Contribution of Highly Industrially Processed Foods to the Nutrient Intakes and Patterns of Middle-Aged Populations in the Europ

Home , Bouillon cube

ORIGINAL ARTICLE Contribution of highly industrially processed foods to the nutrient intakes and patterns of middle-aged populations in the European Prospective Investigation into Cancer and Nutrition study

{ N Slimani1, G Deharveng1, DAT Southgate2, , C Biessy1, V Chaje`s1,3, MME van Bakel1, MC Boutron-Ruault4, A McTaggart5, S Grioni6, J Verkaik-Kloosterman7, I Huybrechts1, P Amiano8, M Jenab9, J Vignat1, K Bouckaert1, C Casagrande1, P Ferrari1,28, P Zourna10, A Trichopoulou10, E Wirfa¨lt11, G Johansson12, S Rohrmann13, A-K Illner14, A Barricarte15, L Rodrı´guez16, M Touvier4,17, M Niravong4, A Mulligan5, F Crowe18, MC Ocke´7, YT van der Schouw19, B Bendinelli20, C Lauria21, M Brustad22, A Hjarta˚ker23, A Tjønneland24, { AM Jensen25, E Riboli26 and S Bingham5,27,

1Dietary Exposure Assessment Group, International Agency for Research on Cancer, Lyon, France; 28 Penryn Close, Norwich, Norfolk, UK; 3Institut Gustave Roussy, CNRS FRE 2939, Villejuif, France; 4Inserm, ERI 20, Institut Gustave Roussy, Villejuif, France; 5Department of Public Health and Primary Care, MRC Centre for Nutritional Epidemiology in Cancer Prevention and Survival, University of Cambridge, Cambridge, UK; 6Department of Preventive & Predictive Medicine, Nutritional Epidemiology Unit, Fondazione IRCCS Istituto Nazionale dei Tumori, Milan, Italy; 7National Institute for Public Health and the Environment (RIVM), Bilthoven, The Netherlands; 8Public Health Department of Gipuzkoa, Basque Government, San Sebastian and CIBER Epidemiologı´ay Salud Pu´blica (CIBERESP), Spain; 9Lifestyle and Cancer Group, International Agency for Research on Cancer, Lyon, France; 10Department of Hygiene, Epidemiology and Medical Statistics, University of Athens Medical School, Athens, Greece; 11Department of Clinical Sciences, Lund University, Malmo¨, Sweden; 12Department of Nutritional Research, University of Umea˚, Umea˚, Sweden; 13Division of Cancer Epidemiology, German Cancer Research Center (DKFZ), Heidelberg, Germany; 14Department of Epidemiology, German Institute of Human Nutrition, Potsdam-Rehbru¨cke, Germany; 15Institute of Public Health of Navarra, Pamplona and CIBER Epidemiologı´a y Salud Pu´blica (CIBERESP), Spain; 16Public Health and Participation Directorate, Health and Health Care Services Council, Asturias, Spain; 17AFSSA (French Food Safety Agency), DERNS/PASER, Maisons-Alfort, France; 18Cancer Epidemiology Unit, University of Oxford, Oxford, UK; 19Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht, Utrecht, The Netherlands; 20Molecular and Nutritional Epidemiology Unit, ISPO, Florence, Italy; 21Cancer Registry, Azienda Ospedaliera ‘Civile-M.P.Arezzo’, Ragusa, Italy; 22Institute of Community Medicine, University of Tromsø, Tromsø, Norway; 23Cancer Registry of Norway, Oslo, Norway; 24Institute of Cancer Epidemiology, Danish Cancer Society, Copenhagen, Denmark; 25Institute of Preventive Medicine, Copenhagen, Denmark; 26Department of Epidemiology, Public Health and Primary Care, Imperial College, London, UK and 27Diet and Cancer Group, MRC Mitochondrial Biology Unit, Cambridge, UK

Correspondence: Dr N Slimani, Dietary Exposure Assessment Group, International Agency for Research on Cancer (IARC), WHO, Lyon, France. E-mail: [email protected] { The authors are deceased. 28Current address: Data Collection and Exposure Unit (DATEX), European Food Safety Authority, Parma, Italy. Guarantor: N Slimani. Contributors: NS initiated and wrote this paper, taking into account comments from all co-authors, and was the overall coordinator of this project and of the EPIC Nutrient DataBase (ENDB) project. CB carried out the statistical analysis and preparation of tables and figures. GD was in charge of recoding dietary data according to the project-specific food reclassification, under the supervision of NS and DATS. DATS acted as an external expert on food chemistry and helped withthe reclassification according to food processing methods. NS, GD, DATS, CB, VC, MMEvB, MCBR, AMcT, SG, JVK, IH, PA and MJ were members of the writing group and gave input on statistical analysis, drafting of the article and interpretation of results. MCBR, AMcT, SG, JVK, PA, PZ, AT, EW, GJ, SR, AKI, AB, LR, MT, MN, AM, FC, MCO, YTvdS, BB, CL, MB, AH, AT, AMJ and SB were local EPIC collaborators involved in collecting data, checking the project-specific food reclassification and documenting, compiling and evaluating the subset of their national nutrient databases used in the ENDB. ER is the overall coordinator of the EPIC study. All co- authors provided comments and suggestions on the article and approved the final version. Highly industrially processed foods in EPIC N Slimani et al S207 Objectives: To describe the contribution of highly processed foods to total diet, nutrient intakes and patterns among 27 redefined centres in the 10 countries participating in the European Prospective Investigation into Cancer and Nutrition (EPIC). Methods: Single 24-hour dietary recalls were collected from 36 034 individuals (aged 35–74 years) using a standardized computerized interview programme (EPIC-SOFT). Centre-specific mean food intakes (g/day) were computed according to their degree of food processing (that is, highly, moderately and non-processed foods) using a specifically designed classification system. The contribution (%) of highly processed foods to the centre mean intakes of diet and 26 nutrients (including energy) was estimated using a standardized nutrient database (ENDB). The effect of different possible confounders was also investigated. Results: Highly processed foods were an important source of the nutrients considered, contributing between 61% (Spain) and 78–79% (the Netherlands and Germany) of mean energy intakes. Only two nutrients, b-carotene (34–46%) and vitamin C (28–36%), had a contribution from highly processed foods below 50% in Nordic countries, in Germany, the Netherlands and the United Kingdom, whereas for the other nutrients, the contribution varied from 50 to 91% (excluding alcohol). In southern countries (Greece, Spain, Italy and France), the overall contribution of highly processed foods to nutrient intakes was lower and consisted largely of staple or basic foods (for example, bread, pasta/rice, milk, vegetable oils), whereas highly processed foods such as crisp bread, breakfast cereals, margarine and other commercial foods contributed more in Nordic and central European centres. Conclusions: Highly industrially processed foods dominate diets and nutrient patterns in Nordic and central European countries. The greater variations observed within southern countries may reflect both a larger contribution of non/moderately processed staple foods along with a move from traditional to more industrialized dietary patterns. European Journal of Clinical Nutrition (2009) 63, S206–S225; doi:10.1038/ejcn.2009.82

Keywords: 24-h dietary recall; standardisation; processed foods; industrial foods; nutrient patterns; EPIC-SOFT

Introduction sophisticated preservation and processing techniques that changed food structure, nutritional content, texture and Two major historical periods have introduced profound taste. These processing technologies varied according to food changes in human diet and other lifestyle factors. The types, and involved packaging, moisture removal, heat introduction of agriculture and animal husbandry in the treatments, chilling and freezing, acidity control, chemical neolithic period (B10 000 years ago) and more recently additives and irradiation (Karmas and Harris, 1988). New and the industrial revolution (B200 years ago) have led to an complex food products that combined natural and artificial increased consumption of certain foods (for example, dairy ingredients, including additives, thus became widely avail- products, cereals and cereal products, refined sugars, vege- able. Sugars, salt and fats, available at a relatively low table oils, salt) and of a myriad of processed foods that were cost, were also extensively used for preservation purposes, virtually absent from pre-agricultural hunter-gatherer diets to make foods more palatable or as convenient ingredients (Eaton et al., 1997; Cordain et al., 2005; Eaton, 2006). The to prevent rancidity and improve texture (for example, substitution of unprocessed or modestly processed foods by hydrogenated fats and margarine in cakes, biscuits and more complex, refined (highly processed) foodstuffs may bakery products) (van Erp-Baart et al., 1998). In the first have affected several metabolic and nutritional character- instance, these urban-industrialized food systems helped to istics of ancestral human diets that had remained unchanged improve life conditions and life expectancy and, with the over millions of years (for example, glycaemic load, fatty increasing availability of elaborate ready-to-eat foods and acid composition, macronutrient composition, micronutri- dishes, responded to time scarcity in food preparation and ent density, acid–base balance, sodium–potassium ratio and cooking (Jabs and Devine, 2006). Since the mid twentieth fibre content) (Cordain et al., 2005). The inability to adapt century, however, a growing body of scientific evidence genetically to these recent changes is hypothesized to be one suggests that increased consumption of industrialized foods of the possible explanations for the increased incidence of increases the risk of various chronic diseases (WHO/FAO, obesity and chronic diseases (for example, type 2 diabetes, 2003; Schulze et al., 2004; Cordain et al., 2005; Mozaffarian cardiovascular disease, cancer) from the mid twentieth et al., 2006; Pomerleau et al., 2006; Ulijaszek, 2007; WCRF/ century onwards (Eaton and Konner, 1985; Tooby and AICR, 2007). Cosmides, 1990; Kious, 2002; Cordain et al., 2005; Ulijaszek, Although several characteristics of urban-industrialized 2007). food systems have been investigated in relation to disease, The development of intensive food production and there is a scarcity of data to evaluate these specific dietary industrialization was started in the eighteenth century in patterns across populations. A better understanding of Europe and in the United States. The first objectives were to the contribution of (highly) processed foods to current provide reliable food supplies, to improve microbiological diets across Europe will help to improve the design of quality and to devise means of preserving fresh and future studies on the association between chronic diseases perishable foods. This was obtained through increasingly and dietary patterns rich in (highly) processed foods,

European Journal of Clinical Nutrition Highly industrially processed foods in EPIC N Slimani et al S208 and to formulate more targeted public health recommenda- Dietary variables tions. The main objective of this study was to use a Using the EPIC-SOFT computer programme, detailed, highly set of comparable, highly detailed dietary data to investigate standardized dietary information was obtained through a the contribution of (highly) industrially processed foods single 24-HDR interview administered face-to-face in all to nutritional intakes and patterns in 27 middle-aged centres (Slimani et al., 1999), except in Norway where it was European population groups participating in the EPIC obtained by telephone interviews (Brustad et al., 2003). study. Details on the rationale, structure and validity of EPIC-SOFT for between-population comparisons are reported elsewhere (Slimani et al., 2002b, 2003; Al-Delaimy et al., 2005; Ferrari Materials and methods et al., 2009; Saadatian-Elahi et al., 2009).

Study population The population involved in this analysis comes from a Food definition and classification of industrially processed calibration substudy nested in the European Prospective foods. To investigate the consumption of highly indust- Investigation into Diet and Cancer (EPIC) study, a large rially processed foods in EPIC, as opposed to non- and cohort study undertaken in 23 centres in 10 countries: moderately processed foods, common definitions and re- Denmark, France, Germany, Greece, Italy, Norway, Spain, classification of the reported EPIC-SOFT food items were Sweden, the Netherlands and the United Kingdom (Riboli specifically developed with the support of an internationally et al., 2002; Bingham and Riboli, 2004). The main rationale recognized food chemistry expert, Professor DAT Southgate of the calibration study was to correct for measurement (Greenfield and Southgate, 2003). Each reported food was errors in baseline country-specific dietary measurements and recoded according to its degree of processing and was to attenuate bias in relative risk estimates through the use of classified into three main categories for which the food a common standardized 24-h dietary recall (24-HDR) group-specific processes and examples are summarized in computerized interview programme (EPIC-SOFT) (Kaaks Table A1 of the Appendix. et al., 1995; Slimani et al., 1999; Ferrari et al., 2008). For the Highly processed foods: Foods that have been industrially calibration study, a stratified random sample (36 994 parti- prepared, including those from bakeries and catering outlets, cipants, B8% of the total EPIC cohort) recorded a single and which require no or minimal domestic preparation apart 24-HDR with a trained interviewer between 1995 and 2000. from heating and cooking (for example, bread, breakfast Most of the EPIC participants were recruited from the cereals, cheese, commercial sauces, canned foods including general population, except in France (women state employ- jams, commercial cakes, biscuits and sauces). ees recruited from a local health insurance), in Turin and Moderately processed foods: This category includes two sets Ragusa in Italy and Spain (blood donors), Utrecht, the of foods. First, industrial and commercial foods involving Netherlands and Florence, Italy (women participating in relatively modest processing and consumed with no further breast cancer screening), and a British cohort of vegans and cooking such as dried fruits, raw vacuum-packed or under ovo-lacto vegetarians (‘‘health-conscious’’ cohort, recruited controlled atmosphere foods (for example, salads), frozen from around the United Kingdom). In Norway, only women basic foods, extra virgin olive oil, fruits and vegetables from the general population were recruited. The initial 23 canned in water/brine or in own juice. Second, foods EPIC administrative centres were redefined into 27 geogra- processed at home and prepared/cooked from raw or phical regions relevant to the analysis of dietary consump- moderately processed foods (for example, vegetables, meat tion patterns (Slimani et al., 2002a). In this paper, ‘‘central’’ and fish cooked from raw fresh ingredients, or vacuum- (European) centres represent those located in Germany, the packed, deep-frozen, canned in water/brine or in own juice). Netherlands and the United Kingdom, whereas southern Non-processed foods: Foods consumed raw without any centres are those in Greece, Italy, Spain and France, and further processing/preparation, except washing, cutting, northern centres are those in Sweden, Denmark and Norway. peeling, squeezing (for example, fruits, non-processed nuts, More details on the rationale and characteristics of the vegetables, crustaceans, molluscs, fresh juices). calibration study are provided elsewhere (Slimani et al., Foods with unknown process: Foods for which the processing 2002a). involved is unknown, based on the information provided by After a systematic exclusion of individuals under the the study subjects (for example, unknown preservation age of 35 or over 74 years, because of low participation method for vegetables, milk, meat or information missing in these age categories (n ¼ 960), a total of 36 034 individuals in homemade or commercially processed foods such as cakes with 24-HDR data were finally included in this analysis. and cream desserts). This category was relatively marginal, Approval for the study was obtained from the ethical below 5% in most centres. review boards of the International Agency for Research To increase comparability within and between centres, all on Cancer (Lyon, France) and the local EPIC collaborating the recoding and reclassification work was carried out centres. All participants provided written informed at the food/ingredient level, after mixed recipes were broken consent. down into ingredients. All the ingredients of ‘industrial/

European Journal of Clinical Nutrition Highly industrially processed foods in EPIC N Slimani et al S209 commercial’ recipes were coded as industrially processed The contribution of different types of food subgroups to foods, whereas those of homemade recipes were coded the centre mean energy intake from highly processed foods depending on whether the ingredients used were raw, was investigated first (Figures 1a and b). Multi-dimensional moderately or industrially processed. Cakes, biscuits, sauces graphic representations (Figures 2a–l) were then used to and soups, treated as foods in EPIC-SOFT, were broken down illustrate the percentage contribution of highly processed a posteriori using broad estimations of the relative propor- foods to the total centre mean intakes of 26 selected tions of non-industrially versus industrially processed foods, nutrients (including energy) after adjustment for gender, when they were made at home. Water was excluded from the age, energy, height and weight and was weighted by comparisons as it was not systematically recalled in all season and weekday. Unless otherwise specified, the compa- centres. risons within or between centres are indicated in percentage points. Nutrient databases. Nutrient intakes and derived patterns were calculated by means of standardized nutrient databases developed through the EPIC Nutrient DataBase (ENDB) Results project. The rationale and procedures used to improve between-country comparability of the 26 nutrients Mean food and beverage consumption according to degree of food included in this analysis are described elsewhere (Slimani processing et al., 2007). The mean total dietary consumption ranged from 1980 g/day and 1489 g/day (Greece) to 3601 g/day and 3176 g/day (Heidelberg, Germany) for men and women, respectively Other lifestyle variables (Tables 1a–b). In comparison with Nordic and central The other lifestyle variables including education level, total regions, southern countries (Greece, Spain and Italy), and physical activity and smoking history considered in this to a lesser extent France, reported much lower contributions analysis were collected at baseline through standardized of beverages (water excluded) to total dietary consumption. questionnaires and clinical examinations (Riboli et al., 2002), In Spain and Greece, women systematically reported a 4–7% and have been described for the calibration sample elsewhere higher contribution of foods to total intakes as compared (Slimani et al., 2002a). with men, with an equivalent lower contribution of Data on age as well as body weight and height were self- beverages. In the other centres, the gender difference was reported by participants during the 24-HDR interview. less than 4% (Tables 1a and b). The mean time interval between these baseline question- There was a strong geographic gradient for the contribu- naire measures and the 24-HDR interview varied by country, tion of foodstuffs to total mean consumption of foods and from 1 day to 3 years later (Slimani et al., 2002a). beverages according to the degree of processing. In both men and women, the mean contribution of highly processed foods ranged from 35–43% in Murcia (Spain) and Ragusa Statistical methods (Italy) to B60% in the Netherlands, Sweden, Norway, All analyses, unless otherwise specified, were performed Denmark and the UK general population. Moderately after stratification by gender and centre, and ordered processed foods represented 20–25% of total food intake in according to a geographical south–north gradient. Tables Italy, the United Kingdom, the Netherlands, Germany, 1a and b provide the mean consumption of foods and Granada (women, Spain), Sweden, Norway and Denmark. beverages (g/day and standard error (s.e.)) by centre and their Slightly higher percentages (27–38%) were observed in relative contribution as a percentage of the total diet Greece, France and in most Spanish centres. For men, the according to their degree of food processing. Centre-specific contribution of non-processed foods ranged from 11–14% of mean intakes and their percentages were adjusted for age, total intake in the UK general population, Sweden, Denmark height, weight and energy intake and weighted by season and the Netherlands (Bilthoven), to 25–32% in several and day of 24-HDR to control for differences in sampling southern centres in Spain, Italy and Greece. In women, procedures. This was computed in a multivariate regression there was a similar trend, although in almost all centres they model (analysis of covariance—ANCOVA) and weighted had a higher relative contribution of non-processed foods using generalized linear models (‘GENMOD’ procedure in than men, up to 5–7% higher in Varese and Ragusa (Italy), SAS software). The effect of other possible confounders Granada (Spain), the UK general population, Germany, (smoking status, physical activity, BMI and education level) Sweden and Denmark. on mean intakes of highly processed foods (g/day) and Beverage consumption (excluding water) consisted almost centre rankings was also examined, by comparing models exclusively of highly processed commercial beverages (for with or without the variable of interest (data not shown). example, alcoholic beverages, tea and coffee), with values The R-square of the model and the partial R-square of the above 85% in most centres and higher values in men than in additional co-variable were calculated. Significance was women, by up to 11–13% in Granada (Spain) and Greece. assessed by means of the partial F-test. The contribution of non-processed beverages (for example,

European Journal of Clinical Nutrition S210 uoenJunlo lnclNutrition Clinical of Journal European

Table 1a Fully adjusted mean consumption (g/day (s.e.)) of foods and beveragesa and their relative contribution (% (s.e.)) to the total diet, according to their degree of food processing: men (N ¼ 13 025)

Country and N Total dietary Total food Total beverage Foods Beverages centre consumptionb consumption consumption

g/day (s.e.) g/day (s.e.) % (s.e.) g/day (s.e.) % (s.e.) Highly Moderately Non Unknown Highly Moderately Non Un-known processedc processedd processede processedc processedd processede % (s.e.) % (s.e.) % (s.e.) % (s.e.) % (s.e.) % (s.e.) % (s.e.) % (s.e.) ihyidsral rcse od nEPIC in foods processed industrially Highly Greece 1311 1979.5 (20.1) 1490.6 (11.6) 79.7 (0.4) 488.9 (20.2) 20.3 (0.4) 44.4 (0.5) 28.8 (0.4) 25.7 (0.4) 1.1 (0.2) 85.1 (0.4) 4.7 (0.3) 10.2 (0.3) 0.0 (0.0)

Spain 1777 Granada 214 2410.5 (47.6) 1756.1 (27.8) 74.5 (1.0) 654.4 (47.6) 25.5 (1.0) 41.0 (1.2) 27.4 (1.1) 27.0 (1.0) 4.6 (0.4) 89.2 (1.0) 5.3 (0.8) 5.5 (0.7) 0.0 (0.1) Murcia 243 2579.6 (44.8) 1683.6 (26.1) 68.1 (0.9) 896.1 (44.8) 31.9 (0.9) 34.5 (1.1) 26.9 (1.0) 32.4 (0.9) 6.2 (0.4) 89.0 (0.9) 6.1 (0.7) 5.0 (0.6) 0.0 (0.1) Navarra 444 2178.7 (33.3) 1625.6 (19.4) 75.1 (0.7) 553.1 (33.3) 24.9 (0.7) 39.4 (0.8) 35.1 (0.7) 24.0 (0.7) 1.6 (0.3) 94.4 (0.7) 2.8 (0.5) 2.8 (0.5) 0.0 (0.0) San Sebastian 490 2385.4 (31.9) 1783.9 (18.6) 73.9 (0.7) 601.6 (31.9) 26.1 (0.7) 34.5 (0.8) 37.7 (0.7) 25.6 (0.6) 2.1 (0.3) 92.3 (0.7) 4.4 (0.5) 3.2 (0.4) 0.1 (0.0) Asturias 386 2405.8 (35.5) 1725.3 (20.7) 72.9 (0.8) 680.5 (35.5) 27.1 (0.8) 42.1 (0.9) 32.6 (0.8) 23.1 (0.7) 2.3 (0.3) 92.4 (0.8) 4.4 (0.6) 3.2 (0.5) 0.0 (0.1) Slimani N Italy 1442 Ragusa 168 2221.7 (53.7) 1667.0 (31.4) 75.5 (1.1) 554.7 (53.7) 24.5 (1.1) 43.1 (1.3) 24.8 (1.2) 30.1 (1.1) 1.9 (0.5) 97.2 (1.1) 0.4 (0.9) 2.4 (0.7) 0.0 (0.1) Florence 271 2762.1 (42.1) 1675.2 (24.6) 62.1 (0.9) 1086.9 (42.1) 37.9 (0.9) 46.7 (1.0) 23.9 (0.9) 27.7 (0.8) 1.7 (0.4) 97.8 (0.9) 0.9 (0.7) 1.2 (0.6) 0.0 (0.1) Turin 676 2918.6 (26.8) 1631.9 (15.6) 57.8 (0.6) 1286.7 (26.8) 42.2 (0.6) 45.2 (0.7) 23.4 (0.6) 30.5 (0.5) 0.9 (0.2) 97.9 (0.6) 1.0 (0.4) 1.1 (0.4) 0.0 (0.0) al et Varese 327 2653.1 (38.5) 1638.0 (22.4) 62.2 (0.8) 1015.1 (38.5) 37.8 (0.8) 52.7 (1.0) 21.3 (0.9) 23.6 (0.8) 2.3 (0.3) 98.0 (0.8) 1.2 (0.6) 0.8 (0.5) 0.0 (0.1)

Germany 2267 Heidelberg 1034 3601.4 (21.7) 1299.5 (12.7) 37.6 (0.5) 2301.9 (21.7) 62.4 (0.5) 56.6 (0.5) 22.4 (0.5) 17.5 (0.4) 3.5 (0.2) 93.5 (0.5) 5.9 (0.4) 0.6 (0.3) 0.0 (0.0) Potsdam 1233 3207.2 (19.7) 1353.5 (11.5) 43.3 (0.4) 1853.8 (19.7) 56.7 (0.4) 57.0 (0.5) 22.1 (0.4) 19.2 (0.4) 1.6 (0.2) 92.9 (0.4) 6.9 (0.3) 0.2 (0.3) 0.0 (0.0)

The Netherlands Bilthoven 1024 3034.7 (22.5) 1425.7 (13.1) 49.0 (0.5) 1609.0 (22.5) 51.0 (0.5) 62.0 (0.6) 25.2 (0.5) 11.3 (0.5) 1.5 (0.2) 96.4 (0.5) 1.9 (0.4) 1.7 (0.3) 0.1 (0.0)

United Kingdom 516 General 402 3377.6 (34.6) 1481.1 (20.2) 45.1 (0.7) 1896.5 (34.6) 54.9 (0.7) 62.3 (0.9) 24.4 (0.8) 11.5 (0.7) 1.9 (0.3) 99.3 (0.7) 0.6 (0.6) 0.0 (0.5) 0.0 (0.0) population Health- 114 3410.0 (65.2) 1583.7 (38.0) 50.2 (1.4) 1826.3 (65.2) 49.8 (1.4) 55.7 (1.6) 24.0 (1.4) 18.8 (1.3) 1.5 (0.6) 89.5 (1.4) 8.1 (1.1) 2.3 (0.9) 0.0 (0.1) conscious

Denmark 1923 Copenhagen 1356 3337.3 (19.0) 1307.9 (11.1) 40.9 (0.4) 2029.5 (19.0) 59.1 (0.4) 60.6 (0.5) 23.0 (0.4) 13.9 (0.4) 2.5 (0.2) 98.5 (0.4) 1.2 (0.3) 0.3 (0.3) 0.0 (0.0) Aarhus 567 3309.7 (29.2) 1337.0 (17.0) 42.4 (0.6) 1972.7 (29.2) 57.6 (0.6) 60.9 (0.7) 23.6 (0.6) 14.1 (0.6) 1.3 (0.3) 98.6 (0.6) 1.1 (0.5) 0.3 (0.4) 0.0 (0.0)

Sweden 2765 Malmo¨ 1421 2817.0 (19.6) 1405.5 (11.5) 51.1 (0.4) 1411.5 (19.6) 48.9 (0.4) 61.4 (0.5) 22.2 (0.4) 11.6 (0.4) 4.7 (0.2) 99.9 (0.4) 0.0 (0.3) 0.0 (0.3) 0.0 (0.0) Umea˚ 1344 2720.6 (19.1) 1526.7 (11.2) 57.4 (0.4) 1193.9 (19.1) 42.6 (0.4) 64.2 (0.5) 20.5 (0.4) 11.1 (0.4) 4.2 (0.2) 99.5 (0.4) 0.3 (0.3) 0.0 (0.3) 0.2 (0.0)

aAdjusted for age, height, weight, energy intakes and weighted by season and weekday of 24-h recall. bTotal dietary consumption including all foods and beverages (except water). cIndustrial and commercial. dIndustrial and commercial (including homemade cooked foods). eConsumed raw. Table 1b Fully adjusted mean consumption (g/day (s.e.)) of foods and beveragesa and their relative contribution (% (s.e.)) to the total diet, according to their degree of food processing: women (N ¼ 23 008)

Country and N Total dietary Total food Total beverage Foods Beverages center consumptionb consumption consumption

g/day (s.e.) g/day (s.e.) % (s.e.) g/day (s.e.) % (s.e.) Highly Moderately Non Unknown Highly Moderately Non Unknown processed processed processed processed processed processed foodsc foodsd foodse foodsc foodsd foodse % (s.e.) % (s.e.) % (s.e.) % (s.e.) % (s.e.) % (s.e.) % (s.e.) % (s.e.)

Greece 1373 1488.6 (18.7) 1193.7 (9.7) 85.5 (0.4) 294.9 (17.3) 14.5 (0.4) 45.3 (0.5) 26.7 (0.4) 27.3 (0.4) 0.7 (0.1) 74.4 (0.5) 4.9 (0.4) 20.7 (0.3) 0.0 (0.0)

Spain 1443 Granada 300 1982.3 (38.0) 1505.4 (20.1) 78.8 (0.9) 476.9 (35.1) 21.2 (0.9) 43.7 (1.0) 23.0 (0.9) 31.6 (0.9) 1.8 (0.3) 76.4 (1.1) 14.1 (0.9) 9.5 (0.7) 0.0 (0.1) Murcia 304 2073.5 (37.8) 1460.1 (20.0) 74.5 (0.9) 613.4 (34.8) 25.5 (0.9) 36.8 (1.0) 26.7 (0.9) 32.3 (0.9) 4.0 (0.3) 79.6 (1.0) 10.9 (0.8) 9.5 (0.6) 0.0 (0.1) Navarra 271 1819.2 (39.9) 1494.1 (21.1) 82.2 (0.9) 325.1 (36.8) 17.8 (0.9) 43.4 (1.1) 27.4 (0.9) 27.6 (0.9) 1.6 (0.3) 91.4 (1.1) 4.3 (0.9) 4.3 (0.7) 0.0 (0.1) San Sebastian 244 1967.1 (42.0) 1579.4 (22.2) 80.5 (1.0) 387.7 (38.7) 19.5 (1.0) 39.6 (1.1) 30.9 (1.0) 27.7 (1.0) 1.8 (0.3) 87.6 (1.2) 7.1 (0.9) 5.0 (0.7) 0.0 (0.1) Asturias 324 1951.9 (36.5) 1518.0 (19.3) 79.4 (0.8) 433.9 (33.7) 20.6 (0.8) 45.1 (1.0) 26.6 (0.8) 26.7 (0.8) 1.5 (0.3) 86.7 (1.0) 7.2 (0.8) 5.8 (0.6) 0.3 (0.1)

Italy 2511 Ragusa 138 1750.6 (56.0) 1276.9 (29.6) 75.6 (1.3) 473.7 (51.7) 24.4 (1.3) 37.7 (1.5) 25.3 (1.3) 35.1 (1.3) 1.9 (0.4) 95.7 (1.5) 1.7 (1.2) 2.6 (0.9) 0.0 (0.1) Naples 403 1843.1 (32.9) 1229.0 (17.4) 69.2 (0.8) 614.1 (30.3) 30.8 (0.8) 48.3 (0.9) 23.8 (0.8) 26.5 (0.7) 1.4 (0.3) 97.0 (0.9) 1.3 (0.7) 1.7 (0.6) 0.0 (0.0) Florence 784 2321.4 (23.4) 1315.2 (12.4) 59.3 (0.5) 1006.1 (21.6) 40.7 (0.5) 45.3 (0.6) 24.4 (0.5) 29.3 (0.5) 1.1 (0.2) 96.9 (0.6) 1.5 (0.5) 1.6 (0.4) 0.0 (0.0) Turin 392 2444.3 (33.1) 1343.7 (17.5) 57.5 (0.8) 1100.6 (30.5) 42.5 (0.8) 41.8 (0.9) 24.6 (0.8) 33.0 (0.8) 0.6 (0.3) 96.1 (0.9) 1.7 (0.7) 2.1 (0.6) 0.0 (0.0) Varese 794 2144.3 (23.4) 1343.4 (12.4) 63.8 (0.5) 801.0 (21.6) 36.2 (0.5) 48.9 (0.6) 20.9 (0.5) 28.5 (0.5) 1.7 (0.2) 96.1 (0.6) 2.7 (0.5) 1.2 (0.4) 0.0 (0.0)

France 4735 South coast 620 2234.6 (26.4) 1351.0 (14.0) 63.7 (0.6) 883.6 (24.4) 36.3 (0.6) 46.3 (0.7) 29.8 (0.6) 22.2 (0.6) 1.7 (0.2) 91.4 (0.7) 6.4 (0.6) 2.2 (0.4) 0.1 (0.0) South 1425 2321.8 (17.6) 1337.8 (9.3) 60.3 (0.4) 984.0 (16.2) 39.7 (0.4) 47.4 (0.5) 30.0 (0.4) 22.0 (0.4) 0.7 (0.1) 92.0 (0.5) 6.2 (0.4) 1.7 (0.3) 0.1 (0.0) North-East 2059 2461.7 (14.6) 1311.4 (7.7) 56.2 (0.3) 1150.4 (13.5) 43.8 (0.3) 47.8 (0.4) 28.9 (0.3) 21.7 (0.3) 1.6 (0.1) 93.1 (0.4) 5.4 (0.3) 1.5 (0.2) 0.0 (0.0) North-West 631 2656.8 (26.2) 1322.8 (13.8) 51.9 (0.6) 1334.0 (24.2) 48.1 (0.6) 45.8 (0.7) 30.6 (0.6) 22.6 (0.6) 1.0 (0.2) 94.8 (0.7) 3.8 (0.6) 1.4 (0.4) 0.0 (0.0)

Germany 2148 EPIC Slimani in N foods processed industrially Highly Heidelberg 1087 3175.6 (20.1) 1152.9 (10.6) 37.7 (0.5) 2022.8 (18.5) 62.3 (0.5) 52.9 (0.5) 22.1 (0.5) 23.0 (0.5) 1.9 (0.2) 89.4 (0.5) 9.6 (0.4) 1.0 (0.3) 0.0 (0.0) Potsdam 1061 2770.4 (20.2) 1208.4 (10.7) 44.2 (0.5) 1562.0 (18.6) 55.8 (0.5) 53.3 (0.5) 20.3 (0.5) 25.5 (0.5) 1.0 (0.2) 89.3 (0.6) 10.4 (0.4) 0.3 (0.3) 0.0 (0.0)

The Netherlands 2956 al et Bilthoven 1086 2487.7 (20.3) 1201.2 (10.7) 49.7 (0.5) 1286.5 (18.7) 50.3 (0.5) 60.5 (0.5) 23.5 (0.5) 14.5 (0.5) 1.5 (0.2) 94.2 (0.6) 3.2 (0.5) 2.5 (0.3) 0.0 (0.0) Utrecht 1870 2625.9 (15.4) 1303.1 (8.1) 50.6 (0.4) 1322.8 (14.2) 49.4 (0.4) 60.3 (0.4) 22.5 (0.4) 16.4 (0.3) 0.8 (0.1) 94.5 (0.4) 2.5 (0.3) 3.0 (0.3) 0.0 (0.0)

United Kingdom 767 General 570 2887.6 (27.4) 1271.7 (14.5) 45.2 (0.6) 1615.9 (25.3) 54.8 (0.6) 59.4 (0.7) 22.8 (0.6) 16.8 (0.6) 1.0 (0.2) 97.5 (0.7) 2.0 (0.6) 0.4 (0.5) 0.0 (0.0) population Health- 197 2741.5 (46.6) 1342.9 (24.6) 52.2 (1.1) 1398.6 (43.0) 47.8 (1.1) 51.8 (1.2) 25.0 (1.1) 22.5 (1.1) 0.7 (0.4) 90.2 (1.3) 9.0 (1.1) 0.7 (0.8) 0.1 (0.1) conscious

Denmark 1994 Copenhagen 1484 2660.1 (17.1) 1119.2 (9.0) 43.8 (0.4) 1540.9 (15.8) 56.2 (0.4) 56.4 (0.5) 22.1 (0.4) 19.8 (0.4) 1.7 (0.1) 95.8 (0.5) 3.6 (0.4) 0.5 (0.3) 0.0 (0.0) Aarhus 510 2681.2 (29.0) 1169.5 (15.3) 45.8 (0.7) 1511.7 (26.8) 54.2 (0.7) 55.8 (0.8) 21.3 (0.7) 21.8 (0.7) 1.1 (0.2) 94.3 (0.8) 5.4 (0.6) 0.2 (0.5) 0.0 (0.0)

Sweden 3284 Malmo¨ 1710 2299.0 (16.3) 1169.8 (8.6) 52.2 (0.4) 1129.2 (15.0) 47.8 (0.4) 58.3 (0.4) 20.4 (0.4) 17.7 (0.4) 3.5 (0.1) 98.9 (0.4) 0.9 (0.4) 0.2 (0.3) 0.0 (0.0) Umea˚ 1574 2200.0 (16.5) 1253.9 (8.7) 58.2 (0.4) 946.1 (15.3) 41.8 (0.4) 58.9 (0.4) 19.3 (0.4) 18.0 (0.4) 3.8 (0.1) 98.5 (0.5) 1.1 (0.4) 0.3 (0.3) 0.2 (0.0)

uoenJunlo lnclNutrition Clinical of Journal European Norway 1797 South & East 1004 2484.7 (21.2) 1166.3 (11.2) 48.9 (0.5) 1318.4 (19.5) 51.1 (0.5) 58.6 (0.6) 21.2 (0.5) 18.7 (0.5) 1.5 (0.2) 93.6 (0.6) 5.8 (0.5) 0.6 (0.4) 0.0 (0.0) North & West 793 2425.3 (23.6) 1203.9 (12.5) 51.4 (0.5) 1221.4 (21.8) 48.6 (0.5) 57.6 (0.6) 23.9 (0.5) 16.6 (0.5) 1.9 (0.2) 93.2 (0.6) 6.5 (0.5) 0.3 (0.4) 0.0 (0.0)

aAdjusted for age, height, weight, energy intakes and weighted by season and weekday of 24-h recall. bTotal dietary consumption including all foods and beverages (except water). cIndustrial and commercial. dIndustrial and commercial (including homemade cooked foods). eConsumed raw. S211 Highly industrially processed foods in EPIC N Slimani et al S212 60

Sugar, Honey, Jam, Syrup

Butter, animal fats

Vegetable oils

% 30

Milk

20 Pasta, Rice, Grains & Other cereal products (1) 10 Bread

0 ö ea ay

urcia Turin alm reece Um M M G ranada Naples Varese France Utrecht Aarhus Navarra Asturias Ragusa G Florence Potsdam of France Bilthoven Heidelberg population

est of France Copenhagen est of Norw San Sebastian South of eneral

North East of France North W UK Health conscious South EastNorth of NorwayW South Coast UK G

60 Miscellaneous beverages (2)

50 Miscellaneous foods (3)

Soft drinks

40 Alcoholic beverages

Processed Meat, Fish and % 30 Egg products Margarine, deep-frying fats

20 Other dairy products (4)

Breakfast cereals

10 Crisp breads, rusks

Cakes, biscuits 0 s in n n ö sa se ce ce us ea rra ian am ada nce gen urcia alm rway rway Tur ran latio elberg Um M Naple M Greece Vare Utrecht nsciousnha Aarh Nava Asturias Ragu f Francef F Gran Florence Potsd Bilthove Heid of No st of Franst of Fra l popu h co st of No e Cope ast e San Sebast oast o South o h E h C

North Ea North W UK Healt Sout North W Sout UK Genera Figure 1 (a) Contributions of highly processed staple/basic foods to centre mean total energy intakes (%) after adjustment for season, weekday, height, weight, age and gender. (b) Contributions of the other more complex highly processed foods to centre mean total energy intakes (%) after adjustment for season, weekday, height, weight, age and gender. (1) Flour, flakes, dough, pastries. (2) Fruit and vegetable juices, coffee, tea, herbal tea, chicory, non- alcoholic beer, alcohol for cooking. (3) Potato-, vegetable-, legume- and fruit-products (including olives and nuts), chocolate, candies, ice creams, sorbets, salty biscuits, sauces, vegetarian foods, dietetic products, creamers, snacks. (4) Yogurt, cheese, milk beverages, curd, cream dessert, dairy cream.

European Journal of Clinical Nutrition Highly industrially processed foods in EPIC N Slimani et al S213 Greece Total energy Magnesium 100 Total proteins Potassium Total fats Greece Iron 75 Total saturated fatty acids

Phosphorus Total monounsaturated fatty acids 50

Calcium Total polyunsaturated fatty acids

Vitamin D Cholesterol

Vitamin E Carbohydrates

Retinol Starch

Beta-carotene Sugar

Vitamin C Dietary fibre

Cobalamin (B12) Alcohol Vitamin (B6) Thiamin (B1) Riboflavin (B2)

Spain Total energy Magnesium 100 Total proteins Asturias Potassium Total fats Granada Murcia Iron 75 Total saturated fatty acids Navarra San Sebastian Phosphorus Total monounsaturated fatty acids 50

Calcium Total polyunsaturated fatty acids

Vitamin D Cholesterol

Vitamin E Carbohydrates

Retinol Starch

Beta-carotene Sugar

Vitamin C Dietary fibre

Cobalamin (B12) Alcohol Vitamin (B6) Thiamin (B1) Riboflavin (B2) Figure 2 (a–l) Percentage contribution of highly processed foods to the total centre mean intakes of 26 selected nutrients (including energy) after adjustment for gender, age, energy, height, weight and weighted by season and weekday.

European Journal of Clinical Nutrition Highly industrially processed foods in EPIC N Slimani et al S214 Italy Total energy Magnesium 100 Total proteins Florence Potassium Total fats Ragusa Turin Iron 75 Total saturated fatty acids Varese Naples Phosphorus Total monounsaturated fatty acids 50

Calcium Total polyunsaturated fatty acids

Vitamin D Cholesterol

Vitamin E Carbohydrates

Retinol Starch

Beta-carotene Sugar

Vitamin C Dietary fibre

Cobalamin (B12) Alcohol Vitamin (B6) Thiamin (B1) Riboflavin (B2)

France (women only) Total energy Magnesium 100 Total proteins North-East of France Potassium Total fats North-West of France South coast of France Iron 75 Total saturated fatty acids South of France

Phosphorus Total monounsaturated fatty acids 50

Calcium Total polyunsaturated fatty acids

Vitamin D Cholesterol

Vitamin E Carbohydrates

Retinol Starch

Beta-carotene Sugar

Vitamin C Dietary fibre

Cobalamin (B12) Alcohol Vitamin (B6) Thiamin (B1) Riboflavin (B2) Figure 2 Continued.

fresh fruit juices) was marginal in most centres (o3% in both Main food sources of energy intake from highly processed foods genders), except in Spain (men 5%, women 9.5%, in the Energy intake was selected as the most relevant nutritional southern centres of Granada and Murcia) and Greece (men indicator for assessing the contribution of highly processed 10%, women 21%). foods to the overall diet, taking into account both food and

European Journal of Clinical Nutrition Highly industrially processed foods in EPIC N Slimani et al S215 Germany Total energy Magnesium 100 Total proteins

Potassium Total fats Heidelberg Potsdam Iron 75 Total saturated fatty acids

Phosphorus Total monounsaturated fatty acids 50

Calcium Total polyunsaturated fatty acids

Vitamin D Cholesterol

Vitamin E Carbohydrates

Retinol Starch

Beta-carotene Sugar

Vitamin C Dietary fibre

Cobalamin (B12) Alcohol Vitamin (B6) Thiamin (B1) Riboflavin (B2)

The Netherlands Total energy Magnesium 100 Total proteins Potassium Total fats Bilthoven Utrecht (women) Iron 75 Total saturated fatty acids

Phosphorus Total monounsaturated fatty acids 50

Calcium Total polyunsaturated fatty acids

Vitamin D Cholesterol

Vitamin E Carbohydrates

Retinol Starch

Beta-carotene Sugar

Vitamin C Dietary fibre

Cobalamin (B12) Alcohol Vitamin (B6) Thiamin (B1) Riboflavin (B2) Figure 2 Continued. beverage energy sources. We consistently observed that three products were by far the most important energy providers main food groups contributed the greatest amount to energy in all centres, with values ranging from B30% (Potsdam, intake from highly processed foods. Cereals and cereal Germany) to B58% (Ragusa, Italy), followed by dairy

European Journal of Clinical Nutrition Highly industrially processed foods in EPIC N Slimani et al S216 United Kingdom Total energy Magnesium 100 Total proteins Potassium Total fats UK Health conscious UK General population Iron 75 Total saturated fatty acids

Phosphorus Total monounsaturated fatty acids 50

Calcium Total polyunsaturated fatty acids

Vitamin D Cholesterol

Vitamin E Carbohydrates

Retinol Starch

Beta-carotene Sugar

Vitamin C Dietary fibre

Cobalamin (B12) Alcohol Vitamin (B6) Thiamin (B1) Riboflavin (B2)

Denmark Total energy Magnesium 100 Total proteins

Potassium Total fats Aarhus Copenhagen Iron 75 Total saturated fatty acids

Phosphorus Total monounsaturated fatty acids 50

Calcium Total polyunsaturated fatty acids

Vitamin D Cholesterol

Vitamin E Carbohydrates

Retinol Starch

Beta-carotene Sugar

Vitamin C Dietary fibre

Cobalamin (B12) Alcohol Vitamin (B6) Thiamin (B1) Riboflavin (B2) Figure 2 Continued.

products (10–23%) or alternatively fats (5–27%), depending to energy intake (o1–12%). Two complementary figures on centre. Greater differences were observed across centres (Figures 1a–b) show comparisons of the contribution to the in the types of the next food groups contributing most centre mean energy intake of highly processed foods from

European Journal of Clinical Nutrition Highly industrially processed foods in EPIC N Slimani et al S217 Sweden Total energy Magnesium 100 Total proteins Potassium Total fats Malmö Umea Iron 75 Total saturated fatty acids

Phosphorus Total monounsaturated fatty acids 50

Calcium Total polyunsaturated fatty acids

Vitamin D Cholesterol

Vitamin E Carbohydrates

Retinol Starch

Beta-carotene Sugar

Vitamin C Dietary fibre

Cobalamin (B12) Alcohol Vitamin (B6) Thiamin (B1) Riboflavin (B2)

Norway (women only) Total energy Magnesium 100 Total proteins Potassium Total fats North and West of Norway South and East of Norway Iron 75 Total saturated fatty acids

Phosphorus Total monounsaturated fatty acids 50

Calcium Total polyunsaturated fatty acids

Vitamin D Cholesterol

Vitamin E Carbohydrates

Retinol Starch

Beta-carotene Sugar

Vitamin C Dietary fibre

Cobalamin (B12) Alcohol Vitamin (B6) Thiamin (B1) Riboflavin (B2) Figure 2 Continued. relatively staple/basic foods (for example, bread, pasta, rice, milk beverages, yoghurt, cheese, cream desserts, margarines milk, vegetable oils) (Figure 1a) versus other more highly and other hardened fats and alcoholic beverages) (Figure 1b). processed foods (for example, cakes, biscuits, breakfast It seemed that B31–45% of the mean energy intake in the cereals, crisp bread, confectionery, processed meat and fish, southern centres (Greece, Spain, Italy) was provided by

European Journal of Clinical Nutrition Highly industrially processed foods in EPIC N Slimani et al S218 All nordic and central European countries Countries

Denmark Total energy Sweden Magnesium 100 Total proteins Norway Potassium Total fats Germany Iron 75 Total saturated fatty acids The Netherlands UK General population Phosphorus Total monounsaturated fatty acids 50

Calcium Total polyunsaturated fatty acids 25

Vitamin D Cholesterol

Vitamin E Carbohydrates

Retinol Starch

Beta-carotene Sugar

Vitamin C Dietary fibre

Cobalamin (B12) Alcohol Vitamin (B6) Thiamin (B1) Riboflavin (B2)

All southern European countries Total energy Countries Magnesium 100 Total proteins Potassium Total fats Italy Spain Iron Total saturated fatty acids 75 Greece France Phosphorus Total monounsaturated fatty acids 50

Calcium Total polyunsaturated fatty acids

Vitamin D Cholesterol

Vitamin E Carbohydrates

Retinol Starch

Beta-carotene Sugar

Vitamin C Dietary fibre

Cobalamin (B12) Alcohol Vitamin (B6) Thiamin (B1) Riboflavin (B2) Figure 2 Continued.

European Journal of Clinical Nutrition Highly industrially processed foods in EPIC N Slimani et al S219 highly processed staple/basic foods. In contrast, in the similar overlapping shapes among all Nordic and central Nordic and central European countries, 50–56% of energy European countries (Figure 2k). was provided by other more highly processed foods. France (44–46%) and the UK health-conscious group (48%) reported Southern European centres. In contrast to Nordic and patterns similar to the Nordic and central European central European countries, southern centres in Spain, Italy, countries. Greece and, to a lesser extent, France showed different nutrient patterns and greater variability within and between countries (Figures 2a–d and l). Overall, we observed a Contribution of highly processed foods to centre mean nutrient much lower contribution to the mean nutrient intake from intake and patterns highly processed foods, particularly in Spain and Italy, The contribution (%) of highly industrially and commer- and for vitamin C (3–24%), b-carotene (14–27%), vitamin cially processed foods to centre mean nutrient intakes and B6 (26–43%), iron (29–47%), fibre (28–53%), cholesterol patterns is summarized by country in Figures 2a–j. After (26–60%) and potassium (26–60%). adjustment for age, gender, weekday, season, energy, height For energy, highly processed foods contributed 61–65% in and weight, a strong geographical gradient was observed, all Spanish centres and Ragusa (Italy) and 72–74% in the with two distinct sets of nutrient patterns, namely, the French regions and Greece. For macronutrients, alcohol came Nordic and central European countries versus the southern essentially from highly processed foods (94–98%), whereas European countries. greater variability within and between countries was reported for the other macronutrients. For protein, the range varied Nordic and central European centres. In the UK (general from 38% (San-Sebastian, Spain) to 61% (Naples, Italy). For population), Germany, the Netherlands, Sweden, Denmark total fat (53–86%) and fat subtypes (48–89%), the contribu- and Norway (Figure 2e–j), highly processed foods provided tion of highly processed foods varied widely across southern less than 50% of total intake for only two nutrients: countries, with values in Greece (81–89%) similar to those b-carotene (34–46%) and vitamin C (28–36%). For all other reported in Nordic and central European centres. However, nutrients and energy, highly processed foods provided bet- highly processed foods contributed much less to cholesterol ween 50 and 99% of intake. A large proportion (76–79%) of the intakes, with values ranging from 26% (San Sebastian, Spain) intake of energy, fat and carbohydrate components, magne- to 60% (Varese, Italy). In contrast to iron (35–56%) and sium, phosphorus, potassium, calcium, retinol and alcohol potassium (29–47%), larger contributions from highly pro- was provided by highly processed foods, and a moderate cessed foods were observed for calcium (58–79%), magne- proportion for vitamin B6 (50–55%), dietary fibre (53–66%), sium (43–67%) and phosphorus (43–64%), although they potassium (57–62%), vitamin B1 (58–68%), cholesterol were still significantly lower than those reported in Northern (58–69%), iron (60–70%) and protein (62–68%). Greater vari- and Central Europe. Retinol (54–86%), vitamin D (39–67%) ability was observed among the Nordic and central European and vitamin B1, B2 and B12 (36–68%) showed greater centres for vitamin D (62–80%) and vitamin E (62–78%). contributions from highly processed foods and greater There were virtually no differences in the overall variability across southern countries, as compared with nutrient patterns (p3% for most nutrients) between centres vitamin C (3–24%), b-carotene (14–27%) and vitamin B6 within Denmark, Sweden or Norway (Figures 2h–j). More (26–43%). In contrast to Nordic and central centres notable between–centre differences (3–9%) were observed for (Figure 2k), nutrient patterns in southern countries do not certain nutrients in the Netherlands (for example, simple overlap, but show distinct local dietary habits and greater sugars, fibre, vitamin E, vitamin B12 and b-carotene) and in heterogeneity in the contribution of highly processed foods Germany (for example, b-carotene, vitamin D, vitamin B12 within and between countries (Figure 2l). and cholesterol). For the United Kingdom, the health- conscious group and the general population sample had a Gender differences. When men and women were considered similar contribution of highly processed foods for a number of together, the P-value of the interaction between gender and nutrients such as total fat, fatty acids and phosphorus (centre the contribution of highly processed foods to centre mean difference p1%), and for energy, starch, fibre, iron, magne- intakes was statistically significant for energy and all nutrients sium, alcohol, retinol, b-carotene and most B-vitamins (centre except thiamine. In most cases, these gender differences did difference 2–5%). Differences were greater between these two not exceed 10%. For fibre and b-carotene, for example, British population groups for vitamin B2 (6%), potassium women had statistically lower contributions from highly (7%), vitamin C (8%), calcium and carbohydrates (10%) and processed foods than did men in most centres, with sugar (16%), with higher contributions from highly processed differences of 2–9% for fibre and 1–13% for b-carotene. foods in the general population sample, whereas for protein (6%), vitamin D (11%) and cholesterol (20%), the proportion of intake was higher for the health-conscious group. Influence of covariates When considered together, the overall contribution of The influence of possible confounders (smoking status, highly processed foods to mean nutrient intakes showed physical activity and education level) on centre rankings

European Journal of Clinical Nutrition Highly industrially processed foods in EPIC N Slimani et al S220 and mean intakes of highly processed foods (g/day) was also and central European countries for a large series of nutrients, examined by comparing the baseline model (adjusted for including energy. Greater variability was reported in south- age, energy, height, weight, weekdays, seasons) with and ern centres, with women in France showing contributions in without variables of interest. For BMI, the effect was between southern and Nordic/central regions. Furthermore, estimated using a model without height and weight to when considering the qualitative types of highly processed remove the dependence between these co-variables. In men, foods, we observed that B31–45% of total energy intake in level of education, physical activity, BMI and smoking, each Greece, Spain and Italy comes from basic or staple foods (for considered separately, had no effect on the centre mean example, bread, pasta/rice, milk, vegetable oils). Similar intakes of highly processed foods (o1%) and ranking, and results have been reported elsewhere (Karamanos et al., could not explain the total variability in the mean intake of 2002; Tur et al., 2004; Garcia-Closas et al., 2006). This highly processed foods. In women, smoking and physical suggests that staple foods still make important contributions activity had a higher impact on centre ranking, although the to the diet in these countries, although dietary patterns are effect on centre mean intakes of highly processed foods was observed to be changing, with a move away from traditional relatively modest for smoking (up to 2–3% in the Spanish to commercial foods, particularly among younger genera- centres) and for physical activity (p1%). tions (Cruz, 2000; Parizkova, 2000). In contrast, (variants of) the so-called ‘Western’ dietary patterns, characterized by a diet rich in (saturated) fats, red meat, sugary desserts and Discussion refined grains and low in fresh fruits and vegetables, poultry and/or fish, were clearly established in Nordic and central The main objective of our analysis was to investigate how European countries, with relatively modest contributions highly industrially processed foods actually contribute to to total energy intake from highly processed staple foods overall diet and nutrient intakes in middle-aged European (21–29%) and non-processed foods. In Umea˚ (Sweden), for populations, using a unique dataset of detailed, standardized example, more highly processed foods such as crispbread, 24-HDR measurements. This study showed that highly breakfast cereals, cakes, biscuits, margarine, dairy products industrially processed foods dominate the diets and particu- excluding milk, processed meats, alcoholic drinks and soft larly nutrient patterns in Nordic and central European drinks comprise up to 56% of total energy intakes. countries, whereas non-processed and highly processed Currently, insufficient specifically designed epidemiologi- staple foods contribute more to the dietary and nutrient cal and intervention studies have been carried out to draw intakes in southern countries. firm conclusions on the effects of industrially processed For this study, we devoted much effort to developing ad foods on disease risk. Increasingly, however, direct and hoc common definitions and food classifications of processed indirect evidence points to adverse effects of industrially foods to enable dietary comparisons across the 27 centres processed foods on the pandemic of obesity (Swinburn et al., participating in the EPIC study. In this detailed descriptive 2004; Astrup et al., 2008) and of various chronic diseases analysis, the definitions and terminology used for highly such as type II diabetes (van Dam et al., 2002), cardiovascular industrially processed foods, as opposed to raw or moder- disease (Hu et al., 2000; Fung et al., 2004) and cancer (Kesse ately processed foods, were conservative and independent of et al., 2006; Ambrosini et al., 2008; Campbell et al., 2008; any a priori knowledge of diet–disease associations. Only the Chajes et al., 2008). Of particular concern are specific food processes involved and criteria used to discriminate features of highly processed foods such as the fact that they between the different degrees of food processing were are high in energy, fats, sugar and salt and poor in dietary considered. For example, staples such as cereal-based foods fibre (Astrup et al., 2008), that other compounds may be (for example, bread, pasta, white rice), milk and vegetable added or generated during food processing (for example, oils that undergo relatively complex food processing are colourants, additives, acrylamide, trans-fatty acids) (Dybing included among the highly processed foods, although their et al., 2005; Astrup et al., 2008; McCarthy et al., 2008) and consumption tends to be inversely associated with several that they are usually consumed in large portion sizes chronic diseases (Hu and Willett, 2002; Mann, 2007; van (Matthiesen et al., 2003). Trans-fatty acid formation, for Dam and Seidell, 2007). example, results from industrial partial hydrogenation of This study shows that in middle-aged populations from vegetable oils (hardened vegetable oils) (Sommerfeld, 1983). Nordic and central European regions, highly processed These partially hydrogenated oils are consumed in marga- foods provide less than 50% of total intake of only two rine, fast foods and highly processed foods (cakes, rolls, of the nutrients considered in the analysis (vitamin C and confectionery, biscuits, chocolate, potato crips and chips) b-carotene), whereas the figures range from 50 to 91% for the and have been associated with different chronic diseases others (excluding alcohol). Surprisingly, despite large differ- (Mozaffarian et al., 2006; Chajes et al., 2008). Many processed ences in the qualitative and quantitative dietary patterns foods, including foods from major companies, still contain reported elsewhere in the same populations (Slimani et al., high levels of industrially produced trans-fatty acids, despite 2002b), highly industrially processed foods contribute in efforts to reduce them (Aro et al., 1998; McCarthy et al., relatively similar proportions within and between the Nordic 2008). A cross-sectional study on plasma fatty acid levels

European Journal of Clinical Nutrition Highly industrially processed foods in EPIC N Slimani et al S221 among a sub-sample of our study population (N ¼ 3003) processing and cancer (WCRF/AICR, 2007). Only certain shows a high correlation (r ¼ 0.72, Po0.01) between marga- specific domestic or industrial food processing methods (for rines and elaidic acid (trans 18:1 n-9), a specific biomarker example, processed meat preserved by smoking, curing, salting of hardened fats and their related industrial foodstuffs or addition of chemicals, Cantonese-style salted preserved fish, (Saadatian-Elahi et al., 2009). Higher levels of plasma elaidic frying/grilling, barbecuing and refining), known to be asso- acid were observed in northern and central than in southern ciated with certain cancers, were reported. More specific regions, which is compatible with the geographical differ- targeted studies on processed foods should be promoted. This ences in the consumption of highly processed foods reported research should also include gene–gene and gene–environ- in our analysis. ment interaction studies, as different hypotheses suggest that a More research is required to understand the effects of possible explanation for the epidemic of chronic diseases is the industrially processed foods on health. Furthermore, rapid inability of humans to adapt genetically to the major recent dietary changes and the massive introduction of industrial changes in diet (Eaton and Konner, 1985; Baschetti, 1998; foods to diet raise questions on how to measure and monitor Cordain et al., 2005; Ulijaszek, 2007). them properly. These include the following: In conclusion, this study shows that highly industrially (1) The shortcomings of dietary assessment methods processed foods dominate diets and nutrient patterns in used so far in nutritional studies. There is cumulating Nordic and central European countries. The wider range in methodological evidence that traditional food frequency the consumption of highly processed foods observed in questionnaires used in nutritional epidemiology have several southern countries and the greater contribution of non- or limitations for measuring current dietary exposure and moderately processed staple foods may reflect changes in its association with diseases compared with open-ended dietary patterns moving from traditional to more industria- methods such as repeated food records and recalls (Bingham lized diets. However, our study, conducted from 1995 to et al., 2003; Kristal et al., 2005). New approaches that use 2000 in middle-aged populations, most likely underesti- a combination of dietary assessment techniques, including mates the current situation in strictly representative popula- specific biomarkers and calibration, are increasingly recom- tions, particularly among younger and poorer, more mended for estimating dietary exposures (Kaaks et al., 1997; vulnerable populations. In view of the high consumption Subar et al., 2006, Ferrari et al., 2008). of industrialized foods in western European populations, the (2) The difficulty in obtaining reliable information from quality of these foods, particularly in terms of possibly the food industry on the composition of foodstuffs, includ- harmful components, should be better monitored and ing commercial recipes, new products such as food supple- evaluated through concerted actions between policy makers, ments and functional foods, is currently a major limitation public health actors, scientists and the food industry. for estimating and monitoring the consumption of processed foods and their association with diseases. Supplementary information (3) Subsequently, and despite important efforts to con- Contribution of staple versus more complex highly pro- solidate existing data and generate new data on commercial cessed foods for all nutrients, except energy, which is foods (www.eurofir.net), current food composition tables are provided in this paper, is available on the EPIC website inadequate for accurately measuring the levels of exposure to (http://epic.iarc.fr). industrially processed foodstuffs or to bioactive components added or generated during food processing (for example, hardened and other fats, trans-fatty acids, acrylamide, colour- Conflict of interest ants and additives). However, although imprecise, nutrient databases remain essential for identifying the main sources of M Jenab has received grant support from the World Cancer food components of interest and providing relevant public Research Fund. S Bingham has received grant support from health recommendations. The increasing use of specific MRC Centre. The remaining authors have declared no biomarkers of industrial food exposures (for example, to financial interests. trans-fatty acids, acrylamide, colourants, etc. for example) should be privileged and further investigated in addition to, or as a substitute for, dietary exposure measurements. Acknowledgements (4) The lack of specifically designed epidemiological studies to investigate the role of food processing—broadly The work described in this paper was carried out with the defined as including all the processes involved in transform- financial support of the European Commission: Public ing basic ingredients into manufactured foods and beverages Health and Consumer Protection Directorate 1993–2004; (production, processing and preservation methods)—in the Research Directorate-General 2005; Ligue contre le Cancer development of chronic diseases. The limited evidence from (France); Socie´te´ 3M (France); Mutuelle Ge´ne´rale de l’Educa- epidemiological studies was the main reason provided tion Nationale; Institut National de la Sante´ et de la recently by a World Cancer Research Fund panel for not Recherche Me´dicale (INSERM); Institut Gustave Roussy; drawing firm conclusions on the relation between food German Cancer Aid; German Cancer Research Center;

European Journal of Clinical Nutrition Highly industrially processed foods in EPIC N Slimani et al S222 German Federal Ministry of Education and Research; Danish Cordain L, Eaton SB, Sebastian A, Mann N, Lindeberg S, Watkins BA Cancer Society; Health Research Fund (FIS) of the Spanish et al. (2005). Origins and evolution of the Western diet: health Ministry of Health; Spanish Regional Governments of implications for the 21st century. Am J Clin Nutr 81, 341–354. Cruz JA (2000). Dietary habits and nutritional status in adolescents Andalucı´a, Asturias, Basque Country, Murcia and Navarra over Europe–Southern Europe. Eur J Clin Nutr 54(Suppl 1), S29–S35. and the Catalan Institute of Oncology; and ISCIII RETIC Dybing E, Farmer PB, Andersen M, Fennell TR, Lalljie SPD, (RD06/0020), Spain; Cancer Research UK; Medical Research Mu¨ller DJG et al. (2005). Human exposure and internal dose Council, UK; the Stroke Association, UK; British Heart assessments of acrylamide in food. Food Chem Toxicol 43, 365–410. Eaton SB, Konner M (1985). Paleolithic nutrition. A consideration Foundation; Department of Health, UK; Food Standards of its nature and current implications. N Engl J Med 312, 283–289. Agency, UK; the Wellcome Trust, UK; Greek Ministry of Eaton SB, Eaton III SB, Konner MJ (1997). Paleolithic nutrition Health; Hellenic Health Foundation; Italian Association for revisited: a twelve-year retrospective on its nature and implica- Research on Cancer; Italian National Research Council, tions. Eur J Clin Nutr 51, 207–216. Eaton SB (2006). The ancestral human diet: what was it and should it Regione Sicilia (Sicilian government); Associazione Iblea be a paradigm for contemporary nutrition? Proc Nutr Soc 65, 1–6. per la Ricerca Epidemiologica—ONLUS (Hyblean association Ferrari P, Day NE, Boshuizen HC, Roddam A, Hoffmann K, for epidemiological research, NPO); Dutch Ministry of Thiebaut A et al. (2008). The evaluation of the diet/disease relation Health, Welfare and Sport; Dutch Prevention Funds; LK in the EPIC study: considerations for the calibration and the disease models. Int J Epidemiol 37, 368–378. Research Funds; Dutch ZON (Zorg Onderzoek Nederland); Ferrari P, Roddam A, Fahey MT, Jenab M, Bamia C, Ocke´ M et al. World Cancer Research Fund (WCRF); Swedish Cancer (2009). A bivariate measurement error model for nitrogen and Society; Swedish Research Council; Regional Government potassium intakes to evaluate the performance of regression of Skane and the County Council of Vasterbotten, Sweden; calibration in the European Prospective Investigation into Cancer Eur J Clin Nutr Norwegian Cancer Society; the Norwegian Research Council and Nutrition study. 63(Suppl 4), S179–S187. Fung TT, Stampfer MJ, Manson JE, Rexrode KM, Willett WC, Hu FB and the Norwegian Foundation for Health and Rehabilita- (2004). Prospective study of major dietary patterns and stroke risk tion. We thank Sarah Somerville, Nicole Suty and Karima in women. Stroke 35, 2014–2019. Abdedayem for assistance with editing and Kimberley Garcia-Closas R, Berenguer A, Gonzalez CA (2006). Changes in food Bouckaert and Heinz Freisling for technical assistance. supply in Mediterranean countries from 1961 to 2001. Public Health Nutr 9, 53–60. Greenfield G, Southgate DAT (2003). Food Composition Data: Produc- tion, Management and Use. FAO: Rome. Hu FB, Rimm EB, Stampfer MJ, Ascherio A, Spiegelman D, References Willett WC (2000). Prospective study of major dietary patterns and risk of coronary heart disease in men. Am J Clin Nutr 72, 912–921. Al-Delaimy WK, Slimani N, Ferrari P, Key T, Spencer E, Johansson I Hu FB, Willett WC (2002). Optimal diets for prevention of coronary et al. (2005). Plasma carotenoids as biomarkers of intake of fruits heart disease. JAMA 288, 2569–2578. and vegetables: ecological-level correlations in the European Jabs J, Devine CM (2006). Time scarcity and food choices: an Prospective Investigation into Cancer and Nutrition (EPIC). Eur J overview. Appetite 47, 196–204. Clin Nutr 59, 1397–1408. Kaaks R, Riboli E, van Staveren W (1995). Calibration of dietary Ambrosini GL, Fritschi L, de Klerk NH, Mackerras D, Leavy J (2008). intake measurements in prospective cohort studies. Am J Epidemiol Dietary patterns identified using factor analysis and prostate 142, 548–556. cancer risk: a case control study in Western Australia. Ann Kaaks R, Riboli E, Sinha R (1997). Biochemical markers of dietary Epidemiol 18, 364–370. intake. In: P Toniolo, P Boffetta, D Shuker, N Rothman, B Hulka, Aro A, Van Amelsvoort J, Becker W, van Erp-Baart M-A, Kafatos A, Leth T N Pearce (eds) Application of Biomarkers in Cancer Epidemiology. et al. (1998). Transfatty acids in dietary fats and oils from 14 European IARC Sci Publ No 142. International Agency for Research on countries: The TRANSFAIR Study. J Food Comp Anal 11, 137–149. Cancer: Lyon. pp 103–126. Astrup A, Dyerberg J, Selleck M, Stender S (2008). Nutrition Karamanos B, Thanopoulou A, Angelico F, Assaad-Khalil S, Barbato A, transition and its relationship to the development of obesity and Del Ben M et al. (2002). Nutritional habits in the Mediterranean related chronic diseases. Obes Rev 9(Suppl 1), S48–S52. Basin. The macronutrient composition of diet and its relation with Baschetti R (1998). Diabetes epidemic in newly westernized popula- the traditional Mediterranean diet. Multi-centre study of the tions: is it due to thrifty genes or to genetically unknown foods? Mediterranean Group for the Study of Diabetes (MGSD). Eur J Clin J R Soc Med 91, 622–625. Nutr 56, 983–991. Bingham SA, Luben R, Welch A, Wareham N, Khaw KT, Day N (2003). Karmas E, Harris RS (1988). Nutritional Evaluation of Food Processing Are imprecise methods obscuring a relation between fat and breast 3rd edn. Van Nostrand Reinhold: New York. cancer? Lancet 362, 212–214. Kesse E, Clavel-Chapelon F, Boutron-Ruault MC (2006). Dietary patterns Bingham S, Riboli E (2004). Diet and cancer–the European Prospec- and risk of colorectal tumors: a cohort of French women of the tive Investigation into Cancer and Nutrition. Nat Rev Cancer 4, National Education System (E3N). Am J Epidemiol 164, 1085–1093. 206–215. Kious BM (2002). Hunter-gatherer nutrition and its implications Brustad M, Skeie G, Braaten T, Slimani N, Lund E (2003). Comparison for modern societies. eScholarship Repositories, UCLA Biol Chem of telephone vs face-to-face interviews in the assessment of dietary (http://repositories.cdlib.org/uclabiolchem/nutritionnoteworthy/ intake by the 24 h recall EPIC SOFT program–the Norwegian vol5/iss1/art3/). Nutrition Noteworthy 5, Article 3, pp 1–5. calibration study. Eur J Clin Nutr 57, 107–113. Kristal AR, Peters U, Potter JD (2005). Is it time to abandon the food fre- Campbell PT, Sloan M, Kreiger N (2008). Dietary patterns and risk of quency questionnaire? Cancer Epidemiol Biomarkers Prev 14, 2826–2828. incident gastric adenocarcinoma. Am J Epidemiol 167, 295–304. Mann J (2007). Dietary carbohydrate: relationship to cardiovascular Chajes V, Thiebaut AC, Rotival M, Gauthier E, Maillard V, Boutron- disease and disorders of carbohydrate metabolism. Eur J Clin Nutr Ruault MC et al. (2008). Association between serum trans- 61(Suppl 1), S100–S111. monounsaturated fatty acids and breast cancer risk in the Matthiesen J, Fagt S, Biltoft-Jensen A, Beck AM, Ovesen L (2003). E3N-EPIC Study. Am J Epidemiol 167, 1312–1320. Size makes a difference. Public Health Nutr 6, 65–72.

European Journal of Clinical Nutrition Highly industrially processed foods in EPIC N Slimani et al S223 McCarthy J, Barr D, Sinclair A (2008). Determination of trans fatty Slimani N, Bingham S, Runswick S, Ferrari P, Day NE, Welch AA et al. acid levels by FTIR in processed foods in Australia. Asia Pac J Clin (2003). Group level validation of protein intakes estimated by 24-h Nutr 17, 391–396. diet recall and dietary questionnaires against 24-h urinary nitrogen in Mozaffarian D, Katan MB, Ascherio A, Stampfer MJ, Willett WC the European Prospective Investigation into Cancer and Nutrition (2006). Trans fatty acids and cardiovascular disease. N Engl J Med (EPIC) calibration study. Cancer Epidemiol Biomarkers Prev 12, 784–795. 354, 1601–1613. Slimani N, Deharveng G, Unwin I, Southgate DA, Vignat J, Skeie G et al. Parizkova J (2000). Dietary habits and nutritional status in adoles- (2007). The EPIC nutrient database project (ENDB): a first attempt to cents in Central and Eastern Europe. Eur J Clin Nutr 54(Suppl 1), standardize nutrient databases across the 10 European countries S36–S40. participating in the EPIC study. Eur J Clin Nutr 61, 1037–1056. Pomerleau J, Lock K, McKee M (2006). The burden of cardiovascular Sommerfeld M (1983). Trans unsaturated fatty acids in natural disease and cancer attributable to low fruit and vegetable intake in products and processed foods. Prog Lipid Res 22, 221–233. the European Union: differences between old and new member Subar AF, Dodd KW, Guenther PM, Kipnis V, Midthune D, states. Public Health Nutr 9, 575–583. McDowell M et al. (2006). The Food Propensity Questionnaire: Riboli E, Hunt KJ, Slimani N, Ferrari P, Norat T, Fahey M et al. (2002). concept, development, and validation for use as a covariate in a European Prospective Investigation into Cancer and Nutrition model to estimate usual food intake. J Am Diet Assoc 106, 1556–1563. (EPIC): study populations and data collection. Public Health Nutr Swinburn BA, Caterson I, Seidell JC, James WP (2004). Diet, nutrition 5, 1113–1124. and the prevention of excess weight gain and obesity. Public Health Saadatian-Elahi M, Slimani N, Chaje`s V, Jenab M, Goudable J, Nutr 7, 123–146. Biessy C et al (2009). Plasma phospholipid fatty acid profiles and Tooby J, Cosmides L (1990). The past explains the present— their association with food intakes: results from a cross-sectional emotional adaptations and the structure of ancestral environ- study within the European Prospective Investigation into Cancer ments. Ethol Sociobiol 11, 375–424. and Nutrition (EPIC). Am J Clin Nutr 89, 331–346. Tur JA, Romaguera D, Pons A (2004). Food consumption patterns in Schulze MB, Manson JE, Ludwig DS, Colditz GA, Stampfer MJ, a Mediterranean region: does the Mediterranean diet still exist? Willett WC et al. (2004). Sugar-sweetened beverages, weight gain, Ann Nutr Metab 48, 193–201. and incidence of type 2 diabetes in young and middle-aged Ulijaszek SJ (2007). Obesity: a disorder of convenience. Obes Rev women. JAMA 292, 927–934. 8(Suppl 1), S183–S187. Slimani N, Deharveng G, Charrondiere RU, van Kappel AL, Ocke MC, van Dam RM, Rimm EB, Willett WC, Stampfer MJ, Hu FB (2002). Welch A et al. (1999). Structure of the standardized computerized Dietary patterns and risk for type 2 diabetes mellitus in U. S. men. 24-h diet recall interview used as reference method in the Ann Intern Med 136, 201–209. 22 centers participating in the EPIC project. European Prospective van Dam RM, Seidell JC (2007). Carbohydrate intake and obesity. Investigation into Cancer and Nutrition. Comput Methods Programs Eur J Clin Nutr 61(Suppl 1), S75–S99. Biomed 58, 251–266. van Erp-Baart MA, Couet C, Cuadrado C, Kafatos A, Stanley J, van Poppel Slimani N, Kaaks R, Ferrari P, Casagrande C, Clavel-Chapelon F, G (1998). Trans fatty acids in bakery products from 14 European Lotze G et al. (2002a). European Prospective Investigation into countries: The TRANSFAIR Study. J Food Compos Anal 11, 161–169. Cancer and Nutrition (EPIC) calibration study: rationale, design WCRF/AICR (2007). Food, Nutrition, Physical Activity, and the Preven- and population characteristics. Public Health Nutr 5, 1125–1145. tion of Cancer: A Global Perspective. American Institute for Cancer Slimani N, Fahey M, Welch AA, Wirfalt E, Stripp C, Bergstrom E et al. Research/World Cancer Research Fund: Washington DC. (2002b). Diversity of dietary patterns observed in the European WHO/FAO (2003). Diet, Nutrition and the Prevention of Chronic Prospective Investigation into Cancer and Nutrition (EPIC) Diseases. Report of a Joint WHO/FAO Expert Consultation WHO project. Public Health Nutr 5, 1311–1328. Technical Report 916. World Health Organization: Geneva.

Appendix

Table A1 Summary of the terminology and definitions of the classification used for industrially and commercially processed foods and ingredientsa

Food group Highly processed foodsb Moderately processed foodsb Non processed foods,b consumed raw No further cooking Cooked foodsb,c from raw or moderately processed foods

Potatoes and other Process: Drying, flaking, industrial deep frying Examples: Vacuum- Examples: Fresh vacuum- tubers Examples: Potato flakes/powder; commercial packed potato packed or frozen cooked french fries; commercial potato products potato (including homemade French fries) Vegetables, legumes Process: Salting, pickling, concentration, Examples: Vegetables Examples: Fresh or frozen Examples: Fresh raw fermentation, drying, canning in a canned in own juice cooked vegetables; vegetables; fresh raw commercial sauce or in fat or in water/brine; Dried boiled legumes grated vegetables Examples: Onion, roasted (commercial); Legumes canned in own vegetables, dried, in oil; garlic or tomato puree; juice or in water/brine sauerkraut; beans canned in tomato sauce Fruit, including Process: Roasting, coating, use of industrial Examples: Dried or Examples: Fresh fruit, Examples: Raw fruit nuts and olives ingredients, salting semi-dried fruits; compote, boiled; cooked Examples: Olives; peanut butter; nut spreads; walnuts; Brazil-nuts; fruit coated nuts; roasted peanuts; commercial almonds; coconuts; compotes; fruit cocktail in syrup; cashew nuts; pine nuts; seeds pistachio nuts

European Journal of Clinical Nutrition Highly industrially processed foods in EPIC N Slimani et al S224 Table A1 Continued

Food group Highly processed foodsb Moderately processed foodsb Non processed foods,b consumed raw No further cooking Cooked foodsb,c from raw or moderately processed foods

Dairy productsd Process: Heat treatment, drying, Examples: Farmer’s homogenization, concentration/ evaporation, milk, fresh, not cheese-making, use of industrial ingredients, enriched; whole fortification fresh cream Examples: Milk, UHT-treated or pasteurized, enriched, condensed; milk beverages reconstituted from powder; buttermilk beverages; Kefir; yogurt in general, unless homemade; cheese in general; commercial cream dessert; dairy cream, UHT-treated or pasteurized or reconstituted from powder; coffee creamer; thickened milk for coffee

Cereal products, Process: Intense milling, use of industrial Examples: Boiled grain; breadd ingredients, canning in a commercial sauce, wholemeal boiled rice drying, bread making, extrusion, fortification Examples: Starch; flakes; flour; wheat germ; wheat bran; ravioli canned in tomato sauce; pasta, enriched or not, fresh or dried, boiled; cooked couscous; white boiled rice; bread; bread crumbs; cream crackers; crispbread; rusks; breakfast cereals; salty biscuits; popcorn, plain; commercial baked dough

Red meat þ Process: Use of industrial ingredients, Examples: Frozen or Examples: Fresh or Examples: Raw meat poultry þ game salting, smoking, curing, canning in a vacuum-packed raw vacuum-packed commercial sauce or in fat meat cooked meat Examples: Meat canned in gravy

Processed meat Process: Use of industrial ingredients, salting, smoking, curing, fermentation Examples: ham, sausages, bacon, paˆte´

Offal Process: Use of industrial ingredients, Examples: Fresh or salting, smoking, canning in a commercial frozen cooked offal sauce or in fat Examples: Gizzard confit; pigs’ or sheep’s trotters, battered and fried

Fish þ crustaceans, Process: Use of industrial ingredients, Examples: Fresh or molluscs drying, salting, smoking, canning in a frozen cooked fish commercial sauce or in fat Examples: Fish canned in oil, in tomato sauce, in vinegar or pickled; marinated salmon; salted anchovy Fish products, Process: Use of industrial ingredients, fish in crumbsd salting, smoking, curing Examples: Cod roe, crabsticks, fish paˆte´

Egg and egg Process: Drying, pickling Examples: Whole Examples: Raw egg products Examples: Egg powder cooked egg white

Fat Process: Oil extraction and purification, Examples: Virgin olive oil; hydrogenation, butter making, fortification fat from cooked fish/ Examples: Oils; butter; margarine; deep meat; dripping frying fat; cooking fat

Sugar and Process: Sugar extraction and purification, Examples: Honey confectionery use of industrial ingredients, cocoa bean fermentation, roasting and grinding Examples: Sugar; candied fruit/peel; jam; marmalade; chocolate products; confectionery; ice cream; sorbet; syrup

European Journal of Clinical Nutrition Highly industrially processed foods in EPIC N Slimani et al S225 Table A1 Continued

Food group Highly processed foodsb Moderately processed foodsb Non processed foods,b consumed raw No further cooking Cooked foodsb,c from raw or moderately processed foods

Cakes, pies, pastries, Process: Use of industrial ingredients puddings (non milk Examples: Commercial cakes based) þ dry cakes, biscuitsd Non alcoholic Process: Use of industrial ingredients, Examples: Green tea; Examples: Freshly beverages (fruit and fermentation, brewing, roasting, drying, camomile tea prepared fruit juice; vegetable juices, concentration, freeze-drying, pasteurization, tap water; ice cubes soft drinks, diluted fortification syrups coffee, tea Examples: Alcohol-free beer; commercial juice; and herbal teas, commercial juice reconstituted from condensed watere) fruits; soft drinks with artificial sugar (light) or sweetened; diluted syrup; instant and brewed coffee; flavoured tea, ready-to-drink iced tea; powdered tea; black tea; powdered sweetened fruit tea; chicory; mineral water; sparkling water Alcoholic beveragesd Process: Use of industrial ingredients, fermentation, brewing Examples: Wine; beer; cider; spirits; brandy; aniseed drinks (Pastis); liquors Saucesd Process: Use of industrial ingredients, drying, concentration, fermentation Examples: Product name; sauce reconstituted from powder or commercial Yeast, spices, herbs, Process: Use of industrial ingredients, Examples: Dried parsley Examples: Fresh condiments drying, fermentation, fortification parsley Examples: Bouillon cube/powder; salt; yeast; vinegar; spices Soupsd Process: Use of industrial ingredients, drying, concentration Examples: Commercial soup; soup reconstituted from condensed tomatoes; soup reconstituted from powder Bouillonsd Process: Use of industrial ingredients, drying Examples: Broth from a cube Vegetarian products Process: Use of industrial ingredients, drying, and dishesd, artificial heat treatment, concentration, fortification, sweeteners and chemical production, extrusion dietetic products Examples: Commercial vegetarian food; aspartame; saccharine; meal replacements; muesli bars, protein powder Snacksd Process: Use of industrial ingredients Examples: Commercial snacks Non-dairy creams, Process: Chemical production, use of creamers industrial ingredients, drying, concentration Examples: Product name Amphibians Process: Use of industrial ingredients Examples: Fresh cooked Examples: Frogs’ legs, marinated frogs’ legs aTo make the EPIC-Soft 24-h dietary recall data comparable across centres, it was decided to break down all the recipes and compare them at the food and ingredient level. Foods and ingredients of recipes were then classified in three main categories depending on the type of processing undergone, as described in the table shown above. bThe term ‘foods’ refers both to foods and to ingredients broken down from recipes. cThis includes food processed (cooked) at home, in restaurants and in cafeterias. dRecipes were broken down into their ingredients for analysis at the ingredient level, so a homemade cake may end up as 80% ‘highly processed industrial/commercial’ and 20% ‘moderately processed’ (egg) ingredients, and a commercial cake will be treated as 100% ‘highly processed industrial/commercial’ ingredients. The drying process may be considered as moderate, close to the natural process, for some foods such as raisins, legumes, green tea, walnut, parsley, or considered as high for potatoes or when combined with salting, canning in oil, etc. eWater was coded but not included in this analysis.

European Journal of Clinical Nutrition