- Email: [email protected]
Evaluation of a Modified Italian European Prospective Investigation into Cancer and Nutrition Food Frequency Questionnaire for Individuals with Celiac Disease
RESEARCH
Original Research: Brief
Evaluation of a Modified Italian European Prospective Investigation into Cancer and Nutrition Food Frequency Questionnaire for Individuals with Celiac Disease Teresa Mazzeo, PhD; Leda Roncoroni, PhD; Vincenza Lombardo, MS; Carolina Tomba, PhD; Luca Elli, PhD; Sabina Sieri, PhD; Sara Grioni; Maria T. Bardella, MD; Carlo Agostoni, MD; Luisa Doneda, PhD; Furio Brighenti, PhD; Nicoletta Pellegrini, PhD ARTICLE INFORMATION Article history: Submitted 26 May 2015 Accepted 20 April 2016 Available online 28 May 2016
Keywords: Individuals with celiac disease Gluten-free diet Validity Dietary assessment Food frequency questionnaire 2212-2672/Copyright ª 2016 by the Academy of Nutrition and Dietetics. http://dx.doi.org/10.1016/j.jand.2016.04.013
ABSTRACT Background To date, it is unclear whether individuals with celiac disease following a gluten-free (GF) diet for several years have adequate intake of all recommended nutrients. Lack of a food frequency questionnaire (FFQ) for individuals with celiac disease could be partly responsible for this still-debated issue. Objective The aim of the study is to evaluate the performance of a modified European Prospective Investigation into Cancer and Nutrition (EPIC) FFQ in estimating nutrient and food intake in a celiac population. Design In a cross-sectional study, the dietary habits of individuals with celiac disease were reported using a modified Italian EPIC FFQ and were compared to a 7-day weighed food record as a reference method. Participants/setting A total of 200 individuals with histologically confirmed celiac disease were enrolled in the study between October 2012 and August 2014 at the Center for Prevention and Diagnosis of Celiac Disease (Milan, Italy). Main outcome measures Nutrient and food category intake were calculated by 7-day weighed food record using an Italian food database integrated with the nutrient composition of 60 GF foods and the modified EPIC FFQ, in which 24 foods were substituted with GF foods comparable for energy and carbohydrate content. Statistical analyses performed An evaluation of the modified FFQ compared to 7-day weighed food record in assessing the reported intake of nutrient and food groups was conducted using Spearman’s correlation coefficients and weighted k. Results One hundred individuals completed the study. The Spearman’s correlation coefficients of FFQ and 7-day weighed food record ranged from .13 to .73 for nutrients and from .23 to .75 for food groups. A moderate agreement, which was defined as a weighted k value of .40 to .60, was obtained for 30% of the analyzed nutrients, and 40% of the nutrients showed values between .30 and .40. The weighted k exceeded .40 for 60% of the 15 analyzed food groups. Conclusions The modified EPIC FFQ demonstrated moderate congruence with a weighed food record in ranking individuals by dietary intakes, particularly food groups. J Acad Nutr Diet. 2016;116:1810-1816.
C
ELIAC DISEASE (CD) IS A CHRONIC IMMUNEmediated disease. It begins with damage to the small bowel mucosa when genetically predisposed individuals with human leukocyte antigen haplotypes DQ2 or DQ8 ingest food containing gluten.1 Damage to the small mucosa can be patchy and progressive, eventually leading to villous atrophy and malabsorption. This condition can be reversible upon elimination of foods containing gluten from the diet.2 The resulting gluten-free (GF) diet excludes many cereal-based staple foods, such as wheat, rye, or barley bread, flour, and pasta, which are important sources of
1810
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
energy, protein, carbohydrate, iron, calcium, niacin, and thiamine.3 Removing these staple foods and their derived products from the diet can impact the nutritional status of individuals with CD.3 Although it is still difficult to draw a conclusion about the nutritional adequacy of a GF diet because of conflicting results reported in available studies, several studies have found that individuals with CD have a different intake of macro- and micronutrients compared with healthy control subjects.3-7 For this reason, assessing the dietary intake of individuals with CD is important for dietary surveillance and treatment of these individuals.
ª 2016 by the Academy of Nutrition and Dietetics.
RESEARCH Several dietary assessment methods are used to quantify both short- and long-term (habitual) dietary intake. They represent essential tools in the area of nutritional epidemiology for assessing the relationship between diet and health in both general living and clinical settings.8 Food records, 24hour recalls, and food frequency questionnaires (FFQ) are the three most common methods used to measure dietary intake.9 The 7-day weighed food record, which involves weighing all foods and drinks consumed during a 7-day period, is widely used in validity studies10,11 and has often been referenced as the “gold standard” against which less detailed and demanding methods can be compared.12,13 However, the 7-day weighed food record is now recognized as having limitations, as it requires a high level of motivation and effort by both participants and researchers. As a consequence, its use is limited to recording dietary habits in individual patients or small groups of individuals. On the other hand, FFQs are retrospective assessment tools that require respondents to report the frequency of consumption of a predefined list of foods over a prolonged period of time, typically the previous 6 or 12 months.14 FFQs are less expensive to process and can be self-administered electronically, making them suitable for online interviews. For these reasons, FFQs are most commonly used in large-scale epidemiologic and intervention studies to determine habitual food and nutrient intake.14 Because they are less accurate than the 7-day weighed food record, FFQs need to be evaluated in the population where they will be applied and for the nutrient(s) of specific interest. This can be done by either measuring specific biological markers of exposure or, more commonly, comparing responses from the FFQ with those derived from a more accurate instrument, such as the weighed food record.14 In the case of individuals with CD, dietary intake has been assessed through the use of either the weighed food record3-7,15-17 or FFQs16,18 in several studies. However, the principal limitations of these studies are, in the case of weighed food records, the small number of individuals with CD investigated and, in the case of FFQs, the use of instruments that have not been evaluated in the celiac population. Therefore, the aim of the present study was to evaluate a specific FFQ, the modified Italian EPIC FFQ, completed in a large number of individuals with CD using the 7-day weighed food record as a reference method.
MATERIALS AND METHODS Participants and Study Design Participants were recruited from among the patients referred to the Center of Prevention and Diagnosis of Celiac Disease at the IRCCS Cà Granda Foundation, Policlinico Hospital, Milan. The exclusion criteria were diagnosis of CD <2 years before, age younger than 18 years or older than 70 years, metabolic or chronic disease (eg, diabetes mellitus, Crohn’s disease, cardiovascular and neurovascular diseases, cancer, neurodegenerative diseases, and rheumatoid arthritis), pregnancy or lactation, or being vegetarian. All individuals were recruited between October 2012 and August 2014 and the data were collected during the same period. In that time period, during their annual medical examination, patients were first screened for the adherence to the GF diet and 1,800 of the approximately 2,500 patients referred to the Center adhered to a GF diet for at November 2016 Volume 116 Number 11
least 2 years. Of those 1,800 patients, 400 were eligible for the study and they were invited to participate. Two hundred individuals with histologically confirmed CD, all adhering to a strict GF diet as confirmed by a negative CD serology (ie, antitransglutaminase IgA antibodies), signed a written informed consent and were enrolled in the study. At the end of their annual medical examination, study participants were interviewed about their usual nutritional intake consumed during the previous year using the modified Italian EPIC FFQ. They also received a 7-day weighed food record. Additional data on age, duration of disease, date of birth, and self-reported anthropometric measures (weight, height) were collected. The local Ethical Committee for Human Research of the City of Milan approved the protocol. The study was registered at ClinicalTrials.gov (ID NCT01975155).
Dietary Records Total food and beverage consumption was assessed by means of a food diary filled out daily for a total of 7 days, as described previously.4 The food diary was a booklet that included columns for the food description, the amount consumed, and the preparation method and recipes. A dietitian trained the participants on how to record all of the food consumed. Participants were asked to weigh all food and drink consumed and to provide a detailed description of each food, including methods of preparation and recipes used, whenever possible. In the case of GF foods, participants were asked to precisely note the name of the manufacturer or to provide the food label. Participants were asked to send their completed 7-day weighed food record to the Department of Food Science of the University of Parma. A dietitian reviewed the diaries and when the dietitian had concerns regarding possible errors or omissions, the participants were contacted by phone to clarify the issues. Nutrient intake was calculated using the Microsoft Access application (version 2003, Microsoft Corp) linked to the European Institute of Oncology’s food database, covering the nutrient composition of >900 Italian foods,19 integrated with the nutrient composition of 60 GF foods present in the Italian market.20 When a food recorded in the 7-day weighed food record was not in the database, an appropriate alternative food was chosen based on similarities in energy and nutrient composition. The output consisted of the mean daily intake of macro- and micronutrients and food items for each subject. Food items consumed were grouped into the following food categories: pasta; breads (including crackers and salty snacks); other cereals (including corn, quinoa, buckwheat, and rice); fruit; vegetables and legumes; meat and preserved meats; dairy products (including milk, yogurt, cream, and cheese); eggs; fish; oils and fats; sweets (including biscuits, sweet snacks, breakfast cereals, ice cream, candies, and chocolate); soft drinks; juices; coffee and tea; and alcoholic beverages. For each subject, the mean daily intake of each food category was then calculated.
FFQ and Evaluation The electronic version of the EPIC FFQ developed for NorthCentral Italy and specifically adapted for the celiac population, which included 188 food items, was used to determine the usual intake of foods and beverages consumed during the previous year.21 The FFQ does not ask about the frequency of JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
1811
RESEARCH intake and dosages of commonly consumed dietary supplements. On the questionnaire, the respondent indicated the number of times a given food item was consumed (per day, week, month, or year). The participants selected an image of a food portion or a predefined standard portion was used when no image was available to quantify portion size. The composition of the food items listed in the EPIC FFQ was modified to include the recipes of composite GF foods and generic GF commercial foods. Complex foods were split into their ingredients and all food ingredients containing gluten were replaced by appropriate GF product. The definition of an appropriate alternative GF food was based on similarities in energy and carbohydrate composition. For the modified EPIC FFQ, 24 foods containing gluten were replaced with 24 GF foods. They were four breads (white for sandwich, white sliced, whole-meal, and with added fats); bread crumbs; focaccia; four pastas (dry, dry with eggs, filled with meat and vegetables, and for broth); pizza with mozzarella and tomatoes; saltine crackers; four breakfast products (plain and chocolate biscuits, melba toast, and muesli); six sweet products (brioches, pastry with cream filling, tiramisu cake, margherita cake, chocolate cake, and chocolate ice cream); sweet dough; and tart filled with vegetables. The interviewer-administered EPIC FFQ was completed during an interview the day before the start of the 7-day weighed food record and a trained researcher interviewed the participants. An ad-hoc computer program (Nutrition Analysis of Food Frequency Questionnaire) developed by the Epidemiology and Prevention Unit of the IRCCS Foundation, National Cancer Institute of Milan, was used to convert the questionnaire’s dietary data into frequencies of consumption and mean daily quantities of foods (grams per day), energy, and nutrients consumed.21 The food items contained in the modified EPIC FFQ were grouped into the same food groups identified for the 7-day weighed food record, based on similarities in nutrient profile and culinary usage.
Data Analysis Food and nutrient intakes derived from the modified EPIC FFQ were compared with those derived from the 7-day weighed food record. The dietary data referred only to foods and beverages consumed and not to supplements. Means and standard deviations were calculated for all
nutrients and food groups for the FFQ and for the 7-day weighed food record. A paired t-test was used to compare the daily nutrient and food group intake reported by the two dietary instruments. To evaluate the degree of association between the rank of modified EPIC FFQ and that of 7-day weighed food record, Spearman’s correlation coefficients were calculated, computed either as unadjusted values or as values adjusted for energy intake. The weighted k statistic22 was used to assess the interquartile agreement between the two dietary instruments. Quadratic weightings were used to decrease small error scores and increase large error scores. Outliers were systematically checked and participants for whom the ratio of total energy intake (determined from the FFQ and the 7-day weighed food record) to basal metabolic rate (determined by the Harris-Benedict equation)23 was in the first or the last percentile of the distribution were excluded in order to reduce the effect of implausible extreme values on the analysis. All of the analyses were performed using the STATA statistical package (StataCorp, 2009, release 11).
RESULTS Of the 200 participants enrolled in the study, 66 communicated to study investigators either orally or in writing that they no longer wanted to participate. Twenty-seven of the remaining 134 participants did not mail back their food records, and 107 completed both the FFQ and the 7-day weighed food record. Of these, seven were excluded from the study because of reported implausible extreme values of the ratio of total energy intake to basal metabolic rate. The final dataset included 100 participants. Mean age, duration of disease, height, weight, and body mass index (BMI; calculated as kg/m2) of the participants with CD who completed the study (21 males, 79 females) are provided in Table 1. Based on the international classification of adult underweight, overweight, and obesity according to BMI,24 16% of the participants were classified as overweight (BMI25), 3% as obese (BMI30), and 12% as underweight (BMI<18.5). In Table 2, the mean daily reported intakes of energy and nutrients calculated from the two dietary methods are presented. The intakes of nutrients estimated from 7-day weighed food record were, in general, higher than those estimated from FFQ, with the exception of sugars, total fat,
Table 1. Descriptive characteristics of 100 individuals with celiac disease who completed a study evaluating a modified Italian European Prospective Investigation into Cancer and Nutrition food frequency questionnaire for individuals with celiac disease Women (n[79)
Men (n[21) Characteristics
Mean–SDa
Range
Mean–SD
Range
Age (y)
41.612.2
18-62
43.313.6
20-70
Duration of disease (y)
14.011.6
2-39
13.110.0
2-37
Weight (kg) Height (cm) BMIb
55-96
57.28.8
39-80
175.16.6
73.810.2
160-193
161.86.7
147-175
24.02.6
19.0-30.4
21.93.3
16.9-32.9
a
SD¼standard deviation. BMI¼body mass index; calculated as kg/m2.
b
1812
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
November 2016 Volume 116 Number 11
RESEARCH Table 2. Daily reported energy and nutrient intake of 100 individuals with celiac disease assessed by a 7-day weighed food record and the modified Italian European Prospective Investigation into Cancer and Nutrition food frequency questionnairea
Variable
7-Day weighed food record (n[100)
Modified EPICb FFQc (n[100)
Paired t test, P value
Spearmand correlation coefficient
Spearmane correlation
Weighted k
meanstandard deviation! Energy (kcal) Protein (g)
<0.001
.37***
—
.40***
78.515.2
70.718.8
<0.001
.34***
.38***
.30**
2,144.0269.4
1,890.7535.7
Total carbohydrate (g)
258.043.1
207.567.4
<0.001
.32**
.33***
.31***
Sugars (g)
100.224.6
96.331.2
0.268
.26**
.29**
.21*
Starch (g)
157.033.9
110.845.6
<0.001
.44***
.39***
.38***
Fiber (g)
23.78.3
20.26.5
<0.001
.45***
.57***
.44***
Total fat (g)
89.516.4
85.925.6
0.192
.21*
.19
.22*
Saturated fats (g)
30.17.3
27.49.2
0.005
.45***
.42***
.38***
Monounsaturated fats (g)
43.69.1
38.512.1
<0.001
.14
.27**
.18*
Polyunsaturated fats (g)
15.64.5
12.24.4
<0.001
.35***
.41***
.36***
287.897.5
282.886.9
0.585
.48***
.37***
.46***
Cholesterol (mg) Alcohol (g) Iron (mg) Calcium (mg)
5.27.6
5.29.1
0.997
.73***
.64***
.74***
11.94.0
9.22.3
<0.001
.36***
.46***
.32***
878.3291.1
824.6346.4
0.071
Potassium (mg)
3,386.5683.4
2,850.4660.6
<0.001
Phosphorus (mg)
1,268.4253.7
1,146.1310.0
.52***
.52***
.49***
.29**
.45***
.31***
<0.001
.44***
.51***
.41***
10.32.0
9.02.4
<0.001
.36***
.36***
.30**
Thiamin (mg)
1.20.3
0.90.2
<0.001
.32**
.46***
.34***
Riboflavin (mg)
1.80.5
1.50.4
<0.001
.55***
.58***
.51***
21.45.2
18.05.1
<0.001
.38***
.39***
.39***
2.40.5
1.90.5
<0.001
.25**
.42***
.24**
Zinc (mg)
Niacin (mg) Vitamin B-6 (mg) Folate (mg) Beta carotene (mg)
326.5111.9
239.566.7
<0.001
.47***
.57***
.45***
4,699.12,505.0
2,629.81,512.3
<0.001
.32**
.41***
.29**
Vitamin C (mg)
143.668.2
122.846.1
Vitamin E (mg)
17.83.7
10.93.8
<0.001
0.002
Vitamin D (mg)
2.61.8
2.81.4
0.349
.44***
.45***
.37***
.13
.22*
.15
.27**
.18
.20*
To compare the daily nutrient and energy intake recorded by the modified EPIC FFQ and the 7-day weighed food record, a paired t test was used. To evaluate the degree of association between the rank of FFQ and that of 7-day weighed food record, Spearman’s correlation coefficients, unadjusted and adjusted for energy intake, were calculated. To assess the interquartile agreement between the two dietary instruments, the weighted k statistic was applied. b EPIC¼European Prospective Investigation into Cancer and Nutrition. c FFQ¼food frequency questionnaire. d Unadjusted values. e Values adjusted for energy. *P<0.05. **P<0.01. ***P<0.001. a
cholesterol, alcohol, calcium, and vitamin D. The Spearman coefficients varied from .13 (vitamin E) to .73 (alcohol). The lowest correlations were reported for the intake of vitamin E and monounsaturated fats, total fat, vitamin B-6, sugars, and vitamin D. Adjustment for energy intake did not consistently improve the correlations. A moderate agreement, which was defined as a weighted k value of .40 to .60, was obtained for 30% of the analyzed November 2016 Volume 116 Number 11
nutrients, and 40% of the nutrients showed values between .30 and .40. The weighted k was not significant only for vitamin E. Mean daily consumption of food groups reported by both dietary tools is presented in Table 3. Compared to 7-day weighed food record, the modified EPIC FFQ underestimated the consumption of other cereals, vegetables and legumes, fruit, and oils and fats, but not that of food groups of JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
1813
RESEARCH Table 3. Daily reported food group intake of 100 individuals with celiac disease assessed by a 7-day weighed food record and the modified Italian European Prospective Investigation into Cancer and Nutrition food frequency questionnairea
Variable
7-Day weighed food record (n[100)
Modified EPICb FFQc (n[100)
Paired t test, P value
Spearman correlation coefficient
Weighted k
meanstandard deviation! Pasta (g)
48.633.1
44.736.0
0.328
.52***
.48***
Breads (g)
90.546.2
82.976.5
0.278
.54***
.50***
47.334.7
24.616.6
<0.001
.31**
.26*
128.965.3
118.158.8
0.165
.25*
.30**
Fish (g)
41.939.7
41.426.1
0.904
.41***
.34***
Eggs (g)
14.814.1
15.410.9
0.664
.36***
.36***
Dairy products (g)
200.6139.9
206.6154.3
0.653
.67***
.58***
Vegetables and legumes (g)
303.0153.8
228.8101.7
<0.001
.50***
.49***
Fruit (g)
261.1138.9
234.9118.5
0.040
.54***
.47***
Sweets (g)
Other cereals (g) Meat and preserved meat (g)
101.846.9
100.158.7
0.765
.60***
.54***
Oils and fats (g)
32.112.5
27.911.9
0.008
.23*
.25**
Juices (g)
35.353.6
46.957.9
0.099
.26**
.25**
Soft drinks (g)
35.355.7
40.081.1
0.532
.48***
.48***
59.893.7
58.3100.5
0.824
.75***
.71***
176.4150.3
160.9111.8
0.183
.73***
.69***
Alcoholic beverages (g) Coffee and teas (g)
a To compare the daily food group intake recorded by the modified EPIC FFQ and the 7-day weighed food record, a paired t test was used. To evaluate the degree of association between the rank of FFQ and that of 7-day weighed food record, Spearman’s correlation coefficients were calculated. To assess the interquartile agreement between the two dietary instruments, the weighted k statistic was applied. b EPIC¼European Prospective Investigation into Cancer and Nutrition. c FFQ¼food frequency questionnaire. *P<0.05. **P<0.01. ***P<0.001.
particular interest in the celiac diet, such as bread, pasta, and sweets. Spearman correlation coefficients (Table 3) for the food categories varied from .23 to .75. The weighted k measured in the food groups exceeded .40 for 60% of the 15 analyzed food groups.
DISCUSSION In the present study, the ability of the modified EPIC FFQ to assess the dietary habits of individuals with CD was evaluated relative to a 7-day weighed food record. Based on the correlation coefficients, the modified EPIC FFQ correctly classified individuals with CD according to the distribution of intake of some nutrients and most food groups. The coefficient values observed in the present study were comparable to those observed in the previous validation study evaluating the FFQ performance in the general population.25 Several studies3,4,15,26 have shown that the dietary habits of individuals with CD do not differ in terms of food choice from those of the general healthy population. However, the celiac population must adhere to a lifelong GF diet in which the gluten-containing foods must be replaced with GF products 1814
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
and other cereals (eg, corn). It has been argued that GF foods tend to be less nutritious than their gluten-containing counterparts.16 Therefore, an evaluation of the FFQ performance in recording the grain-based foods, such as bread, pasta, others cereals, and sweets, is pivotal to the assessment of the nutritional adequacy of the GF diet, for which there is still controversy in the research.20 Mean daily intakes of food groups of particular interest in the celiac diet (ie, bread, pasta, and sweets) reported by the modified EPIC FFQ were in agreement with those reported by the 7-day weighed food record. In fact, significant Spearman correlations and a moderate agreement, indicated by the weighted k values between .48 and .54, were obtained between the two dietary tools. For the “other cereals” food category, the correlation and the weighted k calculated tended to be lower than those reported for bread, pasta, and sweets. This was probably due to a lack of specific questions in the modified EPIC FFQ on dishes including “other cereals” (eg, buckwheat and quinoa), which were occasionally consumed by the participants. As the staple foods and most other major food groups (ie, dairy products and beverages) were satisfactorily recorded by November 2016 Volume 116 Number 11
RESEARCH the FFQ, significant Spearman correlations between the two dietary tools were found for protein, total carbohydrates— including starch and fiber—and polyunsaturated and saturated fats. On the contrary, a low correlation coefficient and weighted k for the intake of monounsaturated fats were observed between the two dietary tools; as a consequence, a low but significant Spearman’s correlation (unadjusted value) and a low but significant weighted k were also observed for total fats. Fat is often added when preparing main dishes for families. Therefore, the amount of fat added to main dishes can be difficult for individuals to estimate, especially if he or she was not involved in the food preparation. As a consequence, some misclassification due to inaccurate estimation of consumed fat by both instruments cannot be ruled out. Because fats are the major source of vitamin E, a very low agreement of intake data between the two dietary tools was also observed for this vitamin. Regarding micronutrients, although the modified EPIC FFQ significantly underestimated their intake, except for calcium and vitamin D, significant correlation coefficients were reported between the dietary tools, especially for micronutrients that are known to be more sensitive for the celiac subjects (ie, iron, calcium, and folate). Some limitations of the present study should be highlighted. Because the modified EPIC FFQ significantly underreported the food groups and the nutrients consumed by the participants, it is not a useful tool to counsel a patient, but it might be more appropriate for population research. In addition, the highly selective nature of respondents that were willing and able to do a weighed record might have introduced a selection bias, possibly reducing the generalizability of the results. However, individuals with CD are known for being health conscious, thus this potential selection bias might have had minimal effects on the present results. Strengths of the present study include the comparison of the modified EPIC FFQ with the gold standard of weighed dietary records and the relatively high number of participants. Furthermore, we used a dietary FFQ that has been used in the Italian EPIC cohort. As the original EPIC FFQ revealed a link between dietary habits and disease risk (ie, colorectal cancer,27 stroke,28 and coronary heart disease29), the modified FFQ is expected to be valuable for studying similar relationships among individuals with CD.
3.
Kinsey L, Burden ST, Bannerman E. A dietary survey to determine if patients with celiac disease are meeting current healthy eating guidelines and how their diet compares to that of the British general population. Eur J Clin Nutr. 2008;62(11):1333-1342.
4.
Dall’Asta C, Scarlato AP, Galaverna G, Brighenti F, Pellegrini N. Dietary exposure to fumonisins and evaluation of nutrient intake in a group of adult celiac patients on a gluten-free diet. Mol Nutr Food Res. 2012;56(4):632-640.
5.
Martin J, Geisel T, Maresch C, Stein KKJ. Inadequate nutrient intake in patients with celiac disease: Results from a German dietary survey. Digestion. 2013;87(4):240-246.
6.
Grehn S, Fridell K, Lilliecreutz M, Hallert C. Dietary habits of Swedish adult coeliac patient treated by a gluten-free diet for 10 years. Scand J Nutr. 2001;45:178-182.
7.
Wild D, Robins GG, Burley VJ, Howdle PD. Evidence of high sugar intake, and low fibre and mineral intake, in the gluten-free diet. Aliment Pharmacol Ther. 2010;32(4):573-581.
8.
Johnson RK. Dietary intake—How do we measure what people are really eating? Obes Res. 2002;10(suppl 1):63S-68S.
9.
Biró G, Hulshof KF, Ovesen L, Amorim Cruz JA. Selection of methodology to assess food intake. Eur J Clin Nutr. 2002;56(2):25-32.
10.
Tabacchi G, Filippi AR, Breda J, et al. Comparative validity of the Asso food frequency questionnaire for the web based assessment of food and nutrients intake in adolescents. Food Nutr Res; 2015. 59-26.
11.
Farukuoye M, Strassburger K, Kacerousky-Bielesz G, Giani G, Roden M. Validity and reproducibility of an intervieweradministered food frequency questionnaire in Austrian adults at risk of or with overt diabetes mellitus. Nutr Res. 2014;34(5): 410-419.
12.
Persson PG, Carlsson S, Grill V, et al. Food frequency questionnaire versus 7-day weighed dietary record information on dietary fibre and fat intake in middle-aged Swedish men. Scand J Soc Med. 1998;26(1):75-80.
13.
Willett WC, Sampson L, Stampfer MJ, et al. Reproducibility and validity of a semiquantitative food frequency questionnaire. Am J Epidemiol. 1985;122(1):51-65.
14.
Willett W. Nutritional Epidemiology. Oxford, UK: Oxford University Press; 2013.
15.
Hopman EGD, Le Cessie S, von Blomberg ME, Mearin ML. Nutritional management of the gluten-free diet in young people with celiac disease in the Netherlands. J Pediatr Gastroenterol Nutr. 2006;43(1): 102-108.
16.
Miranda J, Lasa A, Bustamante MA, Churruca I, Simon E. Nutritional differences between a gluten-free diet and a diet containing equivalent products with gluten. Plant Foods Hum Nutr. 2014;69(2): 182-187.
17.
Shepherd SJ, Gibson PR. Nutritional inadequacies of the gluten-free diet in both recently-diagnosed and long-term patients with coeliac disease. J Hum Nutr Diet. 2013;26(4):349-358.
18.
Black JL, Orfilia C. Impact of coeliac disease on dietary habits and quality of life. J Hum Nutr Diet. 2011;24(6):582-587.
19.
European Institute of Oncology. Food composition database for epidemiological studies in Italy. http://www.ieo.it/bda. Published 2008. Accessed December 1, 2014.
20.
Mazzeo T, Cauzzi S, Brighenti F, Pellegrini N. The development of a composition database of gluten-free products. Public Health Nutr. 2015;18(8):1353-1357.
21.
Pala V, Sieri S, Palli D, et al. Diet in the Italian Epic cohorts: Presentation of data and methodological issues. Tumori. 2003;89(6): 594-607.
22.
Cohen J. Weighted kappa: Nominal scale agreement with provision for scaled disagreement or partial credit. Psychol Bull. 1968;70(4): 213-220.
23.
Harris JA, Benedict FG. A biometric study of human basal metabolism. Proc Natl Acad Sci. 1918;4:370-373.
24.
World Health Organisation. BMI classifications. http://www.who.int/ bmi/index.jsp?introPage¼intro_3.html. Published 2006. Accessed January 15, 2015.
25.
Pisani P, Faggiano F, Krogh V, Palli D, Vineis P, Berrino F. Relative validity and reproducibility of a food frequency dietary questionnaire for use in the Italian EPIC centres. Int J Epidemiol. 1999;26(1): 152-160.
CONCLUSIONS The EPIC FFQ, primarily designed for general populations and then revised with specific products, demonstrated moderate congruence with a weighed food record in ranking individuals by dietary intakes, particularly food groups. Consequently, this modified EPIC FFQ might be a useful instrument for assessing the dietary intake in the Italian population with CD. The present evaluation of this tool will allow for extending research on the dietary habits of individuals with CD from nutritional surveillance to assessing the potential impact of diet on later disease risk.
References 1.
Rubio-Tapia A, Hill ID, Kelly CP, Calderwood AH, Murray JA. ACG clinical guidelines: Diagnosis and management of celiac disease. Am J Gastroenterol. 2013;108(5):656-676.
2.
Farrell RJ, Kelley CP. Celiac sprue. N Engl J Med. 2002;346(3):180-188.
November 2016 Volume 116 Number 11
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
1815
RESEARCH 26.
Hallert C, Grant C, Grehn S, et al. Evidence of poor vitamin status in coeliac patients on a gluten-free diet for 10 years. Aliment Pharmacol Ther. 2002;16(7):1333-1339.
28.
Agnoli C, Krogh V, Grioni S, et al. A priori-defined dietary patterns are associated with reduced risk of stroke in a large Italian cohort. J Nutr. 2011;141(8):1552-1558.
27.
Agnoli C, Grioni S, Sieri S, et al. Italian Mediterranean Index and risk of colorectal cancer in the Italian section of the EPIC cohort. Int J Cancer. 2013;132(6):1404-1411.
29.
Bendinelli B, Masala G, Saieva C, et al. Fruit, vegetables, and olive oil and risk of coronary heart disease in Italian women: The EPICOR Study. Am J Clin Nutr. 2011;93(2):275-283.
AUTHOR INFORMATION T. Mazzeo is a postdoctoral researcher, F. Brighenti is a full professor, and N. Pellegrini is an associate professor, Human Nutrition Unit, Department of Food Science, University of Parma, Parma, Italy. L. Roncoroni is a researcher and nutritionist, Center for Prevention and Diagnosis of Celiac Disease, IRCCS Cà Granda Foundation, Ospedale Maggiore Policlinico, Milan, Italy, and Department of Biomedical, Surgical and Dental Sciences, University of Milan, Milan, Italy. V. Lombardo is a scholarship holder, C. Tomba is a researcher in gastroenterology, and L. Elli is a gastroenterologist, Center for Prevention and Diagnosis of Celiac Disease, and C. Agostoni is a full professor of pediatrics, Department of Clinical Sciences and Community Health, all at the University of Milan, Fondazione IRCCS Cà Granda Foundation, Ospedale Maggiore Policlinico, Milan, Italy. S. Sieri is a senior researcher in nutritional epidemiology and S. Grioni is a research associate in nutritional epidemiology, Epidemiology and Prevention Unit, Fondazione IRCCS Istituto Nazionale dei Tumori, Milan, Italy. M. T. Bardella is a consultant gastroenterologist, Center for Prevention and Diagnosis of Celiac Disease, Gastroenterology and Endoscopy Unit, IRCCS Cà Granda Foundation, Ospedale Maggiore Policlinico, Milan, Italy. L. Doneda is an associate professor in applied biology, Department of Biomedical, Surgical and Dental Sciences, University of Milan, Milan, Italy. Address correspondence to: Nicoletta Pellegrini, PhD, Department of Food Science, University of Parma, Parco Area delle Scienze, 47/A, 43124 Parma, Italy. E-mail: [email protected]
STATEMENT OF POTENTIAL CONFLICT OF INTEREST No potential conflict of interest was reported by the authors.
FUNDING/SUPPORT This work has been partly supported by Dr. Schär GmbH/Srl. Dr. Schär GmbH/Srl had no role in the design, analysis, and writing of this article. The study was registered at ClinicalTrials.gov (ID NCT01975155).
1816
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
November 2016 Volume 116 Number 11
Original Research: Brief
Evaluation of a Modified Italian European Prospective Investigation into Cancer and Nutrition Food Frequency Questionnaire for Individuals with Celiac Disease Teresa Mazzeo, PhD; Leda Roncoroni, PhD; Vincenza Lombardo, MS; Carolina Tomba, PhD; Luca Elli, PhD; Sabina Sieri, PhD; Sara Grioni; Maria T. Bardella, MD; Carlo Agostoni, MD; Luisa Doneda, PhD; Furio Brighenti, PhD; Nicoletta Pellegrini, PhD ARTICLE INFORMATION Article history: Submitted 26 May 2015 Accepted 20 April 2016 Available online 28 May 2016
Keywords: Individuals with celiac disease Gluten-free diet Validity Dietary assessment Food frequency questionnaire 2212-2672/Copyright ª 2016 by the Academy of Nutrition and Dietetics. http://dx.doi.org/10.1016/j.jand.2016.04.013
ABSTRACT Background To date, it is unclear whether individuals with celiac disease following a gluten-free (GF) diet for several years have adequate intake of all recommended nutrients. Lack of a food frequency questionnaire (FFQ) for individuals with celiac disease could be partly responsible for this still-debated issue. Objective The aim of the study is to evaluate the performance of a modified European Prospective Investigation into Cancer and Nutrition (EPIC) FFQ in estimating nutrient and food intake in a celiac population. Design In a cross-sectional study, the dietary habits of individuals with celiac disease were reported using a modified Italian EPIC FFQ and were compared to a 7-day weighed food record as a reference method. Participants/setting A total of 200 individuals with histologically confirmed celiac disease were enrolled in the study between October 2012 and August 2014 at the Center for Prevention and Diagnosis of Celiac Disease (Milan, Italy). Main outcome measures Nutrient and food category intake were calculated by 7-day weighed food record using an Italian food database integrated with the nutrient composition of 60 GF foods and the modified EPIC FFQ, in which 24 foods were substituted with GF foods comparable for energy and carbohydrate content. Statistical analyses performed An evaluation of the modified FFQ compared to 7-day weighed food record in assessing the reported intake of nutrient and food groups was conducted using Spearman’s correlation coefficients and weighted k. Results One hundred individuals completed the study. The Spearman’s correlation coefficients of FFQ and 7-day weighed food record ranged from .13 to .73 for nutrients and from .23 to .75 for food groups. A moderate agreement, which was defined as a weighted k value of .40 to .60, was obtained for 30% of the analyzed nutrients, and 40% of the nutrients showed values between .30 and .40. The weighted k exceeded .40 for 60% of the 15 analyzed food groups. Conclusions The modified EPIC FFQ demonstrated moderate congruence with a weighed food record in ranking individuals by dietary intakes, particularly food groups. J Acad Nutr Diet. 2016;116:1810-1816.
C
ELIAC DISEASE (CD) IS A CHRONIC IMMUNEmediated disease. It begins with damage to the small bowel mucosa when genetically predisposed individuals with human leukocyte antigen haplotypes DQ2 or DQ8 ingest food containing gluten.1 Damage to the small mucosa can be patchy and progressive, eventually leading to villous atrophy and malabsorption. This condition can be reversible upon elimination of foods containing gluten from the diet.2 The resulting gluten-free (GF) diet excludes many cereal-based staple foods, such as wheat, rye, or barley bread, flour, and pasta, which are important sources of
1810
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
energy, protein, carbohydrate, iron, calcium, niacin, and thiamine.3 Removing these staple foods and their derived products from the diet can impact the nutritional status of individuals with CD.3 Although it is still difficult to draw a conclusion about the nutritional adequacy of a GF diet because of conflicting results reported in available studies, several studies have found that individuals with CD have a different intake of macro- and micronutrients compared with healthy control subjects.3-7 For this reason, assessing the dietary intake of individuals with CD is important for dietary surveillance and treatment of these individuals.
ª 2016 by the Academy of Nutrition and Dietetics.
RESEARCH Several dietary assessment methods are used to quantify both short- and long-term (habitual) dietary intake. They represent essential tools in the area of nutritional epidemiology for assessing the relationship between diet and health in both general living and clinical settings.8 Food records, 24hour recalls, and food frequency questionnaires (FFQ) are the three most common methods used to measure dietary intake.9 The 7-day weighed food record, which involves weighing all foods and drinks consumed during a 7-day period, is widely used in validity studies10,11 and has often been referenced as the “gold standard” against which less detailed and demanding methods can be compared.12,13 However, the 7-day weighed food record is now recognized as having limitations, as it requires a high level of motivation and effort by both participants and researchers. As a consequence, its use is limited to recording dietary habits in individual patients or small groups of individuals. On the other hand, FFQs are retrospective assessment tools that require respondents to report the frequency of consumption of a predefined list of foods over a prolonged period of time, typically the previous 6 or 12 months.14 FFQs are less expensive to process and can be self-administered electronically, making them suitable for online interviews. For these reasons, FFQs are most commonly used in large-scale epidemiologic and intervention studies to determine habitual food and nutrient intake.14 Because they are less accurate than the 7-day weighed food record, FFQs need to be evaluated in the population where they will be applied and for the nutrient(s) of specific interest. This can be done by either measuring specific biological markers of exposure or, more commonly, comparing responses from the FFQ with those derived from a more accurate instrument, such as the weighed food record.14 In the case of individuals with CD, dietary intake has been assessed through the use of either the weighed food record3-7,15-17 or FFQs16,18 in several studies. However, the principal limitations of these studies are, in the case of weighed food records, the small number of individuals with CD investigated and, in the case of FFQs, the use of instruments that have not been evaluated in the celiac population. Therefore, the aim of the present study was to evaluate a specific FFQ, the modified Italian EPIC FFQ, completed in a large number of individuals with CD using the 7-day weighed food record as a reference method.
MATERIALS AND METHODS Participants and Study Design Participants were recruited from among the patients referred to the Center of Prevention and Diagnosis of Celiac Disease at the IRCCS Cà Granda Foundation, Policlinico Hospital, Milan. The exclusion criteria were diagnosis of CD <2 years before, age younger than 18 years or older than 70 years, metabolic or chronic disease (eg, diabetes mellitus, Crohn’s disease, cardiovascular and neurovascular diseases, cancer, neurodegenerative diseases, and rheumatoid arthritis), pregnancy or lactation, or being vegetarian. All individuals were recruited between October 2012 and August 2014 and the data were collected during the same period. In that time period, during their annual medical examination, patients were first screened for the adherence to the GF diet and 1,800 of the approximately 2,500 patients referred to the Center adhered to a GF diet for at November 2016 Volume 116 Number 11
least 2 years. Of those 1,800 patients, 400 were eligible for the study and they were invited to participate. Two hundred individuals with histologically confirmed CD, all adhering to a strict GF diet as confirmed by a negative CD serology (ie, antitransglutaminase IgA antibodies), signed a written informed consent and were enrolled in the study. At the end of their annual medical examination, study participants were interviewed about their usual nutritional intake consumed during the previous year using the modified Italian EPIC FFQ. They also received a 7-day weighed food record. Additional data on age, duration of disease, date of birth, and self-reported anthropometric measures (weight, height) were collected. The local Ethical Committee for Human Research of the City of Milan approved the protocol. The study was registered at ClinicalTrials.gov (ID NCT01975155).
Dietary Records Total food and beverage consumption was assessed by means of a food diary filled out daily for a total of 7 days, as described previously.4 The food diary was a booklet that included columns for the food description, the amount consumed, and the preparation method and recipes. A dietitian trained the participants on how to record all of the food consumed. Participants were asked to weigh all food and drink consumed and to provide a detailed description of each food, including methods of preparation and recipes used, whenever possible. In the case of GF foods, participants were asked to precisely note the name of the manufacturer or to provide the food label. Participants were asked to send their completed 7-day weighed food record to the Department of Food Science of the University of Parma. A dietitian reviewed the diaries and when the dietitian had concerns regarding possible errors or omissions, the participants were contacted by phone to clarify the issues. Nutrient intake was calculated using the Microsoft Access application (version 2003, Microsoft Corp) linked to the European Institute of Oncology’s food database, covering the nutrient composition of >900 Italian foods,19 integrated with the nutrient composition of 60 GF foods present in the Italian market.20 When a food recorded in the 7-day weighed food record was not in the database, an appropriate alternative food was chosen based on similarities in energy and nutrient composition. The output consisted of the mean daily intake of macro- and micronutrients and food items for each subject. Food items consumed were grouped into the following food categories: pasta; breads (including crackers and salty snacks); other cereals (including corn, quinoa, buckwheat, and rice); fruit; vegetables and legumes; meat and preserved meats; dairy products (including milk, yogurt, cream, and cheese); eggs; fish; oils and fats; sweets (including biscuits, sweet snacks, breakfast cereals, ice cream, candies, and chocolate); soft drinks; juices; coffee and tea; and alcoholic beverages. For each subject, the mean daily intake of each food category was then calculated.
FFQ and Evaluation The electronic version of the EPIC FFQ developed for NorthCentral Italy and specifically adapted for the celiac population, which included 188 food items, was used to determine the usual intake of foods and beverages consumed during the previous year.21 The FFQ does not ask about the frequency of JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
1811
RESEARCH intake and dosages of commonly consumed dietary supplements. On the questionnaire, the respondent indicated the number of times a given food item was consumed (per day, week, month, or year). The participants selected an image of a food portion or a predefined standard portion was used when no image was available to quantify portion size. The composition of the food items listed in the EPIC FFQ was modified to include the recipes of composite GF foods and generic GF commercial foods. Complex foods were split into their ingredients and all food ingredients containing gluten were replaced by appropriate GF product. The definition of an appropriate alternative GF food was based on similarities in energy and carbohydrate composition. For the modified EPIC FFQ, 24 foods containing gluten were replaced with 24 GF foods. They were four breads (white for sandwich, white sliced, whole-meal, and with added fats); bread crumbs; focaccia; four pastas (dry, dry with eggs, filled with meat and vegetables, and for broth); pizza with mozzarella and tomatoes; saltine crackers; four breakfast products (plain and chocolate biscuits, melba toast, and muesli); six sweet products (brioches, pastry with cream filling, tiramisu cake, margherita cake, chocolate cake, and chocolate ice cream); sweet dough; and tart filled with vegetables. The interviewer-administered EPIC FFQ was completed during an interview the day before the start of the 7-day weighed food record and a trained researcher interviewed the participants. An ad-hoc computer program (Nutrition Analysis of Food Frequency Questionnaire) developed by the Epidemiology and Prevention Unit of the IRCCS Foundation, National Cancer Institute of Milan, was used to convert the questionnaire’s dietary data into frequencies of consumption and mean daily quantities of foods (grams per day), energy, and nutrients consumed.21 The food items contained in the modified EPIC FFQ were grouped into the same food groups identified for the 7-day weighed food record, based on similarities in nutrient profile and culinary usage.
Data Analysis Food and nutrient intakes derived from the modified EPIC FFQ were compared with those derived from the 7-day weighed food record. The dietary data referred only to foods and beverages consumed and not to supplements. Means and standard deviations were calculated for all
nutrients and food groups for the FFQ and for the 7-day weighed food record. A paired t-test was used to compare the daily nutrient and food group intake reported by the two dietary instruments. To evaluate the degree of association between the rank of modified EPIC FFQ and that of 7-day weighed food record, Spearman’s correlation coefficients were calculated, computed either as unadjusted values or as values adjusted for energy intake. The weighted k statistic22 was used to assess the interquartile agreement between the two dietary instruments. Quadratic weightings were used to decrease small error scores and increase large error scores. Outliers were systematically checked and participants for whom the ratio of total energy intake (determined from the FFQ and the 7-day weighed food record) to basal metabolic rate (determined by the Harris-Benedict equation)23 was in the first or the last percentile of the distribution were excluded in order to reduce the effect of implausible extreme values on the analysis. All of the analyses were performed using the STATA statistical package (StataCorp, 2009, release 11).
RESULTS Of the 200 participants enrolled in the study, 66 communicated to study investigators either orally or in writing that they no longer wanted to participate. Twenty-seven of the remaining 134 participants did not mail back their food records, and 107 completed both the FFQ and the 7-day weighed food record. Of these, seven were excluded from the study because of reported implausible extreme values of the ratio of total energy intake to basal metabolic rate. The final dataset included 100 participants. Mean age, duration of disease, height, weight, and body mass index (BMI; calculated as kg/m2) of the participants with CD who completed the study (21 males, 79 females) are provided in Table 1. Based on the international classification of adult underweight, overweight, and obesity according to BMI,24 16% of the participants were classified as overweight (BMI25), 3% as obese (BMI30), and 12% as underweight (BMI<18.5). In Table 2, the mean daily reported intakes of energy and nutrients calculated from the two dietary methods are presented. The intakes of nutrients estimated from 7-day weighed food record were, in general, higher than those estimated from FFQ, with the exception of sugars, total fat,
Table 1. Descriptive characteristics of 100 individuals with celiac disease who completed a study evaluating a modified Italian European Prospective Investigation into Cancer and Nutrition food frequency questionnaire for individuals with celiac disease Women (n[79)
Men (n[21) Characteristics
Mean–SDa
Range
Mean–SD
Range
Age (y)
41.612.2
18-62
43.313.6
20-70
Duration of disease (y)
14.011.6
2-39
13.110.0
2-37
Weight (kg) Height (cm) BMIb
55-96
57.28.8
39-80
175.16.6
73.810.2
160-193
161.86.7
147-175
24.02.6
19.0-30.4
21.93.3
16.9-32.9
a
SD¼standard deviation. BMI¼body mass index; calculated as kg/m2.
b
1812
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
November 2016 Volume 116 Number 11
RESEARCH Table 2. Daily reported energy and nutrient intake of 100 individuals with celiac disease assessed by a 7-day weighed food record and the modified Italian European Prospective Investigation into Cancer and Nutrition food frequency questionnairea
Variable
7-Day weighed food record (n[100)
Modified EPICb FFQc (n[100)
Paired t test, P value
Spearmand correlation coefficient
Spearmane correlation
Weighted k
meanstandard deviation! Energy (kcal) Protein (g)
<0.001
.37***
—
.40***
78.515.2
70.718.8
<0.001
.34***
.38***
.30**
2,144.0269.4
1,890.7535.7
Total carbohydrate (g)
258.043.1
207.567.4
<0.001
.32**
.33***
.31***
Sugars (g)
100.224.6
96.331.2
0.268
.26**
.29**
.21*
Starch (g)
157.033.9
110.845.6
<0.001
.44***
.39***
.38***
Fiber (g)
23.78.3
20.26.5
<0.001
.45***
.57***
.44***
Total fat (g)
89.516.4
85.925.6
0.192
.21*
.19
.22*
Saturated fats (g)
30.17.3
27.49.2
0.005
.45***
.42***
.38***
Monounsaturated fats (g)
43.69.1
38.512.1
<0.001
.14
.27**
.18*
Polyunsaturated fats (g)
15.64.5
12.24.4
<0.001
.35***
.41***
.36***
287.897.5
282.886.9
0.585
.48***
.37***
.46***
Cholesterol (mg) Alcohol (g) Iron (mg) Calcium (mg)
5.27.6
5.29.1
0.997
.73***
.64***
.74***
11.94.0
9.22.3
<0.001
.36***
.46***
.32***
878.3291.1
824.6346.4
0.071
Potassium (mg)
3,386.5683.4
2,850.4660.6
<0.001
Phosphorus (mg)
1,268.4253.7
1,146.1310.0
.52***
.52***
.49***
.29**
.45***
.31***
<0.001
.44***
.51***
.41***
10.32.0
9.02.4
<0.001
.36***
.36***
.30**
Thiamin (mg)
1.20.3
0.90.2
<0.001
.32**
.46***
.34***
Riboflavin (mg)
1.80.5
1.50.4
<0.001
.55***
.58***
.51***
21.45.2
18.05.1
<0.001
.38***
.39***
.39***
2.40.5
1.90.5
<0.001
.25**
.42***
.24**
Zinc (mg)
Niacin (mg) Vitamin B-6 (mg) Folate (mg) Beta carotene (mg)
326.5111.9
239.566.7
<0.001
.47***
.57***
.45***
4,699.12,505.0
2,629.81,512.3
<0.001
.32**
.41***
.29**
Vitamin C (mg)
143.668.2
122.846.1
Vitamin E (mg)
17.83.7
10.93.8
<0.001
0.002
Vitamin D (mg)
2.61.8
2.81.4
0.349
.44***
.45***
.37***
.13
.22*
.15
.27**
.18
.20*
To compare the daily nutrient and energy intake recorded by the modified EPIC FFQ and the 7-day weighed food record, a paired t test was used. To evaluate the degree of association between the rank of FFQ and that of 7-day weighed food record, Spearman’s correlation coefficients, unadjusted and adjusted for energy intake, were calculated. To assess the interquartile agreement between the two dietary instruments, the weighted k statistic was applied. b EPIC¼European Prospective Investigation into Cancer and Nutrition. c FFQ¼food frequency questionnaire. d Unadjusted values. e Values adjusted for energy. *P<0.05. **P<0.01. ***P<0.001. a
cholesterol, alcohol, calcium, and vitamin D. The Spearman coefficients varied from .13 (vitamin E) to .73 (alcohol). The lowest correlations were reported for the intake of vitamin E and monounsaturated fats, total fat, vitamin B-6, sugars, and vitamin D. Adjustment for energy intake did not consistently improve the correlations. A moderate agreement, which was defined as a weighted k value of .40 to .60, was obtained for 30% of the analyzed November 2016 Volume 116 Number 11
nutrients, and 40% of the nutrients showed values between .30 and .40. The weighted k was not significant only for vitamin E. Mean daily consumption of food groups reported by both dietary tools is presented in Table 3. Compared to 7-day weighed food record, the modified EPIC FFQ underestimated the consumption of other cereals, vegetables and legumes, fruit, and oils and fats, but not that of food groups of JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
1813
RESEARCH Table 3. Daily reported food group intake of 100 individuals with celiac disease assessed by a 7-day weighed food record and the modified Italian European Prospective Investigation into Cancer and Nutrition food frequency questionnairea
Variable
7-Day weighed food record (n[100)
Modified EPICb FFQc (n[100)
Paired t test, P value
Spearman correlation coefficient
Weighted k
meanstandard deviation! Pasta (g)
48.633.1
44.736.0
0.328
.52***
.48***
Breads (g)
90.546.2
82.976.5
0.278
.54***
.50***
47.334.7
24.616.6
<0.001
.31**
.26*
128.965.3
118.158.8
0.165
.25*
.30**
Fish (g)
41.939.7
41.426.1
0.904
.41***
.34***
Eggs (g)
14.814.1
15.410.9
0.664
.36***
.36***
Dairy products (g)
200.6139.9
206.6154.3
0.653
.67***
.58***
Vegetables and legumes (g)
303.0153.8
228.8101.7
<0.001
.50***
.49***
Fruit (g)
261.1138.9
234.9118.5
0.040
.54***
.47***
Sweets (g)
Other cereals (g) Meat and preserved meat (g)
101.846.9
100.158.7
0.765
.60***
.54***
Oils and fats (g)
32.112.5
27.911.9
0.008
.23*
.25**
Juices (g)
35.353.6
46.957.9
0.099
.26**
.25**
Soft drinks (g)
35.355.7
40.081.1
0.532
.48***
.48***
59.893.7
58.3100.5
0.824
.75***
.71***
176.4150.3
160.9111.8
0.183
.73***
.69***
Alcoholic beverages (g) Coffee and teas (g)
a To compare the daily food group intake recorded by the modified EPIC FFQ and the 7-day weighed food record, a paired t test was used. To evaluate the degree of association between the rank of FFQ and that of 7-day weighed food record, Spearman’s correlation coefficients were calculated. To assess the interquartile agreement between the two dietary instruments, the weighted k statistic was applied. b EPIC¼European Prospective Investigation into Cancer and Nutrition. c FFQ¼food frequency questionnaire. *P<0.05. **P<0.01. ***P<0.001.
particular interest in the celiac diet, such as bread, pasta, and sweets. Spearman correlation coefficients (Table 3) for the food categories varied from .23 to .75. The weighted k measured in the food groups exceeded .40 for 60% of the 15 analyzed food groups.
DISCUSSION In the present study, the ability of the modified EPIC FFQ to assess the dietary habits of individuals with CD was evaluated relative to a 7-day weighed food record. Based on the correlation coefficients, the modified EPIC FFQ correctly classified individuals with CD according to the distribution of intake of some nutrients and most food groups. The coefficient values observed in the present study were comparable to those observed in the previous validation study evaluating the FFQ performance in the general population.25 Several studies3,4,15,26 have shown that the dietary habits of individuals with CD do not differ in terms of food choice from those of the general healthy population. However, the celiac population must adhere to a lifelong GF diet in which the gluten-containing foods must be replaced with GF products 1814
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
and other cereals (eg, corn). It has been argued that GF foods tend to be less nutritious than their gluten-containing counterparts.16 Therefore, an evaluation of the FFQ performance in recording the grain-based foods, such as bread, pasta, others cereals, and sweets, is pivotal to the assessment of the nutritional adequacy of the GF diet, for which there is still controversy in the research.20 Mean daily intakes of food groups of particular interest in the celiac diet (ie, bread, pasta, and sweets) reported by the modified EPIC FFQ were in agreement with those reported by the 7-day weighed food record. In fact, significant Spearman correlations and a moderate agreement, indicated by the weighted k values between .48 and .54, were obtained between the two dietary tools. For the “other cereals” food category, the correlation and the weighted k calculated tended to be lower than those reported for bread, pasta, and sweets. This was probably due to a lack of specific questions in the modified EPIC FFQ on dishes including “other cereals” (eg, buckwheat and quinoa), which were occasionally consumed by the participants. As the staple foods and most other major food groups (ie, dairy products and beverages) were satisfactorily recorded by November 2016 Volume 116 Number 11
RESEARCH the FFQ, significant Spearman correlations between the two dietary tools were found for protein, total carbohydrates— including starch and fiber—and polyunsaturated and saturated fats. On the contrary, a low correlation coefficient and weighted k for the intake of monounsaturated fats were observed between the two dietary tools; as a consequence, a low but significant Spearman’s correlation (unadjusted value) and a low but significant weighted k were also observed for total fats. Fat is often added when preparing main dishes for families. Therefore, the amount of fat added to main dishes can be difficult for individuals to estimate, especially if he or she was not involved in the food preparation. As a consequence, some misclassification due to inaccurate estimation of consumed fat by both instruments cannot be ruled out. Because fats are the major source of vitamin E, a very low agreement of intake data between the two dietary tools was also observed for this vitamin. Regarding micronutrients, although the modified EPIC FFQ significantly underestimated their intake, except for calcium and vitamin D, significant correlation coefficients were reported between the dietary tools, especially for micronutrients that are known to be more sensitive for the celiac subjects (ie, iron, calcium, and folate). Some limitations of the present study should be highlighted. Because the modified EPIC FFQ significantly underreported the food groups and the nutrients consumed by the participants, it is not a useful tool to counsel a patient, but it might be more appropriate for population research. In addition, the highly selective nature of respondents that were willing and able to do a weighed record might have introduced a selection bias, possibly reducing the generalizability of the results. However, individuals with CD are known for being health conscious, thus this potential selection bias might have had minimal effects on the present results. Strengths of the present study include the comparison of the modified EPIC FFQ with the gold standard of weighed dietary records and the relatively high number of participants. Furthermore, we used a dietary FFQ that has been used in the Italian EPIC cohort. As the original EPIC FFQ revealed a link between dietary habits and disease risk (ie, colorectal cancer,27 stroke,28 and coronary heart disease29), the modified FFQ is expected to be valuable for studying similar relationships among individuals with CD.
3.
Kinsey L, Burden ST, Bannerman E. A dietary survey to determine if patients with celiac disease are meeting current healthy eating guidelines and how their diet compares to that of the British general population. Eur J Clin Nutr. 2008;62(11):1333-1342.
4.
Dall’Asta C, Scarlato AP, Galaverna G, Brighenti F, Pellegrini N. Dietary exposure to fumonisins and evaluation of nutrient intake in a group of adult celiac patients on a gluten-free diet. Mol Nutr Food Res. 2012;56(4):632-640.
5.
Martin J, Geisel T, Maresch C, Stein KKJ. Inadequate nutrient intake in patients with celiac disease: Results from a German dietary survey. Digestion. 2013;87(4):240-246.
6.
Grehn S, Fridell K, Lilliecreutz M, Hallert C. Dietary habits of Swedish adult coeliac patient treated by a gluten-free diet for 10 years. Scand J Nutr. 2001;45:178-182.
7.
Wild D, Robins GG, Burley VJ, Howdle PD. Evidence of high sugar intake, and low fibre and mineral intake, in the gluten-free diet. Aliment Pharmacol Ther. 2010;32(4):573-581.
8.
Johnson RK. Dietary intake—How do we measure what people are really eating? Obes Res. 2002;10(suppl 1):63S-68S.
9.
Biró G, Hulshof KF, Ovesen L, Amorim Cruz JA. Selection of methodology to assess food intake. Eur J Clin Nutr. 2002;56(2):25-32.
10.
Tabacchi G, Filippi AR, Breda J, et al. Comparative validity of the Asso food frequency questionnaire for the web based assessment of food and nutrients intake in adolescents. Food Nutr Res; 2015. 59-26.
11.
Farukuoye M, Strassburger K, Kacerousky-Bielesz G, Giani G, Roden M. Validity and reproducibility of an intervieweradministered food frequency questionnaire in Austrian adults at risk of or with overt diabetes mellitus. Nutr Res. 2014;34(5): 410-419.
12.
Persson PG, Carlsson S, Grill V, et al. Food frequency questionnaire versus 7-day weighed dietary record information on dietary fibre and fat intake in middle-aged Swedish men. Scand J Soc Med. 1998;26(1):75-80.
13.
Willett WC, Sampson L, Stampfer MJ, et al. Reproducibility and validity of a semiquantitative food frequency questionnaire. Am J Epidemiol. 1985;122(1):51-65.
14.
Willett W. Nutritional Epidemiology. Oxford, UK: Oxford University Press; 2013.
15.
Hopman EGD, Le Cessie S, von Blomberg ME, Mearin ML. Nutritional management of the gluten-free diet in young people with celiac disease in the Netherlands. J Pediatr Gastroenterol Nutr. 2006;43(1): 102-108.
16.
Miranda J, Lasa A, Bustamante MA, Churruca I, Simon E. Nutritional differences between a gluten-free diet and a diet containing equivalent products with gluten. Plant Foods Hum Nutr. 2014;69(2): 182-187.
17.
Shepherd SJ, Gibson PR. Nutritional inadequacies of the gluten-free diet in both recently-diagnosed and long-term patients with coeliac disease. J Hum Nutr Diet. 2013;26(4):349-358.
18.
Black JL, Orfilia C. Impact of coeliac disease on dietary habits and quality of life. J Hum Nutr Diet. 2011;24(6):582-587.
19.
European Institute of Oncology. Food composition database for epidemiological studies in Italy. http://www.ieo.it/bda. Published 2008. Accessed December 1, 2014.
20.
Mazzeo T, Cauzzi S, Brighenti F, Pellegrini N. The development of a composition database of gluten-free products. Public Health Nutr. 2015;18(8):1353-1357.
21.
Pala V, Sieri S, Palli D, et al. Diet in the Italian Epic cohorts: Presentation of data and methodological issues. Tumori. 2003;89(6): 594-607.
22.
Cohen J. Weighted kappa: Nominal scale agreement with provision for scaled disagreement or partial credit. Psychol Bull. 1968;70(4): 213-220.
23.
Harris JA, Benedict FG. A biometric study of human basal metabolism. Proc Natl Acad Sci. 1918;4:370-373.
24.
World Health Organisation. BMI classifications. http://www.who.int/ bmi/index.jsp?introPage¼intro_3.html. Published 2006. Accessed January 15, 2015.
25.
Pisani P, Faggiano F, Krogh V, Palli D, Vineis P, Berrino F. Relative validity and reproducibility of a food frequency dietary questionnaire for use in the Italian EPIC centres. Int J Epidemiol. 1999;26(1): 152-160.
CONCLUSIONS The EPIC FFQ, primarily designed for general populations and then revised with specific products, demonstrated moderate congruence with a weighed food record in ranking individuals by dietary intakes, particularly food groups. Consequently, this modified EPIC FFQ might be a useful instrument for assessing the dietary intake in the Italian population with CD. The present evaluation of this tool will allow for extending research on the dietary habits of individuals with CD from nutritional surveillance to assessing the potential impact of diet on later disease risk.
References 1.
Rubio-Tapia A, Hill ID, Kelly CP, Calderwood AH, Murray JA. ACG clinical guidelines: Diagnosis and management of celiac disease. Am J Gastroenterol. 2013;108(5):656-676.
2.
Farrell RJ, Kelley CP. Celiac sprue. N Engl J Med. 2002;346(3):180-188.
November 2016 Volume 116 Number 11
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
1815
RESEARCH 26.
Hallert C, Grant C, Grehn S, et al. Evidence of poor vitamin status in coeliac patients on a gluten-free diet for 10 years. Aliment Pharmacol Ther. 2002;16(7):1333-1339.
28.
Agnoli C, Krogh V, Grioni S, et al. A priori-defined dietary patterns are associated with reduced risk of stroke in a large Italian cohort. J Nutr. 2011;141(8):1552-1558.
27.
Agnoli C, Grioni S, Sieri S, et al. Italian Mediterranean Index and risk of colorectal cancer in the Italian section of the EPIC cohort. Int J Cancer. 2013;132(6):1404-1411.
29.
Bendinelli B, Masala G, Saieva C, et al. Fruit, vegetables, and olive oil and risk of coronary heart disease in Italian women: The EPICOR Study. Am J Clin Nutr. 2011;93(2):275-283.
AUTHOR INFORMATION T. Mazzeo is a postdoctoral researcher, F. Brighenti is a full professor, and N. Pellegrini is an associate professor, Human Nutrition Unit, Department of Food Science, University of Parma, Parma, Italy. L. Roncoroni is a researcher and nutritionist, Center for Prevention and Diagnosis of Celiac Disease, IRCCS Cà Granda Foundation, Ospedale Maggiore Policlinico, Milan, Italy, and Department of Biomedical, Surgical and Dental Sciences, University of Milan, Milan, Italy. V. Lombardo is a scholarship holder, C. Tomba is a researcher in gastroenterology, and L. Elli is a gastroenterologist, Center for Prevention and Diagnosis of Celiac Disease, and C. Agostoni is a full professor of pediatrics, Department of Clinical Sciences and Community Health, all at the University of Milan, Fondazione IRCCS Cà Granda Foundation, Ospedale Maggiore Policlinico, Milan, Italy. S. Sieri is a senior researcher in nutritional epidemiology and S. Grioni is a research associate in nutritional epidemiology, Epidemiology and Prevention Unit, Fondazione IRCCS Istituto Nazionale dei Tumori, Milan, Italy. M. T. Bardella is a consultant gastroenterologist, Center for Prevention and Diagnosis of Celiac Disease, Gastroenterology and Endoscopy Unit, IRCCS Cà Granda Foundation, Ospedale Maggiore Policlinico, Milan, Italy. L. Doneda is an associate professor in applied biology, Department of Biomedical, Surgical and Dental Sciences, University of Milan, Milan, Italy. Address correspondence to: Nicoletta Pellegrini, PhD, Department of Food Science, University of Parma, Parco Area delle Scienze, 47/A, 43124 Parma, Italy. E-mail: [email protected]
STATEMENT OF POTENTIAL CONFLICT OF INTEREST No potential conflict of interest was reported by the authors.
FUNDING/SUPPORT This work has been partly supported by Dr. Schär GmbH/Srl. Dr. Schär GmbH/Srl had no role in the design, analysis, and writing of this article. The study was registered at ClinicalTrials.gov (ID NCT01975155).
1816
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
November 2016 Volume 116 Number 11