Validation of Dietary Intake Estimated by Web-Based Dietary Assessment Methods and Usability Using Dietary Records or 24-h Dietary Recalls: A Scoping Review

The goal was to summarize studies comparing the accuracy of web-based dietary assessments with those of conventional face-to-face or paper-based assessments using 24-h dietary recall or dietary record methods in the general population. Using two databases, mean differences and correlation coefficients (CCs) for intakes of energy, macronutrients, sodium, vegetables, and fruits were extracted from each study independently by the authors. We also collected information regarding usability from articles reporting this. From 17 articles included in this review, the mean dietary intake differences in the web-based dietary assessment compared to conventional methods, were −11.5–16.1% for energy, −12.1–14.9% for protein, −16.7–17.6% for fat, −10.8–8.0% for carbohydrates, −11.2–9.6% for sodium, −27.4–3.9% for vegetables, and −5.1–47.6% for fruits. The CC was 0.17–0.88 for energy, protein, fat, carbohydrates, and sodium, and 0.23–0.85 for vegetables and fruits. In three out of four studies reporting usability, more than half of the participants preferred the web-based dietary assessment. In conclusion, % difference and CC of dietary intake were acceptable in both web-based dietary records and 24-h dietary recalls. The findings from this review highlight the possibility of wide-spread application of the web-based dietary assessment in the future.


Introduction
Accurate measurements of dietary intake are fundamental for health and nutrition research. To date, the dietary record or 24-h dietary recall methods have been used for national nutrition surveys worldwide [1][2][3][4]. New technologies in dietary intake assessment, including web-based dietary assessments, have been shown to reduce issues associated with the conventional collection of dietary data (i.e., paper-based or face-to-face), such as cost, participation rates and accuracy of data collected [5,6]. Recently, the internet has become widely used [7,8], and there has been increasing interest in developing web-based dietary assessment methods [9]. Therefore, it is important to examine the possibility of applying web-based dietary assessment.
Moreover, it has become difficult to conduct in-person interviews due to the COVID-19 pandemic in 2019. In Japan, the National Health and Nutrition Survey has been annually

Inclusion Criteria
Articles eligible for inclusion: (1) were peer-reviewed original articles written in English or Japanese; (2) evaluated the validity of web-based dietary assessment compared to conventional methods (paper-based or face-to-face) as a gold standard using dietary records or 24-h dietary recalls; (3) assessed whole day intake of energy, protein, fat, carbohydrates, sodium, vegetables, and fruit; and (4) were conducted on individuals aged ≥1 years consuming a usual diet. We included studies in which parents responded on behalf of their children, and those where participants were different between web-based and conventional dietary assessments but belonged to the general population.
Articles excluded were: (1) reviews, case studies, conference reports, or abstracts; (2) focusing on specific meal occasions (e.g., breakfast); (3) using biomarkers as a gold standard; (4) focusing on patients with any diseases; (5) focusing on participants who did not eat usual diets, such as obesity, athletes, soldiers, vegetarians, and pregnant or lactating women. We included all eligible studies regardless of the study design.
This review scoped the validity of the estimated intake of energy, macronutrients, and sodium because these nutrients were important for health conditions, and vegetable and fruit intake were suggested to be important to prevent noncommunicable diseases by the World Health Organization [18]. We excluded studies using biomarkers as a gold standard because these assessment methods could not estimate representative values of dietary intake. The validity of web-based dietary assessment summarized in this study was defined as (1) an ability to estimate representative values (mean or median) of dietary intakes, and (2) an ability to rank individuals based on dietary intakes.

Study Selection
The titles and abstracts of all articles were screened (UM, AF, AN). After removing duplicate articles, the full texts of the articles were assessed for eligibility (UM, AF, YS, SH, EK, AN). Any disagreements were discussed and resolved by consensus or by another reviewer, if necessary (HT, RT). The reference lists of the articles identified were manually searched and screened (UM, YS, SH). The reference lists of the articles identified during this process were also examined by hand search (UM) to further identify potentially relevant articles and previous review articles [5,6,14,[19][20][21][22][23][24][25][26][27][28][29][30][31]. All articles were screened or evaluated independently by at least two reviewers.

Data Collection
Data were extracted by two independent reviewers using an electronic spreadsheet. The information extracted included author names, publication year, study year, country, characteristics of participants (sex, age, sample size, and exclusion criteria), number of survey days, tool name, food, recipe, or dish database incorporated into the tool, and devices for web-based dietary assessment method, number of survey days of conventional dietary assessment method, mean or median intake of total energy, protein, fat, carbohydrate, sodium, and vegetables and fruit, and their correlation coefficients (CCs) both of web-based and conventional dietary assessments. When articles reported on usability such as accessibility of web-based dietary assessments, we collected this information.

Evaluation of the Relative Validity of Web-Based Dietary Assessment
Using the extracted information, the percentage difference of mean or median intake was calculated as the following formula: (mean or median dietary intake calculated by the web-based-mean or median dietary intake calculated by the conventional)/mean or median dietary intake calculated by the conventional × 100. According to the criteria proposed by Lombard et al. [32], the percent differences were categorized ≤10 as good; 11.0-20.0 as acceptable; >20.0 as poor, and the CCs between dietary intake estimated using the web-based and the conventional dietary assessment methods was evaluated: ≥0.50 was categorized as good; 0·20-0·49 as acceptable; <0.20 as poor. We assessed the quality score (points) of each validation study using the definition of the previous study [33]. A non-homogeneous sample (e.g., sex) was 0.5, sample size over 100 was 0.5, comparison of dietary intake was 1, correlations (0.5 for crude, 1 for energy-adjusted, and 1.5 for deattenuated or intraclass), 0.5 for agreement of dietary intake (e.g., Bland-Altman plot), 1 for data correction (face-to-face interview), 0.5 for seasonality, and 1.5 for considering supplement intake. Scores ranged from 0 to 7, and validation studies were classified as very good (>5.0), good (5.0-3.5), acceptable (3.0-2.5), and poor (<2.5). We did not conduct an adjustment/weighting of the correlation coefficient according to the quality score and a rating of the adjusted/weighted correlation, because we assessed the crude dietary intake as possible.

Difference of Dietary Intake between the Web-Based and the Conventional Dietary Assessment
Among the seven studies which used the web-based dietary record method [35,37,38,44,46,48,49], the one study [44] which compared to the 24-h dietary recall had the largest difference regarding fat intake (17.6%). Among the three studies [34,36,47] that compared the web-based 24-h dietary recall to the dietary record, the differences between intakes of energy, protein, fat, or carbohydrates were similar to the differences observed between web-based and conventional 24-h dietary recalls. One study [36] showed relatively large differences regarding vegetable (−27.4%) and fruits (47.6%) intakes, compared with the three studies [39][40][41] that applied the conventional 24-h dietary recall for comparisons. Among the seven studies which used the web-based dietary record method [35,37,38,44,46,48,49], there was only one study [44] that applied a 24-h dietary recall as the conventional method. Among the 10 studies which used the web-based 24-h dietary recall method [34,36,[39][40][41][42][43]45,47,50], there were 3 studies [34,36,47] that applied the dietary record as the conventional method. Table 2 showed the mean (or median) difference in dietary intake estimated by the web-based dietary assessment method compared to those estimated by the conventional method. In web-based dietary records, the range of difference was −68.0 to 125.4 kcal (−3.1% to 6.6% difference) for energy intake, −10.4 to 3.1 g (−11.8 to 3.9% difference) for protein intake, −10.1 to 11.2g (−16.7 to 17.6% difference) for fat intake, −35.0 to 19.9 g (−10.7 to 7.8% difference) for carbohydrate intake, and 43 to 400mg (1.4 to 5.5% difference) for sodium intake. One study reported vegetable and fruit intake, and the difference was −8.0 g (−14.0% difference) for vegetable intake and 1.0 g (0.8% difference) for fruit intake. In web-based 24-h dietary recall, the range of differences was −241.0 to 342.0 kcal (−11.5% to 16.1% difference) for energy intake, −11.5 to 11.0 g (−12.1 to 14.9% difference) for protein intake, −15.0 to 10.0 g (−15.4 to 13.2% difference) for fat intake, −21.0 to 18.0 g (−7.9 to 8.0% difference) for carbohydrate intake, and −287 to 305 mg (−11.2 to 9.6% difference) for sodium intake. The range of difference was −65.0 to 3.3 g (−27.4 to 3.9% difference) for vegetable intake and −14.0 to 120.0 g for fruit intake (−5.1 to 47.6% difference).

Difference of Dietary Intake between the Web-Based and the Conventional Dietary Assessment
Among the seven studies which used the web-based dietary record method [35,37,38,44,46,48,49], the one study [44] which compared to the 24-h dietary recall had the largest difference regarding fat intake (17.6%). Among the three studies [34,36,47] that compared the web-based 24-h dietary recall to the dietary record, the differences between intakes of energy, protein, fat, or carbohydrates were similar to the differences observed between web-based and conventional 24-h dietary recalls. One study [36] showed relatively large differences regarding vegetable (−27.4%) and fruits (47.6%) intakes, compared with the three studies [39][40][41] that applied the conventional 24-h dietary recall for comparisons.    Table 3 shows the Spearman, Pearson, or Intraclass CC between dietary intake estimated by the web-based and the conventional dietary assessment methods. Overall, 12 out of 17 studies reported Spearman, Pearson, or Intraclass CC between dietary intakes. In the web-based dietary records, the CC of dietary intake was 0.37 to 0.87 for energy, 0.41 to 0.78 for protein, 0.33 to 0.75 for fat, 0.31 to 0.82 for carbohydrates, and 0.59 for sodium. There were no reports about CC of vegetable and fruit intake. In the web-based 24-h dietary recalls, the CC of dietary intake was 0.44 to 0.88 for energy, 0.41 to 0.83 for protein, 0.33 to 0.75 for fat, 0.36 to 0.82 for carbohydrate, and 0.17 and 0.75 for sodium. In food groups, there were two reported studies: the CCs were 0.23 to 0.84 for vegetable intake, and 0.56 to 0.85 for fruit intake. Among the four studies which used the web-based dietary record method [35,38,44,49], the CCs for energy, protein, fat, or carbohydrate were low in one study [38] which applied different number of survey days for web-based and conventional methods. Among the three studies [34,36,47] that compared the web-based 24-h dietary recall to the dietary record, one study [47] showed the lowest CCs for energy, protein, carbohydrate, and sodium intakes.

Usability of the Web-Based Dietary Assessment Methods
Of 17 articles, 5 reported the usability of web-based dietary assessment methods [36,37,48,49], although details were not described in one study (Table 4) [44]. More than half of the participants in the three studies preferred the web-based dietary assessment [36,37,49]. However, according to Vereecken et al. [48], in the study where 58 parents were asked to either complete a web-based dietary assessment or a conventional (paper-based) dietary record, only 5 of them chose the web-based assessment. On the other hand, in the same study [48], the majority of 164 parents who were requested to answer the web-based dietary assessment indicated that the web-based dietary assessment was user-friendly (79%), attractive (68%), well-liked (66%), provided adequate instructions (92%), clear information (93%), clear pictures (91%), and easy-to-find food items (88%). Regarding registration in the dietary intake using web-based dietary assessments, participants in two of three studies answered the web-based dietary assessment was easy to report their dietary intakes on site compared with the conventional method [36,37], while a study reported that participants preferred the conventional method (paper-based) [49], and one study did not collect that information [48]. In the number of survey days, only one study reported the period for mastering the web-based dietary assessment and the most common (48.4%) answer was taking two days [37].

Discussion
This scoping review of 17 studies summarized the representative values and the CC of the intakes of energy, macronutrients, sodium, vegetables, and fruits between the webbased and conventional dietary assessments in the general population. Although the % difference in energy, protein and sodium intake was slightly smaller in web-based dietary records than in those 24-h dietary recalls, the mean difference in dietary intake estimated from both methods was within approximately ± 20% difference for energy and nutrient intakes. The CC between the web-based dietary records and conventional methods were acceptable (0.37-0.87 for energy and 0.31-0.82 for nutrients and sodium, but NA for food groups), and slightly better for web-based 24-h dietary recalls regarding energy (0.44-0.88) but not for nutrients and sodium (0.17-0.83).
We evaluated the representative dietary intakes for a whole day using web-based dietary records, and our results supported the previous study [15]. In a previous study, web-based dietary records underestimated dietary intake compared with the conventional method [15]. However, this review assessed only mobile phone applications and included studies of not assessing dietary intake for the whole day, face-to-face conventional dietary assessments, or compared with biomarkers as conventional methods [15]. In addition, we showed that the validity of dietary intake using web-based 24-h dietary recalls was also reasonable. A previous review evaluated the web-based 24-h dietary recall methods [13], and another review reported the validity of mobile phone applications or web-based dietary assessments in childhood and adolescence [14]. As this scoping review focused on the representative value of dietary intake in the general population, we excluded several studies included in previous studies [13,14]. The reasons for the excluded studies were listed in Table S1 [5,6,14,[19][20][21][22][23][24][25][26][27][28][29][30][31].

Characteristics of Included Studies
In this review, only one study included participants over 70 years old [45]. In addition, included studies mainly recruited individuals who regularly used the internet or computers [34][35][36][37]40,41,[43][44][45]50], although most of the included studies did not describe devices. In the national dietary survey in the UK conducted using a web-based 24-h dietary recall, participants who were not able to access through the internet or felt uncomfortable using it, or unconfident completing the dietary recalls, were supported by the survey team via telephone or internet video (Zoom) conferencing [12]. Therefore, when conducting a web-based dietary assessment for the general population including a wide range of age groups or people without internet access, it may be necessary to consider a support system. Ten studies used databases based on foods or meals [34][35][36][39][40][41][42]46,48,49]. Although two of the remaining studies used food composition tables [45,50] instead of food items as the database, five studies did not mention the database [37,38,43,44,47]. Because the database is important to estimate dietary intakes, it needed to be described in the validation studies.
As a result of this review study, the majority of studies were conducted in developed countries and no studies involving immigrants or ethnic language minority groups were included. As the number of immigrants and people living overseas increases and racial diversity expands, web-based dietary survey methods for researching these groups will also be needed. In the UK, for participants who were not able to access through the internet or felt uncomfortable using it, for example the elderly, they were helped by using devices to assist [12]. Similarly, the National Health and Nutrition Examination Survey in the US hired interpreters of various languages when necessary [104]. It is important to pay attention to the diversification of languages or emigrations when conducting web-based national dietary surveys in the future.

The Differences and Correlation Coefficients between Dietary Intake between the Web-Based and the Conventional Dietary Assessment
Overall, the mean or median nutrient intake calculated from the web-based dietary assessment method was underestimated compared to those from the conventional method, and it was more frequent with the web-based 24-h dietary recall method than with the web-based dietary records. Although the number of studies that described the database was limited, the number of food and beverage items was larger in the web-based dietary records than in web-based 24-h dietary recalls, and it could affect accuracy when participants registered food or beverage items.
The mean difference in dietary intake estimated by the web-based dietary assessment compared to those estimated by the conventional method was within approximately ± 20% difference for energy and nutrient intake except for fat intake in both web-based dietary records and those 24-h dietary recalls (Table 2). Regarding fat intake, sources of error between the web-based food record and the conventional method suggested that food portion estimation was the greatest source of error (50%) [44]. Another reason may be the differences in food selection, such as selecting similar but not exactly the same foods [44]. In another study [49], the database using web-based dietary assessment included 3 million items, and it seemed difficult for the subject to select the exact foods. On the other hand, Timon et al. [36] indicated a small number of databases (751 food and beverage items) as a reason for less agreement. This should be taken into account when constructing databases for web-based dietary assessments. CC of dietary intake was similar results to % difference (Table 3). Additionally, CC for sodium was lower compared to other nutrients. The webbased dietary surveys extracted in this study included dish-based dietary records [35]. In the case of the dish-based survey, the differences from the dietary records may be observed because the detailed seasonings were not checked. In addition, some systems recruited a method of selecting several portion sizes for intake [40,41]. These would have the potential to affect the estimating of the intake of salt and other seasonings.
Vegetable or fruit intake has been reported in only one study in web-based dietary records [48] and five studies in those 24-h dietary recalls. In the web-based 24-h dietary recalls, intakes of vegetables and fruits were mean different by over 20% [36], considering that dietary intakes over two different time points have been noted. CCs of vegetable and fruit intake were not reported in the web-based dietary record and only three studies were represented in those 24-h dietary recalls. Therefore, further studies are needed to confirm this association.
In the two studies, the comparison between the web-based and conventional dietary assessment methods was based on the two different populations [43,48]. Although CCs were not reported in these studies [43,48], the difference in energy intake was as small as 5% between web-based and conventional methods. As the validity of the representative values was reasonable even when the subjects of the web-based and conventional dietary assessments were different, web-based dietary assessments may be useful for a dietary assessment for the general population.

Usability of the Web-Based Dietary Assessment
Participants, recruited through the internet or e-mail, reported the web-based dietary assessment method made it easy to register their dietary intake on-site compared with the conventional method [36,37]. The web-based dietary assessments may be useful for people who are used to the internet which may be possible to improve the response rates of dietary assessments by using web-based dietary assessments. In a study of participants who preferred the conventional method (paper-based) [49], participants needed to register the web-based dietary record from over 3 million food items and in portion size details (e.g., grams and milliliters), which was difficult to complete, and estimate portion sizes. In the previous study, it was also noted to be difficult to identify the correct food in web-based 24-h dietary recalls [13]. According to Vereecken et al. [48], only 8% of parents who were asked to either complete a web-based or paper-based dietary assessment chose the webbased dietary assessment compared to the conventional method. The authors considered that parents did not want to spend time on the new web-based dietary assessment tools. However, in the same study [48], the majority of parents who were requested to answer the web-based dietary assessment indicated that they were favorable to the web-based method. Participants do not prefer to use the web-based dietary assessment for the first time. However, they may find it easy when they use it. Therefore, it should be considered for the recruitment and collecting of information for each generation or people who are unfamiliar with the internet. In addition, for accurate estimated dietary intake in web-based dietary assessments, careful database construction is required, such as the number of foods, dish composition, and portion size setting in both web-based dietary records and 24-h dietary recalls.

Strength and Limitation
The strengths of this scoping review are its investigation of the validity of representative dietary intake using the web-based dietary assessment methods compared to the conventional methods by dietary records or 24-h dietary recalls, which are used worldwide, and % difference and CC of dietary intake were reasonable in the web-based dietary records and 24-h dietary recalls. Although discussion for some nutrients is warranted, this result indicated that web-based dietary assessments may be effective as dietary assessments for the general population, especially for people who are familiar with the internet or e-mail. However, there are several limitations to this scoping review. First, despite our use of wide search terms and hand-search of reference lists, we could not capture all relevant publications. In addition, we used only two independent literature searches and searched in only two languages. Second, although we considered including the latest study, only one study included participants over 70 years old [45]. Further studies on dietary assessment methods for older people and people without familiarity with the internet are needed. Since some eligible studies were recruiting participants on the internet or who could use e-mail, we could not consider such as the participation rate of participants who do not use the internet and their usability of them. Finally, we used the criteria employed by Lombard et al. [32] to evaluate the difference between the web-based and conventional methods. This criterion also used by Luevano-Contreras et al. [105,106], is used to identify differences between two survey instruments when examining the validity and reliability of the FFQ. However, caution may need to be exercised in its interpretation, as it is not a criterion developed to compare between web-based and conventional methods.

Conclusions
The validity of web-based dietary assessment methods compared to conventional methods using dietary records and 24-h dietary recalls for dietary intake was reasonable. Although the databases and web tools used differed for most studies, the representative values of dietary intakes were consistent to some extent between the web-based and conventional dietary assessment, and it may be possible to use the web-based dietary assessment. Three of four studies reported usability, and more than half of the participants preferred the web-based dietary assessment. The results of this review provided the information needed to examine the validity of a web-based dietary evaluation method compared to conventional methods. Moreover, our results serve as both a reference and an indication for further research, as well as for the development of non-face-to-face dietary assessment methods. This review highlights findings that may be applied in designing and investigating the performance of the web-based dietary survey in the future.
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/nu15081816/s1, Table S1: List of excluded studies and reason for exclusion; Table S2: Score of validity study.