Advertisement: Preregister now for the Medicine 2.0 Congress
Online Dietary Intake Estimation: The Food4Me Food Frequency Questionnaire
Hannah Forster1, BSc (Hons); Rosalind Fallaize2, BSc (Hons); Caroline Gallagher1, BSc (Hons); Clare B O’Donovan1, MSc; Clara Woolhead1, MSc; Marianne C Walsh1, PhD; Anna L Macready2, PhD; Julie A Lovegrove2, PhD; John C Mathers3, PhD; Michael J Gibney1, PhD; Lorraine Brennan1, PhD; Eileen R Gibney1, PhD
1UCD Institute of Food and Health, University College Dublin, Dublin, Ireland
2Hugh Sinclair Unit of Human Nutrition and Institute for Cardiovascular and Metabolic Research, University of Reading, Reading, United Kingdom
3Human Nutrition Research Centre, Institute for Ageing and Health, Newcastle University, Newcastle, United Kingdom
UCD Institute of Food and Health
University College Dublin
Science Centre South,
Phone: 353 1 716 2819
Fax: 353 1 716 1147
Background: Dietary assessment methods are important tools for nutrition research. Online dietary assessment tools have the potential to become invaluable methods of assessing dietary intake because, compared with traditional methods, they have many advantages including the automatic storage of input data and the immediate generation of nutritional outputs.
Objective: The aim of this study was to develop an online food frequency questionnaire (FFQ) for dietary data collection in the “Food4Me” study and to compare this with the validated European Prospective Investigation of Cancer (EPIC) Norfolk printed FFQ.
Methods: The Food4Me FFQ used in this analysis was developed to consist of 157 food items. Standardized color photographs were incorporated in the development of the Food4Me FFQ to facilitate accurate quantification of the portion size of each food item. Participants were recruited in two centers (Dublin, Ireland and Reading, United Kingdom) and each received the online Food4Me FFQ and the printed EPIC-Norfolk FFQ in random order. Participants completed the Food4Me FFQ online and, for most food items, participants were requested to choose their usual serving size among seven possibilities from a range of portion size pictures. The level of agreement between the two methods was evaluated for both nutrient and food group intakes using the Bland and Altman method and classification into quartiles of daily intake. Correlations were calculated for nutrient and food group intakes.
Results: A total of 113 participants were recruited with a mean age of 30 (SD 10) years (40.7% male, 46/113; 59.3%, 67/113 female). Cross-classification into exact plus adjacent quartiles ranged from 77% to 97% at the nutrient level and 77% to 99% at the food group level. Agreement at the nutrient level was highest for alcohol (97%) and lowest for percent energy from polyunsaturated fatty acids (77%). Crude unadjusted correlations for nutrients ranged between .43 and .86. Agreement at the food group level was highest for “other fruits” (eg, apples, pears, oranges) and lowest for “cakes, pastries, and buns”. For food groups, correlations ranged between .41 and .90.
Conclusions: The results demonstrate that the online Food4Me FFQ has good agreement with the validated printed EPIC-Norfolk FFQ for assessing both nutrient and food group intakes, rendering it a useful tool for ranking individuals based on nutrient and food group intakes.
(J Med Internet Res 2014;16(6):e150)
food frequency questionnaire; online dietary assessment tool; Food4Me; dietary assessment; Web-based
Associations between dietary behaviors and chronic health risks have been established on numerous occasions [1,2], making the ability to measure dietary intake crucial for researchers and public health practitioners to improve health outcomes [3,4]. Dietary assessment methods are used for quantification of both short- and long-term (habitual) dietary intakes, and are essential tools in epidemiological investigations and intervention studies assessing relationships between diet and health in both population and clinical settings [3,5].
Food records (or diaries), 24-hour recalls, and food frequency questionnaires (FFQ) are the three principle assessment methods that are used traditionally to measure dietary intake [6,7]. Food records require respondents to record all foods and beverages consumed over a specified period of time, generally between 3 and 7 days prospectively [8-10]. The 24-hour recalls are interviewer led and entail asking the respondent to remember and record all foods consumed in the preceding 24-hour period [7,10,11]. FFQs are also retrospective assessment tools and 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 .
Internet availability and usage has increased globally over the past decade and, as a result, traditional methods of dietary assessment have been modified for online and electronic use in both research and industry . Compared with traditional methods of dietary assessment, online methods allow for the automatic storage of input data and the automatic generation of nutritional outputs . While traditional methods of dietary assessment can be supplemented with food photograph atlases to aid portion size recognition and estimation , online methods of dietary assessment can be designed to incorporate food photographs, making them more convenient for users to complete [15-17]. Furthermore, compared with traditional methods, online dietary assessment methods can be used to target specific geographical population groups, can be accessed remotely, and can be designed to be easy to complete.
The strengths and weaknesses of traditional dietary assessment methods are well documented [8,18-21]. The quality of data from any dietary assessment method, traditional or online, will depend ultimately on the respondent’s accuracy in recording the required details . As a result, online food diaries may have greater use commercially or for small dietary studies as they require respondents to be highly motivated to record data .
FFQs and 24-hour recalls are the most commonly used approaches for assessing dietary intake for large population studies . Although Web-based 24-hour recalls have been demonstrated to show good agreement with traditional methods on numerous occasions [23-25,26], they are largely limited by the day-to-day variability in dietary intake and may not accurately assess intakes of foods that are eaten infrequently (eg, oily fish). As a result, multiple 24-hour recalls over several non-consecutive days are required to reflect usual dietary intake . Unlike food records and 24-hour recalls, FFQs can capture long-term dietary intake in a single administration and are less cumbersome to complete [27-29]. Although FFQs have often been reported to bear the greatest amount of measurement error, not only when considering under-reporting but also over-estimation of dietary intakes , they have been shown to have good validity for ranking nutrient intakes on numerous occasions [20-22,31]. As a result, they can be used to categorize nutrient intakes as “low”, “recommended”, or “high” compared with recommended intakes, rendering them invaluable tools for assessing nutritional intake status [20-21].
In recent years, many well-established FFQs have been developed into Web-based versions and there is a growing body of evidence demonstrating that data from Web-based FFQs are comparable with data from printed versions and/or have good validity with reference methods such as 24-hour recalls and food diaries [15-17,32]. Web-based FFQs possess many benefits over printed questionnaires: they are more cost-effective, they can be pre-programmed to ensure all questions are answered, and photographs can be incorporated to enhance food recognition and portion size estimation [15,17,32].
The present study was conducted as part of the EU 7th European Framework Programme “Food4Me” project . The Food4Me project aims to investigate the potential of, and public attitudes toward, personalized nutrition and is the first study of its kind designed to emulate an entirely Web-based, personalized nutrition service .
The objectives of the present study were to develop an online FFQ for dietary data collection in the Food4Me study and to compare estimates of intakes obtained using this tool with those obtained from the validated European Prospective Investigation of Cancer (EPIC) Norfolk printed FFQ .
Development of the Online Food4Me Food Frequency Questionnaire
The online Food4Me FFQ was designed to assess food and nutritional intake across seven centers in Europe, as part of a dietary intervention study within the Food4Me project . The design and development of the novel online Food4Me FFQ was led by researchers at University College Dublin and software company Creme Global (Dublin, Ireland).
The well-validated EPIC-Norfolk FFQ (version CAMB/PQ/6/1205)  was used as a guide for food items and food group categories. In developing the Food4Me FFQ, the original 130 food items presented in the EPIC-Norfolk FFQ were expanded upon to incorporate an additional 27 commonly consumed food items that were considered nutritionally important across the seven EU countries in the Food4me study. In expanding the food list, some food items were added (eg, tortillas, wraps) to existing food categories; some new foods were added to existing food items (eg, “noodles and cous cous” were added to “white pasta or green pasta”), and, in some cases, existing food items were split into more defined types (eg, “oily fish, fresh or canned” was split into “non-smoked oily fish, canned” and “non-smoked oily fish, fresh”). Standardized color photographs were incorporated into the online Food4Me FFQ to facilitate accurate portion size estimation. As in the printed EPIC-Norfolk FFQ, the 157 food items were divided into 11 categories viz, “cereal”, “bread and savory biscuits”, “potatoes, rice, and pasta”, “meat and fish”, “dairy products”, “fats and spreads”, “sweets and snacks”, “soups, sauces, and spreads”, “drinks”, “fruit”, and last, “vegetables”. In addition, the Food4Me FFQ included an additional section on dietary habits with further questions relating to additional foods consumed, the addition of salt to foods, consumption of fried foods, and supplement use (as in the EPIC-Norfolk FFQ).
Frequency of intake was estimated by asking, “How often would you have consumed each of the following in the past month?” and participants could select their frequency from nine categories of intake ranging from “never (<1 per month)” to “6+ per day”. After selecting their frequency of consumption, participants were asked to choose their usual serving size from a range of portion size pictures for each food item (see Figure 1). The online Food4Me FFQ was pre-programmed to ensure that a frequency of consumption was reported for every food item before the participant could submit the FFQ and was designed so that participants could check and/or modify previous responses before submitting the FFQ. Illustrations of the Food4Me FFQ are shown in Figure 1 and Multimedia Appendix 1.
[view this figure]
|Figure 1. Screenshot of the online Food4Me Food Frequency Questionnaire.|
Nutritional Composition and Portion Sizes
Nutritional composition and portion sizes were calculated from the 2008-2010 National Adult Nutrition Survey (NANS) database, which consists of detailed dietary intake data for 1500 Irish adults .
The nutritional composition of the 157 food items, listed in the Food4Me FFQ, was derived from the wider list of corresponding foods within the NANS database. From these, the most frequently consumed foods were identified and used to calculate the composition of the list of foods in the Food4Me FFQ. For example, for all pizzas consumed in the NANS dataset, the nutritional composition per 100g was computed for the three most frequently consumed and the mean of these was then calculated to give a single nutritional composition for pizza. The nutritional composition for the NANS dataset was analyzed using WISP (Tinuviel Software, Anglesey, UK). WISP uses the 5thand 6theditions of McCance and Widdowson’s The Composition of Foods, plus all nine supplemental volumes to generate nutrient intake data [37,38]. In addition, the nutritional composition dataset was modified previously and updated to include recipes of composite dishes, generic commercial foods, new foods on the market, and current manufacturers’ information. For additional foods unique to a specific European country in the Food4Me study, such as Greek baklava, the relevant national food composition database was used.
For the calculation of portion sizes, the food codes for each of the frequently consumed foods identified from the NANS database were merged and recoded into a single food code for each food item listed in the FFQ. Using the newly assigned food code, PASW Statistics version 18 (SPSS Inc Chicago, IL, USA) was used to calculate the 25th, 50th, and 75thpercentiles of daily intake, which represent small, medium, and large portion sizes of these foods when consumed in the free living population. Options for portion sizes above, below, and in between these percentiles were also provided to accommodate the wide variability in portion size across populations.
Foods were purchased from local supermarkets and bakeries in Dublin. All foods were prepared and photographs taken in the Institute of Food and Health, University College Dublin. Food photographs were taken by a professional photographer over 5 days/sessions. A standard dining set of plates and cutlery, positioned uniformly with the same lighting, was used for each session. The calculated NANS portion sizes were used as a guideline and all foods were weighed out using calibrated portable food scales (Tanita, Japan).
Study Sample and Study Design
The comparison of the online Food4Me FFQ with the printed EPIC-Norfolk FFQ was conducted in two centers involved in the Food4Me study (University College Dublin and University of Reading) using the English language version of the questionnaire (consisting of 157 food items). The study received approval from the ethical committees at both universities and was carried out between March and October 2012 (LS-11-118-Gibney-Walsh and 01/12-Lovegrove respectively). Participants (n=177) aged ≥18yrs were recruited at both centers. Participants were provided with an information sheet explaining the study, signed a consent form, and completed a brief screening questionnaire either in person or by post. Weight and height were self-reported. Individuals with self-reported/diagnosed food intolerances/allergies or those receiving dietary advice were ineligible to participate.
Eligible participants completed both the online Food4Me FFQ and the printed EPIC-Norfolk FFQ in random order. The printed EPIC-Norfolk FFQ was modified to ask participants’ about their food intake over the past month rather than over the past year. To minimize possible effects of temporal changes in dietary intake, participants with more than 4 weeks between completing both FFQs were excluded from analyses. The printed EPIC-Norfolk FFQ was delivered to the participant in person or by post. The Food4Me FFQ was accessed via a hyperlink to the website sent in an email containing the participant’s individual username and password.
The Henry equation was used to calculate basal metabolic rate (BMR) and BMR was multiplied by 1.1 to calculate the lowest possible estimated energy requirements (EER) for each participant . Participants reporting energy intakes lower than their EER were classified as under-reporters.
Dietary Intake Analysis
Printed EPIC-Norfolk questionnaires were coded using the specified template format and cross-checked before sending to Strangeways Research Laboratory (University of Cambridge) for processing using FETA software . The nutritional composition database used in the EPIC-Norfolk FFQ is based on the revised and extended 5thedition of McCance and Widdowson’s The Composition of Foods plus supplemental volumes . The Food4Me FFQ nutritional intake data was generated automatically by the online Food4Me programmed system, as described above. For the purpose of the current study, consumption of dietary supplements was not included in the analyses.
Data were imported into SPSS for analysis and descriptive statistics were computed to describe the general characteristics of participants. Mean nutrient intakes and standard deviations were determined for both FFQs. General linear model analysis controlling for energy was used to compare nutrient intakes between the FFQs. Correlation coefficients were computed to assess the association between the two methods. The relative agreement between the two FFQs was assessed using cross-classification of nutrient intakes to estimate the percentage of participants who were classified by the two methods into quartiles of “exact agreement”, “exact agreement plus adjacent”, “disagreement”, and “extreme disagreement”. Bland and Altman analysis was performed for the macronutrients to assess the limits of agreement between the two FFQs. For each macronutrient, the differences of the mean between the two methods (EPIC−Food4Me) were plotted against the average of the two methods ([EPIC+Food4Me]/2). Methods were considered comparable if greater than 95% of data plots lay within the limits of agreement (mean ± 2SD).
Differences in food group intakes between the two methods were examined. To do this, the food items in the EPIC-Norfolk and Food4Me FFQs were arranged into 35 food groups. Independent samples t tests were used to compare daily food group intakes between the two FFQs. Bland and Altman analysis was performed for the food groups to assess the limits of agreement between the two FFQs. Spearman’s correlation coefficients (SCC) were computed to assess the associations between the two methods for the daily intake of each of the 35 food groups. To assess the relative agreement between the two methods for daily food group intake, food group intakes were cross-classified to estimate the percentage of participants classified by the two methods into quartiles of “exact agreement”, “exact plus adjacent agreement”, and “disagreement”. All data were analyzed using PASW Statistics version 18 (SPSS Inc Chicago, IL, USA). P<.05 was considered statistically significant. GraphPad PRISM version 6 was used to produce the Bland and Altman plots (GraphPad Software, Inc California, USA).
Overview of the Study Population
A total of 177 participants were screened to participate in the study with 159 eligible for inclusion. Following initiation of the study, 27 participants dropped out, as shown in Figure 2. Reasons for dropouts included external commitments, for example, holidays and technical issues. A further 19 participants were excluded from the analysis: 16 had >4 weeks between completion of the two FFQs and 3 reported energy intakes >4500 kcal per day with the Food4Me FFQ (considered to be unrealistically high) . The final data set, therefore, consisted of 113 participants as illustrated in Figure 2. The results presented here are for the dietary assessment of 67 females (59.3%) and 46 males (40.7%) who completed both the printed EPIC-Norfolk FFQ and online Food4Me FFQ in random order.
Demographic characteristics of the study population are presented in Table 1. Overall, for all participants, there were no significant differences in age for males and females; however, self-reported body mass index (BMI) was significantly lower for females compared with males. As shown in Table 1, significant differences in age, weight, and BMI were observed across the two centers. The participants recruited in Dublin were significantly older (P<.005), were significantly heavier (P<.005), and had higher BMI (P<.05) than those recruited in Reading.
[view this table]
|Table 1. Demographic characteristics of the study population in total and across centers, by gender.|
[view this figure]
|Figure 2. Flow of participants through the study. FFQ: food frequency questionnaire.|
Comparison of Nutrient Intakes Between the Two Questionnaires
Mean energy and nutrient intakes estimated from the two FFQs are presented in Table 2. Energy intakes were significantly higher for the Food4Me FFQ in comparison with the EPIC-Norfolk FFQ (P<.001). However, when the occurrence of under-reporting was examined, 58 participants were classified as under-reporters with the EPIC-Norfolk FFQ compared with only 20 participants for the Food4Me FFQ. Of the 20 participants’ under-reporting in the Food4me FFQ, 18 also under-reported using the EPIC-Norfolk FFQ (90%).
After controlling for energy, intakes of macronutrients showed relatively good agreement with no significant differences in energy derived from total fat, saturated fatty acids (SFA), monounsaturated fatty acids (MUFA), or carbohydrates. Significant differences were observed in both total polyunsaturated fatty acid intake and % energy derived from polyunsaturated fatty acids (PUFA), at P<.001. Examination of differences at a micronutrient level demonstrated no significant differences between the two FFQs for intakes of calcium, vitamin B12, and vitamin D. Significant differences were observed for intakes of folate, iron, carotene, riboflavin, thiamin, vitamin B6, vitamin C, vitamin A (RE), retinol, vitamin E, and sodium (Table 2). However, after controlling for energy and, where appropriate, center, randomization group and/or gender, vitamin C and vitamin A (RE) were no longer significantly different (Table 2). The removal of under-reporters reduced the agreement between the two FFQs for total fat (% total energy [TE]), but improved the agreement for protein (% TE) showing no significant difference when controlling for energy, center, and gender (both alone and in combination), as shown in Table 3.
Bland and Altman plots for mean daily energy, total fat, protein, and carbohydrate intakes are presented in Figure 3. The Bland and Altman plot for energy indicated broad limits of agreement. The mean difference in estimated energy intake between FFQs was relatively high (676 kcal/d) with greater energy intakes recorded using the Food4Me FFQ compared with the EPIC-Norfolk FFQ. In addition, the difference in estimates of energy intakes between FFQs became progressively greater with higher mean intakes (Figure 3). Despite this, less than 5% of cases fell outside the limits of agreement for energy (n=5), confirming an acceptable level of agreement between the two methods.
As illustrated in Figure 3, similar to energy, the difference in estimates of total fat (g) and protein (g) intakes between the two FFQs became progressively greater with higher mean intakes. Poor levels of agreement between the methods were observed for the macronutrients, with greater than 5% of cases falling outside the limits of agreement for protein (g) (n=8), carbohydrate (g) (n=6), and total fat (g) (n=6). After adjusting for energy, the agreement between the two methods improved, as less than 5% of cases fell outside the limits of agreement for both carbohydrate (% TE) (n=4) and total fat (% TE) (n=4) (Figure 3). Although energy adjustment improved the distribution of cases for protein, more than 5% of cases (n=7) remained outside the limits of agreement , indicating a poorer level of agreement between the methods for protein (% TE). Correlation coefficients for estimates of nutrient intakes between the two FFQs are presented in Table 4. The mean correlation coefficient for nutrient intake between the two FFQs was .60. Correlations varied from .43 (polyunsaturated fatty acids % TE) to .86 (alcohol). For micronutrients, correlation coefficients were lowest for thiamin (.46) and highest for vitamin C (.69). The cross-classification of mean daily intakes between the two FFQs are also shown in Table 4. The percentage of participants classified into quartiles of “exact agreement” varied from 37% (polyunsaturated fatty acids %TE) to 63% (alcohol). The percentage of participants cross-classified into quartiles of “exact agreement plus adjacent” was lowest for polyunsaturated fatty acids (77%) and highest for alcohol (97%). Although the mean percentage of participants classified into quartiles of “disagreement” (opposite plus extreme quartiles) was 15%, the mean percentage categorized into quartiles of “extreme disagreement” was 2.5%.
[view this table]
|Table 2. Mean daily nutrient intakes estimated by printed EPIC-Norfolk FFQaand online Food4Me FFQ (n=113).|
[view this table]
|Table 3. Mean daily nutrient intakes estimated by printed EPIC FFQa(n=55) and online Food4Me FFQ (n=93) with under-reporters removed.|
[view this table]
|Table 4. Unadjusted correlation coefficients and cross-classification of quartiles of mean energy and nutrient intakes derived from the online Food4Me FFQaand printed EPIC-Norfolk FFQ.|
[view this figure]
|Figure 3. Bland and Altman plots with mean difference and limits of agreement (solid line represents mean difference and dotted lines represent limits of agreement). TE: total energy.|
Comparison of Food Group Intakes Between the Two Questionnaires
To examine differences in food group intakes between the two FFQs, the food items in the EPIC-Norfolk FFQ and Food4Me FFQ were aggregated into 35 food groups. Mean daily intakes estimated from the two FFQs for the 35 food groups are presented in Table 5. Mean daily intakes for 17 of the 35 food groups analyzed were significantly higher for the online Food4Me FFQ in comparison with the printed EPIC-Norfolk FFQ.
Bland and Altman analysis for daily food group intake was performed to examine the agreement between both methods. Overall, acceptable levels of agreement were observed for 15 of the 35 food groups, with less than 5% of cases (n<6) falling outside the limits of agreement. The Bland and Altman plots for mean daily intakes for six of the 35 food groups (“rice, pasta, grains, and starches”, “yoghurts”, “eggs and egg dishes”, “other vegetables”, “fish and fish products”, “red meat”) are presented in Figure 4. The mean difference between the methods (bias) was small for all six food groups. Overall, these plots indicate good agreement between the two methods for the assessment of “rice, pasta, grains, and starches”, “yoghurts”, “eggs and egg dishes”, and “red meat”. Poorer agreement was observed for “other vegetables” and “fish and fish products/dishes”, with more than 5% of cases (n=7) outside the limits of agreement for both food groups .
SCC and cross-classification of mean daily food group intakes are presented in Table 6. The mean SCC for food group intake between the two FFQs was .68 and SCCs ranged from .41 for “savories” (lasagna, pizza) to .90 for “other fruits” (apples, pears, citrus fruits). High correlations (SCCs≥.5) were observed for 32 of the 35 food groups. Overall, ranking participants into quartiles of “exact agreement” was lowest for “cakes, buns, and pastries” (35%) and highest for “bananas” (73%). The percentage of participants cross-classified into quartiles of “exact agreement plus adjacent” was lowest for “cakes, buns, and pastries” (77%) and highest for “other fruits” (apples, pears, citrus fruits) (99%). The mean percentage misclassified for all food groups was 12%.
[view this table]
|Table 5. Mean daily food group intakes estimated by printed EPIC-Norfolk FFQaand online Food4Me FFQ (n=113).|
[view this table]
|Table 6. Spearman’s correlation coefficients (SCC) and cross-classification of quartiles of food group intake.|
[view this figure]
|Figure 4. Bland and Altman plots for selected food groups with mean difference and limits of agreement (solid line represents mean difference and dotted lines represent the limits of agreement).|
Principal Results and Comparisons With Other Work
The present study demonstrates the development of a novel online FFQ and its comparison with the validated printed EPIC-Norfolk FFQ . The Food4Me FFQ was developed to capture dietary intake in an online fashion. It was designed to include a range of portion size options and incorporated food photographs to aid food recognition and portion size estimation.
Overall, the current results demonstrate good agreement between the online Food4Me FFQ and the printed EPIC-Norfolk FFQ for both nutrient and food group intakes. Cross-classification of daily energy and nutrient intakes showed good agreement between the two FFQs indicating that the online Food4Me FFQ generates ranks of dietary intakes that are highly comparable with the previously validated printed EPIC-Norfolk FFQ. Similar to many previous published studies, classification into exact plus adjacent quartiles ranged from 77% to 97% and exact disagreement/misclassification ranged between 0% and 5% [21,28,44,45]. Furthermore, Bland and Altman plots demonstrated an acceptable level of agreement between the two methods for energy and total fat and carbohydrate intakes as a percentage of total energy.
In the present study, there were moderate to high correlation coefficients, ranging from .43 to .86, between the methods for individual nutrients. For the majority of nutrients analyzed, the correlation coefficients were within the ranges recommended by Willet et al  and Masson et al ; 25 out of 29 of the nutrients had a correlation ≥.50 of which four had a correlation >.70, indicating that estimates of intake obtained using the Food4Me FFQ are strongly correlated with those from the printed EPIC-Norfolk FFQ. The mean correlation coefficient was .60, which is within the range of those reported in the literature [16,32]. Furthermore, the ranges of correlation coefficients obtained in the present study are similar to those reported by Kristiansen et al . The latter authors compared estimates of nutrient intake obtained at two time points by a semi-quantitative FFQ that had undergone slight modifications and obtained correlations ranging from .43 to .75.
Analyses at the food group level also showed good agreement between the EPIC-Norfolk and Food4Me FFQs. Validation studies of FFQs examining food group intakes have reported SCC ranging from .46 to .87  and .14 to .90 . In the current study, correlation coefficients >.50 were obtained for the majority of food groups showing that the Food4Me FFQ has reasonable ranking ability for food group intake estimates, and is comparable with the EPIC-Norfolk FFQ. Good agreement between the two methods for quartile classification of food group intakes was also observed with more than 75% of participants correctly classified into the same or adjacent quartiles for each of the food groups analyzed.
While we have demonstrated that the online Food4Me FFQ shows good agreement with the printed EPIC-Norfolk FFQ overall, some disagreement was observed between the two FFQs, particularly in relation to energy intakes. Similar to findings by Beasley et al , in the current study, the online FFQ yielded higher estimates for mean nutrient intakes in comparison to the printed FFQ. However, the percentage of participants classified as under-reporters was 33.6% higher with the printed EPIC-Norfolk FFQ compared with the online Food4Me FFQ, indicating that the Food4Me FFQ may have a greater ability to capture usual dietary intake than the EPIC-Norfolk FFQ.
Multiple factors could have contributed to the discrepancies observed in mean daily intakes and the occurrence of under-reporting between the two methods. First, numerous differences in food group consumption were observed between the two FFQs. For 17 of the 35 food groups (49%), mean daily intakes were significantly higher with the online Food4Me FFQ compared with the printed EPIC-Norfolk FFQ. In addition, Bland and Altman plots demonstrated a poor level of agreement between the methods for 20 of the 35 food groups (57%). The low level of agreement for food group consumption between the two methods is most likely attributable to differences in reported food intake between both FFQs. The Food4Me FFQ included an additional 27 food items not present in the printed EPIC-Norfolk FFQ and as a result several of the 35 food groups analyzed in the current study did not contain equal numbers of food items for both methods. For example, the food group “fish and fish products/dishes” consisted of 10 food items for the Food4Me FFQ compared with six food items from the EPIC-Norfolk FFQ. Such variances in the number of food items aggregated into food groups (for both FFQs) could partly explain the differences observed in mean daily food group intakes between the two methods.
The additional 27 food items listed in the Food4Me FFQ would have offered participants a greater selection of food items and, as a result, both food group consumption and subsequent nutrient intakes may be more reflective of true dietary intake. However, it is also possible that the additional food items included in the Food4Me FFQ may have resulted in an overestimation of consumption frequency for particular nutrients and food groups. This can occur when several food items of a single food group are listed in a questionnaire . Furthermore, despite both FFQs being conducted within 4 weeks (to minimize temporal changes in dietary intakes), there is also the potential that the consumption of a particular food item/group was recalled/reported with one FFQ and not the other (impacting on nutrient and energy intakes).
A second factor potentially contributing to the differences observed between the two FFQs is portion size estimation. The online Food4Me FFQ incorporates a selection of portion sizes for the majority of food items, as opposed to applying a standard one to each food item. Presuming all participants consume standard portion sizes is a generalization and the use of standard portion sizes in heterogeneous population groups is likely to result in additional inaccuracies . Printed questionnaires are limited in their ability to collect complex information (eg, portion size consumption) due to practical restrictions in-built in the questionnaire format . It is more feasible to obtain complex information with online questionnaires, which can be designed to embed multiple photographs to aid with portion size estimation. However, questionnaires with multiple portion size options will exhibit more variability compared to those without variable portion sizes (which may be a truer estimate of actual intakes) . When examining food group intakes, Ocke et al  found that the use of photographs for dairy desserts resulted in an overestimation of milk and milk products. Other studies have reported that photographs have a positive effect on the respondents’ ability to accurately estimate portion sizes [14,32] and further investigation is needed to determine whether the use of photographs in the Food4Me FFQ aids portion size estimation and food recognition.
Third, the differences we observed between the two methods in relation to daily nutrient intakes may have been related to variances in the nutritional composition databases utilized in both methods, as the nutritional composition database used to calculate daily nutrient intakes is more up to date than that used for analysis of the printed EPIC-Norfolk FFQ . The nutritional composition database used in the EPIC-Norfolk FFQ is based on the revised and extended 5thedition of McCance and Widdowson’s The Composition of Foods plus supplemental volumes, while the nutritional composition data for the Food4Me FFQ is based on the 6thand 5theditions of McCance and Widdowson’s The Composition of Foods plus all nine supplemental volumes [38,41].
Strengths and Limitations
The strengths of the current study include the cross-over design and adequate sample size [46,54]. In addition, all participants included in the analyses completed both questionnaires within 1 month, minimizing the likelihood of changes in dietary intake. Limitations of the current study include the use of self-reported weight and height measurements for all participants. BMR was estimated using the Henry equations and therefore any errors in these self-reported measurements could have impacted on the frequency of under-reporting that we observed in the results. Another limitation of the current study is that the majority of participants involved were recruited from within the universities and are therefore representative of a convenient sample rather than a nationally representative sample. As the accuracy of any dietary assessment method using self-report depends ultimately on the cooperation and motivation of the participants, further testing of the Food4Me FFQ will be necessary to establish its wider utility.
In conclusion, the online Food4Me FFQ has good agreement with the previously validated printed EPIC-Norfolk FFQ for assessing both nutrient and food group intakes of healthy young adults. While some differences were observed between the methods, particularly in relation to mean daily nutrient and food group intakes, good agreement was observed at both the nutrient and food group level using a variety of analyses. The most common use of an FFQ is not to measure absolute intake but to rank individuals by their food and nutrient intakes . In the current study, the Food4Me FFQ was able to generate ranks of nutrient and food group intakes that were highly comparable with the validated EPIC-Norfolk FFQ, with levels of agreement from quartile cross-classification similar to many previous published studies. Therefore, the good agreement with the printed EPIC-Norfolk FFQ combined with its ease of use make the online Food4Me FFQ a useful tool for ranking individuals based on their nutrient intake and could be potentially valuable for use in other epidemiological studies.
This work was funded by Food4Me (KBBE.2010.2.3-02, Project no. 265494). The authors would like to thank Creme Global (Dublin, Ireland) for their considerable involvement in the development of the online Food4Me FFQ and the participants for their time and dedication throughout the study.
All authors participated in the design of the study. HF, RF, CG, and ALM carried out the data collection. HF, MCW, LB, and ERG carried out the statistical analyses and drafted the manuscript. All authors critically reviewed and approved the final manuscript submitted for publication.
Conflicts of Interest
Multimedia Appendix 1
Food4Me Food Frequency Questionnaire screenshots.[PDF File (Adobe PDF File), 240KB]
- Roberts CK, Barnard RJ. Effects of exercise and diet on chronic disease. J Appl Physiol (1985) 2005 Jan;98(1):3-30 [FREE Full text] [CrossRef] [Medline]
- Kant AK. Dietary patterns: biomarkers and chronic disease risk. Appl Physiol Nutr Metab 2010 Apr;35(2):199-206. [CrossRef] [Medline]
- Fialkowski MK, McCrory MA, Roberts SM, Tracy JK, Grattan LM, Boushey CJ. Evaluation of dietary assessment tools used to assess the diet of adults participating in the Communities Advancing the Studies of Tribal Nations Across the Lifespan cohort. J Am Diet Assoc 2010 Jan;110(1):65-73 [FREE Full text] [CrossRef] [Medline]
- Beechy L, Galpern J, Petrone A, Das SK. Assessment tools in obesity - psychological measures, diet, activity, and body composition. Physiol Behav 2012 Aug 20;107(1):154-171. [CrossRef] [Medline]
- Jain M, Howe GR, Rohan T. Dietary assessment in epidemiology: comparison on food frequency and a diet history questionnaire with a 7-day food record. Am J Epidemiol 1996 May 1;143(9):953-960 [FREE Full text] [Medline]
- Biró G, Hulshof KF, Ovesen L, Amorim Cruz JA, EFCOSUM Group. Selection of methodology to assess food intake. Eur J Clin Nutr 2002 May;56 Suppl 2:S25-S32 [FREE Full text] [CrossRef] [Medline]
- Thompson FE, Subar AF, Loria CM, Reedy JL, Baranowski T. Need for technological innovation in dietary assessment. J Am Diet Assoc 2010 Jan;110(1):48-51 [FREE Full text] [CrossRef] [Medline]
- Wrieden W, Peace H, Armstrong J, Barton K. A short review of dietary assessment methods used in National and Scottish Research Studies. 2003. URL: http://multimedia.food.gov.uk/multimedia/pdfs/scotdietassessmethods.pdf [accessed 2014-03-05] [WebCite Cache]
- Rutishauser IH. Dietary intake measurements. Public Health Nutr 2005 Oct;8(7A):1100-1107. [Medline]
- Tucker KL. Assessment of usual dietary intake in population studies of gene-diet interaction. Nutr Metab Cardiovasc Dis 2007 Feb;17(2):74-81. [CrossRef] [Medline]
- Johnson RK. Dietary intake--how do we measure what people are really eating? Obes Res 2002 Nov;10 Suppl 1:63S-68S. [CrossRef] [Medline]
- Falomir Z, Arregui M, Madueño F, Corella D, Coltell Ó. Automation of food qestionnaires in medical studies: a state-of-the-art review and future prospects. Comput Biol Med 2012 Oct;42(10):964-974. [CrossRef] [Medline]
- Ngo J, Engelen A, Molag M, Roesle J, García-Segovia P, Serra-Majem L. A review of the use of information and communication technologies for dietary assessment. Br J Nutr 2009 Jul;101 Suppl 2:S102-S112. [CrossRef] [Medline]
- Foster E, Matthews JN, Nelson M, Harris JM, Mathers JC, Adamson AJ. Accuracy of estimates of food portion size using food photographs--the importance of using age-appropriate tools. Public Health Nutr 2006 Jun;9(4):509-514. [Medline]
- Matthys C, Pynaert I, De Keyzer W, De Henauw S. Validity and reproducibility of an adolescent web-based food frequency questionnaire. J Am Diet Assoc 2007 Apr;107(4):605-610. [CrossRef] [Medline]
- González Carrascosa R, García Segovia P, Martínez Monzó J. Paper and pencil vs online self-administered food frequency questionnaire (FFQ) applied to university population: a pilot study. Nutr Hosp 2011 Dec;26(6):1378-1384 [FREE Full text] [CrossRef] [Medline]
- Labonté MÈ, Cyr A, Baril-Gravel L, Royer MM, Lamarche B. Validity and reproducibility of a web-based, self-administered food frequency questionnaire. Eur J Clin Nutr 2012 Feb;66(2):166-173. [CrossRef] [Medline]
- Thompson FE, Byers T. Dietary assessment resource manual. J Nutr 1994 Nov;124(11 Suppl):2245S-2317S [FREE Full text] [Medline]
- Segovia-Siapco G, Singh P, Jaceldo-Siegl K, Sabaté J. Validation of a food-frequency questionnaire for measurement of nutrient intake in a dietary intervention study. Public Health Nutr 2007 Feb;10(2):177-184. [CrossRef] [Medline]
- Araujo MC, Yokoo EM, Pereira RA. Validation and calibration of a semiquantitative food frequency questionnaire designed for adolescents. J Am Diet Assoc 2010 Aug;110(8):1170-1177. [CrossRef] [Medline]
- Dehghan M, del Cerro S, Zhang X, Cuneo JM, Linetzky B, Diaz R, et al. Validation of a semi-quantitative Food Frequency Questionnaire for Argentinean adults. PLoS One 2012;7(5):e37958 [FREE Full text] [CrossRef] [Medline]
- Tucker KL, Smith CE, Lai CQ, Ordovas JM. Quantifying diet for nutrigenomic studies. Annu Rev Nutr 2013 Jul;33:349-371. [CrossRef] [Medline]
- Vereecken CA, Covents M, Sichert-Hellert W, Alvira JM, Le Donne C, De Henauw S, HELENA Study Group. Development and evaluation of a self-administered computerized 24-h dietary recall method for adolescents in Europe. Int J Obes (Lond) 2008 Nov;32 Suppl 5:S26-S34. [CrossRef] [Medline]
- Arab L, Tseng CH, Ang A, Jardack P. Validity of a multipass, web-based, 24-hour self-administered recall for assessment of total energy intake in blacks and whites. Am J Epidemiol 2011 Dec 1;174(11):1256-1265 [FREE Full text] [CrossRef] [Medline]
- Touvier M, Kesse-Guyot E, Méjean C, Pollet C, Malon A, Castetbon K, et al. Comparison between an interactive web-based self-administered 24 h dietary record and an interview by a dietitian for large-scale epidemiological studies. Br J Nutr 2011 Apr;105(7):1055-1064. [CrossRef] [Medline]
- Liu B, Young H, Crowe FL, Benson VS, Spencer EA, Key TJ, et al. Development and evaluation of the Oxford WebQ, a low-cost, web-based method for assessment of previous 24 h dietary intakes in large-scale prospective studies. Public Health Nutr 2011 Nov;14(11):1998-2005. [CrossRef] [Medline]
- Kroke A, Klipstein-Grobusch K, Voss S, Möseneder J, Thielecke F, Noack R, et al. Validation of a self-administered food-frequency questionnaire administered in the European Prospective Investigation into Cancer and Nutrition (EPIC) Study: comparison of energy, protein, and macronutrient intakes estimated with the doubly labeled water, urinary nitrogen, and repeated 24-h dietary recall methods. Am J Clin Nutr 1999 Oct;70(4):439-447 [FREE Full text] [Medline]
- Goulet J, Nadeau G, Lapointe A, Lamarche B, Lemieux S. Validity and reproducibility of an interviewer-administered food frequency questionnaire for healthy French-Canadian men and women. Nutr J 2004 Sep 13;3:13 [FREE Full text] [CrossRef] [Medline]
- Christensen SE, Möller E, Bonn SE, Ploner A, Wright A, Sjölander A, et al. Two new meal- and web-based interactive food frequency questionnaires: validation of energy and macronutrient intake. J Med Internet Res 2013;15(6):e109 [FREE Full text] [CrossRef] [Medline]
- Adamson AJ, Collerton J, Davies K, Foster E, Jagger C, Stamp E, Newcastle 85+ Study Core Team. Nutrition in advanced age: dietary assessment in the Newcastle 85+ study. Eur J Clin Nutr 2009 Feb;63 Suppl 1:S6-18. [CrossRef] [Medline]
- Liu L, Wang PP, Roebothan B, Ryan A, Tucker CS, Colbourne J, et al. Assessing the validity of a self-administered food-frequency questionnaire (FFQ) in the adult population of Newfoundland and Labrador, Canada. Nutr J 2013 Apr;12:49 [FREE Full text] [CrossRef] [Medline]
- Beasley JM, Davis A, Riley WT. Evaluation of a web-based, pictorial diet history questionnaire. Public Health Nutr 2009 May;12(5):651-659 [FREE Full text] [CrossRef] [Medline]
- Food4Me. 2011. URL: http://www.food4me.org/ [accessed 2013-08-23] [WebCite Cache]
- Walsh M, Kuhn S. Developments in personalised nutrition. Nutrition Bulletin 2012 Nov 08;37(4):380-383. [CrossRef]
- Bingham SA, Gill C, Welch A, Cassidy A, Runswick SA, Oakes S, et al. Validation of dietary assessment methods in the UK arm of EPIC using weighed records, and 24-hour urinary nitrogen and potassium and serum vitamin C and carotenoids as biomarkers. Int J Epidemiol 1997;26 Suppl 1:S137-S151 [FREE Full text] [Medline]
- IUNA. National Adult Nutrition Survey. 2011. URL: http://www.iuna.net/?P=106 [accessed 2013-08-23] [WebCite Cache]
- Tinuviel Software. Food Composition Databank. 2003. URL: http://www.tinuvielsoftware.com/wisp41.htm [accessed 2013-08-23] [WebCite Cache]
- IUNA. National Adults Nutrition Survey (NANS) methodology. 2011. URL: http://www.iuna.net/wp-content/uploads/2011/04/methods.pdf [accessed 2013-08-23] [WebCite Cache]
- Goldberg GR, Black AE, Jebb SA, Cole TJ, Murgatroyd PR, Coward WA, et al. Critical evaluation of energy intake data using fundamental principles of energy physiology: 1. Derivation of cut-off limits to identify under-recording. Eur J Clin Nutr 1991 Dec;45(12):569-581. [Medline]
- European Prospective Investigation of Cancer. EPIC-Norfolk food frequency questionnaire (FFQ) entry and processing documentation. 2013. URL: http://www.srl.cam.ac.uk/epic/epicffq/websitedocumentation_2111.html [accessed 2013-10-09] [WebCite Cache]
- Welch AA, Luben R, Khaw KT, Bingham SA. The CAFE computer program for nutritional analysis of the EPIC-Norfolk food frequency questionnaire and identification of extreme nutrient values. J Hum Nutr Diet 2005 Apr;18(2):99-116. [CrossRef] [Medline]
- Hébert JR, Peterson KE, Hurley TG, Stoddard AM, Cohen N, Field AE, et al. The effect of social desirability trait on self-reported dietary measures among multi-ethnic female health center employees. Ann Epidemiol 2001 Aug;11(6):417-427. [Medline]
- Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986 Feb 8;1(8476):307-310. [Medline]
- Torheim LE, Barikmo I, Hatløy A, Diakité M, Solvoll K, Diarra MM, et al. Validation of a quantitative food-frequency questionnaire for use in Western Mali. Public Health Nutr 2001 Dec;4(6):1267-1277. [Medline]
- Chiu TH, Huang HY, Chen KJ, Wu YR, Chiu JP, Li YH, et al. Relative validity and reproducibility of a quantitative FFQ for assessing nutrient intakes of vegetarians in Taiwan. Public Health Nutr 2013 Jun 20:1-8. [CrossRef] [Medline]
- Willett W. Nutritional epidemiology. New York: Oxford University Press; 1998.
- Masson LF, McNeill G, Tomany JO, Simpson JA, Peace HS, Wei L, et al. Statistical approaches for assessing the relative validity of a food-frequency questionnaire: use of correlation coefficients and the kappa statistic. Public Health Nutr 2003 May;6(3):313-321. [CrossRef] [Medline]
- Kristiansen AL, Lillegaard IT, Lande B, Andersen LF. Effect of changes in an FFQ: comparing data from two national dietary survey instruments among 2-year-olds. Br J Nutr 2013 Jan 28;109(2):363-369. [CrossRef] [Medline]
- Wong JE, Parnell WR, Black KE, Skidmore PM. Reliability and relative validity of a food frequency questionnaire to assess food group intakes in New Zealand adolescents. Nutr J 2012;11:65 [FREE Full text] [CrossRef] [Medline]
- Bohlscheid-Thomas S, Hoting I, Boeing H, Wahrendorf J. Reproducibility and relative validity of food group intake in a food frequency questionnaire developed for the German part of the EPIC project. European Prospective Investigation into Cancer and Nutrition. Int J Epidemiol 1997;26 Suppl 1:S59-S70 [FREE Full text] [Medline]
- Subar AF, Thompson FE, Kipnis V, Midthune D, Hurwitz P, McNutt S, et al. Comparative validation of the Block, Willett, and National Cancer Institute food frequency questionnaires: the Eating at America's Table Study. Am J Epidemiol 2001 Dec 15;154(12):1089-1099 [FREE Full text] [Medline]
- Kristal AR, Peters U, Potter JD. Is it time to abandon the food frequency questionnaire? Cancer Epidemiol Biomarkers Prev 2005 Dec;14(12):2826-2828 [FREE Full text] [CrossRef] [Medline]
- Ocké MC, Bueno-de-Mesquita HB, Goddijn HE, Jansen A, Pols MA, van Staveren WA, et al. The Dutch EPIC food frequency questionnaire. I. Description of the questionnaire, and relative validity and reproducibility for food groups. Int J Epidemiol 1997;26 Suppl 1:S37-S48 [FREE Full text] [Medline]
- Cade J, Thompson R, Burley V, Warm D. Development, validation and utilisation of food-frequency questionnaires - a review. Public Health Nutr 2002 Aug;5(4):567-587. [CrossRef] [Medline]
|BMI: body mass index|
|BMR: basal metabolic rate|
|EER: estimated energy requirements|
|EPIC: European Prospective Investigation of Cancer|
|FFQ: food frequency questionnaire|
|MUFA: monounsaturated fatty acids|
|NANS: National Adult Nutrition Survey|
|PUFA: polyunsaturated fatty acids|
|SCC: Spearman’s correlation coefficient|
|SFA: saturated fatty acids|
|TE: total energy|
|Vitamin A (RE): vitamin A retinol equivalents|
|Edited by G Eysenbach; submitted 19.11.13; peer-reviewed by P Stumbo, L Korkalo; comments to author 11.02.14; revised version received 07.03.14; accepted 05.05.14; published 09.06.14|
Please cite as:
Forster H, Fallaize R, Gallagher C, O’Donovan CB, Woolhead C, Walsh MC, Macready AL, Lovegrove JA, Mathers JC, Gibney MJ, Brennan L, Gibney ER
Online Dietary Intake Estimation: The Food4Me Food Frequency Questionnaire
J Med Internet Res 2014;16(6):e150
END, compatible with Endnote
BibTeX, compatible with BibDesk, LaTeX
RIS, compatible with RefMan, Procite, Endnote, RefWorks
Add this article to your Mendeley library
Add this article to your CiteULike library
Copyright©Hannah Forster, Rosalind Fallaize, Caroline Gallagher, Clare B O’Donovan, Clara Woolhead, Marianne C Walsh, Anna L Macready, Julie A Lovegrove, John C Mathers, Michael J Gibney, Lorraine Brennan, Eileen R Gibney. Originally published in the Journal of Medical Internet Research (http://www.jmir.org), 09.06.2014.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www.jmir.org/, as well as this copyright and license information must be included.