Appendix 2. Development of a test database on food composition:
Main assumptions
Summary
This document contains the detailed assumptions that were made for the preparation of a European Test Database (EU TD). The EU TD is based on a selection of food composition databases. Making these assumptions (on product groups, missing values, and serving sizes) is an important step in making this EU TD suitable to be used for the criteria setting (nutrient profiling) for health Logo use. This was done for the International Scientific Committee of Choices.
Index:
I Selection of databases…………………………………………………………………..…..1
II The approach in words.….…………………………………………………………………2
III Appendix: detailed overview of assumptions for missing values (TFA, added sugar)……3
i. TFA…………..…………………………………………………………………………3
ii. Added sugar…………………………………………………………………………….9
I Selection of databases
The Choices International Scientific Committee has requested an evaluation of the preliminary Choices International Criteria using objective datasets. The most suitable sets for this exercise are internationally available Food Composition Databases (FCD). To be able to make a workable selection from the globally available FCDs we have defined several selection criteria. These are geographic representation, completeness of data, availability in the English language, electronically available & costs. All databases were assessed with these criteria. Table 1 shows the resulting selection for European use.
Table 1. Food composition databases included in the EU TD:a) NEVO 1 (Food Composition Database from The Netherlands)
b) NUBEL 2(Food Composition Database from Belgium)
c) Danish FCD 3 (Food Composition Database from Denmark)
d) McCance&Widdowson’s 4 (Food Composition Database from UK)
e) Polish FCD 5 (Food Composition Database from Poland)
f) EFSA Food Basket limited version 6 (EU Limited Food Basket by EFSA)
- TFA values are assumed in NUBEL, Polish FCD and EFSA Limited Food Basket
- Added sugar values are assumed for all the databases, except for the Danish FCD
- Sodium values in Soups in the Polish FCD are including the table salt (390mg sodium/100g5)
Some food items from the Danish FCD 3 and McCance&Widdowson’s 4 are already included in the EFSA Limited Food Basket 6 (EFSA FB), and are therefore double present in the EU TD. In case of Choices International Criteria evaluation with the complete EU TD, these double products will be taken out of the EFSA FB. In case of local Choices International Criteria evaluation (per database), the products are left in the database and thereby included in the analysis.
II The approach in words
Preparation of the EU TD for evaluation against proposed criteria implies deciding on the:
1. Food product group classification based on:
Products are classified according to the description of the product groups in the 2007 criteria document. If there is doubt on the right assignment of a product to a product group, the decision tree will be used (see Figure 1). This ensures, in a transparent way, a standardized and pragmatic classification of all foods according to defined prioritized criteria. The main principle of the decision tree is to minimize the number of product groups while maximally promoting healthy product innovations / reformulations.
2. Trans Fatty Acid content based on:
a) Total fat content in relation to the origin, in line with TRANSFAIR* Study 7
b) Availability of TFA data for comparable products in other FCDs
c) Recipe (ingredients, amounts & food processing)
(see appendix for a detailed overview of the assumptions)
3. Added sugar content based on:
a) (Assumed) intrinsic fruit and milk sugars subtracted from the total sugar content
b) Availability of added sugar data for comparable products in the Danish FCD 3
c) Recipes & existing food regulations for fruit based food items like jam 10 or juice 9 (ingredients and amounts)
(see appendix for a detailed overview of the assumptions)
4. Serving sizes:
Pragmatic application of existing serving sizes as indicated by various nutrition organisations (FNLI / Voedingscentrum / Choices / CIAA), while minimizing the variation of serving sizes within each product group (see Figure 2).
Used approach to assume the missing values in words**:
1. First priority in making assumptions for all foods was to use when possible the most reliable approach: to assume the missing values based on the amount and origin of the ingredients. The missing TFA values are therefore where possible assumed based on the amount and origin of total fat 7, while added sugars are assumed based on the amount and origin of natural sugars.
2. If these data were not available, imputed TFA and added sugar data based on assumptions are used assumed from the similar foods from another database. Imputation is the process of assuming that a nutrient value in one food can represent that in another similar food (from another database). Imputation also includes process of assuming that some foods contain none of a particular nutrient, based on knowledge of composition of that food. For this purpose different FCDs were used in a fixed preference order as a reference.
3. In case no reference of a similar food was found in another FCD, recipe*** based calculations are applied. Overall nutrient profile from that food is calculated from the nutrient data for the individual recipe ingredients, where possible taking into account, loss or gain of moisture and nutrients during processing. Assumptions have not been made for (processed) products with limited information on ingredients and/or processing.
* TRANSFAIR study is used as a golden standard for it is assumed to provide reliable and comparable data on trans fatty acid (TFA) content of foods in Europe 7
** A similar approach is used to estimate nutrient values in Food Composition Databases, as described in The Journal of Food Composition and Analysis 11
*** Recipes used are common used standard recipes as published in the Food Composition Databases and documented in an excel file used for estimated calculations for this purpose 12
III Appendix: A detailed overview of the assumptions made
Assumptions are made in order to prepare the EU TD (including FCD as listed in Table 1) for the evaluation of the proposed Choices International criteria.
i TFA content
Overall TFA content assumptions
While assuming TFA content, Product Groups are defined that by default contain 0g TFA:
- All products with 0g total fat.
- Fresh or fresh frozen fruit, vegetables and legumes [fruits and vegetables by default do not contain TFA]
- Fruit juices [fruits do not contain TFA]
- Potatoes (unprocessed) [potatoes do not contain TFA]
- Fresh or fresh frozen fish, shellfish and crustaceans [fish does not contain TFA]
- Beverages without dairy [these beverages do not contain TFA]
TFA content is calculated with the fixed total fat: TFA ratio from a similar food product from another food composition database and expressed as % total fat. Where other FCDs are used as a reference for similar food products, the database preference, based on data availability, is agreed upon and given in Table 2. Similar food products are identified by their name, ingredients, energy content and usage. The NEVO 2006 1, Danish FCD 3 and McCance& Widd 2002 4 contain TFA values and to assure the consistency of the data within a database, the missing TFA values in these databases were not assumed.
Table 2. The food composition database preference used as reference for TFA content of similar products.1st preference NEVO 2006 1
2nd preference Danish FCD 2005 3
3rd preference McCance & Widd 2002 4
4th preference USDA 8
Product Group specific TFA content assumptions
Oils and fats
1. Oils [plant fats] do not contain TFA
2. TFA content of various animal and hardened industrial fats is calculated as % total fat based on:
a. TRANSFAIR study 7 concerning fats like lard and butter
b. Data from similar food products taken from other FCD where available (see Table 2)
Milk and cheese products
3. TFA content of various dairy products is calculated as % total fat using following Transfair 7 based values:
a. TFA in milk products is 4.3% total fat
b. TFA in butter is 4.8% total fat
c. TFA in cheese products 4.5% total fat.
(Processed) Meat products
4. TFA content of various meats is calculated as % total fat using following Transfair 7 based values:
a. TFA in Veal and Beef meat products is 4.3% total fat
b. TFA in Lamb and Mutton meat products is 6.4% total fat
c. TFA in Pork meat products is 0.7% total fat
d. TFA in Chicken and Turkey meat products is 0.8% total fat
5. TFA content of various combined or breaded meat products is calculated as % total fat from data of similar food products taken from other FCD where available (see Table 2):
a. For example: TFA in Chicken nuggets6 is 12% total fat
6. Assumption is made that smoking, cooking, braising, frying, roasting and canning processes do not influence the TFA content relative to total fat
7. TFA content of plain or vegetarian eggs is assumed to be 0% total fat
8. TFA content of egg products with bacon is calculated assuming 50% of total fat is pork fat and the only TFA source
9. TFA content of sausages with undefined origin of meat is not assumed
Processed potatoes
10. TFA content of various processed potatoes is calculated as % total fat from the data from similar food products taken from other FCD where available (see Table 2)
Snacks
11. TFA content of Nuts and Seeds is 0% total fat
12. TFA content of various sweet and savoury snacks is calculated as an estimated average of % total fat from data for similar food products taken from other FCD where available (see Table 2)
a. TFA in chocolate (milk and dark) is 0.3% total fat, also in chocolate-based snacks, e.g. Chocolate wafel2, Rijstwafel, chocolate2,
b. TFA in Biscuits is 1.5% total fat
c. TFA ins Cakes and Pies is 5% total fat
d. TFA in Cookies is 12.2% total fat1
e. TFA in Corn crisps is 0.6% total fat
f. TFA in Graan repen is 4% total fat1
g. TFA in Wafels is 10% total fat1
h. TFA in Filled rolls (w3ith meat or dairy based filling) 5 is 4.3% total fat
i. TFA in plain rice crackers and rice wafels is 0% total fat1
j. TFA in other snacks e.g. Mars1 (7.5%), Nougat6 (0%), Toffee1 (8.3%), Marzipan1,3,4 (0%), Bounty1 (0%), Snickers1 (2.4%), Twix1 (8.6%), Popcorn (0%), Liquorice (0%), Amandel broodje1 (8.7%), Mueslireep1 (0.51%), Chips1 (0.6%), Tompoes1 (11.5%), Doughnuts1 (5%)
13. TFA content in other sweet snacks is only calculated based on dairy fat as a source
- TFA in dairy fat is 4.3% of total fat, e.g. Milkshakes, Cream Ice creams
Filled sandwiches / rolls
14. TFA content of filled sandwiches and rolls is calculated based on common sandwich recipes and TFA content of the ingredients (see Table 2)
Soups
15. TFA content of soups is estimated based on recipe calculations with TFA delivering ingredients and standard recipes. As there is a limited amount of soups included in the EU TD (see Figure 3), the Unilever soup database 13 (n= ~5000 soups) can be used as a reference for TFA content calculations expressed as % total fat
- TFA in vegetable, chicken and fish soups is 0g assumed there are no TFA delivering ingredients in vegetable soups (ruminant meat, dairy or partially hardened fats)
- TFA in soups with ruminant meat is estimated 0.1g per 100g soup assumed that there is a maximum of 30g meat in meat soups
- TFA in powdered soups is 1.56% SAFA (estimated average of TFA in UL DB13 powdered soups)
- TFA in cream soups 3.7% total fat or 0.1g TFA per 100g soup assumed all the fat in cream soups is dairy fat
Main dish
16. TFA content of various dishes is calculated as % total fat from the data from similar dishes taken from other FCD, where TFA and total fat values were available (see Table 2)
- Pizza with meat1 (2.7%), Pizza margherita1 (1.3%), Pizza with vegetables6 (0%), Quiche average1 (5.6%)
- TFA in Pizza’s without a similar reference food in other FCD is calculated with the average TFA value Pizza average1 (2%), e.g. Pizza, 4 cheese6
- TFA in pie with vegetables is 0.1g per 100g product 1
17. TFA in common Main dishes
a. content of common main dishes is calculated based on common recipes and TFA content of the ingredients (see Table 2)
- e.g. Gemengde salade, garnalen2 has no TFA as salad and shrimps are not a TFA source
- Gemengde salade,kip,ananas2 has 0.02g TFA, 30g portion size of chicken in the salad and no other TFA sources assumed
- Gemengde salade,ham, kaas, kip2 has 0.214g TFA, portion sizes assumed (30g portion size of chicken, 15g of ham and 15g of cheese in the salad) and no other TFA sources assumed (only chicken, ham and cheese)
- TFA in Paëlla2 is 5% total fat, assumed the TFA content is equal to Risotto6
18. TFA in Main dishes with Vegetables, Fish, Pasta or Rice (no meat, cheese, breaded fish or puffed pastry) is 0g, assumed that
a. the ingredients do not contain TFA (vegetables, fish, vegetable oils, pasta and rice)
b. food processing (canning, etc.) and preparation (braise, cook, steam, bake) has a minor influence on the TFA content
c. e.g. TFA in Ratatouille2 is 0% total fat, assumed vegetable and vegetables oils are the only ingredients which do not contain TFA and food processing did not influence the TFA content
d. e.g. Gemengde salade, garnalen2 has no TFA as salad and shrimps are not a TFA source
19. TFA in Main dishes with meat
a. TFA in main dish with meat of undefined origin is 3% total fat (average TFA value of various meats)
b. Main dish with meat of defined origin is % total fat of that meat according to the Transfair study assuming that meat is the only fat (and TFA) source, e.g. TFA in Rice with chicken is 0.8% total fat
c. TFA in Meat pies is 4.3% total fat (average TFA content meat pies4)
20. No TFA is assumed where no similar reference food or a recipe is available, resulting in a missing value
a. Vol-au-vent2, Dumplings5, Rice,egg-fried,takeaway6, Tartiflette6 are not found in other FCD