European Journal of Clinical Nutrition (2015) 69, 579–584 © 2015 Macmillan Publishers Limited All rights reserved 0954-3007/15 www.nature.com/ejcn
ORIGINAL ARTICLE
Development of the HELIUS food frequency questionnaires: ethnic-specific questionnaires to assess the diet of a multiethnic population in The Netherlands MH Beukers1,2, LH Dekker1, EJ de Boer2, CWM Perenboom3, S Meijboom3, M Nicolaou1, JHM de Vries3 and HAM Brants2 OBJECTIVES: Ethnic minorities are often not included in studies of diet and health because of a lack of validated instruments to assess their habitual diets. Given the increased ethnic diversity in many high-income countries, insight into the diets of ethnic minorities is needed for the development of nutritional policies and interventions. In this paper, we describe the development of ethnic-specific food frequency questionnaires (FFQs) to study the diets of Surinamese (African and South Asian), Turkish, Moroccan and ethnic Dutch residents of The Netherlands. METHODS: An existing Dutch FFQ was adapted and formed the basis for three new FFQs. Information on food intake was obtained from single 24 h recalls. Food items were selected according to their percentage contribution to and variance in absolute nutrient intake of the respective ethnic groups. A nutrient database for each FFQ was constructed, consisting of data from the Dutch Food Composition table; data on ethnic foods were based on new chemical analyses and available international data. RESULTS: We developed four ethnic-specific FFQs using a standardised approach that included ~ 200 food items each and that covered more than 90% of the intake of the main nutrients of interest. CONCLUSIONS: The developed FFQs will enable standardised and comparable assessment of the diet of five different ethnic groups and provide insight into the role of diet in differences in health between ethnic groups. The methodology described in this paper and the choices made during the development phase may be useful in developing similar FFQs in other settings. European Journal of Clinical Nutrition (2015) 69, 579–584; doi:10.1038/ejcn.2014.180; published online 17 September 2014
INTRODUCTION Assessing habitual dietary intake is one of the greatest challenges in dietary research, especially in ethnic minority groups. To date, ethnic minorities have been underrepresented in studies on diet and disease. Issues such as language, culture and the relatively low socioeconomic status of ethnic minorities have formed barriers for their inclusion in research. Furthermore, differences in dietary habits, a lack of food composition data on ethnic foods1 as well as a lack of validated dietary assessment tools further complicate the collection of accurate and representative dietary intake data.1,2 European populations have become increasingly ethnically diverse. In The Netherlands, the predominant ethnic minority groups, that is, African-origin Surinamese, South Asian Surinamese, Turkish and Moroccans, form ~ 23% of the population in Amsterdam.3 Diet is an important modifiable risk factor for chronic disease including cardiovascular disease.4 In The Netherlands, the prevalence of cardiovascular disease risk factors is higher among ethnic minorities than in the ethnic Dutch population.5–9 The role of dietary patterns in these differences is unclear. Therefore, it is important to include these groups in monitoring surveys and in studies on diet and health to guide the development of public health strategies to promote healthy diets and reduce health inequalities. Food frequency questionnaires (FFQs) are an acceptable method for assessing usual dietary intake.10 FFQs can provide data on habitual dietary intake, while their administration does
not require highly trained interviewers and may be selfadministered in the participant’s preferred language. A problem is that FFQs should be adapted to the food habits of the target population, which may hamper comparability of collected intake data between different populations. We aimed to develop comparable ethnic-specific FFQs to assess the habitual dietary intake of five ethnic groups living in The Netherlands: Moroccan, Turkish, South Asian-origin Surinamese, African-origin Surinamese and ethnic Dutch. These FFQs are intended to enable research into the role of diet in ethnic differences in chronic disease risk, with a focus on cardiovascular disease. MATERIALS AND METHODS Basis for the FFQs We adapted an existing validated 183-item semiquantitative FFQ developed for the ethnic Dutch group.11 This questionnaire was adjusted to reflect the foods usually consumed by the different ethnic groups. A single FFQ for the two Surinamese groups, African origin and South Asian, was developed because of similarities in commonly eaten foods.12 Nutrients and food groups that have been recognised to have a role in chronic disease risk were the focus of the developed FFQs. In addition, we aimed to measure the intake of nutrients that have previously been identified as potentially inadequate by the ethnic minority groups.13,14 Table 1 provides an overview of the nutrients and food groups prioritised in each of the newly developed FFQs; the original, Dutch, FFQ was adapted to measure the same nutrients and food groups.
1 Department of Public Health, Academic Medical Centre, Amsterdam, The Netherlands; 2Centre for Nutrition, Prevention and Health Services, National Institute for Public Health and the Environment (RIVM), Bilthoven, The Netherlands and 3Division of Human Nutrition, Wageningen University, Wageningen, The Netherlands. Correspondence: Dr M Nicolaou, Academic Medical Centre, University of Amsterdam, PO Box 22660, 1100 Amsterdam, The Netherlands. E-mail: [email protected] Received 21 June 2013; revised 30 June 2014; accepted 18 July 2014; published online 17 September 2014
Food frequency questionnaires for ethnic minorities MH Beukers et al
580 Table 1.
Nutrients and food groups prioritised in the four ethnic-specific FFQs Nutrients
Food groups
Total energy intake Macronutrients Mono- and disaccharides Polysaccharides Dietary fibre Animal protein Vegetable protein Total fat Fatty acid clusters (saturated, monounsaturated, polyunsaturated and trans-fatty acids) Alcohol
Fruits and vegetables Milk and dairy Meat Poultry Fish Soya and other meat substitutes Nuts and seeds
Micronutrients Calcium Iron (haem and non-haem) Vitamin A Vitamin B2 Folate Vitamin B2 Vitamin C Vitamin D Abbreviation: FFQ, food frequency questionnaire.
The food items for the ethnic-specific FFQs were selected on the basis of single 24 h recalls. Data originated from the LASER study, including Moroccan and Turkish participants,15 the SUNSET study, which included African-origin and South Asian Surinamese participants,12 and a study among Moroccan and Turkish women in The Netherlands.16 These studies were conducted between 1998 and 2005 and included, in total, 154 Moroccan, 275 Turkish and 109 Surinamese adults.
Selection of food items The selection of food items to be included in each of the ethnic-specific FFQs was based on two criteria: (1) the percentage of contribution of that food item to absolute nutrient intake and its percentage contribution to the variance in absolute nutrient intake and (2) comparability with the FFQs for other ethnic groups. 1. Foods were selected on their percentage contribution to and variance in absolute nutrient intake of the specific ethnic group based on the methodology described by Molag et al.17 The single foods reported in the 24-h recalls were arranged into food groups at different levels. The 24 food groups in the Dutch Food Composition Table were regarded as level 0, the highest level of aggregation. However, the food groups on this level are too broad to be questioned as a food item in an FFQ. Therefore, these 24 food groups were further subdivided into three hierarchical levels of aggregation. The grouping was based on similarity in the types of food and nutrient composition and represented increased detail in the specification of food groups—for example: level 0, milk products; level 1, yoghurt; level 2, natural yoghurt; and level 3, low-fat natural yoghurt. The reported foods in the 24-h dietary recalls were classified in the food groups at all hierarchical levels.17 Consumption in grams/per day of each food group at levels 1, 2 and 3 and the contribution of these food groups to absolute nutrient intake were calculated. At each level, the food groups were ranked according to their percentage contribution to the absolute intake of the individual nutrients of interest. The set of foods that yielded at least 80% of intake was determined at each level and selected for inclusion in the FFQ. This procedure was repeated for each of the selected nutrients. As suggested, we also performed calculations based on variance of nutrient intake within each ethnic group; these calculations revealed no additional food groups to those already selected for percentage contribution.17 To minimise the FFQ length, food items were grouped at the highest possible hierarchical level. However, we chose to split these groups further if the variation in nutrient content of the different food items within one group was too large or if there were differences in consumption units or how food items are consumed. European Journal of Clinical Nutrition (2015) 579 – 584
2. Comparable food groups were included in each of the ethnic-specific FFQs to maximise their comparability. Thus, food items were considered for selection if they were included in one of the other FFQs. When this was not feasible, highly specific food items—for example, antroewa, a Surinamese vegetable—were included as a single food. Other reasons for inclusion of food items were an increasing trend in consumption in recent years, based on the DNFCS 2007–2010 (Dutch National Food Consumption Survey)18—for example, breakfast drinks, as well as face validity, as detailed in the paragraph pretesting FFQs (see below). Finally, to account for the consumption of foods not included in the FFQ, an open question was included at the end of the questionnaire. Food items added by respondents in this open field are individually evaluated by the researchers and, where appropriate, are included in an existing questionnaire item.
Layout of the FFQs The questionnaire layout was based on the existing Dutch FFQ. Where possible, food items are asked together based on the order that they are usually consumed during the day; thus, the questionnaire begins with foods commonly consumed at breakfast. Ethnic-specific food habits were also accounted for in ordering the questions; for example, in the Turkish FFQ, foods related to breakfast also include questions about sliced raw vegetables and olives, which are consumed as part of a traditional Turkish breakfast. In addition, foods consumed at a particular meal are ordered so that less commonly used foods are mentioned before more common items. For example, a question on different types of crackers precedes the question on bread rolls and croissants, which in turn comes before the question on sliced bread. In this way, we hope to minimise the chance of overreporting, as participants are forced to think about all bread products before the question about sliced bread, which they presumably eat more frequently, is asked. Respondents were asked to report the frequency of consumption over the last 4 weeks in 10 categories, followed by the usual quantity eaten on a consumption day, based on standardised units, household measures or colour photos. For many food items participants were asked to specify the types of foods usually eaten—for example, orange juice or other fruit juice and/or the preparation methods used, for example, boiled or fried. Box 1 illustrates the layout of questions, using soup as an example. The differences in the types of soup eaten per ethnic group are also illustrated here. The language used in all FFQs is Dutch. The Turkish and Moroccan FFQs were also translated into Turkish and Arabic–Moroccan, respectively.
Pretesting of the FFQs The newly developed FFQs were tested to supplement information on food items. Several methods were applied to identify potential problems: (1) expert judgement by 13 dieticians of the specific ethnic origins; (2) 30 plus-minus interviews conducted with participants;19,20 (3) 9 focus groups consisting of 6–12 participants of the same ethnic origin; and (4) cognitive interviews among 7 Moroccan participants. Dieticians from the respective ethnic groups evaluated the FFQs for face validity, specifically to check for missing ethnic-specific food items, inaccurate portion sizes, incorrect spelling or incorrect description of foods. Plus-minus interviews were conducted with potential participants from each ethnic group to identify areas of the FFQs that required improvement. This method requires participants to add a plus or a minus in the margin of the questionnaire while filling it out. A plus indicates a well-understood question, whereas a minus indicates a difficult or unclear question. Consequently, the interviewer probes for information from the participant on the basis of the pluses or minuses they note. In the focus groups, similar issues were covered. In addition, two different options for specifying the types of foods consumed were tested, one using the grid format, in which four answer categories ‘never’ until ‘always’ were included, and one in which the participants could indicate which foods they ‘usually’ ate. Finally, cognitive interviews were conducted using the ‘think aloud’ method.21–23 The latter was applied in testing questions from the Moroccan FFQ, as it was expected that this group would best represent the potential difficulties in filling out the questionnaire. We obtained information on design and comprehension of the questions, portion sizes, food products, frequency of intake, order of food groups and composed meals. Once the FFQs were fully adapted, the original Dutch FFQ underwent a final language check for comprehensibility; this was used as the basis for the formulation of the other FFQs. © 2015 Macmillan Publishers Limited
Food frequency questionnaires for ethnic minorities MH Beukers et al
581 Box 1
Questionnaire layout and illustration of differences in questions between the three ethnic-specific FFQs—question about soup
All: In the past 4 weeks, how often did you eat soup (homemade, ready-made, cup-a-soup and soup based on bouillon)? Please note: plain bouillon does not count. Not in the past 4 weeks
1 day per 2–3 days 4 weeks per 4 weeks
□ ↓ go to …
□
1 day per week
2 days per week
3 days per week
4 days per week
□
□
□
□
□
5 days per 6 days per week week □
7 days per week
□
□
How many bowls of soup have you mostly eaten on a day that you ate soup? Less than ½
½
1
1½
2
2½
3
4
5
6
More than 6
□
□
□
□
□
□
□
□
□
□
□
Dutch:What types of soup you usually eat? (You can check multiple answers) & Soup with legumes, such as pea soup, brown bean soup and lentil soup & Other soups Moroccan:What types of soup you usually eat? (You can check multiple answers) & Soup with legumes, such as lentil soup, harira, pea soup and brown bean soup & Other soups Turkish:What types of soup you usually eat? (You can check multiple answers) & Soup with legumes, such as mercimek (lentil soup), pea soup and brown bean soup & Other soups Surinamese:What types of soup you usually eat? (You can check multiple answers) & Soup with legumes, such as karhi, lentil soup, pea soup and brown bean soup & Soup with root vegetables, such as bravoe, okra soup, cassava soup, bananas griti, tajersoep or soups with tomtom or napi & Peanut soup (peanut bravoe) & Other soups
Nutrient database A nutrient database for each FFQ was constructed on the basis of data in the Dutch Food Composition Table 2011.24 For ethnic-specific products, the database was supplemented with international food composition data, information from food packaging and data from recipe calculations. Six commonly consumed Turkish foods were chemically analysed: two types of yoghurt, cow cheese, goat cheese, Turkish pizza and Turkish brown bread. Multiple samples of each food were mixed for analysis; these samples were based on brand and different sale locations.
Conversion into nutrients Each food item in the questionnaire was linked with one or more foods from the food composition table. For food items consisting of more than a single food, we calculated the nutrient values by weighting the individual foods according to their frequency of consumption (as percentage consumption days). Data from the DNFCS18 were used where appropriate, but, when the contribution of the individual foods to the food item differed from that in the general Dutch population, weighting was based on the ethnic-specific 24-h recall data. Foods contributing o2.5% to the amount of intake of the food item were not included in the composition of the food item, as it was decided that at this level of consumption those food items would have little impact on the overall nutrient intake. In the case of fortified food products, this limit was set at 1%, as these products are likely to have a greater impact on nutrient intake. The choice to use the cutoffs of 2.5 and 1% was subjective. The composite portion size for each food item was also based on a weighted average of the portion sizes of the individual foods, using the same weight factors as for the nutrient values. SAS version 9.2 (SAS Institute Inc., Cary, NC, USA) was used for all calculations.
RESULTS FFQ development The ethnic-specific FFQs range from 189 (Moroccan) to 238 (Dutch) food items. Differences in the number of items are due to inclusion of ethnic-specific food items. Table 2 presents an © 2015 Macmillan Publishers Limited
overview of the main differences in the foods specified between the FFQs. For example, in the Surinamese FFQ, we included questions relating to the intake of roti (a flat bread usually eaten as part of a meal), but as this food is not commonly consumed by the other ethnic groups it was not included in the other FFQs; chocolate milk was included in the Dutch and Surinamese FFQs but not in the Turkish and Moroccan FFQs. There are also differences in the specification of food items; Box 1 illustrates this with an example of differences in the question about soup. Overall, the FFQs cover 490% of the absolute intake of each of the nutrients that are of interest in this study. Table 3 shows the minimum contribution to absolute nutrient intake and the minimum proportion of between-person variance explained by the selected food items in the various ethnic-specific FFQs for each of these nutrients. These calculations were not made for the Dutch FFQ, as it was based on an existing questionnaire that was developed using similar methodology. With regard to absolute intake, the nutrients with the highest contribution are alcohol (100% in all FFQs), haem iron and animal protein. The lowest contribution is 90.3% for polysaccharides in the Turkish FFQ, 92.9% for non-haem iron in the Surinamese FFQ and 93.8% for polysaccharides in the Moroccan FFQ. However, the actual contribution to absolute intake is likely to be somewhat higher than that shown in this table because of the contribution of foods that are not mentioned separately but are part of mixed dishes– for example, flour is an ingredient in bread or cakes. With regard to their contribution to explaining the between-person variation, the nutrients with the highest contribution percentage explained are alcohol and haem iron (100%), the lowest is 81.6% for vitamin A in the Moroccan FFQ, 89.5% for polysaccharides in the Turkish FFQ and 92.5% for fibre in the Surinamese FFQ. The lower values for vitamin A in the Moroccan FFQ and for polysaccharides in the Turkish FFQ are due to carrot juice (rarely consumed) and flour (frequently consumed in 24-h recalls but included in mixed food items like bread and cake in FFQs), respectively. European Journal of Clinical Nutrition (2015) 579 – 584
Food frequency questionnaires for ethnic minorities MH Beukers et al
582 Table 2.
Overview of the main differences in food items included in the four ethnic-specific FFQs Ethnic Dutch
Surinamese
Turkish
Moroccan
Total number of items 238 per FFQ
228
209
189
Items common to some of the FFQs
Gingerbread
Gingerbread
Turkish/Moroccan bread
Turkish/Moroccan bread
Chocolate milk Salty biscuits or cheese biscuits/cheese sticks Cordial/Syrup diluted with water
Chocolate milk Salty biscuits or cheese biscuits/cheese sticks Cordial/Syrup diluted with water
Olive oil to dip bread Olives
Olive oil to dip bread Olives
Roti (flat bread)
Börek andpogaca (pastry-based savouries) Ayran (a yoghurt drink)
Couscous
Items that are unique, Pancakes per FFQ Applesauce Gravy Cream in warm dish, e.g., sour cream Nuts and seeds in warm dish Salad based on mayonnaise, e.g., Russian salad Small toast with cheese, fish, pate, salad, etc. Cheese as an in-between meal snack Sausage as an in-between meal snack
Pom (mixed dish with chicken and root vegetables) Other tubers: pomtajer, cassava, sweet yam, cooking banana Coconut milk/cream Soya sauce Avocado Hindu sweets, e.g., barfi, gullab jamun Fried banana
Bulgur
Rghaif/lemsemmen (Moroccan pancakes) Beghrir (Moroccan donut)
Sarma and Dolma (filled grape leaves) Yogurt sauce Leblebi (dried chickpeas, eaten as a snack) Sunflower seeds (eaten as a snack)
Abbreviation: FFQ, food frequency questionnaire.
Testing phase Results from the testing phase indicated that participants did not always comprehend questions in the FFQ. Common problems included the use of unfamiliar terminology and long phrases. Participants also had difficulties describing portion sizes for items such as couscous and Turkish or Moroccan bread. Input from dieticians and the focus groups gave additional insight into the suitability of portion sizes and missing food items. In testing the two types of questions used to establish the foods eaten within composite food item questions, the grid format was perceived as more difficult by most participants. These findings led to refinement, including addition of descriptions of food items, rephrasing long sentences, adaptation of portion size ranges, extra colour photos and dropping the grid format in favour of multiple answer questions (Box 1). Issues detected by the dieticians, like missing ethnic-specific food items, were discussed by the researchers. The decision to add missing items was made on the basis of the evaluation that the specific food was important in the diet of the group in question. DISCUSSION Using the standardised methodology previously described by Molag et al.,17 we developed three ethnic-specific FFQs and adapted an existing Dutch FFQ to assess the habitual diet of five ethnic groups. The FFQs consist of 185–238 food items and cover 490% of the intake of the nutrients of interest. This discussion focuses on a number of challenges that were faced during the process of development. Food consumption data underlying the FFQs The development of FFQs usually relies on up-to-date food consumption data from the populations for which the questionnaire is intended. The baseline food consumption data used to develop the FFQs described in this paper originated from European Journal of Clinical Nutrition (2015) 579 – 584
relatively small studies that conducted single 24 h recalls. Although the use of single 24 h recalls for this purpose is not unusual, most studies have tended to include larger groups, from 200 to 500 participants25 to 4800.26;27, In the literature, one other study was found that used single 24 h recalls from small numbers of participants to adapt an existing questionnaire.27 However, most studies have compensated for small numbers by using multiple 24 h dietary recalls and food records.28–30 In this study, we compensated for small numbers by extensively testing the questionnaire for completeness and comprehension. However, it remains necessary to regularly collect detailed dietary data of these ethnic groups using an open-ended method such as multiple 24 h recalls. Ideally, the ethnic groups would be included in sufficient numbers in national food consumption surveys, which would provide insight into food consumption patterns and allow updates of the newly developed FFQs. FFQ data are useful in ranking individuals on the basis of their dietary intake and are therefore suitable for epidemiological research of the link between diet and health. However, the closed nature of the data limits its utility as a tool for establishing trends in the food consumption of populations. This is important for the development of strategies to deal with dietary inadequacies or excesses. Selection of food items There are several methods to select food items from existing food consumption data.10 We selected food items on the basis of their percentage contribution to and variances in absolute nutrient intake in a population. Food items can also be selected on the basis of variance while taking covariance between nutrient intakes of food items into account, using forward regression. Because of the small numbers of participants in our underlying data, the data set was less suitable for selection based on regression analysis. In addition, the availability of just a single 24-h dietary recall per participant meant that calculations of variances were unlikely to be optimal. Nonetheless, analyses were performed on variances, © 2015 Macmillan Publishers Limited
Food frequency questionnaires for ethnic minorities MH Beukers et al
583 Table 3. Percent contribution of the FFQs to the % absolute intake of nutrients of interest and to explaining the between-person variation in the intake of these nutrients Contribution to absolute intake of nutrients (%) Contribution to percentage of between-person variation explained (%) Nutrient
Surinamese FFQ
Moroccan FFQ
Turkish FFQ
Surinamese FFQ
Moroccan FFQ
Turkish FFQ
97.8 96.7 97.4 94.8 99.2 96.6 99.2 98.3 98.8 99.2 97.8 100 96.7 99.8 92.9 98.2 97.4 97.3 99.2 98.5 99.0
96.2 95.5 93.8 96.7 99.2 94.6 98.3 97.2 98.6 99.0 97.0 100 97.3 100 95.2 94.4 97.2 97.5 99.1 97.9 98.9
95.1 96.6 90.3 94.5 99.1 91.9 98.1 97.7 98.5 98.5 98.4 100 96.8 99.9 93.4 96.5 95.3 94.9 98.9 96.6 97.0
99.6 97.0 98.3 92.5 99.9 99.2 99.9 99.7 99.9 100 99.9 100 97.7 100 96.4 99.9 99.0 99.6 100 99.9 99.9
96.9 95.2 95.7 98.8 99.9 96.8 98.5 97.0 99.1 99.9 97.8 100 98.9 100 98.3 81.6 98.3 99.3 99.9 98.8 99.9
94.6 97.9 89.5 95.5 99.8 92.0 99.1 98.7 98.8 99.6 99.6 100 97.4 100 95.8 99.9 94.3 97.7 100 99.2 95.9
Total energy intake Mono- and disaccharides Polysaccharides Dietary fibre Animal protein Vegetable protein Total fat Saturated fatty acids Mono-unsaturated fatty acids Poly-unsaturated fatty acids Trans fatty acids Alcohol Calcium Iron haem Iron non-haem Vitamin A Vitamin B2 Folate Vitamin B12 Vitamin C Vitamin D
Abbreviation: FFQ, food frequency questionnaire.
but this did not lead to the selection of additional food items for the FFQs. This may be explained by the fact that selection according to the contribution of absolute intake was performed for a large number of nutrients (van Lemmen, paper submitted for publication). To confirm the suitability of the chosen food items in the Surinamese questionnaire, the available 24-h recall data (2003) were supplemented with data from a more recent study at the municipal health service of The Hague (2007).31 The latter data consisted of 3-day food records of 82 South Asian Surinamese women. Food items that were not selected from the 24-h recall data but were eaten regularly according to the food records were added to the Surinamese FFQ. Unfortunately, there were no such data available to enable a similar process in the development of the Turkish and Moroccan questionnaires. To allow for the reporting of foods not included in the FFQ, an open question relating to this was included at the end of the questionnaire. Comparability FFQs The underlying aim in developing the ethnic-specific FFQs is to conduct research into ethnic differences in the association between diet and health. A simple approach would be to apply a single FFQ in all ethnic groups. However, this option would either result in a very long questionnaire or in a compromise of its face validity. Therefore, we chose to develop separate FFQs per ethnic group and strived to make them as similar as possible to each other by applying rigorous methodology in their development and testing. The resulting four questionnaires have the same layout and consist of similar, comparable food items. However, differences are inevitable because of differences in the diet of the groups studied. Questions relating to mixed dishes One of the difficulties encountered when developing the FFQs was in deciding how to deal with mixed dishes. We chose not to name specific mixed dishes but to ask participants about their intake of individual food components. For instance, the FFQ does © 2015 Macmillan Publishers Limited
not have a question about ‘tajine’, a Moroccan dish consisting of meat, potatoes and vegetables, but in the questions about meat, vegetables and potatoes participants are reminded to include the amounts of these individual items that are eaten in tajine and other mixed dishes. This method was chosen as it was expected that the large variation in ingredients used in such mixed dishes would complicate the use of standard recipes. A possible consequence of this choice may be that participants have problems reporting frequencies and quantities of ingredients that are eaten as part of a mixed dish. This issue was addressed during the testing phase; it was found that asking mixed dishes in this way did not pose a problem. Inclusion of fortified products Fortified products that were eaten in the DNFCS 2007–2010 and contributed 1% or more to the consumption of composite food items were included in the weighting. The most frequently consumed fortified products were dairy products, margarines and nonalcoholic beverages. Assessing the intake of fortified products with an FFQ is difficult: the nutritive value of fortified products varies as fortification level differs between brands and between products within one brand; people often do not know whether they consume fortified products. We therefore chose not to specify these items in the FFQs. As a result, the intake of fortified products will reflect intake at the group level and not at the individual level. Furthermore, as the DNFCS data are based on the general Dutch population, the population level intake may not apply to the ethnic minority groups. This further underscores the need for including ethnic minorities in future national food consumption studies. Future perspective The FFQs described in this paper were primarily designed for the HELIUS-Dietary Patterns study.32,33 In this study, data will be gathered on the habitual intake of participants of ethnic Dutch, African Surinamese, South Asian Surinamese, Turkish and Moroccan origin, aiming for ~ 1000 participants per ethnic group. The European Journal of Clinical Nutrition (2015) 579 – 584
Food frequency questionnaires for ethnic minorities MH Beukers et al
584 ethnic-specific FFQs will be validated using nutritional biomarkers in a subsample of the main study. Fasting blood samples from ~ 200 participants per ethnic group will be analysed for exogenous fatty acid composition and for the major carotenoids found in human plasma. CONCLUSION To our knowledge, this is the first study in Europe that developed four comparable ethnic-specific FFQs to explore the dietary intake of different ethnic groups living in one setting. The developed FFQs will enable standardised and comparable dietary assessment and can be used to provide insight into the role of diet in differences in health between ethnic groups. The methodology described in this paper and the choices made during the development phase may be useful in developing similar FFQs in other settings. CONFLICT OF INTEREST The authors declare no conflict of interest.
ACKNOWLEDGEMENTS This study was funded by The Netherlands Organisation for Health Research and Development, grant number 115100005.
REFERENCES 1 Gilbert PA, Khokhar S. Changing dietary habits of ethnic groups in Europe and implications for health. Nutr Rev 2008; 66: 203–215. 2 Stronks K. Public health research among immigrant populations: still a long way to go. Eur J Epidemiol 2003; 18: 841–842. 3 Research and Statistics. Population on the Basis of Origin. Amsterdam Research and Statistics Service 2012. Available from: http://www.os.amsterdam.nl/grafiek/ 6119/. (accessed 21 June 2013). 4 Mozaffarian D, Wilson PW, Kannel WB. Beyond established and novel risk factors: lifestyle risk factors for cardiovascular disease. Circulation 2008; 117: 3031–3038. 5 Agyemang C, Bindraban N, Mairuhu G, Montfrans G, Koopmans R, Stronks K. Prevalence, awareness, treatment, and control of hypertension among Black Surinamese, South Asian Surinamese and White Dutch in Amsterdam, The Netherlands: the SUNSET study. J Hypertens 2005; 23: 1971–1977. 6 Bos V, Kunst AE, Keij-Deerenberg IM, Garssen J, Mackenbach JP. Ethnic inequalities in age- and cause-specific mortality in The Netherlands. Int J Epidemiol 2004; 33: 1112–1119. 7 Nicolaou M. Diet and Overweight Perception: An Explorative Study among Turkish, Moroccan and Surinamese Migrants Living in the Netherlands. Amsterdam, The Netherlands: Vrije Universiteit, 2009. 8 Uitewaal PJ, Manna DR, Bruijnzeels MA, Hoes AW, Thomas S. Prevalence of type 2 diabetes mellitus, other cardiovascular risk factors, and cardiovascular disease in Turkish and Moroccan immigrants in North West Europe: a systematic review. Prev Med 2004; 39: 1068–1076. 9 Ujcic-Voortman JK, Schram MT, Jacobs-van der Bruggen MA, Verhoeff AP, Baan CA. Diabetes prevalence and risk factors among ethnic minorities. Eur J Public Health 2009; 19: 511–515. 10 Willett W. Nutritional Epidemiology2nd ednNew York, NY: Oxford University Press, 1998. 11 Siebelink E, Geelen A, de Vries JH. Self-reported energy intake by FFQ compared with actual energy intake to maintain body weight in 516 adults. Br J Nutr 2011; 106: 274–281. 12 van Dam RM, Nicolaou M, Stronks K. Study of the diet of Surinamese Amsterdammers. Ned Tijdschr Diet 2005; 60: 98–102.
European Journal of Clinical Nutrition (2015) 579 – 584
13 Ministry of Health WaS. Healthy Diet, from the Begining Till the End. Ministry of Health, Welfare and Sport and Ministry of Agriculture, Nature and Food Quality: The Hague 2008. 14 Health Council of The Netherlands. Towards an Adequate Intake of Vitamins and Minerals. Health Council of The Netherlands: The Hague; April 27 2009. Report No.: 2009/06E 2009. 15 Palsma AH, Nicolaou M, van Dam RM, Stronks K. (De Voeding van Turkse en Marokkaanse Nederlanders in de leeftijd van 18-30 jaar: prioriteiten voor interventies). Tijdschr Gez Wetenschap 2006; 84: 415–422. 16 Brussaard JH, van Erp-Baart MA, Brants HA, Hulshof KF, Lowik MR. Nutrition and health among migrants in The Netherlands. Public Health Nutr 2001; 4: 659–664. 17 Molag ML, de Vries JH, Duif N, Ocke MC, Dagnelie PC, Goldbohm RA et al. Selecting informative food items for compiling food-frequency questionnaires: comparison of procedures. Br J Nutr 2010; 104: 446–456. 18 van Rossum CTM, Fransen HP, Verkaik-Kloosterman J, Buurma-Rethans EJM, Ocke MC. Dutch National Food Composition Survey 2007-2010. Bilthoven, The Netherlands: National Institute for Public Health and the Environment, 2011. 19 de Jong M. Reader Feedback in Text Design. Validity of the Plus-Minus Method for the Pretesting of Public Information Brochures. Amsterdam, Atlanta (GA): Rodopi, 1998. 20 Lentz L, de Jong M. The evaluation of text quality: expert-focused and readerfocused methods compared. IEEE Trans Prof Commun 1997; 40: 224–234. 21 Subar AF, Thompson FE, Smith AF, Jobe JB, Ziegler RG, Potischman N et al. Improving food frequency questionnaires: a qualitative approach using cognitive interviewing. J Am Diet Assoc 1995; 95: 781–788. 22 Caspar RA, Lessler JT, Willis GB. Cognitive interviewing: a 'how to' guide. Reducing survey error through research on the cognitive and decision processes in surveys. Short course presented at the 1999 Meeting of the American Statistical Association–Research Triangle Institute, 1999. Available at http://appliedresearch. cancer.gov/archive/cognitive/interview.pdf. 23 Beatty PC, Willis GB. Research synthesis: the practice of cognitive interviewing. Public Opin Q 2007; 71: 287–311. 24 Dutch Food Composition Table 2011. National Institute for Public Health and the Netherlands Nutrition Centre: The Hague 2011. 25 Shahar D, Shai I, Vardi H, Brener-Azrad A, Fraser D. Development of a semi-quantitative Food Frequency Questionnaire (FFQ) to assess dietary intake of multiethnic populations. Eur J Epidemiol 2003; 18: 855–861. 26 Kim SH, Choi HN, Hwang JY, Chang N, Kim WY, Chung HW et al. Development and evaluation of a food frequency questionnaire for Vietnamese female immigrants in Korea: the Korean Genome and Epidemiology Study (KoGES). Nutr Res Pract 2011; 5: 260–265. 27 Tucker KL, Maras J, Champagne C, Connell C, Goolsby S, Weber J et al. A regional food-frequency questionnaire for the US Mississippi Delta. Public Health Nutr 2005; 8: 87–96. 28 Hankin JH, Stram DO, Arakawa K, Park S, Low SH, Lee HP et al. Singapore Chinese Health Study: development, validation, and calibration of the quantitative food frequency questionnaire. Nutr Cancer 2001; 39: 187–195. 29 Kelemen LE, Anand SS, Vuksan V, Yi Q, Teo KK, Devanesen S et al. Development and evaluation of cultural food frequency questionnaires for South Asians, Chinese, and Europeans in North America. J Am Diet Assoc 2003; 103: 1178–1184. 30 Kolonel LN, Henderson BE, Hankin JH, Nomura AM, Wilkens LR, Pike MC et al. A multiethnic cohort in Hawaii and Los Angeles: baseline characteristics. Am J Epidemiol 2000; 151: 346–357. 31 Karamali N, Tamsma J, de Groot C, Ariens G, Kanbai H, Middelkoop B. The risk of diabetes and heart disease among Hindustanis: Resuilts of the India-study, a study among Hindustani pregnant women and their newborn baby. Epidemiol Bull 2012; 47: 2–12. 32 Dekker LH, Snijder MB, Beukers MH, de Vries JH, Brants HA, de Boer EJ et al. A prospective cohort study of dietary patterns of non-western migrants in the Netherlands in relation to risk factors for cardiovascular diseases: HELIUS-Dietary Patterns. BMC Public Health 2011; 11: 441. 33 Stronks K, Snijder MB, Peters RJ, Prins M, Schene AH, Zwinderman AH. Unravelling the impact of ethnicity on health in Europe: the HELIUS study. BMC Public Health 2013; 13: 402.
© 2015 Macmillan Publishers Limited
Copyright of European Journal of Clinical Nutrition is the property of Nature Publishing Group and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.
ORIGINAL ARTICLE
Development of the HELIUS food frequency questionnaires: ethnic-specific questionnaires to assess the diet of a multiethnic population in The Netherlands MH Beukers1,2, LH Dekker1, EJ de Boer2, CWM Perenboom3, S Meijboom3, M Nicolaou1, JHM de Vries3 and HAM Brants2 OBJECTIVES: Ethnic minorities are often not included in studies of diet and health because of a lack of validated instruments to assess their habitual diets. Given the increased ethnic diversity in many high-income countries, insight into the diets of ethnic minorities is needed for the development of nutritional policies and interventions. In this paper, we describe the development of ethnic-specific food frequency questionnaires (FFQs) to study the diets of Surinamese (African and South Asian), Turkish, Moroccan and ethnic Dutch residents of The Netherlands. METHODS: An existing Dutch FFQ was adapted and formed the basis for three new FFQs. Information on food intake was obtained from single 24 h recalls. Food items were selected according to their percentage contribution to and variance in absolute nutrient intake of the respective ethnic groups. A nutrient database for each FFQ was constructed, consisting of data from the Dutch Food Composition table; data on ethnic foods were based on new chemical analyses and available international data. RESULTS: We developed four ethnic-specific FFQs using a standardised approach that included ~ 200 food items each and that covered more than 90% of the intake of the main nutrients of interest. CONCLUSIONS: The developed FFQs will enable standardised and comparable assessment of the diet of five different ethnic groups and provide insight into the role of diet in differences in health between ethnic groups. The methodology described in this paper and the choices made during the development phase may be useful in developing similar FFQs in other settings. European Journal of Clinical Nutrition (2015) 69, 579–584; doi:10.1038/ejcn.2014.180; published online 17 September 2014
INTRODUCTION Assessing habitual dietary intake is one of the greatest challenges in dietary research, especially in ethnic minority groups. To date, ethnic minorities have been underrepresented in studies on diet and disease. Issues such as language, culture and the relatively low socioeconomic status of ethnic minorities have formed barriers for their inclusion in research. Furthermore, differences in dietary habits, a lack of food composition data on ethnic foods1 as well as a lack of validated dietary assessment tools further complicate the collection of accurate and representative dietary intake data.1,2 European populations have become increasingly ethnically diverse. In The Netherlands, the predominant ethnic minority groups, that is, African-origin Surinamese, South Asian Surinamese, Turkish and Moroccans, form ~ 23% of the population in Amsterdam.3 Diet is an important modifiable risk factor for chronic disease including cardiovascular disease.4 In The Netherlands, the prevalence of cardiovascular disease risk factors is higher among ethnic minorities than in the ethnic Dutch population.5–9 The role of dietary patterns in these differences is unclear. Therefore, it is important to include these groups in monitoring surveys and in studies on diet and health to guide the development of public health strategies to promote healthy diets and reduce health inequalities. Food frequency questionnaires (FFQs) are an acceptable method for assessing usual dietary intake.10 FFQs can provide data on habitual dietary intake, while their administration does
not require highly trained interviewers and may be selfadministered in the participant’s preferred language. A problem is that FFQs should be adapted to the food habits of the target population, which may hamper comparability of collected intake data between different populations. We aimed to develop comparable ethnic-specific FFQs to assess the habitual dietary intake of five ethnic groups living in The Netherlands: Moroccan, Turkish, South Asian-origin Surinamese, African-origin Surinamese and ethnic Dutch. These FFQs are intended to enable research into the role of diet in ethnic differences in chronic disease risk, with a focus on cardiovascular disease. MATERIALS AND METHODS Basis for the FFQs We adapted an existing validated 183-item semiquantitative FFQ developed for the ethnic Dutch group.11 This questionnaire was adjusted to reflect the foods usually consumed by the different ethnic groups. A single FFQ for the two Surinamese groups, African origin and South Asian, was developed because of similarities in commonly eaten foods.12 Nutrients and food groups that have been recognised to have a role in chronic disease risk were the focus of the developed FFQs. In addition, we aimed to measure the intake of nutrients that have previously been identified as potentially inadequate by the ethnic minority groups.13,14 Table 1 provides an overview of the nutrients and food groups prioritised in each of the newly developed FFQs; the original, Dutch, FFQ was adapted to measure the same nutrients and food groups.
1 Department of Public Health, Academic Medical Centre, Amsterdam, The Netherlands; 2Centre for Nutrition, Prevention and Health Services, National Institute for Public Health and the Environment (RIVM), Bilthoven, The Netherlands and 3Division of Human Nutrition, Wageningen University, Wageningen, The Netherlands. Correspondence: Dr M Nicolaou, Academic Medical Centre, University of Amsterdam, PO Box 22660, 1100 Amsterdam, The Netherlands. E-mail: [email protected] Received 21 June 2013; revised 30 June 2014; accepted 18 July 2014; published online 17 September 2014
Food frequency questionnaires for ethnic minorities MH Beukers et al
580 Table 1.
Nutrients and food groups prioritised in the four ethnic-specific FFQs Nutrients
Food groups
Total energy intake Macronutrients Mono- and disaccharides Polysaccharides Dietary fibre Animal protein Vegetable protein Total fat Fatty acid clusters (saturated, monounsaturated, polyunsaturated and trans-fatty acids) Alcohol
Fruits and vegetables Milk and dairy Meat Poultry Fish Soya and other meat substitutes Nuts and seeds
Micronutrients Calcium Iron (haem and non-haem) Vitamin A Vitamin B2 Folate Vitamin B2 Vitamin C Vitamin D Abbreviation: FFQ, food frequency questionnaire.
The food items for the ethnic-specific FFQs were selected on the basis of single 24 h recalls. Data originated from the LASER study, including Moroccan and Turkish participants,15 the SUNSET study, which included African-origin and South Asian Surinamese participants,12 and a study among Moroccan and Turkish women in The Netherlands.16 These studies were conducted between 1998 and 2005 and included, in total, 154 Moroccan, 275 Turkish and 109 Surinamese adults.
Selection of food items The selection of food items to be included in each of the ethnic-specific FFQs was based on two criteria: (1) the percentage of contribution of that food item to absolute nutrient intake and its percentage contribution to the variance in absolute nutrient intake and (2) comparability with the FFQs for other ethnic groups. 1. Foods were selected on their percentage contribution to and variance in absolute nutrient intake of the specific ethnic group based on the methodology described by Molag et al.17 The single foods reported in the 24-h recalls were arranged into food groups at different levels. The 24 food groups in the Dutch Food Composition Table were regarded as level 0, the highest level of aggregation. However, the food groups on this level are too broad to be questioned as a food item in an FFQ. Therefore, these 24 food groups were further subdivided into three hierarchical levels of aggregation. The grouping was based on similarity in the types of food and nutrient composition and represented increased detail in the specification of food groups—for example: level 0, milk products; level 1, yoghurt; level 2, natural yoghurt; and level 3, low-fat natural yoghurt. The reported foods in the 24-h dietary recalls were classified in the food groups at all hierarchical levels.17 Consumption in grams/per day of each food group at levels 1, 2 and 3 and the contribution of these food groups to absolute nutrient intake were calculated. At each level, the food groups were ranked according to their percentage contribution to the absolute intake of the individual nutrients of interest. The set of foods that yielded at least 80% of intake was determined at each level and selected for inclusion in the FFQ. This procedure was repeated for each of the selected nutrients. As suggested, we also performed calculations based on variance of nutrient intake within each ethnic group; these calculations revealed no additional food groups to those already selected for percentage contribution.17 To minimise the FFQ length, food items were grouped at the highest possible hierarchical level. However, we chose to split these groups further if the variation in nutrient content of the different food items within one group was too large or if there were differences in consumption units or how food items are consumed. European Journal of Clinical Nutrition (2015) 579 – 584
2. Comparable food groups were included in each of the ethnic-specific FFQs to maximise their comparability. Thus, food items were considered for selection if they were included in one of the other FFQs. When this was not feasible, highly specific food items—for example, antroewa, a Surinamese vegetable—were included as a single food. Other reasons for inclusion of food items were an increasing trend in consumption in recent years, based on the DNFCS 2007–2010 (Dutch National Food Consumption Survey)18—for example, breakfast drinks, as well as face validity, as detailed in the paragraph pretesting FFQs (see below). Finally, to account for the consumption of foods not included in the FFQ, an open question was included at the end of the questionnaire. Food items added by respondents in this open field are individually evaluated by the researchers and, where appropriate, are included in an existing questionnaire item.
Layout of the FFQs The questionnaire layout was based on the existing Dutch FFQ. Where possible, food items are asked together based on the order that they are usually consumed during the day; thus, the questionnaire begins with foods commonly consumed at breakfast. Ethnic-specific food habits were also accounted for in ordering the questions; for example, in the Turkish FFQ, foods related to breakfast also include questions about sliced raw vegetables and olives, which are consumed as part of a traditional Turkish breakfast. In addition, foods consumed at a particular meal are ordered so that less commonly used foods are mentioned before more common items. For example, a question on different types of crackers precedes the question on bread rolls and croissants, which in turn comes before the question on sliced bread. In this way, we hope to minimise the chance of overreporting, as participants are forced to think about all bread products before the question about sliced bread, which they presumably eat more frequently, is asked. Respondents were asked to report the frequency of consumption over the last 4 weeks in 10 categories, followed by the usual quantity eaten on a consumption day, based on standardised units, household measures or colour photos. For many food items participants were asked to specify the types of foods usually eaten—for example, orange juice or other fruit juice and/or the preparation methods used, for example, boiled or fried. Box 1 illustrates the layout of questions, using soup as an example. The differences in the types of soup eaten per ethnic group are also illustrated here. The language used in all FFQs is Dutch. The Turkish and Moroccan FFQs were also translated into Turkish and Arabic–Moroccan, respectively.
Pretesting of the FFQs The newly developed FFQs were tested to supplement information on food items. Several methods were applied to identify potential problems: (1) expert judgement by 13 dieticians of the specific ethnic origins; (2) 30 plus-minus interviews conducted with participants;19,20 (3) 9 focus groups consisting of 6–12 participants of the same ethnic origin; and (4) cognitive interviews among 7 Moroccan participants. Dieticians from the respective ethnic groups evaluated the FFQs for face validity, specifically to check for missing ethnic-specific food items, inaccurate portion sizes, incorrect spelling or incorrect description of foods. Plus-minus interviews were conducted with potential participants from each ethnic group to identify areas of the FFQs that required improvement. This method requires participants to add a plus or a minus in the margin of the questionnaire while filling it out. A plus indicates a well-understood question, whereas a minus indicates a difficult or unclear question. Consequently, the interviewer probes for information from the participant on the basis of the pluses or minuses they note. In the focus groups, similar issues were covered. In addition, two different options for specifying the types of foods consumed were tested, one using the grid format, in which four answer categories ‘never’ until ‘always’ were included, and one in which the participants could indicate which foods they ‘usually’ ate. Finally, cognitive interviews were conducted using the ‘think aloud’ method.21–23 The latter was applied in testing questions from the Moroccan FFQ, as it was expected that this group would best represent the potential difficulties in filling out the questionnaire. We obtained information on design and comprehension of the questions, portion sizes, food products, frequency of intake, order of food groups and composed meals. Once the FFQs were fully adapted, the original Dutch FFQ underwent a final language check for comprehensibility; this was used as the basis for the formulation of the other FFQs. © 2015 Macmillan Publishers Limited
Food frequency questionnaires for ethnic minorities MH Beukers et al
581 Box 1
Questionnaire layout and illustration of differences in questions between the three ethnic-specific FFQs—question about soup
All: In the past 4 weeks, how often did you eat soup (homemade, ready-made, cup-a-soup and soup based on bouillon)? Please note: plain bouillon does not count. Not in the past 4 weeks
1 day per 2–3 days 4 weeks per 4 weeks
□ ↓ go to …
□
1 day per week
2 days per week
3 days per week
4 days per week
□
□
□
□
□
5 days per 6 days per week week □
7 days per week
□
□
How many bowls of soup have you mostly eaten on a day that you ate soup? Less than ½
½
1
1½
2
2½
3
4
5
6
More than 6
□
□
□
□
□
□
□
□
□
□
□
Dutch:What types of soup you usually eat? (You can check multiple answers) & Soup with legumes, such as pea soup, brown bean soup and lentil soup & Other soups Moroccan:What types of soup you usually eat? (You can check multiple answers) & Soup with legumes, such as lentil soup, harira, pea soup and brown bean soup & Other soups Turkish:What types of soup you usually eat? (You can check multiple answers) & Soup with legumes, such as mercimek (lentil soup), pea soup and brown bean soup & Other soups Surinamese:What types of soup you usually eat? (You can check multiple answers) & Soup with legumes, such as karhi, lentil soup, pea soup and brown bean soup & Soup with root vegetables, such as bravoe, okra soup, cassava soup, bananas griti, tajersoep or soups with tomtom or napi & Peanut soup (peanut bravoe) & Other soups
Nutrient database A nutrient database for each FFQ was constructed on the basis of data in the Dutch Food Composition Table 2011.24 For ethnic-specific products, the database was supplemented with international food composition data, information from food packaging and data from recipe calculations. Six commonly consumed Turkish foods were chemically analysed: two types of yoghurt, cow cheese, goat cheese, Turkish pizza and Turkish brown bread. Multiple samples of each food were mixed for analysis; these samples were based on brand and different sale locations.
Conversion into nutrients Each food item in the questionnaire was linked with one or more foods from the food composition table. For food items consisting of more than a single food, we calculated the nutrient values by weighting the individual foods according to their frequency of consumption (as percentage consumption days). Data from the DNFCS18 were used where appropriate, but, when the contribution of the individual foods to the food item differed from that in the general Dutch population, weighting was based on the ethnic-specific 24-h recall data. Foods contributing o2.5% to the amount of intake of the food item were not included in the composition of the food item, as it was decided that at this level of consumption those food items would have little impact on the overall nutrient intake. In the case of fortified food products, this limit was set at 1%, as these products are likely to have a greater impact on nutrient intake. The choice to use the cutoffs of 2.5 and 1% was subjective. The composite portion size for each food item was also based on a weighted average of the portion sizes of the individual foods, using the same weight factors as for the nutrient values. SAS version 9.2 (SAS Institute Inc., Cary, NC, USA) was used for all calculations.
RESULTS FFQ development The ethnic-specific FFQs range from 189 (Moroccan) to 238 (Dutch) food items. Differences in the number of items are due to inclusion of ethnic-specific food items. Table 2 presents an © 2015 Macmillan Publishers Limited
overview of the main differences in the foods specified between the FFQs. For example, in the Surinamese FFQ, we included questions relating to the intake of roti (a flat bread usually eaten as part of a meal), but as this food is not commonly consumed by the other ethnic groups it was not included in the other FFQs; chocolate milk was included in the Dutch and Surinamese FFQs but not in the Turkish and Moroccan FFQs. There are also differences in the specification of food items; Box 1 illustrates this with an example of differences in the question about soup. Overall, the FFQs cover 490% of the absolute intake of each of the nutrients that are of interest in this study. Table 3 shows the minimum contribution to absolute nutrient intake and the minimum proportion of between-person variance explained by the selected food items in the various ethnic-specific FFQs for each of these nutrients. These calculations were not made for the Dutch FFQ, as it was based on an existing questionnaire that was developed using similar methodology. With regard to absolute intake, the nutrients with the highest contribution are alcohol (100% in all FFQs), haem iron and animal protein. The lowest contribution is 90.3% for polysaccharides in the Turkish FFQ, 92.9% for non-haem iron in the Surinamese FFQ and 93.8% for polysaccharides in the Moroccan FFQ. However, the actual contribution to absolute intake is likely to be somewhat higher than that shown in this table because of the contribution of foods that are not mentioned separately but are part of mixed dishes– for example, flour is an ingredient in bread or cakes. With regard to their contribution to explaining the between-person variation, the nutrients with the highest contribution percentage explained are alcohol and haem iron (100%), the lowest is 81.6% for vitamin A in the Moroccan FFQ, 89.5% for polysaccharides in the Turkish FFQ and 92.5% for fibre in the Surinamese FFQ. The lower values for vitamin A in the Moroccan FFQ and for polysaccharides in the Turkish FFQ are due to carrot juice (rarely consumed) and flour (frequently consumed in 24-h recalls but included in mixed food items like bread and cake in FFQs), respectively. European Journal of Clinical Nutrition (2015) 579 – 584
Food frequency questionnaires for ethnic minorities MH Beukers et al
582 Table 2.
Overview of the main differences in food items included in the four ethnic-specific FFQs Ethnic Dutch
Surinamese
Turkish
Moroccan
Total number of items 238 per FFQ
228
209
189
Items common to some of the FFQs
Gingerbread
Gingerbread
Turkish/Moroccan bread
Turkish/Moroccan bread
Chocolate milk Salty biscuits or cheese biscuits/cheese sticks Cordial/Syrup diluted with water
Chocolate milk Salty biscuits or cheese biscuits/cheese sticks Cordial/Syrup diluted with water
Olive oil to dip bread Olives
Olive oil to dip bread Olives
Roti (flat bread)
Börek andpogaca (pastry-based savouries) Ayran (a yoghurt drink)
Couscous
Items that are unique, Pancakes per FFQ Applesauce Gravy Cream in warm dish, e.g., sour cream Nuts and seeds in warm dish Salad based on mayonnaise, e.g., Russian salad Small toast with cheese, fish, pate, salad, etc. Cheese as an in-between meal snack Sausage as an in-between meal snack
Pom (mixed dish with chicken and root vegetables) Other tubers: pomtajer, cassava, sweet yam, cooking banana Coconut milk/cream Soya sauce Avocado Hindu sweets, e.g., barfi, gullab jamun Fried banana
Bulgur
Rghaif/lemsemmen (Moroccan pancakes) Beghrir (Moroccan donut)
Sarma and Dolma (filled grape leaves) Yogurt sauce Leblebi (dried chickpeas, eaten as a snack) Sunflower seeds (eaten as a snack)
Abbreviation: FFQ, food frequency questionnaire.
Testing phase Results from the testing phase indicated that participants did not always comprehend questions in the FFQ. Common problems included the use of unfamiliar terminology and long phrases. Participants also had difficulties describing portion sizes for items such as couscous and Turkish or Moroccan bread. Input from dieticians and the focus groups gave additional insight into the suitability of portion sizes and missing food items. In testing the two types of questions used to establish the foods eaten within composite food item questions, the grid format was perceived as more difficult by most participants. These findings led to refinement, including addition of descriptions of food items, rephrasing long sentences, adaptation of portion size ranges, extra colour photos and dropping the grid format in favour of multiple answer questions (Box 1). Issues detected by the dieticians, like missing ethnic-specific food items, were discussed by the researchers. The decision to add missing items was made on the basis of the evaluation that the specific food was important in the diet of the group in question. DISCUSSION Using the standardised methodology previously described by Molag et al.,17 we developed three ethnic-specific FFQs and adapted an existing Dutch FFQ to assess the habitual diet of five ethnic groups. The FFQs consist of 185–238 food items and cover 490% of the intake of the nutrients of interest. This discussion focuses on a number of challenges that were faced during the process of development. Food consumption data underlying the FFQs The development of FFQs usually relies on up-to-date food consumption data from the populations for which the questionnaire is intended. The baseline food consumption data used to develop the FFQs described in this paper originated from European Journal of Clinical Nutrition (2015) 579 – 584
relatively small studies that conducted single 24 h recalls. Although the use of single 24 h recalls for this purpose is not unusual, most studies have tended to include larger groups, from 200 to 500 participants25 to 4800.26;27, In the literature, one other study was found that used single 24 h recalls from small numbers of participants to adapt an existing questionnaire.27 However, most studies have compensated for small numbers by using multiple 24 h dietary recalls and food records.28–30 In this study, we compensated for small numbers by extensively testing the questionnaire for completeness and comprehension. However, it remains necessary to regularly collect detailed dietary data of these ethnic groups using an open-ended method such as multiple 24 h recalls. Ideally, the ethnic groups would be included in sufficient numbers in national food consumption surveys, which would provide insight into food consumption patterns and allow updates of the newly developed FFQs. FFQ data are useful in ranking individuals on the basis of their dietary intake and are therefore suitable for epidemiological research of the link between diet and health. However, the closed nature of the data limits its utility as a tool for establishing trends in the food consumption of populations. This is important for the development of strategies to deal with dietary inadequacies or excesses. Selection of food items There are several methods to select food items from existing food consumption data.10 We selected food items on the basis of their percentage contribution to and variances in absolute nutrient intake in a population. Food items can also be selected on the basis of variance while taking covariance between nutrient intakes of food items into account, using forward regression. Because of the small numbers of participants in our underlying data, the data set was less suitable for selection based on regression analysis. In addition, the availability of just a single 24-h dietary recall per participant meant that calculations of variances were unlikely to be optimal. Nonetheless, analyses were performed on variances, © 2015 Macmillan Publishers Limited
Food frequency questionnaires for ethnic minorities MH Beukers et al
583 Table 3. Percent contribution of the FFQs to the % absolute intake of nutrients of interest and to explaining the between-person variation in the intake of these nutrients Contribution to absolute intake of nutrients (%) Contribution to percentage of between-person variation explained (%) Nutrient
Surinamese FFQ
Moroccan FFQ
Turkish FFQ
Surinamese FFQ
Moroccan FFQ
Turkish FFQ
97.8 96.7 97.4 94.8 99.2 96.6 99.2 98.3 98.8 99.2 97.8 100 96.7 99.8 92.9 98.2 97.4 97.3 99.2 98.5 99.0
96.2 95.5 93.8 96.7 99.2 94.6 98.3 97.2 98.6 99.0 97.0 100 97.3 100 95.2 94.4 97.2 97.5 99.1 97.9 98.9
95.1 96.6 90.3 94.5 99.1 91.9 98.1 97.7 98.5 98.5 98.4 100 96.8 99.9 93.4 96.5 95.3 94.9 98.9 96.6 97.0
99.6 97.0 98.3 92.5 99.9 99.2 99.9 99.7 99.9 100 99.9 100 97.7 100 96.4 99.9 99.0 99.6 100 99.9 99.9
96.9 95.2 95.7 98.8 99.9 96.8 98.5 97.0 99.1 99.9 97.8 100 98.9 100 98.3 81.6 98.3 99.3 99.9 98.8 99.9
94.6 97.9 89.5 95.5 99.8 92.0 99.1 98.7 98.8 99.6 99.6 100 97.4 100 95.8 99.9 94.3 97.7 100 99.2 95.9
Total energy intake Mono- and disaccharides Polysaccharides Dietary fibre Animal protein Vegetable protein Total fat Saturated fatty acids Mono-unsaturated fatty acids Poly-unsaturated fatty acids Trans fatty acids Alcohol Calcium Iron haem Iron non-haem Vitamin A Vitamin B2 Folate Vitamin B12 Vitamin C Vitamin D
Abbreviation: FFQ, food frequency questionnaire.
but this did not lead to the selection of additional food items for the FFQs. This may be explained by the fact that selection according to the contribution of absolute intake was performed for a large number of nutrients (van Lemmen, paper submitted for publication). To confirm the suitability of the chosen food items in the Surinamese questionnaire, the available 24-h recall data (2003) were supplemented with data from a more recent study at the municipal health service of The Hague (2007).31 The latter data consisted of 3-day food records of 82 South Asian Surinamese women. Food items that were not selected from the 24-h recall data but were eaten regularly according to the food records were added to the Surinamese FFQ. Unfortunately, there were no such data available to enable a similar process in the development of the Turkish and Moroccan questionnaires. To allow for the reporting of foods not included in the FFQ, an open question relating to this was included at the end of the questionnaire. Comparability FFQs The underlying aim in developing the ethnic-specific FFQs is to conduct research into ethnic differences in the association between diet and health. A simple approach would be to apply a single FFQ in all ethnic groups. However, this option would either result in a very long questionnaire or in a compromise of its face validity. Therefore, we chose to develop separate FFQs per ethnic group and strived to make them as similar as possible to each other by applying rigorous methodology in their development and testing. The resulting four questionnaires have the same layout and consist of similar, comparable food items. However, differences are inevitable because of differences in the diet of the groups studied. Questions relating to mixed dishes One of the difficulties encountered when developing the FFQs was in deciding how to deal with mixed dishes. We chose not to name specific mixed dishes but to ask participants about their intake of individual food components. For instance, the FFQ does © 2015 Macmillan Publishers Limited
not have a question about ‘tajine’, a Moroccan dish consisting of meat, potatoes and vegetables, but in the questions about meat, vegetables and potatoes participants are reminded to include the amounts of these individual items that are eaten in tajine and other mixed dishes. This method was chosen as it was expected that the large variation in ingredients used in such mixed dishes would complicate the use of standard recipes. A possible consequence of this choice may be that participants have problems reporting frequencies and quantities of ingredients that are eaten as part of a mixed dish. This issue was addressed during the testing phase; it was found that asking mixed dishes in this way did not pose a problem. Inclusion of fortified products Fortified products that were eaten in the DNFCS 2007–2010 and contributed 1% or more to the consumption of composite food items were included in the weighting. The most frequently consumed fortified products were dairy products, margarines and nonalcoholic beverages. Assessing the intake of fortified products with an FFQ is difficult: the nutritive value of fortified products varies as fortification level differs between brands and between products within one brand; people often do not know whether they consume fortified products. We therefore chose not to specify these items in the FFQs. As a result, the intake of fortified products will reflect intake at the group level and not at the individual level. Furthermore, as the DNFCS data are based on the general Dutch population, the population level intake may not apply to the ethnic minority groups. This further underscores the need for including ethnic minorities in future national food consumption studies. Future perspective The FFQs described in this paper were primarily designed for the HELIUS-Dietary Patterns study.32,33 In this study, data will be gathered on the habitual intake of participants of ethnic Dutch, African Surinamese, South Asian Surinamese, Turkish and Moroccan origin, aiming for ~ 1000 participants per ethnic group. The European Journal of Clinical Nutrition (2015) 579 – 584
Food frequency questionnaires for ethnic minorities MH Beukers et al
584 ethnic-specific FFQs will be validated using nutritional biomarkers in a subsample of the main study. Fasting blood samples from ~ 200 participants per ethnic group will be analysed for exogenous fatty acid composition and for the major carotenoids found in human plasma. CONCLUSION To our knowledge, this is the first study in Europe that developed four comparable ethnic-specific FFQs to explore the dietary intake of different ethnic groups living in one setting. The developed FFQs will enable standardised and comparable dietary assessment and can be used to provide insight into the role of diet in differences in health between ethnic groups. The methodology described in this paper and the choices made during the development phase may be useful in developing similar FFQs in other settings. CONFLICT OF INTEREST The authors declare no conflict of interest.
ACKNOWLEDGEMENTS This study was funded by The Netherlands Organisation for Health Research and Development, grant number 115100005.
REFERENCES 1 Gilbert PA, Khokhar S. Changing dietary habits of ethnic groups in Europe and implications for health. Nutr Rev 2008; 66: 203–215. 2 Stronks K. Public health research among immigrant populations: still a long way to go. Eur J Epidemiol 2003; 18: 841–842. 3 Research and Statistics. Population on the Basis of Origin. Amsterdam Research and Statistics Service 2012. Available from: http://www.os.amsterdam.nl/grafiek/ 6119/. (accessed 21 June 2013). 4 Mozaffarian D, Wilson PW, Kannel WB. Beyond established and novel risk factors: lifestyle risk factors for cardiovascular disease. Circulation 2008; 117: 3031–3038. 5 Agyemang C, Bindraban N, Mairuhu G, Montfrans G, Koopmans R, Stronks K. Prevalence, awareness, treatment, and control of hypertension among Black Surinamese, South Asian Surinamese and White Dutch in Amsterdam, The Netherlands: the SUNSET study. J Hypertens 2005; 23: 1971–1977. 6 Bos V, Kunst AE, Keij-Deerenberg IM, Garssen J, Mackenbach JP. Ethnic inequalities in age- and cause-specific mortality in The Netherlands. Int J Epidemiol 2004; 33: 1112–1119. 7 Nicolaou M. Diet and Overweight Perception: An Explorative Study among Turkish, Moroccan and Surinamese Migrants Living in the Netherlands. Amsterdam, The Netherlands: Vrije Universiteit, 2009. 8 Uitewaal PJ, Manna DR, Bruijnzeels MA, Hoes AW, Thomas S. Prevalence of type 2 diabetes mellitus, other cardiovascular risk factors, and cardiovascular disease in Turkish and Moroccan immigrants in North West Europe: a systematic review. Prev Med 2004; 39: 1068–1076. 9 Ujcic-Voortman JK, Schram MT, Jacobs-van der Bruggen MA, Verhoeff AP, Baan CA. Diabetes prevalence and risk factors among ethnic minorities. Eur J Public Health 2009; 19: 511–515. 10 Willett W. Nutritional Epidemiology2nd ednNew York, NY: Oxford University Press, 1998. 11 Siebelink E, Geelen A, de Vries JH. Self-reported energy intake by FFQ compared with actual energy intake to maintain body weight in 516 adults. Br J Nutr 2011; 106: 274–281. 12 van Dam RM, Nicolaou M, Stronks K. Study of the diet of Surinamese Amsterdammers. Ned Tijdschr Diet 2005; 60: 98–102.
European Journal of Clinical Nutrition (2015) 579 – 584
13 Ministry of Health WaS. Healthy Diet, from the Begining Till the End. Ministry of Health, Welfare and Sport and Ministry of Agriculture, Nature and Food Quality: The Hague 2008. 14 Health Council of The Netherlands. Towards an Adequate Intake of Vitamins and Minerals. Health Council of The Netherlands: The Hague; April 27 2009. Report No.: 2009/06E 2009. 15 Palsma AH, Nicolaou M, van Dam RM, Stronks K. (De Voeding van Turkse en Marokkaanse Nederlanders in de leeftijd van 18-30 jaar: prioriteiten voor interventies). Tijdschr Gez Wetenschap 2006; 84: 415–422. 16 Brussaard JH, van Erp-Baart MA, Brants HA, Hulshof KF, Lowik MR. Nutrition and health among migrants in The Netherlands. Public Health Nutr 2001; 4: 659–664. 17 Molag ML, de Vries JH, Duif N, Ocke MC, Dagnelie PC, Goldbohm RA et al. Selecting informative food items for compiling food-frequency questionnaires: comparison of procedures. Br J Nutr 2010; 104: 446–456. 18 van Rossum CTM, Fransen HP, Verkaik-Kloosterman J, Buurma-Rethans EJM, Ocke MC. Dutch National Food Composition Survey 2007-2010. Bilthoven, The Netherlands: National Institute for Public Health and the Environment, 2011. 19 de Jong M. Reader Feedback in Text Design. Validity of the Plus-Minus Method for the Pretesting of Public Information Brochures. Amsterdam, Atlanta (GA): Rodopi, 1998. 20 Lentz L, de Jong M. The evaluation of text quality: expert-focused and readerfocused methods compared. IEEE Trans Prof Commun 1997; 40: 224–234. 21 Subar AF, Thompson FE, Smith AF, Jobe JB, Ziegler RG, Potischman N et al. Improving food frequency questionnaires: a qualitative approach using cognitive interviewing. J Am Diet Assoc 1995; 95: 781–788. 22 Caspar RA, Lessler JT, Willis GB. Cognitive interviewing: a 'how to' guide. Reducing survey error through research on the cognitive and decision processes in surveys. Short course presented at the 1999 Meeting of the American Statistical Association–Research Triangle Institute, 1999. Available at http://appliedresearch. cancer.gov/archive/cognitive/interview.pdf. 23 Beatty PC, Willis GB. Research synthesis: the practice of cognitive interviewing. Public Opin Q 2007; 71: 287–311. 24 Dutch Food Composition Table 2011. National Institute for Public Health and the Netherlands Nutrition Centre: The Hague 2011. 25 Shahar D, Shai I, Vardi H, Brener-Azrad A, Fraser D. Development of a semi-quantitative Food Frequency Questionnaire (FFQ) to assess dietary intake of multiethnic populations. Eur J Epidemiol 2003; 18: 855–861. 26 Kim SH, Choi HN, Hwang JY, Chang N, Kim WY, Chung HW et al. Development and evaluation of a food frequency questionnaire for Vietnamese female immigrants in Korea: the Korean Genome and Epidemiology Study (KoGES). Nutr Res Pract 2011; 5: 260–265. 27 Tucker KL, Maras J, Champagne C, Connell C, Goolsby S, Weber J et al. A regional food-frequency questionnaire for the US Mississippi Delta. Public Health Nutr 2005; 8: 87–96. 28 Hankin JH, Stram DO, Arakawa K, Park S, Low SH, Lee HP et al. Singapore Chinese Health Study: development, validation, and calibration of the quantitative food frequency questionnaire. Nutr Cancer 2001; 39: 187–195. 29 Kelemen LE, Anand SS, Vuksan V, Yi Q, Teo KK, Devanesen S et al. Development and evaluation of cultural food frequency questionnaires for South Asians, Chinese, and Europeans in North America. J Am Diet Assoc 2003; 103: 1178–1184. 30 Kolonel LN, Henderson BE, Hankin JH, Nomura AM, Wilkens LR, Pike MC et al. A multiethnic cohort in Hawaii and Los Angeles: baseline characteristics. Am J Epidemiol 2000; 151: 346–357. 31 Karamali N, Tamsma J, de Groot C, Ariens G, Kanbai H, Middelkoop B. The risk of diabetes and heart disease among Hindustanis: Resuilts of the India-study, a study among Hindustani pregnant women and their newborn baby. Epidemiol Bull 2012; 47: 2–12. 32 Dekker LH, Snijder MB, Beukers MH, de Vries JH, Brants HA, de Boer EJ et al. A prospective cohort study of dietary patterns of non-western migrants in the Netherlands in relation to risk factors for cardiovascular diseases: HELIUS-Dietary Patterns. BMC Public Health 2011; 11: 441. 33 Stronks K, Snijder MB, Peters RJ, Prins M, Schene AH, Zwinderman AH. Unravelling the impact of ethnicity on health in Europe: the HELIUS study. BMC Public Health 2013; 13: 402.
© 2015 Macmillan Publishers Limited
Copyright of European Journal of Clinical Nutrition is the property of Nature Publishing Group and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.
