Accuracy of an online meal-based diet questionnaire for assessing overall diet quality and each meal type in Japanese adults

This cross-sectional study was based on the data collected from fourteen (of the forty-seven) prefectures between August and October 2021. Recruitment of participants and data collection were conducted by our research dietitians (n 60) with expertise in collecting DR data^(32,33). First, healthy women aged 30–69 years who were willing to participate and were living with their husbands were recruited for this study. For each prefecture, two women from each 10-year age category (30–39, 40–49, 50–59 and 60–69 years) were selected. Their husbands were then recruited (irrespective of age), resulting in 112 individuals by sex. The sample size was determined primarily based on the recommendation made by Cade et al. that for validation studies, a sample size of at least 50 and preferably much larger (e.g. 100 or more subjects) is desirable⁽¹⁴⁾. To minimise the dropout rate, the potential participants were restricted to individuals who had full understanding of the procedure and showed willingness to complete the entire survey. Meanwhile, dietitians, individuals living with a dietitian, those who had received dietary counselling from a doctor or dietitian, those taking insulin treatment for diabetes, those undergoing dialysis treatment, those without sufficient Internet access, those who had difficulty answering the web-based questionnaires and pregnant or lactating women were excluded. Only participation in the study as a couple (one woman and one man) was permitted.

The study schedule is shown in Fig. 1. Each participant was asked to answer the web version of the MDHQ (web MDHQ). After an interval of 7–10 d (to ensure the completion of the web MDHQ), a 4-non-consecutive-day weighed DR was conducted for 2 weeks. Finally, after an interval of at least 1 d, the paper version of the MDHQ (paper MDHQ) was completed. We designed this schedule because the main purpose of this study was to evaluate the validity of the web MDHQ; thus, a web MDHQ survey was performed prior to the conduct of DR. A total of 111 women aged 30–69 years and 111 men aged 30–76 years completed the study. As a financial incentive, each couple received a voucher worth 5000 Japanese Yen (31 British Pound as of 1 October 2022) after the study.

The study was conducted in accordance with the guidelines of the Declaration of Helsinki, and all procedures involving humans were approved by the Ethics Committee of the University of Tokyo Faculty of Medicine (protocol code: 2020326NI; date of approval: 29 January 2021). Written informed consent was obtained from all participants.

Details of the MDHQ have been published elsewhere^(16,17). Briefly, the MDHQ is a self-administered questionnaire designed to estimate dietary intake in the previous month for each meal type (breakfast, morning snack, lunch, afternoon snack, dinner and night snack). The MDHQ comprises three parts. Part 1 of the MDHQ includes quantitative questions on the consumption frequency of generic food groups (Tier 1 food groups) for each meal type, with potential answers of 0–7 d/week. Part 2 of the MDHQ includes questions on the relative consumption frequency of sub-food groups (Tier 2 food group) within one of the generic food groups (Tier 1 food group), with possible answers of ‘always, often, sometimes, rarely, and never’. By combining the information derived from Parts 1 and 2, the number of foods that can be estimated efficiently can be increased but within a limited number of questions. Part 3 of the MDHQ enquires about the general eating behaviours, including the amount of brown rice consumed, the relative consumption frequency of wholegrain bread and whether bread was consumed with jam, honey, etc. or with fat spread. Finally, the MDHQ includes the assessment of basic characteristics (sex, age, body height, body weight, education level and current smoking status).

In the MDHQ, information on portion sizes was not collected (except for alcoholic beverages, for which the overall consumption frequency and portion size were assessed in Part 2). This decision was based on our previous observation that the Brief-type Diet History Questionnaire (BDHQ), which assesses the consumption frequency of fifty-eight food items but does not collect information on portion sizes and applies fixed portion sizes for dietary intake calculation, had a similar efficacy in estimating the food and nutrient intake as the Diet History Questionnaire (DHQ), which assesses not only the consumption frequency but also the portion size of 150 food items^(27,34,35). The limited usefulness of portion size information has also been supported by several previous studies^(36,37). All the food groups included in the MDHQ (see online Supplementary Table S1↗), as well as the sex-specific and meal-type-specific portion sizes, were determined based on the 16-d weighed DR data collected from 121 Japanese women and 121 Japanese men, comprising 206 837 food item entries⁽¹⁶⁾.

In the present study, two delivery modes of MDHQ, which are identical in terms of content, were used: web MDHQ and paper MDHQ. The web MDHQ was prepared using Google Forms. Each question was answered by each participant, with non-response not permitted. All responses to the web MDHQ automatically allocated into a spreadsheet format were downloaded from Google Drive. The paper MDHQ used in this study was an A4 21-page questionnaire. Responses to all questions were checked by the research dietitians and staff at the study centre. If any responses were missing, the participants were asked to answer the questions again in person or by phone. All answers in the paper MDHQ were manually entered into a spreadsheet in duplicate, and any discrepancies were checked and corrected. Data obtained using the web MDHQ and paper MDHQ were converted to a dataset suitable for dietary intake calculation.

On the basis of a series of ad hoc computer algorithms in the MDHQ⁽¹⁶⁾, estimated intakes of Tier 1 and 2 food groups were calculated. Estimated intakes of energy and nutrients were calculated using food intake information and the 2015 version of the Standard Tables of Food Composition in Japan⁽³⁸⁾. Component scores needed for the calculation of HEI-2015 were calculated using the Japanese version⁽²⁷⁾ of the US Food Patterns Equivalents Database⁽³⁹⁾. These calculations were done for each meal type, and the overall intake was calculated as the sum of the intake of each meal type.

The 4-non-consecutive-day weighed DR was selected as the reference method in this validation study. Each recording period consisted of three weekdays (Monday–Friday, except for national holidays) and one weekend day (Saturday, Sunday or national holidays). For each couple, a recording day was allocated within 2 weeks by research dietitians. Each couple was provided with recording sheets and a digital scale (KS-274, Dretec, Japan; ±2 g precision for 0–500 g and ±3 g precision for 500–2000 g). After receiving written and verbal instructions from the assigned research dietitian, as well as an example of a completed diary sheet, each participant was requested to document and weigh all consumed foods and drinks, both inside and outside of their homes, on each recording day. On certain occasions when weighing was inconvenient to carry out (e.g. dining out), they were instructed to document as much information as possible, including the brand name of the food and the consumed portion size (based on typical household measures), as well as the details of the leftovers.

The recording sheets used in each survey day were submitted directly to the research dietitian after the survey was completed, who then reviewed the forms and, whenever necessary, sought additional information or modified the record via phone or in-person interview. All collected records were then reviewed by the research dietitians and trained staff at the study centre. In accordance with a standardised procedure, the portion sizes estimated using household measures were converted into weights, and the individual food items were coded based on the 2015 version of the Standard Tables of Food Composition in Japan⁽³⁸⁾. A total of 1297 food codes were used in the DR.

The structure of the food diary sheet used was based on a typical Japanese eating pattern, which comprised breakfast, lunch, dinner and snacks; these meal types were prescribed in the diary. For DR data, the name of the meal type used in the present analysis was based on this classification. As was the case in the MDHQ, estimated intakes of energy and nutrients and component scores needed for the calculation of HEI-2015 were calculated using the 2015 version of the Standard Tables of Food Composition in Japan⁽³⁸⁾ and the Japanese version⁽²⁷⁾ of the US Food Patterns Equivalents Database⁽³⁹⁾, respectively. These calculations were done for each meal type, and the overall intake was calculated as the sum of the intake of each meal type. For all dietary variables, the mean daily values within the 4-d period were used for each individual.

As described elsewhere^(24–27), HEI-2015 is a composite measure of compliance with the 2015–2020 Dietary Guidelines for Americans⁽⁴⁰⁾. The HEI-2015 is a 100-point scale, with a higher score indicating a better quality of diet. The HEI-2015 consists of nine adequacy components, namely, total fruits (maximum score: 5), whole fruits (5), total vegetables (5), greens and beans (5), whole grains (10), dairy products (10), total protein foods (5), seafood and plant proteins (5) and fatty acids as the ratio of the sum of PUFA and MUFA to SFA (10), and four moderation components, namely, refined grains (10), Na (10), added sugars (10) and saturated fats (10). For each meal type and overall diet for each participant for each dietary assessment method, we calculated the HEI-2015 component and total scores based on energy-adjusted values of dietary intake, namely, amount per 4184 kJ (1000 kcal) of energy or percentage of energy, except for the fatty acids component⁽²⁷⁾.

The overall diet quality was also assessed using the NRF9.3, as described in detail elsewhere^(27–31). The NRF9.3 is a composite measure of the nutrient density of the diet, calculated as the sum of the percentage of reference daily values for nine qualifying nutrients, namely, protein, dietary fibre, vitamin A, vitamin C, vitamin D, Ca, Fe, K and Mg, minus the sum of the percentage of reference daily values for three disqualifying nutrients, namely, added sugars, saturated fats and Na. Reference daily values were determined for sex and age categories, based on the Dietary Reference Intakes for Japanese, 2020⁽⁴¹⁾, namely, the RDA for protein, vitamin A, vitamin C, Ca, Fe and Mg and tentative dietary goal for preventing lifestyle-related diseases for dietary fibre, K, saturated fats and Na. For added sugars, the conditional recommendation advocated by the WHO (i.e. upper limit of 5 % of energy)⁽⁴²⁾ was used because of the lack of a recommended value for added sugars in Japan, as well as their low intake levels⁽⁴³⁾. We calculated the NRF9.3 component and total scores based on the daily intake of each nutrient for each participant, which was adjusted for energy intake by the density method and then normalised for the sex- and age-specific Estimated Energy Requirement for a moderate level of physical activity (from the Dietary Reference Intakes for Japanese, 2020⁽⁴¹⁾) and expressed as a percentage of the reference daily value⁽²⁷⁾. These calculations were done for each meal type and for overall diet. Higher NRF9.3 scores indicated a better quality of the diet. A maximum possible score of 900 indicated a diet in which intakes per given amount of energy were above the reference daily values for the nine qualifying nutrients but below the reference daily values for the three disqualifying nutrients. In this study, dietary supplements were not considered during the nutrient intake calculation in any of the dietary assessment methods because it was our intention to assess nutrient intake from foods and beverages only.

Statistical analyses were performed using the SAS statistical software (version 9.4; SAS Institute Inc.). A two-tailed P value of < 0·05 was considered significant. Analyses were stratified by sex and conducted to determine the overall intake and intake for each meal type (breakfast, lunch, dinner and snacks). The dietary variables examined in this study included the total and component scores of HEI-2015 and NRF9.3, in addition to energy intake (MJ/d) and percentage of energy intake from each meal type. The amounts of snacks consumed were combined for analysis due to their relatively low intake in both methods. All dietary data were expressed as median and 25th and 75th percentiles. To assess the estimation ability at the group level, the median values of estimates derived from the MDHQ were compared with those derived from the DR using the Wilcoxon signed-rank test. The Spearman correlation coefficients between the MDHQ and DR estimates were used to assess the ability of the MDHQ to rank individuals in a population. In addition, agreement of the total scores of HEI-2015 and NRF9.3 between the MDHQ and DR was assessed using the Bland–Altman analysis⁽⁴⁴⁾. To examine the proportional bias between the MDHQ and DR, the Bland–Altman analysis was accompanied by the linear regression analysis⁽⁴⁵⁾. Identical analyses were conducted to assess the web MDHQ and paper MDHQ. The findings (tables and figures) on the web MDHQ are provided in the “Results” section, whereas those on the paper MDHQ are provided as online Supplementary Materials.

Identical analyses of the paper MDHQ were conducted (online Supplementary Table S2↗ for median estimations, online Supplementary Table S3↗ for Spearman correlations, online Supplementary Fig. S2↗ for Bland–Altman plots for the HEI-2015 total score and online Supplementary Fig. S3↗ for Bland–Altman plots for the NRF9.3 total score). The results for the paper MDHQ were generally similar to those for the web MDHQ, except for somewhat high Spearman correlation coefficients between the paper MDHQ and the DR.