Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2023 May 1.
Published in final edited form as: J Am Nutr Assoc. 2021 Mar 11;41(4):360–382. doi: 10.1080/07315724.2021.1887775

Comparison of the Diet ID platform to the Automated Self-Administered 24-Hour (ASA24) Dietary Assessment Tool for assessment of dietary intake

Gabrielle Turner-McGrievy 1, Brent Hutto 2, John A Bernhart 3, Mary Wilson 4
PMCID: PMC8634522  NIHMSID: NIHMS1757612  PMID: 33705267

Abstract

Background:

Collecting multiple 24-hour recalls (24HR) can be burdensome necessitating alternative methods to assess dietary intake in the research setting.

Methods:

This cross-sectional study compared the use of the Diet ID™ online platform with three unannounced 24HR assessed via the Automated Self-Administered 24-Hour recall among participants in the Nutritious Eating with Soul (NEW Soul) study. NEW Soul participants (n=68; 100% African American, 79% female, mean age 50.7±9.6 years) were randomized to follow one of two healthy soul food diets: vegan or omnivorous. For the present study, data from both groups were combined. Energy intake, dietary quality (Healthy Eating Index), and macro-/micronutrients densities per 1,000 kcals, as assessed by either the averaged values of the three 24HR or the Diet ID. Descriptive statistics (means, standard deviations and Spearman rank correlations) summarized each nutrient as measured by the Diet ID and ASA24. Bland-Altman plots were used as the main method to assess agreement between the two measures.

Results:

Nutrients from the Diet ID were generally higher than the 24HR except for the HEI score (69.6±12.2 ASA24 vs. 51.1±34.5 Diet ID). Diet ID reported 950 kcals higher energy intake than ASA24 with the difference being most pronounced at lower ASA24-reported energy intake. There were significant correlations between measures for HEI score, protein, carbohydrates, cholesterol, potassium, copper, thiamin, and vitamins B12 and E. Diet ID is a rapid way to assess dietary intake. There was higher reporting of nutrients using Diet ID compared to the 24HR.

Conclusions:

Future studies should consider comparing these two methods with objective assessments of energy and nutrient intake and using multiple instruments to ensure the strengths of all methods are included.

Keywords: diet assessment, diet, energy intake, 24-hour dietary recall, food frequency questionnaire

Introduction

Assessing dietary intake is important for nutrition research as it can provide valuable information regarding current intake or changes in intake over time [1, 2]. Dietary assessment methods, such as 24-hour dietary recalls (24HR), food frequency questionnaires (FFQ), and food records aim to collect usual dietary intake in order to provide information on energy intake (kcals), macronutrients, and micronutrients [3]. These tools are important in nutrition research to understand patterns of food intake and how these patterns relate to health [4]. Assessing the diet can be burdensome as it requires participants to either log all meals consumed over the course of several days, or the previous 24-hours, or general patterns of eating over the previous months [5, 6].

Multiple, unannounced 24HR, including both weekend and weekdays, are considered a valid method for assessing dietary intake [7-9]. The 24HR requires users to remember all beverages and foods consumed over the previous 24 hours and can be done either by an in-person or over the phone interview [10] or can be computer assisted [11]. Assessing more than one day of dietary intake, especially at least one weekday and one weekend day, increases the ability of capturing more accurate energy intake [12]. Previous studies, however, have found significant underreporting of energy intake when using the 24HR method. Underreporting of energy intake means that foods were omitted during the 24HR, or portion sizes were not correctly recalled, and can therefore lead to incomplete dietary data. For example, one study using an online, self-administered dietary recall found that participants underreported energy intake by an average of 25% as compared to doubly labeled water [13]. Among adults with overweight and obesity, underreporting may be greater, ranging from 35% to 50% underreporting using dietary recalls versus doubly labeled water [14]. In addition, underreporting may be greater in women compared to men [15]. Most all self-assessment methods (24HR, FFQ, and food records) result in underreporting more often than overreporting of energy and other nutrients [16].

Collecting multiple 24HR from study participants can be challenging. For example, using computer-administered, self-directed 24HR, such as the National Cancer Institute’s Automated Self-administered 24HR (ASA24) requires both computer and reading literacy [17]. In addition, because 24HR should be unannounced [2], participants cannot plan for them and may be unable to complete them on the day assigned. The multiple-pass method should also be employed when a 24HR is administered [18]. This method also results in additional time needed to complete a 24HR. Because of the burden that dietary assessment places on participants, alternative methods for valid and reliable dietary assessment have begun to be explored and developed [1, 19].

The Diet ID™ commercial platform (www.DietID.com)[20] is based on pattern recognition by providing groups of food photos for participants to view on a computer or mobile device screen. After answering a few questions about current dietary habits (e.g., food group exclusions and allergies), participants view two images displaying different composite images of foods that reflect established dietary patterns simultaneously and select which image depicting the group of food best represents their current dietary pattern. Food patterns are informed by identified food patterns common in North America, such as data from the NHANES study [20].Using a “this or that” approach, the program then refines the selections by providing new groups of images until the best possible fit is achieved. Similar in nature to an FFQ, the Diet ID program then takes these food groups and patterns and provides data on estimated energy intake, macro-, and micronutrients, along with a Healthy Eating Index 2015 (HEI-2015) score [21]. Nutrient values calculated by the Diet ID platform are listed in Table 1.

Table 1:

Means and Standard Deviations of energy intake, Healthy Eating Index (HEI) score, and nutrients per 1,000 kcals assessed with the Diet ID platform or three 24-hour dietary recalls using the ASA24 and the correlation between the two instruments

Diet ID ASA24 Spearman
rank
correlation		 P-value for
correlation
Energy intake and HEI Mean ±SD Mean ±SD
Kilocalories 2451.1 443.5 1501.5 527.1 r=0.17 p=0.17
HEI 51.1 34.5 69.6 12.2 r=−0.32 p=0.009
Nutrients per 1,000 kcals
Protein (g) 41.6 9.5 37.8 10.5 r=0.29 p=0.02
Carbohydrates (g) 128.6 16.8 133.1 25 r=0.30 p=0.014
Total Sugar (g) 40.9 8.2 51.2 19.6 r=0.00 p=0.984
Dietary Fiber (g) 20 9.1 16.6 7.3 r=0.20 p=0.10
Total fat (g) 38.8 5.6 37 9.3 r=0.08 p=0.51
Saturated Fat (g) 7.6 3.2 9 3.1 r=0.13 p=0.28
Polyunsaturated Fat (g) 11.5 2.2 13.4 4.2 r=0.02 p=0.86
Monounsaturated Fat (g) 16.5 4.3 11.5 4.1 r=0.20 p=0.10
Cholesterol (mg) 77.8 68.9 85.5 78.6 r=0.46 p<0.001
Calcium (mg) 540.4 157.1 498.3 170.5 r=−0.13 p=0.29
Iron (mcg) 9.6 2 9 4.3 r=0.13 p=0.29
Magnesium (mg) 253.5 98.3 222.2 76 r=0.16 p=0.20
Phosphorus (mg) 750.1 155.9 668 149.8 r=0.05 p=0.67
Potassium (mg) 1952 608.2 1705.5 591.4 r=0.24 p=0.045
Sodium (mg) 1142.4 425.5 1757.8 434.4 r=0.20 p=0.10
Zinc (mg) 5.5 1 5.2 1.6 r=0.08 p=0.88
Copper (mg) 1.1 0.4 1 0.3 r=0.29 p=0.018
Selenium (mcg) 62.6 17.5 56.7 33.9 r=0.14 p=0.27
Vitamin C (mg) 101.6 60.6 71.6 46 r=0.23 p=0.06
Thiamin (mg) 1 0.2 1.1 0.7 r=0.30 p=0.014
Riboflavin (mg) 1 0.2 0.9 0.3 r=−0.13 p=0.30
Niacin (mg) 11.7 2.8 11.3 3.6 r=0.08 p=0.52
Vitamin B6 (mg) 1.2 0.3 1.2 0.4 r=0.06 p=0.63
Total Folate (mcg) 312.8 85.6 293.2 132.1 r=0.12 p=0.32
Vitamin B12 (mcg) 1.9 1.1 1.8 1.2 r=0.32 p=0.008
Vitamin A (mcg) 717.5 263.7 563.5 362.6 r=−0.05 p=0.67
Vitamin E (mg) 9.1 4 6.3 2.4 r=0.31 p=0.01
Vitamin K (mcg) 290 205 268.7 246.9 r=0.15 p=0.22
Vitamin D (mcg) 2.8 1.9 1.8 2.2 r=−0.04 p=0.77
Open in a new tab

The goal of this paper is to examine the use of Diet ID to assess dietary intake among participants in the two-year Nutritious Eating with Soul (NEW Soul) study [22] as compared to three unannounced 24HR using the ASA24 (which is how dietary data are being collected and is typically used as a reference instrument for dietary instrument validation studies [23]). Comparing dietary intake via Diet ID vs. the ASA24 with multiple 24HR was conducted for several reasons. First, because collecting multiple 24HR throughout the NEW Soul study presents a high degree of burden on staff and participants, the Diet ID platform was used in conjunction with the three announced 24HR to assess potential use in assessing dietary intake in future studies. Second, because dietary patterns are of interest in nutrition research, we sought to compare a method of selecting dietary patterns to arrive at nutrient intake and diet quality (Diet ID) vs. asking for specific foods and beverages to arrive at nutrient intake and diet quality (24HR). Third, there has been a call for new methods to assess diet that are less burdensome [20] and more accurate [24]. While we could have compared the Diet ID with other methods, such as an FFQ, we chose to compare it to 24HR since FFQs may overestimate intakes of food groups [25] or have less accuracy in estimating short-term energy intakes as compared to 24HR [9, 16]. Lastly, the NEW Soul study represented diets (soul food, vegan, etc.) and populations (African American adults) that had not been previously tested with Diet ID; therefore, it was important to assess the utility of this method with these unique diets and populations. Previously, Diet ID has been tested among mostly white adults [20]. We hypothesized that there would be high correlation between the instruments, but nutrient values would be higher on the Diet ID vs. ASA24, since FFQs tend to overestimate intakes of food groups [25-27] compared to 24HR.

Methods

The NEW Soul study has been described elsewhere [22]. This study is a two-year, behavioral nutrition intervention study comparing two different soul food dietary patterns (vegan and low-fat omnivorous) on the reduction in cardiovascular disease risk factors among African American adults. Adults with obesity or overweight (BMI 25-49.9 kg/m2) who self-identified as African American were recruited to enroll in one of two cohorts, separated by one year (2018-2019). Two cohorts were used so we could accommodate a certain number of participants in our research teaching kitchen and computer data collection lab. There were no differences in the intervention between cohorts. Participants were randomized to follow one of the two dietary patterns for two years and attend weekly, group-based hands-on cooking classes for six months, followed by bi-weekly classes for six months, and monthly classes for 12 months. Dietary intake is assessed using three unannounced 24HR using the ASA24 system (version 2018) [28]. Participants complete assessments at baseline and three, six, 12, and 24 months. Participants complete their first 24HR in a computer lab where they learn how to use the ASA24 system and can receive assistance with completing the recalls. Participants with literacy issues or difficulty with navigating a computer are provided with interview-assisted recalls over the phone, using the ASA24 to guide the recalls.

In order to assess the relationship between intake collected via three unannounced 24HR and one administration of the Diet ID system, this cross-sectional study utilized dietary intake at the six-month assessment from participants in cohort two of the NEW Soul study. The Diet ID platform was not fully available to test at the time that cohort one completed the study so only one time point with cohort two was utilized. Participants came to a university computer lab where they completed both their first of three 24HR and one Diet ID (participants did not know what assessments they would be completing that day). Prior to completing Diet ID, participants were provided with instructions on how to complete the instrument. Both the ASA24 and Diet ID websites were already loaded on the computers and participants were given unique IDs and passwords to access each site. Participants then completed the remaining unannounced 24HR from their home or work computer, smartphone, or tablet. Participants needing extra assistance with completing the 24HR were given opportunities to return to the research site to complete it in person or complete one over the phone. In order to have complete 24HR data, participants needed to complete at least one weekend day and two weekdays of intake. Because Diet ID assesses usual intake, versus intake over the previous 24HR, only one day is needed and there is no need for a weekend and/or weekend day of collection. Dietary intake from the three 24HR were averaged to get a single mean day of dietary intake. To maintain consistency with data collection and cleaning methods utilized in the main NEW Soul trial, any 24HR that was <500kcal was deemed implausible and incomplete [29] and was excluded from the analysis. The Goldberg method was not used because it tends to have higher sensitivity for FFQs as compared to 24HR [30]. Macro- and micronutrients from both Diet ID and ASA24 were analyzed as nutrient density (amount of nutrient per 1,000 kcals) in order to standardize between the methods. Both the ASA24 and Diet ID pull nutrient data from USDA’s nutrient database [31] and then provide a researcher output of nutrient calculation results for kcals, macro-, and micronutrients. HEI from the ASA24 was calculated using the SAS plug-in provided by the National Cancer Institute.

Energy intake obtained from the ASA24 and from the Diet ID was compared to estimated total daily energy expenditure (TDEE), as calculated by the Mifflin-St Jeor equation, using the participants measured height, weight, and reported age and sex [32]. Because most participants were sedentary (mean moderate-to-vigorous physical activity for cohort 2 participants as assessed by accelerometers was 13.6±12.1 min/day) and previous studies have shown that formulas to calculate energy may overestimate energy needs for both adults with overweight and African Americans [33], an activity factor of 1.0 was used as the physical activity level to calculate TDEE [34]. The University Institutional Review Board approved the study protocol and all participants provided written informed consent.

Statistical analysis

Sample size calculations for the main NEW Soul study have been described elsewhere [22]. The present study was a secondary analysis. Descriptive statistics (means, standard deviations and Spearman rank correlations) summarized each nutrient as measured by the Diet ID and ASA24, along with estimated TDEE. Variables were examined for normality and, because many variables were not normally distributed, Spearman correlations were utilized. Paired samples t-tests also examined the relationship between energy estimated from the ASA24 and Diet ID with estimated TDEE. The main method used to assess the agreement and bias between the two measures was Bland-Altman plots [35]. All analyses were completed using R 3.6.3 [36].

Results

There was a total of 97 participants who enrolled in the NEW Soul study as part of cohort 2. Only participants with complete 24HR (three 24HR with one weekend and two weekdays) and a complete Diet ID were included in the present analysis (n=68). Participants who completed both the ASA24 and Diet ID were all African American and mostly female (79%), employed (84%), and well-educated (75% college or greater), with a mean age of 50.7±9.6 years. Participants took a mean of 3 minutes and 49 seconds (range 3:02 to 11:11; median 3:12) to complete the Diet ID. ASA24 does not provide data for participant completion time.

Table 1 provides the mean intakes of energy intake, macronutrients and micronutrients (all nutrients provided by the Diet ID system compared to corresponding nutrients from ASA24), and HEI, along with the correlations between each pair of nutrients (from Diet ID and from ASA24). Every nutrient from the Diet ID output was higher than the nutrients from the ASA24 except for the HEI score, which was lower for Diet ID compared to the ASA24. Correlations were strongest between the ASA24 and Diet ID for HEI score, protein, carbohydrates, cholesterol, potassium, copper, thiamin, and vitamins B12 and E.

Bland-Altman plots for energy, nutrients, and HEI are presented in Figure 1 and described below. For each nutrient and HEI score, there are top and bottom plots in Figure 1. The top plot in each pair is a plot of differences (Diet ID values minus ASA24 values) vs. the ASA24-reported intake. These differences scores show the amount of bias Diet ID scores have relative to ASA24 scores. The resulting mean of those differences is the mean bias for that component (e.g., the mean raw difference). The bottom plot is the absolute value of those individual difference scores plotted against ASA24 intake, which is a measure of how much disagreement there is in absolute terms. Therefore, the mean of absolute differences demonstrates how different Diet ID is likely to be for a person, on average, as compared to the ASA24.

Figure 1:

Figure 1:

Figure 1:

Figure 1:

Figure 1:

Figure 1:

Figure 1:

Figure 1:

Figure 1:

Figure 1:

Figure 1:

Figure 1:

Figure 1:

Figure 1:

Figure 1:

Figure 1:

Figure 1:

Open in a new tab

Bland-Altman plots for energy, nutrients per 1,000 kcals, and Healthy Eating Index comparing the ASA24 dietary recalls to Diet ID. Scatterplots (grey), fitted locally estimated scatterplot smoothing (LOESS) curves (dark purple) and 95% confidence intervals (light purple) for differences between methods versus ASA24 values.

Energy intake

On average, Diet ID reported 950 kcals higher energy intake than ASA24 with the difference being most pronounced at lower ASA24-reported energy intake. For participants with less than approximately 1,200 kcals, the Diet ID reported more than one to two times the ASA24 report. Only one participant reported lower kcal values for Diet ID than ASA24 so the plots and means of absolute difference and bias were nearly identical. Energy intake was also compared to estimated TDEE. Mean estimated TDEE was 1643.6±262.0 kcals per person with Diet ID reporting a mean difference in energy intake of +807.6±35.8 kcals (p<0.001) and ASA24 reporting a mean difference of −142.1±513.0 kcals (p=0.03). Estimated TDEE was significantly correlated with energy reported on both the Diet ID (r=0.77, p <0.001) and ASA24 (r=0.30, p =0.012). Of note, 75% of the sample had lost weight at six months (and thus should be consuming fewer kcals than their TDEE), while 25% (n=17) had either not lost weight or gained weight.

Macronutrients and cholesterol

All nutrient amounts presented are per 1,000 kcals as assessed by either the ASA24 or Diet ID. Protein intake had a relatively flat pattern with an average of 3.9g higher for Diet ID values than ASA24 and a pattern of larger differences at lower ASA24-reported protein consumption. Likewise, consumption of total fat (Diet ID 1.8g kcals higher on average) and carbohydrates (Diet ID 4.5g kcals higher on average) followed the same pattern of almost no underreporting and having greater bias as lower consumption levels. The plots for total sugars show a trend of higher reporting by Diet ID (relative to ASA24) at lower consumption and lower reporting by Diet ID at higher consumption although bias was fairly small at moderate sugar consumption levels of approximately 25g-75g. Overall, for this sample, the bias was 10g higher for Diet ID with a mean absolute difference of 18g. Dietary fiber reporting was overall slightly higher for Diet ID than ASA24 with a bias of 3g but the differences were predominantly for higher amounts of dietary fiber. Rather than a consistent trend, the disagreement between Diet ID and ASA24 was in both directions except at the highest fiber consumptions where Diet ID was consistently lower. For dietary fiber, the bias was small but non-systematic variation (as indicated by Absolute Difference) was large, especially at the highest levels of fiber consumption.

Cholesterol intake displays a similar pattern to fiber, in which there is large error but less bias. The bias is 8 units higher for Diet ID than ASA24, but the absolute error magnitude averages 58 units. These numbers mean that Diet ID-reported cholesterol consumption has a large amount of error (relative to ASA24 as criterion), but that appears to depend on level of ASA24-reported cholesterol intake. There may be a slight trend for Diet ID to have lower cholesterol output at very high levels, but that is largely an artifact of one participant with very high consumption (according to ASA24). Saturated fat and monounsaturated fat, along with total fat, share a pattern of positive bias in Diet ID reported intake at relatively low levels according to ASA24, with some negative bias at higher ASA24-reported consumptions. Polyunsaturated fat has a pattern more like cholesterol, with high absolute error magnitude at the extremes but little error or bias at moderate intake levels. Overall, the agreement of Diet ID for various components of fat intake seems better than on many of the other nutrients. Saturated fat bias is only 1.4g on average (mean absolute difference 3.6g) and polyunsaturated fat bias is 2g with mean absolute difference of 4g. Total fat and monounsaturated fat bias and error are somewhat larger.

Micronutrients

Various other nutrients (calcium, iron, vitamin K, etc.) produced plots intermediate between the kcal-type and dietary fiber-type patterns. Essentially, these nutrients were a mixture of higher over-reporting by ASA24 at low intake levels with either less overreporting or a mixture of over- and underreporting at higher consumptions.

HEI

HEI had an opposite direction to the consumption nutrients with a negative bias (Diet ID report lower than ASA24) for HEI having a similar interpretation to a positive bias in the others. As such, the plots for HEI are similar to the plots of individual nutrients. At low HEI scores, less than about 50, the biases are all positive. Diet ID tends to have higher HEI at low ASA24-reported HEI. For the rest of the range of ASA24-reported HEI, there are both positive and negative biases although as HEI approaches 90 according to ASA24, the Diet ID values tend to be lower. Examining the absolute difference plot, there is a clear trend of small (absolute) errors below 75 on the ASA24 report then increasing error at the highest HEI levels. While the bias overall is −19 points (i.e. Diet ID underreport of HEI compared to ASA24) the mean absolute difference is twice a large at 38 points (on a 100-point scale).

Discussion

Accurately assessing dietary intake is integral to nutrition research, but methods of dietary assessment can be burdensome [5, 6] or require high levels of literacy [17]. The present study examined the relationship of the Diet ID platform with the current method of assessing diet using three ASA24 recalls in the NEW Soul study. Diet ID mostly relies on images of commonly eaten foods so participants can identify food patterns that best match with their eating styles. The use of photos to identify food patterns allows for greater use by low-literacy populations [37, 38]. This measure of dietary assessment could easily be incorporated into nutrition research/studies to quickly assess dietary patterns and quality. The NEW Soul study provided a unique opportunity to test the Diet ID platform among African American adults in the southeastern United States, participating in a dietary intervention that emphasized consumption of healthy soul food diets. The findings of this study highlight the variations in intake assessed between the two instruments.

When examining the correlations between the two instruments, there were significant relationships for energy intake, HEI score, protein, carbohydrates, cholesterol, sodium, copper, vitamin C, thiamin, and vitamins B12 and E. Participants tended to report higher energy intake on the Diet ID than the ASA24, especially at lower kcal levels on the ASA24. Similar patterns were found when comparing estimated TDEE to kcals reported via Diet ID or AS24, with significantly higher kcals reported on Diet ID compared to estimated TDEE and lower kcals reported on the ASA24. That pattern was also observed for most of the macronutrients, with higher values recorded by Diet ID as opposed to ASA24, especially as values decreased on the ASA24. Previous work has also found higher vitamin C and E values using an FFQ as compared to the ASA24, but also found higher intakes for total sugar, folate, calcium, magnesium, and vitamin K [16], which the present study did not observe. While only certain nutrients were significantly correlated between the two instruments, the mean nutrients per 1,000 kcals of most nutrients were fairly close and tended to go in similar directions (higher for Diet ID and lower for ASA24). The agreement between the measures for micronutrients was varied, but for HEI, the pattern that emerged was opposite of the macronutrients. Just as in the correlations, which were positive for all comparisons except for HEI, the relationship between the ASA24 and Diet ID was in the opposite direction to the nutrient variables with a negative bias (Diet ID reported lower than ASA24).

Overall, with the exception of the HEI score, values assessed with the Diet ID instrument were higher than those assessed with the ASA24. These findings may be explained by several reasons. For one, respondents may have underreported intake on the 24HR. Underreporting of foods seen as undesirable or unhealthy is common, especially among participants with overweight [39, 40]. This could be why overall energy was lower on the ASA24, yet HEI was higher (e.g., fewer undesirable foods were reported). This underreporting could lead to intakes on the ASA24 that are lower than what is actually being consumed by participants. To help account for underreporting, recalls with energy intakes <500 kcals were excluded [29]. In a previous study, the ASA24 had high levels of agreement for energy, micronutrients, and macronutrients as compared to a four-day food record [41]. Difficulty with findings foods in the ASA24 database or low levels of reading or computer literacy could have also resulted in participants not entering all foods consumed on the ASA24, leading to lower reported intakes on the 24HR compared to Diet ID. In a study examining the use of the ASA24 among low-income adults, several usability issues were identified including misunderstanding prompts, lack of knowledge on how to assess the next step in the recall, and difficulties with the search function [17]. The participant sample (African American adults) and dietary patterns (vegan soul food or low-fat, omnivorous soul food) in the NEW Soul study may have been outside the usual profiles included in the Diet ID database, potentially leading to a lack of food pictures relevant to the diets currently being consumed by participants. Lastly, Diet ID has participants self-report their activity level in order to assist in calculating probable energy intake. Approximately 59% of participants estimated their activity as moderate or greater on Diet ID whereas accelerometer-based data indicated participants were getting, on average, less than 14 minutes of moderate-to-vigorous physical activity each day. This indicates participants likely underreported physical activity on the Diet ID.

In the present study, true intake of nutrients (e.g., doubly-labeled water, biomarkers, weighed and observed intake of foods) was not collected and therefore, the dietary intake collected via ASA24 and Diet ID cannot be compared to actual values. While three ASA24 recalls were used as the comparison and standard way to collect dietary data, it is not known if these 24HR reflect intake that is closer to actual intake than the Diet ID. A previous study that examined the use of the ASA24 compared to a 4-day food record or an FFQ found that the ASA24 had the least underreporting of energy intake (assessed via doubly labeled water) compared to the other two methods [16]. There has been a call to assess diet in multiple ways, combining both short-term methods (24HR) and longer-term methods (FFQ, similar to the method employed by Diet ID), in order to have the strengths of both types of assessment methods included in a study [42]. Therefore, future studies may wish to combine these methods, which could perhaps allow for the need for fewer 24HR to be collected, therefore reducing participant burden.

Future studies should consider the advantages to each method of assessing diet. The ASA24 platform provides an extensive output of nutrients, food groups, and individual food items for analysis. ASA24 is free to use and can be administered in person or could potentially be used to guide a phone-administered 24HR. In addition, it has been validated against other dietary assessment methods [8, 41, 43]. The Diet ID method has substantial advantages as well. The platform requires less computer knowledge, spelling/typing skills, and literacy fluency and takes significantly less time. In addition, just one completion of the Diet ID is required (vs. three 24HR), and in the present study, participants completed the assessment in less than four minutes on average, compared to reported mean completion time of 24 minutes (range 17 to 34 minutes) for studies using ASA24 [28].

The present study has several strengths, including examining two dietary assessment methods among African American adults consuming various dietary patterns. In addition, Bland-Altman plots were used to demonstrate the level of agreement and bias between the two methods. The study also has some limitations. While two methods of assessing dietary intake were compared and TDEE was also presented in relationship to reported energy intake, no measure of true intake was included. Therefore, it is not clear which measure best captures actual dietary intake. In addition, 24HR reflect recent intake. In the present study, the ASA24 recalls were collected over a period of 2 weeks. The Diet ID reflects more of a long-term method of intake, assessing dietary patterns that have occurred over the past several weeks. The photos of foods used in both platforms may have differed in portion size, effecting estimated intakes. In addition, the Diet ID is a commercial instrument and methods used to calculate the nutrient data and HEI score from the photos are proprietary. Also, the two methods were assessed together at one timepoint in the NEW Soul study and therefore does not capture how dietary intake may change over time. While the main NEW Soul study was powered to detect differences between two diet groups, only one cohort was used in the present study and so it is not clear if the sample size provided enough power to detect differences between the two methods; however, our sample size is comparable to other previous dietary instrument validation studies [12, 44, 45]. Lastly, the population used in this study was mostly female and all African American adults living in the south, so results may differ based on who is included in future studies comparing these measures. This make-up of mostly women is similar to previous weight-loss trials, which have included primarily females [46]. However, numerous studies have found that women tend to underreport foods more so than males when using 24HR [47], so results of this study may have differed if more males were included.

Diet ID is a new method that holds promise as a way to rapidly assess dietary intake. Diet ID tended to lead to higher values of kcals and nutrients, whereas ASA24 resulted in lower values. Future research should examine these two methods against objective assessments of energy and nutrient intake. Future studies should also continue to acknowledge the potential limitations of the diet assessment methods used and consider using multiple methods, both short- and long-term, as a way to ensure the strengths of all methods are included [42].

Acknowledgements

Research reported in this publication was supported by the National Heart, Lung, and Blood Institute (NHLBI) of the National Institutes of Health under award number R01HL135220. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. The authors do not have any conflicts of interest to declare.

Footnotes

Declaration of interest statement

There are no relevant financial or non-financial competing interests to report among any of the authors.

Contributor Information

Gabrielle Turner-McGrievy, Department of Health Promotion, Education, and Behavior, Arnold School of Public Health, University of South Carolina, 915 Greene Street, Columbia, SC 29208.

Brent Hutto, Prevention Research Center, Arnold School of Public Health, University of South Carolina, 921 Assembly Street, Columbia, SC 29208 USA.

John A. Bernhart, Department of Health Promotion, Education, and Behavior, Arnold School of Public Health, University of South Carolina, 915 Greene Street, Columbia, SC 29208.

Mary Wilson, Department of Health Promotion, Education, and Behavior, Arnold School of Public Health, University of South Carolina, 915 Greene Street, Columbia, SC 29208.

References

  • 1.Archundia Herrera M, Chan CB. Narrative review of new methods for assessing food and energy intake. Nutrients 10(8):1064, 2018. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Johnson RK. Dietary intake—how do we measure what people are really eating? Obesity Research 10(S11):63S–68S, 2002. [DOI] [PubMed] [Google Scholar]
  • 3.Naska A, Lagiou A, Lagiou P. Dietary assessment methods in epidemiological research: Current state of the art and future prospects. F1000Research 6, 2017. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Thompson FE, Subar AF: Dietary assessment methodology. "Nutrition in the prevention and treatment of disease." Elsevier, pp 5–48, 2017. [Google Scholar]
  • 5.Gibson RS, Charrondiere UR, Bell W. Measurement errors in dietary assessment using self-reported 24-hour recalls in low-income countries and strategies for their prevention. Advances in Nutrition 8(6):980–991, 2017. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Shim J-S, Oh K, Kim HC. Dietary assessment methods in epidemiologic studies. Epidemiology and health 36, 2014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Kipnis V, Subar AF, Midthune D, Freedman LS, Ballard-Barbash R, Troiano RP, Bingham S, Schoeller DA, Schatzkin A, Carroll RJ. Structure of dietary measurement error: Results of the open biomarker study. American journal of epidemiology, Research Support, U.S. Gov't, P.H.S, 158(1):14–21; discussion 22-16, 2003. [DOI] [PubMed] [Google Scholar]
  • 8.Thompson FE, Dixit-Joshi S, Potischman N, Dodd KW, Kirkpatrick SI, Kushi LH, Alexander GL, Coleman LA, Zimmerman TP, Sundaram ME. Comparison of interviewer-administered and automated self-administered 24-hour dietary recalls in 3 diverse integrated health systems. American journal of epidemiology 181(12):970–978, 2015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Freedman LS, Commins JM, Moler JE, Arab L, Baer DJ, Kipnis V, Midthune D, Moshfegh AJ, Neuhouser ML, Prentice RL, Schatzkin A, Spiegelman D, Subar AF, Tinker LF, Willett W. Pooled results from 5 validation studies of dietary self-report instruments using recovery biomarkers for energy and protein intake. American journal of epidemiology 180(2):172–188, 2014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Tran KM, Johnson RK, Soultanakis RP, Matthews DE. In-person vs telephone-administered multiple-pass 24-hour recalls in women: Validation with doubly labeled water. Journal of the American Dietetic Association 100(7):777–783, 2000. [DOI] [PubMed] [Google Scholar]
  • 11.Subar AF, Kirkpatrick SI, Mittl B, Zimmerman TP, Thompson FE, Bingley C, Willis G, Islam NG, Baranowski T, McNutt S, Potischman N. The automated self-administered 24-hour dietary recall (asa24): A resource for researchers, clinicians, and educators from the national cancer institute. Journal of the Academy of Nutrition and Dietetics 112(8):1134–1137, 2012. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Ma Y, Olendzki BC, Pagoto SL, Hurley TG, Magner RP, Ockene IS, Schneider KL, Merriam PA, Hébert JR. Number of 24-hour diet recalls needed to estimate energy intake. Annals of epidemiology 19(8):553–559, 2009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Foster E, Lee C, Imamura F, Hollidge SE, Westgate KL, Venables MC, Poliakov I, Rowland MK, Osadchiy T, Bradley JC, Simpson EL, Adamson AJ, Olivier P, Wareham N, Forouhi NG, Brage S. Validity and reliability of an online self-report 24-h dietary recall method (intake24): A doubly labelled water study and repeated-measures analysis. Journal of nutritional science 8:e29, 2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Schoeller DA, Bandini LG, Dietz WH. Inaccuracies in self-reported intake identified by comparison with the doubly labelled water method. Canadian journal of physiology and pharmacology 68(7):941–949, 1990. [DOI] [PubMed] [Google Scholar]
  • 15.Gardner CD, Hartle JC, Garrett RD, Offringa LC, Wasserman AS. Maximizing the intersection of human health and the health of the environment with regard to the amount and type of protein produced and consumed in the united states. Nutr Rev 77(4):197–215, 2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Park Y, Dodd KW, Kipnis V, Thompson FE, Potischman N, Schoeller DA, Baer DJ, Midthune D, Troiano RP, Bowles H, Subar AF. Comparison of self-reported dietary intakes from the automated self-administered 24-h recall, 4-d food records, and food-frequency questionnaires against recovery biomarkers. The American journal of clinical nutrition 107(1):80–93, 2018. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Kupis J, Johnson S, Hallihan G, Olstad DL. Assessing the usability of the automated self-administered dietary assessment tool (asa24) among low-income adults. Nutrients 11(1), 2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.NCI. 24-hour dietary recall (24hr) at a glance. Https://dietassessmentprimer.Cancer.Gov/profiles/recall/index.Html. Accessed 1/3/17.
  • 19.Boushey CJ, Spoden M, Zhu FM, Delp EJ, Kerr DA. New mobile methods for dietary assessment: Review of image-assisted and image-based dietary assessment methods. The Proceedings of the Nutrition Society 76(3):283–294, 2017. [DOI] [PubMed] [Google Scholar]
  • 20.Katz DL, Rhee LQ, Katz CS, Aronson DL, Frank GC, Gardner CD, Willett WC, Dansinger ML. Dietary assessment can be based on pattern recognition rather than recall. Medical Hypotheses 140:109644, 2020. [DOI] [PubMed] [Google Scholar]
  • 21.Guenther PM, Casavale KO, Reedy J, Kirkpatrick SI, Hiza HA, Kuczynski KJ, Kahle LL, Krebs-Smith SM. Update of the healthy eating index: Hei-2010. J Acad Nutr Diet 113(4):569–580, 2013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Turner-McGrievy G, Wilcox S, Frongillo EA, Murphy A, Hutto B, Williams K, Crimarco A, Wilson M, Davey M. The nutritious eating with soul (new soul) study: Study design and methods of a two-year randomized trial comparing culturally adapted soul food vegan vs. Omnivorous diets among african american adults at risk for heart disease. Contemp Clin Trials 88:105897, 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Zanini B, Simonetto A, Bertolotti P, Marullo M, Marconi S, Becchetti C, Gilioli G, Valerio A, Donato F, Ricci C, Castellano M. A new self-administered semi-quantitative food frequency questionnaire to estimate nutrient intake among italian adults: Development design and validation process. Nutrition Research 80:18–27, 2020. [DOI] [PubMed] [Google Scholar]
  • 24.Ravelli MN, Schoeller DA. Traditional self-reported dietary instruments are prone to inaccuracies and new approaches are needed. Frontiers in Nutrition, Mini Review, 7(90), 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Hjartåker A, Andersen LF, Lund E. Comparison of diet measures from a food-frequency questionnaire with measures from repeated 24-hour dietary recalls. The norwegian women and cancer study. Public Health Nutrition 10(10):1094–1103, 2007. [DOI] [PubMed] [Google Scholar]
  • 26.Shu XO, Yang G, Jin F, Liu D, Kushi L, Wen W, Gao YT, Zheng W. Validity and reproducibility of the food frequency questionnaire used in the shanghai women's health study. European Journal of Clinical Nutrition 58(1):17–23, 2004. [DOI] [PubMed] [Google Scholar]
  • 27.Steinemann N, Grize L, Ziesemer K, Kauf P, Probst-Hensch N, Brombach C. Relative validation of a food frequency questionnaire to estimate food intake in an adult population. Food & nutrition research 61(1):1305193, 2017. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Asa24 frequently asked questions. Https://epi.Grants.Cancer.Gov/asa24/resources/faq.Html. Accessed april 6, 2020.
  • 29.Banna JC, McCrory MA, Fialkowski MK, Boushey C. Examining plausibility of self-reported energy intake data: Considerations for method selection. Frontiers in nutrition 4:45, 2017. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Tooze JA, Krebs-Smith SM, Troiano RP, Subar AF. The accuracy of the goldberg method for classifying misreporters of energy intake on a food frequency questionnaire and 24-h recalls: Comparison with doubly labeled water. European journal of clinical nutrition 66(5):569–576, 2012. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Usda. Fooddata central database. Https://fdc.Nal.Usda.Gov/.
  • 32.Mifflin MD, St Jeor ST, Hill LA, Scott BJ, Daugherty SA, Koh YO. A new predictive equation for resting energy expenditure in healthy individuals. The American Journal of Clinical Nutrition 51(2):241–247, 1990. [DOI] [PubMed] [Google Scholar]
  • 33.Douglas CC, Lawrence JC, Bush NC, Oster RA, Gower BA, Darnell BE. Ability of the harris benedict formula to predict energy requirements differs with weight history and ethnicity. Nutrition research (New York, NY) 27(4):194–199, 2007. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Gerrior S, Juan W, Basiotis P. An easy approach to calculating estimated energy requirements. Preventing chronic disease 3(4):A129–A129, 2006. [PMC free article] [PubMed] [Google Scholar]
  • 35.Giavarina D. Understanding bland altman analysis. Biochemia medica: Biochemia medica 25(2):141–151, 2015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.R core team (2020). R: A language and environment for statistical computing. R foundation for statistical computing, vienna, austria. Url http://www.R-project.Org/. [Google Scholar]
  • 37.Park J, Zuniga J. Effectiveness of using picture-based health education for people with low health literacy: An integrative review. Cogent Medicine 3(1):1264679, 2016. [Google Scholar]
  • 38.Chaudry B, Connelly K, Siek KA, Welch JL. The design of a mobile portion size estimation interface for a low literacy population. Int Conf Pervasive Comput Technol Healthc:160–167, 2011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Scagliusi FB, Ferriolli E, Pfrimer K, Laureano C, Cunha CS, Gualano B, Lourenço BH, Lancha AH Jr. Underreporting of energy intake in brazilian women varies according to dietary assessment: A cross-sectional study using doubly labeled water. Journal of the American Dietetic Association 108(12):2031–2040, 2008. [DOI] [PubMed] [Google Scholar]
  • 40.Orcholski L, Luke A, Plange-Rhule J, Bovet P, Forrester TE, Lambert EV, Dugas LR, Kettmann E, Durazo-Arvizu RA, Cooper RS. Under-reporting of dietary energy intake in five populations of the african diaspora. British Journal of Nutrition 113(3):464–472, 2015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Frankenfeld CL, Poudrier JK, Waters NM, Gillevet PM, Xu Y. Dietary intake measured from a self-administered, online 24-hour recall system compared with 4-day diet records in an adult us population. Journal of the Academy of Nutrition and Dietetics 112(10):1642–1647, 2012. [DOI] [PubMed] [Google Scholar]
  • 42.Subar AF, Freedman LS, Tooze JA, Kirkpatrick SI, Boushey C, Neuhouser ML, Thompson FE, Potischman N, Guenther PM, Tarasuk V. Addressing current criticism regarding the value of self-report dietary data. The Journal of nutrition 145(12):2639–2645, 2015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Kirkpatrick SI, Subar AF, Douglass D, Zimmerman TP, Thompson FE, Kahle LL, George SM, Dodd KW, Potischman N. Performance of the automated self-administered 24-hour recall relative to a measure of true intakes and to an interviewer-administered 24-h recall. The American journal of clinical nutrition 100(1):233–240, 2014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Stote KS, Radecki SV, Moshfegh AJ, Ingwersen LA, Baer DJ. The number of 24 h dietary recalls using the us department of agriculture's automated multiple-pass method required to estimate nutrient intake in overweight and obese adults. Public Health Nutrition 14(10):1736–1742, 2011. [DOI] [PubMed] [Google Scholar]
  • 45.Lazarte CE, Encinas ME, Alegre C, Granfeldt Y. Validation of digital photographs, as a tool in 24-h recall, for the improvement of dietary assessment among rural populations in developing countries. Nutr J 11:61–61, 2012. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Pagoto SL, Schneider KL, Oleski JL, Luciani JM, Bodenlos JS, Whited MC. Male inclusion in randomized controlled trials of lifestyle weight loss interventions. Obesity (Silver Spring, Md) 20(6):1234–1239, 2012. [DOI] [PubMed] [Google Scholar]
  • 47.Burrows TL, Ho YY, Rollo ME, Collins CE. Validity of dietary assessment methods when compared to the method of doubly labeled water: A systematic review in adults. Frontiers in endocrinology 10:850, 2019. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES