2nd IMEKOFOODS Promoting Objective and Measurable Food Quality & Safety October, 2nd -5th 2016 Benevento (Italy) FOOD COMPOSITION DATA QUALITY: ASSESSING VARIATION AND QUALITY OF DATA FROM DIFFERENT SOURCES Mark Roe 1, Jenny Plumb 1, Susanne Westenbrink 2, Luisa Oliveira 3, Isabel Castanheira 3, Marine Oseredczuk 4, Aida Turrini 5 and Paul Finglas 1,6 1 Institute of Food Research, Norwich, UK, [email protected] & [email protected] 2 National Institute for Public Health and the Environment (RIVM), Bilthoven, NL; [email protected] 3Coordenadora da Unidade de Observação e Vigilância, Departamento de Alimentação e Nutrição, Lisbon, PT, [email protected] & [email protected] 4French Agency for Food, Environmental and Occupational Health Safety (ANSES), Paris, FR; [email protected] 5Council for Agricultural Research and Economics - Research Center for Food and Nutrition, Rome, IT; [email protected] 6EuroFIR AISBL, Brussels, BE; [email protected] associated with the values. Generic values that are Abstract - Users of food composition data assume intended to represent foods at a population level that all data is good quality and fit for purpose but are published in most countries. However, even quality can vary and depend on a wide range of though they are intended to be representative of factors. Different types of data, e.g. nutrients and the foods described, individual samples may differ bioactive compounds and data derived from considerably because of natural or extrinsic different sources can differ in quality. The EuroFIR variations and also because of apparent differences system for assessing data quality allows sources of caused by analytical variations. Sources of natural variation to be evaluated. Evaluation of food variation in foods include season, growing description, component identification, sampling, conditions and post-harvest storage in plants and number of analytical samples, sample handling, feeding regime and age in animals. Nutrients that analytical method and analytical quality control are associated with structure and metabolic allows users to assess data quality. function tend to be less variable than nutrients that accumulate in particular parts of a plant or animal Keywords : Food composition, Data quality, or nutrients that are unstable. Variation may also be Bioactive components, Food labelling introduced during processing or food manufacturing, e.g. changes in recipes, variation in 1. INTRODUCTION amounts or types of ingredients. Food composition data quality has been a Food composition data is an essential feature of many European and International component of most investigations into links collaborative projects and formed the basis of the between diet and health. Data is used for a wide authoritative guide to the production, management range of purposes including: research, public health and use of food composition data [1]. Quality nutrition and education, policy development, trade, approaches have been implemented and further food product development and clinical nutrition. developed by EuroFIR (European Food Information Developments in information technology systems, Resource) (www.eurofir.org ) and INFOODS particularly the move towards web based (International Network of Food Data Systems) technology, have led to increasing access to food (http://www.fao.org/infoods/infoods/en/ ). The composition data. Recent changes to European EuroFIR quality framework [2] includes: Food Information Regulations mean that all pre- · Guidelines for the assessment of packaged food producers will need to provide methods of analysis; composition data for key nutrients and food · A flow chart describing the food composition data is increasingly available directly to composition compilation process consumers. · Development of a CEN standard ‘ Food Most users take published values at face value data – structure and interchange format and will not be aware of variation that may be (EN 16104:2012)’ [3]; 54 1 www.imekofoods.org 2nd IMEKOFOODS Promoting Objective and Measurable Food Quality & Safety October, 2nd -5th 2016 Benevento (Italy) · Thesauri for food description, 2. DATA QUALITY CRITERIA component identification, units, matrix unit, acquisition type, method type, The EuroFIR system for evaluation of quality of value type and reference; data from scientific publications [4], including · A recommended approach to calculation scientific literature and laboratory analytical reports of composition based on recipe includes six main criteria and was developed based ingredients; on existing European systems. It is also compatible · A system for evaluation of quality of with the well-established USDA quality system [5]. data from scientific literature or The importance of each quality criteria will depend laboratory analytical reports; on the type of data and what it is intended to be · Peer review of data compiler used for. The system allows users to assess data organisations based on an ISO 9001 quality based on common guidelines and a quality approach. score can be attributed to each quality criteria. An INFOODS also actively encourages improvement overall quality score can be calculated but scrutiny in quality of data by: providing international of each quality criteria allows the user to consider guidelines and standards (e.g. on food and the factors that are most important for their component nomenclature; data interchange; intended use. compilation; data checks; conversions; sampling; data quality), assisting countries to compile and 2.1. Food description publish food composition databases and tables and Correct identification of a food is vital to ensure by providing training. that food composition data is fit for purpose. The implementation of standards has enabled Identification is usually based only on the food publication of harmonised data in national food name provided but there are many factors that can composition databases and web technology has influence nutrient content of a food and they enabled publication of meta-data to support should be included in the food name. For example published values. However, many users are not variations may depend on cultivar, cut of meat, aware of quality considerations so will use cooking and processing factors including addition of published values without any regard for variability, ingredients, fortification, removal or reduction of limitations of use or source of data. Published data nutrients (e.g. fat, salt or sugar) and preservation is no longer limited to data from national methods. Food name should be unambiguous so composition datasets and scientific literature and that users can select the data that is most nutrient composition of branded products is appropriate for their need. routinely available via manufacturer or retailer Language can be a problem because data is websites. Data on bioactive components in foods is often translated into English. It is not always widely used for research purposes and is also possible to make a direct translation that is beginning to be made available to consumers accurate and cultural differences can lead to through a variety of media. different interpretations of meaning. Even within These different types of data generally have the the same language there will be instances where same quality issues as nutrient data but certain the same food name can describe different foods aspects, e.g. analytical methods used for analysis of e.g. prawn/shrimp, swede/turnip, or where the bioactive components, may have increased impact name is misleading, e.g. mincemeat does not on data quality. To enable users to make an contain meat and Bombay duck is a fish. informed decision as to the limitations of data, it is Facetted systems have been developed to essential to make available documentation that describe foods and to distinguish between different describes the sources and methods used to obtain foods that may not be fully described by the food data. This paper describes quality considerations name alone. The LanguaL (Langua alimentaria) related to different types of food composition data system was developed by the United States Food and different aspects of the data. and Drug Administration and has been modified and adopted for use in European countries as part of the EuroFIR initiative to better standardise approaches for food description in Europe. The 55 2 www.imekofoods.org 2nd IMEKOFOODS Promoting Objective and Measurable Food Quality & Safety October, 2nd -5th 2016 Benevento (Italy) European Food Safety Authority has also developed important to ensure that samples have been a similar system of facet descriptors for use with the handled correctly prior to analysis. Appropriate FoodEX2 food list that is used as the basis for sample handling could include storage in inert dietary intake, exposure and risk assessments. containers, cooling or freezing samples and minimising storage time or exposure to oxygen. 2.2. Component identification Prevention of contamination and complete Correct identification of food components is homogenisation are also important. usually straight forward but can be complicated in components that are method dependent. For 2.6. Analytical method and quality control example dietary fibre is not a single food Analytical methods may impact on values component but is a combination of compounds that produced and an appropriate method should be are measured by methods intended to capture used, wherever possible by a laboratory that is carbohydrates