) or presumed (e.g. where general knowledge is that the value is zero). A good quality food composition table will have documentation making it clear how each value was derived. FAO have published the FAO/INFOODS Evaluation framework to assess the quality of published food composition tables and databases to help users assess the quality FCT/FCDBs for their work.
In assessing the quality of the analytical data in an FCT the following are important to take into consideration:
Are the foods representative of the foods consumed in the country? High quality data on foods collected from a different country may not be appropriate to use if the variety/growing conditions/brand of the food item differ between the 2 locations.
How old is the data? If the data is more than 5 years old it may not be representative of the foods on the market now, especially for packaged foods.
Is the coverage of nutrients comprehensive?
Are the nutrients clearly defined? e.g. “Available Carbohydrate by difference” not “Carbohydrate”. Are the units, denominators and analytical method given?
Is the food description unambiguous? E.g. “Chicken, breast, roasted without skin” not “Chicken”. It should include the cooking method, preservation methods, colour, form, origin, part of the animal or plant, whether it includes any inedible portion and whether the product has been fortified.
Are the data sources well documented?
Is the analytical method clearly stated and is it appropriate?
A well-documented FCT should contain all of this information.
Ideally you would also want to see a sampling plan including:
How many samples were analysed(preferably more than 5 for each food item).
How many analysis runs were conducted (Should be 2 or 3).
How the samples were transported and handled.
Whether samples of fresh produce been taken during different seasons and from different regions and/or including different varieties
If different brands were sampled for packaged food items to reflect the market share.
What quality control measures were taken.
In assessing the quality of calculated values for cooked foods and recipes it is important to check:
That the full recipe name, ingredients and their weights and preparation steps are documented
That there is a clear description of the nutrient values for the ingredients including definition, units and denominator
That the calculation method and source of any yield and retention factors is well documented and appropriate
The source of the recipe and nutrient data are clearly given.
For borrowed or imputed data there should be:
Information on the comparability of the foods
Details of whether the components have the same definition, analytical method, units and denominators
Details of whether or not fortification has been included
Details of whether correction factors have been applied where water, protein or fat content are different between the source food and the local food
Food composition tables vary in their coverage of nutrients. All food composition tables should include the following key components:
Water – this is useful for identifying whether nutrient analysis is of a food in its dried form or as consumed, if the description of the food is unclear.
Ash (the residue left after the sample is burned) may also be given this along with carbohydrate, protein, fat, alcohol and water are called the proximates. These are the macromolecules within food and their sum should be approximately 100.
Components and conversion factors used in the calculation of other nutrients e.g. amino acids or nitrogen used in the calculation of protein content
Edible coefficient – this is needed for a good description, for transforming total weight into the edible portion and to aid food matching.
Energy, Protein, Fat, Available carbohydrate, Dietary fibre, Iron, Zinc, Calcium, Vitamin A, Vitamin B1, Vitamin B2, Vitamin B12, Folate and Vitamin C.
Recipe calculation – Generating new food composition data
Note - Recipe calculations are time consuming and should only be conducted if a food or dish is frequently consumed.
It is strongly recommended that lesson 5.2 of the FAO e-learning course on Food Composition Data, which covers recipe calculations, is completed prior to conducting any recipe calculations in order to understand the basic concepts and principles.
Recipe calculations can be used to generate information for cooked foods where only the values for the raw food are known and to determine the nutrient content of a prepared dish from the individual ingredients. For example, it can be used to calculate the nutrient content of a food as consumed from the values for the dry content of the food if the change in weight on cooking is known. e.g. cooked pasta from values for raw pasta or the nutrient content of a dried fruit from the fresh values if water loss is known. Recipe calculations can also be used to calculate the average of several food codes, for example if the average of a food from several neighbouring countries is required for a new Intake24 locale. The main use is for calculating the nutrient values of mixed dishes from the individual ingredients.
To conduct a recipe calculation, you will need a recipe which gives full details of the ingredients including the weights, the cooking method and the yield factor (final weight of the dish divided by the weight of the raw ingredients) and, if applicable, fat absorption (this can be prepared and weighed or published yield factors can be used (e.g Bergstrom, 1999, Bognar, 2002)). You will also need the nutrient composition of each of the ingredients and the nutrient retention factors (see below). If the recipe gives the amount in household measures these will need to be converted to grams.
The yield factor is applied at the recipe level, evenly across all ingredients, although in reality some ingredients will lose or gain more than others. The nutrient retention factors are applied to each individual ingredient.
Retention factor
Nutrient retention factors account for the loss of nutrients due to cooking. The retention factor is specific to that nutrient for the given food type and cooking method.
For example, the retention factor for folate in roast meat is 0.7 meaning 30% of the folate content is lost on cooking.
Yield factors are the ratio between the weight of the final cooked dish and the weight of all of the ingredients combined. The differences in weight may be due to water loss or gain e.g. a baked potato will lose water whereas rice or pasta will gain water, or fat loss or gain, e.g. a grilled burger will lose fat and water whereas fried chips will gain fat.
The yield factor = prepared weight of the whole dish
combined weight of all ingredients
e.g. for a baked potato the yield factor is 0.81 as 19% of the water content is lost due to cooking. Therefore 100g of raw potato will result in 81 g of cooked potato, on average.
For macaroni the yield factor is 2.7 so 100g of dry macaroni will result in 270g cooked macaroni.
If a recipe contains dry ingredients such as rice or beans but the liquid is absorbed from another ingredient such as coconut milk or tinned tomatoes then there is no need to include water in the calculations.
Water uptake in g/100g = total quantity of ingredients - weight of the prepared dish edible portion of the dish without cooking water or fat including waste
If total ingredients is greater than the weight of the dish then water loss occurs.
Water content can also be calculated using the following formula:
Water per 100g edible portion = 100 – Σ Protein, Fat, CHO, Dietary Fibre + ASH in edible portion
Fat absorption
Fat uptake can either be analysed or borrowed (e.g. Bognar, 2002). The Bognar report includes values for fat uptake (from page 44 onwards) in g per 100g of initial food.
Where foods are fried, the fat uptake should be accounted for by including the amount of fat absorbed as an ingredient. For items which are shallow fried it is assumed that all of the fat is absorbed. For foods which are deep fried work out % fat uptake and then add as an ingredient. For example the fat uptake for a fried egg is given as 5g per 100g. If the egg weighs 60g then 3g of cooking oil should be added to the recipe (5% of 60g).
If no published values for fat uptake exist then the difference in fat content between raw and fried or grilled and fried values for a similar food could be used. For example in calculating the fat uptake of fried mango, the difference in fat content between fried pineapple and raw pineapple could be used.
For foods inherently high in fat e.g meat and fish the difference in fat content between fried and grilled versions of a similar food can be used.
The National Diet and Nutrition Survey standard recipes database is also a very useful resource in helping determine fat uptake for different prepared dishes. For example here you can see that for fried egg in vegetable oil the assumption is that 1.2% of the prepared food will be the absorbed cooking oil.
Fat loss
The FAO has some published values for fat and water loss from meat and poultry.
Where published data don’t exist the difference between raw and grilled values for a similar food can be used.
Steps to recipe calculation:
Select the recipe and the name of the dish.
Be as descriptive as possible so not “lasagne” but “beef lasagne” or “vegetable lasagne”. Where a recipe varies depending on the region you may want to have one entry which represents an average or the most common version or you could include several versions of the same dish with the region in the description.
Collect information on the ingredients including the amount, the edible portion, if applicable (e.g. if the weight of the ingredient includes the inedible portion such as peel or bones), and the cooking method.
Appropriate sources include recipe books, the food industry and local experts (chefs and nutritionists/ dietitians) – Make sure to document the source.
Check that water and fat used in cooking are included. For fried foods you will need to include the amount absorbed which will need to be estimated or measured.
Collect the nutrient profile for each of the ingredients
Carry out the quality checks including making sure all the nutrient data are complete, analysis method, component definition, units and denominator and dealing with any missing values.
Determine the yield factor
If published include the data source, or prepare the dish following the recipe and measure the weight of the final dish. The yield factor is the weight of the final dish divided by the total weight of the ingredients.
You will need to find the retention factors for each ingredient.
The values are food and cooking method specific. Where the exact food is not available it is appropriate to use the food group value e.g. a value for fresh vegetables could be used for carrot.
Calculate the nutrient values of the recipe following one of the 3 recipe calculation methods:
The Raw ingredient method - this method should be used only for foods which aren’t cooked such as fruit salads, salads and sandwiches. The nutrient values for each ingredient (for the weight included in the recipe) can be simply summed and then divided by the weight of the final dish to give a per 100g value. For these foods no yield or retention factors need to be applied.
The Single ingredient method – this is used for the calculation of the nutrient content of cooked foods based on the values for the raw food, taking into account the yield factor and the nutrient retention factors. Details of the steps are given below.
The Mixed method recipe calculation – this is used for multi-ingredient cooked dishes. Here the retention factors are applied at the ingredient level and the yield factor is applied to the whole dish. Details of the steps are given below
Finally, it is important that the process including the calculation method used, the yield and retention factors are documented.
Full details of how to conduct the single ingredient method recipe calculation and the mixed method recipe calculation are given here as it is useful to understand the process. There is however a spreadsheet template (provided by Fernande Grande and explained in these videos) which automatically performs the calculations. Instructions on using the spreadsheet are given later in this document.
Raw ingredient method
The raw ingredient method is used for mixed dishes which are not cooked, for example, a fruit salad or mixed green salad. For this simply calculate the nutrient value per ingredient:
Nutrient value in ingredient per 100g * weight used in the recipe = NV in dish from Ingredient 1 100
Then calculate the total for that nutrient per dish by summing the amount from each nutrient and divide by the final weight of the dish and multiply by 100 to calculate the nutrient values per 100g:
NV in dish from ingredient 1 + NV in dish from ingredient 2……etc * 100 Total weight of all ingredients
Single ingredient method
Take the nutrient values of the raw food from the reference food composition data
Calculate the cooked weight from the raw weight using the yield factor
Weight of cooked food = Weight of raw food * yield factor
Note – If no published yield factor exists you could prepare the dish and calculate the yield factor by dividing the weight of cooked food by weight of the raw ingredients.
For each nutrient calculate the content of that nutrient in the cooked food
Nutrient values in the prepared food = Nutrient value in the raw food * Retention factor
Note – this gives the value for the portion not per 100g, so if a food has a yield factor of 0.75 then the nutrient values will be the amount in 75g of the prepared food.
Convert the value to per 100g of the cooked food
Repeat for each nutrient (except water) to be included in the final data set
Calculate the water content of the dish per 100g
Water in the raw ingredients – weight loss on cooking * 100
Total weight of the cooked dish
Where weight loss on cooking is the weight of the raw ingredients minus the weight of the cooked dish
Document and enter the nutrient values of the cooked food into the food composition dataset.
Example
e.g. for 50g of dried noodles
100g of dried noodles contains 72.6g of Carbohydrate (therefore 50g contains 36.3g)
The yield factor is 2.33 therefore weight of cooked food = 50g * 2.33 = 117g
No nutrient retention factor is applied for carbohydrate
Nutrient value in 100 g prepared food = 36.3 *100 = 31g 117
Convert the nutrients from per 100g to the per recipe amount i.e. the weight of that individual ingredient.
NV per 100 g × weight of ingredient in dish = NV per weight of ingredient in the dish 100
Calculate the cooked weight of the recipe from the sum of ingredient’s raw weights multiplied by the yield factor.
Weight of cooked dish = ∑ weights of raw ingredients × YF
Use the retention factor to calculate the nutrient values in the dish for each ingredient using the raw ingredient weight.
NV per weight of ingredient in dish (RF corrected) = NV per weight of ingredient in dish × RF
Sum the retention factor adjusted nutrient values for all of the ingredients and multiply them by the yield factor in order to get the nutrient values in the total cooked dish.
NV per weight cooked dish = ∑ NV of ingredients adjusted to nutrient loss
For each nutrient calculate the nutrient value per 100g of the recipe.
NV (100 g cooked dish) = NV (total weight of cooked dish) × 100 total weight of cooked dish
Calculate the water content of the dish per 100g
Water in the raw ingredients – weight loss on cooking * 100
Total weight of the cooked dish
Where weight loss on cooking is the weight of the raw ingredients minus the weight of the cooked dish
Document and enter the nutrient values of the cooked food into the reference dataset.
The documentation should include; a clear description of all ingredients including the weight, the yield factor and whether measured or published data were used (in which case include the source), a description and cooking method (including oil or water added to the recipe) and the retention factors used and their source.
Note – It is strongly recommended that you view the excellent videos on the Food Composition Explained YouTube channel which cover recipe calculations and the use of this spreadsheet.
Description of the recipe calculation spreadsheet
The first 3 sheets of the recipe calculation spreadsheet contain details on the format of the spreadsheet and instructions on it’s correct usage.
Sheet 1 contains details of the copyright information along with a summary of the data contained within each of the following sheets.
Sheet 2 Gives an overview of all components included in the template along with the component description, unit and denominators, Tagnames and where appropriate details of the calculation method used.
Sheet 3 gives detailed instructions on the stages involved in the recipe calculation process.
Sheet 4 is where the nutrient composition of the individual ingredients included in the recipe are to be entered.
Sheet 5 is the recipe calculation sheet which contains the formulas for deriving the nutrient values of the cooked dish from the individual ingredients, the yield and the retention factors.
Conducting recipe calculations using the spreadsheet
Prior to conducting the recipe calculation you should have gathered information on each recipe including the cooking method and all ingredients along with the nutrient profile of each ingredient. The nutrient data should be checked to ensure standardisation of the components, units and denominators and that the Tagnames match the column headings in sheet 4 of the spreadsheet.
Step 1
Copy and paste the nutrient profile for each ingredient into sheet 4 “Nutrient Profile” of the spreadsheet. It is important that you DO NOT change the order or the number of the columns otherwise the calculations in the recipe sheet will not work!
Step 2
In sheet 5 “Recipe Calculation” Copy rows 4-79 and paste in below, leaving the top section blank so the blank template can be reused for the next recipe. The cells highlighted in red indicate where data must be entered.
Step 3
Add in the recipe name and ingredients including their state (raw or cooked) as they are used in the recipe.
Step 4
Scroll to the right and add in the food code and food description for each of the ingredients, matching to the food code and food description as pasted into sheet 4. This will bring in the nutrient profile of each ingredient into columns Q to BM in sheet 5 to be used in the calculation.
Step 5
The nutrient profile section is now complete. Check that all the values have been correctly brought in and that there are no missing values. Delete unused rows down to the next heading row “nutrient retention factors” and then delete the rows containing “#REF!” from the nutrient retention factors and recipe data sections below.
Step 6
Select the nutrient retention factors for each ingredient. In the nutrient retention factors section scroll to the right until you see rows for cooking type, food group and food/cooking method. Select the cooking type.
The cooking method can be dry heat (e.g. oven), moist heat (e.g. steamed), cooked with fat (e.g. fried), RF_not_available (if no RF exists for the food group and cooking method) or RF_not_required (if that individual ingredient is not cooked). In the latter two cases the RF will be assumed to be 1 i.e. no change in nutrient value. If you are unsure about whether a RF exists for a particular ingredient or cooking method you can check in sheet 6 of the spreadsheet.
Next select the food group for each ingredient
(EUROFIR recommends that for sugar the retention factor for fruit and fruit products is used)
Finally select the food/cooking method from the drop-down list
The nutrient retention factors section is now complete. Scroll to the right and check that the nutrient retention factors have been filled for all of the minerals and vitamins.
Step 7
Now enter the mandatory data (red fields) in the recipe data section.
Add the description of the preparation method in column B (it is important to check that the appropriate nutrient retention factors were selected), add the weight of each ingredient as used in the recipe in column J this will automatically populate column K. If you add a weight including inedible parts e.g. bones on meat into column H the edible portion coefficient will automatically be calculated in column L. The spreadsheet will automatically sum the weight of the raw ingredients and if you add in the weight of the cooked dish the spreadsheet will calculate the yield factor.
The recipe calculation is now complete. In the nutrient columns you will have the nutrient content of each ingredient, this is the amount of nutrient in the weight of the ingredient included in the recipe rather than per 100g. In the nutrient content per 100g row you will have the values for each nutrient in 100g of the mixed dish. The energy values will be blank for the ingredients as the energy is calculated from the proximates using the mixed dish data. Where there is missing data in the nutrient profile of the ingredients you will need to delete any values in the nutrient content of the recipe generated for those nutrients. The water content of the cooked dish is automatically adjusted for the weight change on cooking based on the assumption that all of the weight change was due to water loss/gain.
For regions where little or no food composition data exist it will be necessary to compile a new food composition database. As ever the amount of time and resources available will dictate the extent to which this can be done. The following guidelines represent the ideal process which should result in a food composition database of sufficient quality that it could be published and used by other researchers in the country, independent of Intake24. As with the food list it is recommended that priority is given to commonly consumed foods and foods which are of important sources of key nutrients, in terms of the time and effort spent on compiling the nutrient composition data. It may be that this thorough process can be followed only for frequently consumed items with those less frequently consumed foods being flagged as lower quality within the database for future refinement.
It is strongly recommended that you have completed the FAO e-learning course on food composition data and consult the FAO/INFOODS study guide for further reading before starting this process. The following notes represent a summary of the course content.
All of the foods in the newly created food list for the region will need to be linked with food composition data. The full range of nutrients, definitions and units required for Intake24 are given in the Intake24 Nutrients spreadsheet. The core nutrients have been flagged in red and these should be included at a minimum.
FAO data compilation tool
FAO have developed a compilation tool which can be used to facilitate and guide this process. It is recommended that the user guidelines are consulted before starting this process. The compilation tool is an excel spreadsheet with the following worksheets:
Codes This worksheet documents the various codes used within the spreadsheet including progress codes which show how far through the compilation process each individual food record is, food type codes which indicate whether they are recipes or individual foods, codes to indicate whether a value has been calculated, estimated, presumed zero or analysed, the source of the data, the units and denominators used, quality control measures and confidence in the final value. This should be used to document both the process and the values as comprehensively as possible.
Archival database This is where the original data gathered from other food composition tables and scientific reports are entered. They should be standardised in terms of the units and denominators but should otherwise remain unchanged.
Reference database This is where any missing values are completed, average codes are calculated, and the sum of proximates is calculated. Only food records flagged as final should be copied into the user database.
Recipe and ingredients and Recipe calculation Ignore these 2 sheets and instead use the Template for recipe calculation. The data generated in the recipe calculation template can be copied into the user database. The completed recipe template should be stored along with the compilation tool in order that data can be readily checked.
User database this is where the final foods and nutrient values to be published in the new food composition database are to be pasted.
Component this sheet lists all of the components, their Tagnames, definition and units, along with the maximum number of decimal points.
Bibliography this sheet isused to document the references for the data used in the compilation tool.
Value documentation this sheetcontains detailed information on the source of the data on foods and their nutrient values in the reference database including whether data was from original analytical data, aggregation of analytical values, from food tables or food labels. Whether values were imputed, borrowed, estimated or presumed zero. Whether values were based on the same food in a different state (cooked versus raw) or a similar food.
Sampling contains detailed information on the sampling procedure used for data in the reference database which were derived from chemical analysis.
Methods is for recording the analytical methods used for foods where chemical analysis was completed.
Data compilation process
Step 1
Start by grouping the foods on the food list you developed into food groups and allocating each food a unique food code. Decide on a coding system and allocate a unique code to each food entry, the FAO recommends that the food code is a 5-digit code where the first 2 digits indicate the food group and the next 3 the specific food type.
Step 2
Review the existing nutrient composition data for each food. Ideally this would be analytical data based on foods from the national food supply however this is very expensive. Collect published nutrient composition data for local produce and dishes such as published lab reports, scientific articles, manufacturers information and food composition databases. For mixed dishes collect recipes, including ingredient weights and detailed description, bear in mind that cooking practices may vary by region in which case it may be necessary to list the mixed dish more than once in the food list, clearly marked with which region the recipe comes from. For foods which are frequently consumed and no/poor quality nutrient composition data exists chemical analysis is recommended if resources allow.
Step 3
Enter data into the archive database worksheet. This is where all data collected from sources of sufficient quality are stored. Quality of data can be assessed following EuroFIR Quality Index Guidelines which can be used to score analytical data (from scientific literature, lab reports) in 7 categories (e.g. food description, component identification, sample plan, sample handling, lab performance, etc). It may include multiple sources for each food, for example data on the nutrient values of chicken may be collected from several neighbouring countries FCTs with a view to using the average value. Data will need to be standardised in terms of the units and denominators before being entered into the spreadsheet. Where appropriate document whether the data is for the edible portion or if it includes the weight of peel/core or bones.
Enter the original data and document the units, factors, calculations, sampling plan, number of samples, analysis methods and any quality control, if this information is given.
Enter the food code and food group for each food.
Document the data source (reference/ISBN).
Indicate whether the item is a raw food, cooked food or mixed dish.
Document your progress through the journey using column C (e.g copied in, all details and references added, units and denominators checked, component and analysis methods checked).
Complete column J with a match code for borrowed data indicating the quality of the match (e.g. candlenut raw weighed with shell matched with candlenut raw with shell from another FCT would be an exact match, candlenut matched with macadamia nut would be a similar food match and candlenut matched with cashew nuts would be a rough match). Remember – If copying data from another FCT it is important to check that the food, component, unit and denominator are the same and to consider the fortification policy in the country of the source FCT.
When composition data are entered it is important to document the source, complete the quality index, whether it is an exact or similar match and to include the food code and food name from the source. Record number is used when there are several sources of data for one food, e.g. when the average of the food in several neighbouring FCTs is to be used. There is also a priority class column which can be used to prioritise the effort put into different foods, e.g. a frequently consumed food would be high priority. The quality index is used to assess the confidence in the data source.
Step 4
Copy data into the reference database. This is where any missing values are completed, average codes (see lines highlighted yellow) are created from several similar food composition codes, the sum of proximates is calculated and conversion factors are applied to calculate standardised components (highlighted yellow) for those variables for which several different Tagnames exist.
Copy all final data from the Archive database.
Compare the data and select the values which are of acceptable quality to be included, aggregated (if an average of several samples is to be used) and standardised.
Calculate the values which are calculated based on other components; Energy, Nitrogen, Carbohydrate by difference and the Vitamin equivalents. (See https://www.fao.org/3/ap810e/ap810e.pdf Annex 3 for details of energy conversion factors and nitrogen to protein conversion factors).
Fill in any missing values and record whether they are estimated, borrowed, imputed or calculated and document the data source. Complete column J with a match code for borrowed data indicating the quality of the match (e.g. candlenut raw weighed with shell matched with candlenut raw with shell from another FCT would be an exact match, candlenut matched with macadamia nut would be a similar food match and candlenut matched with cashew nuts would be a rough match). Remember – If copying data from another FCT it is important to check that the food, component, unit and denominator are the same and to consider the fortification policy in the country of the source FCT.
Aggregate data where the mean of several foods or foods from several FCTs is to be used (It can be useful to check for any outliers prior to deciding on the final list of foods to include).
Re-check that the correct component definitions, units and denominators are used. Within the recipe template evaluate the quality of data on any ingredients on which mixed dishes are based.
Select the data to be included in the final FCT (including the nutrient data for mixed dishes generated using the recipe template.
It is important to document the source of the data, complete the quality index, indicate whether it is an exact or similar match and to include the food code and food name from the source. Record number is used when there are several sources of data for one food, e.g. when the average of the food in several neighbouring FCTs is to be used. There is also a priority class column which can be used to prioritise the effort put into different foods, e.g. a frequently consumed food would be high priority. The quality index is used to assess the confidence in the data source.
Nomenclature here is really important, give as much detail as is available on the food item. Any characteristics which are likely to impact on the nutrient content should be included e.g. for milk what is the animal source, the processing type, is it dried or fresh, what is the fat content, has it been fortified? Variables should include the food name in the local language, the food name in English (which should not be ambiguous e.g. chips could be fried potato slices served warm, or thinly sliced potato deep fried or dried and sold cold as a packaged food item depending on whether the UK or US definition is used). A food description should be used to give further details such as the colour or ripeness of fruit, the part of the animal for meat and whether skin is included, whether a grain is cooked or raw, the cooking method of cooked items, preservation method if appropriate e.g. canned in brine, frozen or pickled, the origin (country/region), level of fortification and if it is the edible portion only or includes inedible parts such as peel or bones.
Step 5
Create the User-database – This is the final data which will be included in the FCT for users e.g. for foods where the nutrient values are calculated using an average of data from several databases only this average value will be included. Only food records flagged as final should be copied into the user database. It is useful to keep the raw data as is so that it can be referred back to if any of the values need to be checked.
Liquids should be expressed as per 100g and the density should be included in the data.
The user database should be accompanied by documentation which details the methodology and decisions made thoroughly. For each component (nutrient value) the description should include the analytical method (where known) or the calculation along conversion factors (where applied), unit and denominator. If appropriate sample handling and the number of samples analysed should also be given.
Include an introduction with publication year and version number, an index and a section on the foods and components selected and make sure there is standardisation of the number of significant figures and decimal places. Use the FAO/INFOODS Guidelines for Checking Food Composition Data prior to Publication of a User Table/Database and consider asking experienced food composition compilers to check your FCT before publishing.
Chemical analysis
If chemical analysis is to be conducted it is important to create a sampling plan. This will include the number of samples to be collected for each food (this should include the major brands/varieties consumed), how the samples will be stored and transported, which nutrients will be analysed and using which methods and how many analysis runs there will be per sample.
Sampling of foods
Chemical analysis is costly and therefore foods should be prioritised based on the frequency and amount of consumption along with their relative contribution to specific nutrients of interest. As mixed dishes can be calculated from recipe information it is advisable to concentrate efforts on compositional analysis of ingredients. Although chemical analysis is more accurate than recipe calculations the intra and inter individual variation in the preparation of mixed dishes is much greater and therefore due to the high cost of compositional analysis recipe calculation is often sufficient. Samples should be collected to reflect the average content of the food item in the local food supply. For example, if creating a code for a shop bought carrot cake it is important to look at the market share in order to inform the decision on which brands to include. For fresh produce it is important to document the variety, the season and the region from which it was sampled, for some fruits and vegetables the colour will also make a difference. All edible parts of the food should be included in the sample in the correct proportions.
A detailed sampling plan should be developed including the following information:
Which foods will be included and from which regions?
How many samples will be collected for each food type? This should include samples collected during different seasons and different regions for fresh produce, and a range of the popular brands for packaged items.
How will the samples be transported? This should ensure that there is minimal degradation of the nutrients e.g. it may be desirable to keep the samples in the dark or frozen.
How will the samples be processed prior to analysis? e.g. cut/ground/homogenised
Choosing a laboratory
A laboratory should:
Be accredited for the analysis method and the food stuff of interest
Use a certified reference material
Calibrate their instruments on a regular basis
Analyse samples in at least duplicate
Conduct cross checks with other laboratories
Perform recovery studies where they test the method with a substance of known composition
Be willing to provide detailed documentation on all analysis undertaken
Final quality checks of the complied food composition database
Once the user database is complete the data should be checked and cross-checked for completeness and accuracy, referring back to the reference and archive databases if required. Check that no frequently consumed foods contain missing values for any of the core nutrients and that no missing values have been entered as zero.
Check that the sum of the proximates is around 100 (97-103g) and that the energy content has been correctly calculated and the calculation has been documented. The comprehensiveness of the descriptions of foods, components and recipes should be checked. Where data have been combined checks should include making sure that the component definitions and analytical methods match and that foods which have significantly different fat or water contents haven’t been combined. Water values should be given for each food item and should be in accordance with the food description e.g. dried/raw/cooked. Make sure that appropriate conversion factors have been used in calculating the protein content from the nitrogen content. That the correct formula was applied in calculating the available carbohydrate content by difference and that total sugars is equal to or greater than added sugar and the sum of the individual mono- and di-saccharides.
For recipe items check that if all ingredients have a missing value for a nutrient the recipe nutrient content also indicates that the value is missing, it should not be zero.
Box-plots of nutrients by food and food group can be useful in identifying outliers which can then be checked to verify the data.
Range checks should be conducted to make sure that certain nutrients are within the expected range for a particular food group:
Acknowledgements: Huge thanks to Fernande Grande, Doris Rittenschober and Bridget Holmes from the FAO food an nutrition division who provided extensive reviews of this section resulting in a much improved document.
