Both Comprehensive and Brief Self-Administered Diet History Questionnaires Satisfactorily Rank Nutrient Intakes in Japanese Adults

1Dietary questionnaires are useful tools for assessing long-term dietary habits. However, because food culture and dietary habits vary by country, these questionnaires need to be developed specifically for each country. To assess Japanese diets, Sasaki et al developed a comprehensive self-administered diet history questionnaire (DHQ) and a brief self-administered DHQ (BDHQ), which use both the food frequency and diet history methodologies., The DHQ yields information on the dietary intake of 150 food and beverage items but requires about 45 to 60 minutes to answer, whereas the BDHQ provides information on only 58 items but requires only about 15 to 20 minutes to complete.

2Because dietary questionnaires do not necessarily estimate true food intake, their validity should be evaluated. Several US studies have compared the validity of short and long versions of questionnaires,– but no such studies have been reported in other countries. We previously examined the validity of the DHQ and BDHQ in terms of food group intake and found that they had similar and satisfactory validity for a wide range of food groups among adult Japanese women and men in terms of their ability to both estimate median values and rank individuals in a population. However, the validity of these questionnaires with regard to nutrient intake has not been compared. Because a finding of satisfactory validity for food groups does not reflect satisfactory validity for nutrients, we must examine the validity of nutrient intakes for the DHQ and BDHQ despite the existence of a validation study for food group intakes.

3Here, we compared the relative validity of energy and nutrient intakes derived from the DHQ and BDHQ among healthy women and men in Japan against 16-day semi-weighed dietary records (DRs).

4Details of the study design, participant characteristics, and dietary assessment methods have been reported elsewhere., Briefly, the study was conducted in 3 areas of Japan, namely, Osaka (urban), Nagano (rural inland), and Tottori (rural coastal). In each area, we recruited apparently healthy women aged 30 to 69 years who were willing to participate with a cohabitating husband, such that each of the 10-year age class strata had 8 women, regardless of the age of the husbands. Thus, a total of 96 women and 96 men were invited. Before the study, the study purpose and protocol were explained at group orientations. Written informed consent was obtained from each participant. The study did not undergo ethical approval because it was conducted before the advent of ethical guidelines for epidemiology research in Japan. However, the use of the data of this study was approved by the Ethics Committee at the University of Tokyo Faculty of Medicine (No. 3421). Ultimately, 92 women aged 31 to 69 years and 92 men aged 32 to 76 years who completed the protocol were included in the present analysis.

5Between November 2002 and September 2003, the participants completed four 4-nonconsecutive-day semi-weighed DRs, 1 in each season, at intervals of approximately 3 months, ie, November and December 2002 (autumn), February 2003 (winter), May 2003 (spring), and August and September 2003 (summer) (Figure). Each of the 4 recording days consisted of 1 weekend day and 3 weekdays within about 2 weeks. During the orientation session, registered dietitians gave the participants both written and verbal instructions on how to keep the DR, provided recording sheets and a digital scale, and asked the participants to record and weigh all foods and beverages consumed on the recording day. All collected records were checked by trained registered dietitians in the respective local center and then again in the study center. A total of 1299 food and beverage items appeared in the dietary records.

6The participants also answered the DHQ and BDHQ 4 times, once in each season, at intervals of approximately 3 months from November 2002 to September 2003. In each season, the DHQ and BDHQ were answered approximately 2 days before the start of the dietary recording period, in the order of BDHQ before DHQ (Figure). Responses to the DHQ and BDHQ were checked at least twice for completeness by dietitians. When missing answers or logical errors were identified, the participants were asked to complete the questions again.

7The DHQ is a 16-page semi-quantitative questionnaire that asks about the consumption frequency and portion size of selected foods to estimate the dietary intake of 150 food and beverage items during the preceding month.,,– The DHQ consists of 7 sections: (1) general dietary behavior, (2) usual cooking methods, (3) consumption frequency and amount of alcoholic beverages, (4) consumption frequency and semi-quantitative portion size of selected food and nonalcoholic beverage items, (5) dietary supplements, (6) consumption frequency and semi-quantitative portion size of staple foods, soup for noodles, and miso (fermented soybean paste) soup, and (7) open-ended items for foods consumed more than once a week but not appearing in the DHQ.

8The BDHQ is a 4-page fixed-portion questionnaire that asks about the consumption frequency of selected foods, but not about portion size, to estimate the dietary intake of 58 food and beverage items during the preceding month. The BDHQ consists of 5 sections: (1) intake frequency of food and nonalcoholic beverage items, (2) daily intake of rice and miso soup, (3) frequency of drinking and amount per drink for alcoholic beverages, (4) usual cooking methods, and (5) general dietary behavior.

9Food and beverage items contained in the DHQ and BDHQ were selected from foods commonly consumed in Japan, mainly from a food list used in the National Health and Nutrition Survey of Japan, while standard portion sizes and the sizes of bowls for rice and cups for miso soup were derived from several recipe books for Japanese dishes., Because the use of dietary supplements is uncommon in Japan (8% of the general population), we did not considered information on dietary supplements that was derived from the DHQ.

10All statistical analyses were done using SAS statistical software, version 9.2 (SAS Institute Inc., Cary, NC, USA) and were performed separately for women and men. Crude values for the intake of energy and 42 selected nutrients were estimated based on the intake of food items obtained with the DR or respective questionnaire and the corresponding food composition list in the Standard Tables of Food Composition in Japan. Because intakes of most nutrients were positively correlated with energy intake, energy-adjusted values were also calculated by the residual method using a regression model, while the density method was used to compute the amount of each nutrient consumed daily, as a percentage from daily energy intake for energy-containing nutrients or per 10 MJ of daily energy intake for non–energy-containing nutrients. All statistical analyses were performed on log-transformed values to account for non-normality. Relative validity of the DHQ and BDHQ were assessed in terms of their ability to estimate representative values by comparing mean values and ranking ability, using Pearson correlation coefficients. We compared nutrient intakes derived from the first DHQ (DHQ1) and the first BDHQ (BDHQ1) with those from the four 4-day DRs. Although the reference period differed between DHQ1 or BDHQ1 and the four 4-day DRs (1 conducted in each season), our purpose here was to examine whether a single DHQ or BDHQ for dietary habits during the previous month could represent habitual dietary intake over a longer period. Additionally, we also examined the mean of the 4 DHQs and BDHQs (mDHQ and mBDHQ) using the same method to compare annual intake from the questionnaires with that of the DR.

11Energy and nutrient intakes in the crude and energy-adjusted models for DR, DHQ1, BDHQ1, mDHQ, and mBDHQ are presented as means. Statistically significant differences between the DR and each of the questionnaires were determined with the paired t-test using 1-sided values. A P value less than 0.05 was considered to indicate a significant difference. Pearson correlation coefficients between the DR and DHQ1, BDHQ1, mDHQ, and mBDHQ were then calculated. Additionally, because a 16-day DR may be insufficient to account for intra-individual variation, we also calculated deattenuated correlation coefficients using intraindividual and interindividual variances. Correlation coefficients obtained from the DHQ and BDHQ were compared using the Meng-Rosenthal-Rubin method to compare overlapping correlation coefficients, with tested correlation coefficient pairs considered statistically different when z was greater than 1.96, with a significance level of 5%.

12Selected characteristics of the participants are shown in Table 1. The means of energy and nutrient intakes derived from the DR, DHQ, and BDHQ are shown in Table 2 for women and Table 3 for men. Energy intakes estimated by BDHQ1 and mBDHQ for women and by DHQ1, BDHQ1, and mBDHQ for men were significantly lower than those of the DR. Regarding crude intakes of the 42 nutrients, no significant differences between the DR and DHQ1 and between the DR and BDHQ1 were observed for 28 (67%) and 18 (43%) nutrients, respectively, for women and 11 (26%) and 27 (64%) nutrients for men (data not shown). Regarding nutrients that were energy-adjusted by the residual method, no significant differences between the DR and DHQ1 and between the DR and BDHQ1 were observed for 18 (43%) and 14 (33%) nutrients, respectively, for women and 4 (10%) and 21 (50%) nutrients for men. Regarding nutrients that were energy-adjusted by the density method, no significant differences between the DR and DHQ1 and between the DR and BDHQ1 were observed for 21 (50%) and 8 (19%) nutrients, respectively, for women and 13 (31%) and 16 (38%) nutrients for men. For mDHQ and mBDHQ, the numbers of nutrients that did not significantly differ between the DR and mDHQ and between the DR and mBDHQ were similar to those that did not significantly differ between the DHQ1 and BDHQ1. For women, the respective numbers of nutrients for the crude, residual, and density methods were 22 (52%), 15 (36%), and 17 (40%) for the mDHQ and 19 (45%), 12 (29%), and 7 (17%) for the mBDHQ. The respective values for men were 14 (33%), 11 (26%), and 15 (36%) for the mDHQ and 24 (57%), 15 (36%), and 11 (26%) for the mBDHQ.

13Pearson correlation coefficients for estimates of crude and energy-adjusted nutrient intakes and energy intake on the DR versus the DHQ and BDHQ are shown in Table 4 for women and Table 5 for men. For women, the correlation values of energy with the DR were 0.30 for DHQ1 and 0.29 for BDHQ1; for men, the respective values were 0.41 and 0.23. Additionally, correlation values of energy with the DR for women were 0.38 for mDHQ and 0.42 for mBDHQ; for men, the respective values were 0.49 and 0.38. For nutrients expressed as crude values, median (interquartile range) correlation coefficients for women were 0.40 (0.34 to 0.47) for DHQ1 and 0.39 (0.32 to 0.48) for BDHQ1; for men, the respective values were 0.40 (0.33 to 0.45) and 0.34 (0.25 to 0.40). For energy-adjusted nutrient intakes, almost all correlation values were improved as compared with crude values. Regarding energy-adjusted nutrient intakes by the residual method, median correlation values for women were 0.50 (0.43 to 0.58) for DHQ1 and 0.49 (0.38 to 0.57) for BDHQ1; for men, the respective values were 0.45 (0.37 to 0.55) and 0.51 (0.38 to 0.59). The results of correlation coefficients of energy-adjusted nutrients using the density method were similar to those of the residual method. For energy-adjusted nutrients using the density method, median correlation coefficients for women were 0.49 (0.43 to 0.58) for DHQ1 and 0.49 (0.38 to 0.57) for BDHQ1; for men, the respective values were 0.47 (0.36 to 0.56) and 0.49 (0.36 to 0.59) (data not shown). All deattenuated correlation values were 1.0 to 1.4 times the respective original values. For deattenuated correlations of energy-adjusted nutrient intakes by the residual method, median values for women were 0.57 (0.50 to 0.64) for DHQ1 and 0.54 (0.45 to 0.61) for BDHQ1, with respective values for men of 0.50 (0.42 to 0.59) and 0.56 (0.41 to 0.63). Almost none of the correlation coefficients for DR and BDHQ1 differed from the respective values for DR and DHQ1. Medians of correlation coefficients between the DR and mDHQ or mBDHQ were slightly higher (1.1 to 1.4 times) than the respective values for DHQ1 and BDHQ1. Further, almost none of the correlation coefficients for DR and mBDHQ differed from the respective values for DR and mDHQ.

14We examined the relative validity of energy and nutrient intakes estimated by the DHQ and BDHQ, using a 16-day DR as reference. For energy and many (ie, 50%–90%) energy-adjusted nutrients, mean intakes estimated by DHQ1 and BDHQ1 were significantly different from those estimated by the DR. The results for mDHQ and mBDHQ were similar. These results suggest that the DHQ and BDHQ are satisfactory for estimating mean values for only a limited number of nutrients.

15However, for many (ie, 57%–83%) correlations with energy-adjustment and many (ie, 79%–90%) deattenuated correlations with energy-adjustment, values greater than 0.4 were observed between the DR and DHQ1 and between the DR and BDHQ1. Although the correlation with the DR was slightly higher for mDHQ and mBDHQ than for DHQ1 and BDHQ1, the latter questionnaires had reasonable correlations for many nutrient intakes. Furthermore, correlation coefficients between the DR and DHQ1 and between the DR and BDHQ1 did not significantly differ for almost all nutrients, which was also the case for mDHQ and mBDHQ. Thus, the DHQ and BDHQ had satisfactory ranking ability for many nutrients in this population.

16Of the several US studies that have compared short and long versions of the same questionnaire with regard to energy and nutrient intake,– all have shown reasonable correlations between the 2 versions. Our results in a Japanese population are consistent with those findings. Wakai reviewed the validity of dietary questionnaires developed and validated in Japan and showed that medians of coefficients between DRs and questionnaires on energy and nutrients ranged from 0.31 to 0.56 in the studies investigated. Our present DHQ and BDHQ thus performed similarly to these other Japanese dietary questionnaires. Wakai then considered long (97 or more food items) versus short questionnaires (<70 items) and reported that the median correlation coefficient for nutrients in an individual questionnaire ranged from 0.42 to 0.52 for the long form and from 0.31 to 0.45 for the short form. On the basis of those results, he concluded that long questionnaires had slightly higher validity in the estimation of nutrient intake. In the present study, the corresponding values for women were 0.57 for the DHQ and 0.54 for the BDHQ, with respective values for men of 0.50 and 0.56. Our study indicated that, unlike the results of the review, the BDHQ did not necessarily have a worse correlation as compared with the DHQ.

17There are several explanations for why both the 150-item DHQ and the 58-item BDHQ had relatively high correlations. First, the food and beverage items in the BDHQ may accurately reflect the foods commonly consumed in Japan. Second, the DHQ requires more time to complete than the BDHQ, which may reduce the accuracy of responses for the former questionnaire. Third, the BDHQ differs from the DHQ in that it does not ask about portion size. Given that misestimation of portion size is a source of error in answering these questionnaires, some participants might have inaccurately answered at least some of the DHQ questions on portion size. Although it is not clear why the median correlation coefficient of the DHQ was higher than that of the BDHQ in women while the opposite was true in men, the degree of the effects of questionnaire length and misestimation of portion size may differ by sex. Because of its lesser burden, the BDHQ might be preferred in the investigation of many of the nutrients examined here. However, it should be mentioned that correlations of some nutrients (eg, zinc in women and cryptoxanthin in men) were higher for the DHQ than for the BDHQ. Further, the DHQ was better than the BDHQ in estimating mean intakes of vitamin C and some other nutrients. These results indicate that the DHQ is more suitable than the BDHQ in estimating intakes of some nutrients.

18Although the median correlations for both the DHQ and BDHQ were relatively good, in men, correlation coefficients for retinol in the DHQ1 and BDHQ1, α-carotene in the DHQ1, and n-3 polyunsaturated fatty acid in the BDHQ1 were much lower than those of other nutrients in the present study and much lower than those of any nutrient in other studies.– However, the correlations of these nutrients from mDHQ and mBDHQ were slightly higher than those from DHQ1 and BDHQ1. If we were to examine the intakes of these nutrients in an epidemiologic study using the DHQ or BDHQ, multiple DHQs and BDHQs would be needed.

19Several limitations of this study should be mentioned. First, although we assumed that energy and nutrient intakes derived from the DR were the standard, DRs are susceptible to measurement error due to erroneous recording and potential changes in eating behavior. Nevertheless, as compared with 24-hour dietary recall or other instruments that rely on memory, errors in DRs are thought to have a lower correlation with errors in the DHQ and BDHQ. Biomarkers are a better standard for some nutrients, and any errors in such markers are independent of errors in the questionnaire. Validity testing of the DHQ for Japanese adults by using 24-hour urine, serum, or doubly labeled water methods has revealed satisfactory validity for some nutrients.– Validation studies with quantitative biomarkers are also necessary for the BDHQ. Second, because the algorithm used to calculate food intake from the BDHQ was written using previous information and other unpublished observations, it had insufficient reliability. Finally, the generalizability of the present results is limited because the participants were not representative of the general Japanese population and might have been highly health conscious. Additionally, while all the women answered the questionnaires themselves, a considerable number of men (32%) answered with the help of their wives. The validity of the 2 questionnaires in men might have been lower had they answered the questionnaires themselves.

20In conclusion, this study showed that the DHQ and BDHQ had satisfactory ranking ability for energy-adjusted intakes of many nutrients, as compared with DRs, in a population of Japanese men and women, although ability of these instruments to estimate mean values was satisfactory for only a limited number of nutrients. Additionally, although the correlations obtained from mDHQ and mBDHQ were better than the respective values from DHQ1 and BDHQ1, even a single DHQ and BDHQ had reasonable correlations for many nutrient intakes. The validity of the BDHQ was similar to that of the DHQ. These findings support the idea that, for many energy-adjusted nutrients, both the DHQ and BDHQ can be used in large-scale epidemiologic studies in Japan.