Hopman, Erica G.D.*; Pruijn, Ricardo*; Tabben, Eric H.*; le Cessie, Saskia†; Mearin, M. Luisa‡
See “Challenge of Measuring How Much Gluten Is Too Much” by Kaukinen and Lähdeaho on page 719.
In studies in which assessment of food intake is done, various assessment methods can be used depending on the goal, the nutrient of interest, and on the population being studied. For example, recall methods by interview demand the experience and time of both the interviewer and the respondent, whereas written methods (eg, weighed records for which the respondent must be taught to weigh and record the food and its weight immediately before eating and to weigh any leftovers, or estimated records for which the respondent is taught to keep records in portion sizes all of the foods he or she eats and drinks in household measures) ask much effort from the respondent (1). The development of a food frequency questionnaire, however, requires more time from the investigator in preparing the questionnaire beforehand than from the respondent during the study period. Thus, all of the methods have their advantages and disadvantages with regard to adequacy and workload (1,2). The food record (FR) assesses the intake recorded on the requested days, considering the intake on these days as a reflection of what someone normally eats. The food questionnaire (FQ) estimates the frequency of consumed foods or nutrients, relevant to a specific question during a specific period. Therefore, the FR may be the more accurate method, whereas the FQ may be the more representative, making it arguable which one best reflects true dietary intake (2). The FQ and the FR are instruments often used in epidemiological studies, but the FQ is less time consuming than the FR (1). The FQ-gluten to assess gluten intake and consumption of breast-feeding by infants 0 to 12 months old, developed and validated by our group, is already being used in research projects such as PreventCD (3,4); however, for further follow-up, an FQ to assess the gluten intake of older children was needed. The aim of the present study was to develop and validate an FQ for the assessment of gluten consumption in healthy young children ages 1 to 4 years (FQ-gluten4).
Subjects and Ethics
During the spring of 2008, parents from children in the general population ages 1 to 4 years attending 10 child health care centers in the western part of the Netherlands were asked to participate in the present study. In the Netherlands, 91% of infants from 0 to 4 years attend child health care centers (5).
Exclusion criteria were mental retardation or oral-motor dysfunction, diagnosed food allergy, impossibility of oral food intake, and parents with insufficient knowledge of the Dutch language. For this noninvasive study requiring only written data from the participants, no informed consent was required.
Development of the FQ-Gluten4
For the development of the FQ-gluten4, the FQ-gluten for infants 0 to 12 months old (4) was used as a basis. The FQ-gluten for children 0 to 12 months was extended by adding gluten-containing food products consumed by children 1 to 4 years according to the latest national food consumption study among young children in the Netherlands (6). The FQ-gluten4 comprises 81 items on food intake (Fig. 1).
Validation of the FQ-Gluten4
To validate the FQ-gluten4, results were compared with the results of a 2-day FR, which is an accepted method used in food consumption studies (7,8). Parents participating in the present study were asked to fill in the new FQ-gluten4, which is a 7-day method, and a 2-day FR (1 weekday and 1 weekend day) recording their child's food intake in household measures. The parents were also asked to precisely note the name of the manufacturer of the product used. The parents were free to choose the days to fill in the FR and the FQ-gluten4, but in any case on different days and within 6 weeks after they had received it, the forms had to be returned to the researchers.
Assessing Gluten Amount
From the results of the FQ-gluten4 and of the FR, a mean daily intake was calculated. We considered food products containing wheat, rye, and barley as containing gluten. Of all of the food products reported in the FR and the frequency of the food products consumed by the child registered in the FQ-gluten4 containing these cereals, gluten content was calculated. Because there is no information in food composition tables on the gluten content of food products, we used the method of Overbeek et al (9) to calculate it. This assumption is widely accepted and it is the only accepted rule to estimate gluten content in food products for population inquiries. Following this method, we multiplied the grams of vegetable protein according to the Dutch food composition table and derived from gluten-containing cereals according to the composition of the food product by 0.8 (9). In case food composition was not available, the composition of a similar generic food product was used.
Because we know from our former study results that the contribution of “risk products” is only a small percentage of the total gluten consumption (0.21%–0.65%) (4), we did not calculate the gluten content of these products for the present study. Risk products are, for example, ready-to-eat fruit and vegetable mixes, and flavored milk products (eg, fruit yogurt), known to possibly contain gluten, but for which the exact amount of gluten cannot be calculated due to either missing brand information or a rounded number of zero grams protein in the food composition table.
All of the analyses were carried out using SPSS version 17 (SPSS Inc, Chicago, IL). Amounts of gluten consumption were derived from the FQ-gluten4 and the FR. The mean percentages of gluten per day from different food products per child were calculated. The mean daily gluten intake and the mean percentages between FR and FQ-gluten4 were compared using the paired t test. Agreement was assessed by Bland-Altman plot and by calculating limits of agreement, defined as the mean difference ± 2 standard deviations (SD). P < 0.05 were considered significant.
Of the parents invited for the study, 75 agreed to participate. Of them, 71 (95%) filled in both the new FQ-gluten4 and the FR. The parents of 2 children (1 boy aged 40 months and 1 girl aged 18 months) did not fill in the FR and of 2 other children (2 boys ages 35 and 37 months, respectively), only 1 day was completed instead of 2. For that reason, the results of these 4 children were not included in the analyses. Table 1 presents the sex distribution of the children in the different age categories.
Validation of the FQ-Gluten4
The origin of the gluten intake, derived from the FR and the FQ-gluten4, is presented in Table 2. Except for the percentage of gluten intake from porridge in children ages 1 to 3 years, both instruments reported similar gluten origin and consumption of the different food groups. All of the gluten-containing products that were reported in the food consumed by the children by using the FR were contained in the FQ-gluten4.
Figure 2 shows the results of the mean amount of daily gluten consumption calculated from the FQ-gluten4 and the FR. Overall, the FQ-gluten4 detected a mean daily gluten intake comparable with that of the FR (P = 0.43). A significant higher intake, however, was reported by the FR in the youngest age group (P = 0.035). Differences were found in warm meal products used and in the number of slices of bread consumed. The absolute median difference in the assessment of gluten consumption by the FR or the FQ-gluten4 was 1160 mg (range 10–8712).
The Bland-Altman plot with limits of agreement between the data obtained by the FR and the FQ-gluten4 is presented in Figure 3. The SD of the differences was 2600 mg, which means that the differences in the gluten intake using the 2 methods are relatively high in some individual cases. The Bland-Altman limits of agreement for the different food products assessed by the FR and by the FQ-gluten4 are presented in Table 3. The food group with the largest difference in some individual cases is the food group with the highest gluten intake, the bread group (SD 2405 mg), whereas the largest mean difference between the 2 instruments is found in the porridge group (mean 425 mg), which is comparable with the results in Table 2 (significant difference in the percentage of porridge).
An easy-to-use instrument to quantify the gluten consumption of young children up to 4 years of age was lacking. Therefore, we developed such an instrument: the FQ-gluten4. To our knowledge, we present here the first questionnaire to assess the gluten consumption by children 1 to 4 years old. We have validated this FQ-gluten4 by using an accepted method for food consumption studies as a reference: a 2-day FR.
Overall, results show that the FQ-gluten4 compared with the FR method gives similar results in detecting the mean amount of daily gluten consumption, but that the FQ-gluten4 detects a significantly lower amount of gluten intake in 1- to 2-year-old children and a significant lower percentage of gluten from porridge among 1- to 3-year olds.
A possible explanation for these differences is that we compared a 7-day method (FQ-gluten4) to a 2-day method (FR). For example, the youngest-age children had relatively high gluten consumption on the 2 days the FR was filled in caused by eating pancakes or pasta as their warm meal, whereas the FQ-gluten4 only reported that pasta or pancakes were eaten by the children once or twice per week. In averaging these amounts over 7 days, as the FQ-gluten4 does, it will lead to a lower gluten intake from pasta and pancakes. Furthermore, a higher number of slices of bread were consumed according to the FR compared with the number of slices of bread filled in the FQ-gluten4. This caused differences of 2000 to 6000 mg of gluten per day. The Bland-Altman limits of agreement with an SD of 2600 mg were −5118 to 5630 mg. This means that for individual cases, the difference is large, but considering that variances in consumption of basic food products such as bread in young childrens’ daily food intake are common (7), these differences can be acceptable. To illustrate this, a difference in the intake of 1 slice of bread or 1 pancake will lead to a difference in gluten consumption of about 2500 or 2800 mg of gluten, respectively. Had we asked the parents to fill in a 7-day FR, the higher gluten intake on the 2 days would have been averaged with the other 5 days to a lower mean and would probably be more consistent with the results of the FQ-gluten4. A 2-day FR, however, is an accepted method used in food consumption studies because it gives good enough information on the variety of the food products that these young children consume (7,8) and is of lesser burden on the parents than a 7-day FR. Seemingly, in cases in which the nutrient of interest is gluten and in the age category of 1 to 4 years, the 2-day FR is less suitable. For that reason, a 7-day method such as the FQ-gluten4 will be more representative than the 2-day FR specific for gluten intake, averaging it to a mean daily gluten intake but not overestimating the amount of gluten consumption.
Another possible explanation may be that this is the first test of validity. We did not pretest this FQ-gluten4 to rule out possible shortcomings.
For practical reasons, the inclusion period for the study was only 3 months. In that period, 75 parents of healthy young children agreed to participate. The total number of invited parents was not registered because the present study was aimed to validate an FQ on gluten intake against the widely used FR and not aimed to study the representativity of the study population. For that reason we did not register information on the social status or the age of the participating parents. We do not believe that representativity is an issue because the comparison between the FQ and FR is independent of the social status and of the representativity of the population. The number of participating children in the 3 age categories was rather small; however, groups were large enough to detect significant differences in the 1- to 2- and 2- to 3-year-old children.
The previously developed FQ-gluten for quantification of gluten intake by infants 0 to 12 months old (4) was based on the Dutch eating pattern, with specific Dutch brand names; it was suitable to be used only in the Dutch population. On the contrary, this FQ-gluten has been adapted in other countries to the eating patterns and food products used by their young children through obtaining information on the food products and brand names used from food consumption studies. Afterwards, the FQs were validated using a 2-day FR, and are now being used in a multinational study, the PreventCD study (www.preventcd.com) (3). The newly developed and validated FQ-gluten4 can be adapted by researchers in other countries following the same procedure. Where food consumption studies are not available, a possible way to obtain the necessary information is by using a FR for a basis inquiry among young children. From these results an FQ-gluten can be composed and validated.
Another application for the new FQ-gluten4 can be in quantifying the gluten intake in children suspected to have celiac disease, but with dubious specific celiac antibodies or with mild histological changes in small bowel biopsy samples, under which circumstances the diagnosis cannot be confirmed. One possible reason for these results may be a lower-than-normal gluten intake, for example, in children living in families with patients with celiac disease. Using this FQ-gluten4 for this purpose can give an indication of the average daily gluten intake by the child.
An advantage of a preprinted FQ above the use of a FR is that it can be easily computerized, is less burdensome for the patient or the parent, and may generate an outcome in a short time that can be helpful in consultation with his or her doctor.
In conclusion, the new, short, standardized, validated, and easy-to-use FQ-gluten4 may be a useful instrument to assess gluten intake in individual children 1 to 4 years old suspected of having CD to assess gluten consumption and, at the population level, for study purposes. The use of this standardized method by investigators in different countries will provide opportunities for better comparisons of the results of gluten consumption throughout the world.
We thank PreventCD for their request to develop the FQ-gluten4 and providing the opportunity to use the PROMISE computerized database.
1. Methods for data collection at an individual level. In: Cameron ME, van Staveren WA, eds. Manual on Methodology for Food Consumption Studies. New York: Oxford University Press; 1988:53–99.
2. Thomas B. Dietary assessment. In: Manual of Dietetic Practice, 3rd ed. Oxford: Blackwell Science; 2001:30–7.
3. Hogen Esch CE, Rosén A, Auricchio R, et al. The PreventCD study design: towards new strategies for the prevention of coeliac disease. Eur J Gastroenterol Hepatol 2010; 12:1424–1430.
4. Hopman EG, Kiefte- de Jong JC, le Cessie S, et al. Food questionnaire for assessment of infant gluten consumption. Clin Nutr 2007; 26:264–271.
6. Rossum van CT. Dutch National Food Consumption Survey—Young Children 2005/2006 (VCP). Report 350070001/2008. Bilthoven, the Netherlands: RIVN; 2008.
7. Deckers Druk België, eds. Zo eten Jonge Peuters in Nederland 2002; Resultaten van het Voedingsstoffen Inname Onderzoek 2002 (Results of the Study on Nutrient Intake in Young Toddlers 2002). Ir. BC Breedveld, Dutch Nutrition Center, The Hague and Dr. KFAM Hulshof, TNO Nutrition Zeist, The Netherlands; 2002.
8. Zo eet Nederland 1998; Resultaten van de Voedselconsumptiepeiling 1997–1998 (Results of the Food Consumption Study 1997–1998). The Hague: Van Marken Delft Drukkers, Dutch Nutrition Center; 1998.
9. Overbeek FM, Uil-Dieterman IG, Mol IW, et al. The daily gluten intake in relatives of patients with coeliac disease compared with that of the general Dutch population. Eur J Gastroenterol Hepatol 1997; 9:1097–1099.