Protocol of the validation of the experience sampling-based dietary assessment method (ESDAM) against doubly labeled water, urinary protein, and biomarkers

Abstract

Background

The quest towards more feasible, low-cost yet accurate dietary assessment methods has led to the development of the new Experience Sampling-based Dietary Assessment Method (ESDAM). ESDAM is an app-based quantitative dietary assessment method to assess habitual dietary intake over a period of 2 weeks. ESDAM prompts three 2-hour recalls daily requesting to provide dietary intake on meal and food group level. The ESDAM allows to measure dietary intake near real-time in a rapid, low-cost and feasible manner. Following the user experience evaluation of the ESDAM, the validity of the ESDAM will now be assessed against objective biomarkers.

Methods

This protocol describes the validation of the ESDAM against three 24-hour dietary recalls (24-HDR), doubly labeled water, urinary nitrogen, serum carotenoids, and erythrocyte membrane fatty acids. The primary outcomes include energy intake and protein intake measured by the ESDAM in relation to energy expenditure measured by the doubly labeled water method and protein intake derived from urinary nitrogen analysis respectively. Secondary outcomes include nutrient and food group consumption derived from ESDAM and repeated 24-HDR’s, beta-carotenoids as biomarker of fruit and vegetable consumption and erythrocyte membrane fatty acid composition as biomarker of dietary fatty acid composition. Blinded continuous glucose monitoring will serve as method to assess compliance to ESDAM prompts. A target sample of 115 healthy volunteers will be recruited. Validity will be evaluated by mean differences and spearman correlations between nutrient and biomarker reference values. Bland-Altman plots will be developed to assess agreement. Method of triads will allow quantifying measurement error of the ESDAM, 24-HDR’s and biomarkers in relation to the (unknown) ‘true dietary intake’.

Discussion

This protocol describes an extensive validation study of a newly developed Experience Sampling-based Dietary Assessment Method against both self-reported and objective reference methods. The main strength is the use of the state-of-the-art validation techniques using doubly labeled water, urinary nitrogen, carotenoids and erythrocyte membrane fatty acids as biomarkers for dietary intake. A limitation of the study is that the evaluation of reproducibility of the ESDAM is not included.

Registration

This study is registered at ClinicalTrials.gov on (NCT06485960, date: 07/02/2024).

Background

Assessment of dietary intake is a central pillar in the complex evaluation of associations between diet and human health and disease [1]. As a result, dietary assessment provides key information in different settings. For public health nutrition, dietary assessment allows to evaluate the diet at the population level (e.g. surveillance) both in terms of diet quality and adherence to dietary guidelines or dietary patterns [2]. In regards to food safety, dietary assessment provides information on exposure to pathogenic substances (e.g. monitoring of food borne diseases) as part of risk assessment [3]. Additionally, dietary assessment serves a role in evaluating environmental impact with regards to sustainability [4, 5]. In research, dietary intake is evaluated as exposure variable or as treatment in dietary intervention studies [6]. More recently, dietary assessment serves a role in personalized nutrition and metabolomics research as well as in the development of different preventive or quality-of-life enhancing mHealth applications [7]. Lastly, in the clinical setting, dietary assessment allows evaluation of dietary intake of patients to provide personal dietary advice. Altogether, both public health nutrition and research inform nutrition policy to develop strategies enabling people to eat healthy diets [6, 8].

The various applications of dietary assessment represent the versatility of dietary exposure variables. Consequently, dietary intake is a complex concept to assess consisting of different dietary exposure variables that are interrelated and might act synergistically. Especially more recently, since assessment of dietary patterns and dietary intake at the meal-level as opposed to assessment of single nutrients in relation to health and disease outcomes is gaining more attention [9].

Together with the varying shifts in the field of nutrition research, the field of dietary assessment is continuously evolving as well in terms of methodology and tools, in order to meet the need for accurate and feasible dietary assessment methods [1]. In recent years, many traditional dietary assessment tools have been implemented in technology-based versions resulting in web-based or app-based diet records, 24-hour dietary recalls and Food Frequency Questionnaires [10–12]. Additionally, image-based methods (i.e. image-based food records) and automated image recognition have been researched to advance the field of dietary assessment [11]. Incorporation of such technologies has substantially leveraged the collection, analysis and interpretation of dietary intake data and improved feasibility by enabling remote data collection, simplifying user engagement, and allowing automated nutrient analysis [13]. Despite these advancements, substantial measurement errors persist in these web- and app-based approaches [14, 15].

To further improve accuracy and usability, the development of novel methodologies, beyond adopting new technologies, is essential. Approaches that integrate seamlessly into daily routines, are time-efficient, cognitively undemanding, and user-friendly may inherently reduce respondent burden, reporting fatigue, social desirability bias, recall bias, and misreporting.

In this light, Experience Sampling Methodology (ESM), originating from the field of psychology and behavioral sciences, was investigated as an alternative methodology for dietary assessment [16]. ESM uses near real-time data collection at (seemingly) random moments through app-based prompting of questions in order to collate data in real life minimizing recall bias, reactivity bias and misreporting [17]. An Experience Sampling-based Dietary Assessment Method (ESDAM) was developed which prompts three times daily the question to report dietary intake during the past two hours at meal and food-group level. The ESDAM is developed to measure habitual dietary intake over a period of twee weeks. A usability and feasibility study showed the ESDAM to be a low-burden, rapid and easy-to-use tool for dietary assessment [18]. Following this user experience evaluation, assessment of validity of the ESDAM will allow to evaluate capability of the ESDAM to measure true intake.

Therefore, the ESDAM will be validated against repeated 24-hour dietary recalls together with objective reference markers including doubly labeled water (total energy expenditure), urinary nitrogen (protein intake), carotenoids (fruit and vegetable intake) and erythrocyte membrane fatty acids.

Methods

This protocol was written following principles of the Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT) [19]. The SPIRIT checklist is available in a supplementary file.

Aim

The aim of this study is to assess validity of the newly developed Experience Sampling-based Dietary Assessment Method (ESDAM).

The objectives are as follows:

• Assess convergent validity of the ESDAM for energy, nutrient and food group intake against three interviewer-administered 24-hour dietary recalls.

• Assess validity of the ESDAM against objective biomarkers including:

  1. Total energy expenditure determined by the doubly labeled water method as reference for energy intake,
  2. Urinary nitrogen as reference for protein intake,
  3. Erythrocyte membrane fatty acid composition as reference for omega-3 and omega-6 poly unsaturated fatty acids (PUFAs).
  4. Serum carotenoids as reference for vegetable and fruit consumption.

• Assess compliance for registering dietary intake in ESDAM through continuous glucose monitoring (CGM) as reference for eating episodes.

Study design

This study will use a prospective observational non-interventional design to assess validity of the ESDAM compared to three 24-hour dietary recalls (24-HDR’s) and biomarkers in Belgian adults. The study has a duration of four weeks (as presented in Fig. 1). Table 1 presents the SPIRIT figure including the assessments at specified timepoints. The first two weeks will be used to collect baseline data including socio-demographic, biometric data and three 24-HDR’s to assess dietary intake. During the last two weeks the Experience Sampling-based Dietary Assessment Method will be evaluated against urine biomarkers (Doubly labeled water and urinary nitrogen), blood biomarkers (erythrocyte membrane fatty acid composition, beta-carotenoids) together with continuous glucose monitoring.

Fig. 1.

Study design of the validation study of ESDAM (t_1,2,3…= timepoint 1, 2, 3…(confer Table 1), 24 h = 24-hour urine collection, DLW = Doubly labeled water technique)

Table 1.

SPIRIT checklist

	Study period
	Enrolment	Allocation	Post-allocation										Close-out
Timepoint	-t 1	0	t 1	t 2	t 3	t 4	t 5	t 6	t 7	t 8	t 9	t 10	t x
Enrolment:
Eligibility screen	X
Informed consent	X
Allocation		X
Interventions:
24-hour Dietary Recall			X	X	X
DLW administration						X
Assessments:
Sociodemographic questionnaire	X
Anthropometrics						X							X
Continuous glucose monitoring
ESDAM
DLW							X			X	X	X
Urinary nitrogen								X	X
Erythrocyte membrane fatty acid profile													X
Polyphenols													X
Serum carotenoïds													X

Recruitment and eligibility criteria

Participants will be recruited through flyers at the local university and university hospital together with social media calls and the snowball effect. Recruitment will take place between May 2024 and June 2025. Volunteers of age 18 to 65 who have a smartphone and a freezer at home, with sufficient knowledge of the Dutch language and who have a stable body weight (not changed by 5% during last 3 months) not aiming to gain or lose weight during the study period will be eligible. Participants who are pregnant or lactating or who follow a specific diet for medical reasons (i.e. coeliac disease) are excluded. All participants will receive study information in a video call and will be able to ask questions to a study team member. All participants will be requested to read and sign the online informed consent before participation.

Sample size

Sample size calculation was performed using G*Power 3.1.9.7. Based on our previous validation research and the research of Lucassen et al. Spearman’s correlation coefficients of ≥ 0.30 are considered as meaningful for validity of dietary assessment methods [20]. The null hypothesis being tested is that there is no correlation between dietary intake variables obtained from the ESDAM and the corresponding biomarkers. The alternative hypothesis is that there is a correlation between dietary intake variables obtained from the ESDAM and the corresponding biomarkers. A sample size of 83 is needed for detecting correlation coefficients of 0.30 with a power of 80% and alpha error probability of 0.05 (two-tailed). As described by Willet et al., a sample size of at least 100 is recommended in the context of validation studies for dietary assessment methods [21]. According to our previous validation studies, a drop-out rate of 10–15% is expected. Therefore, the aim is to recruit 115 participants.

Measurements

Experience sampling-based dietary assessment method (ESDAM)

The ESDAM is a newly developed method to measure habitual dietary intake (i.e. food group consumption, energy intake, micro- and macronutrient consumption) during a two-week period [18]. ESDAM makes use of the mPath application (KU Leuven), an experience sampling survey application, to send questions by a prompt message on the smartphone [22]. The ESDAM is described elsewhere [23]. In short, according to a fixed schedule, the participants will receive 2 to 3 prompt messages a day between 8 A.M. and 10 P.M. during two weeks. Each prompt message will request to indicate if the participant has consumed any foods or drinks during the last two hours. If any foods or drinks were consumed, follow-up multiple-choice questions will request to specify type and quantity of foods and drinks consumed. In this way, ESDAM collects dietary intake of three two-hour time windows each day and assesses each time window five times during the two-week ESM sampling period. Total intake of energy, nutrients and food groups (g/day) will be calculated using the Belgian Food Composition Database (NUBEL) [22].

24-hour dietary recall

The 24-hour dietary recall is a semi-structured interview performed by a trained dietician to assess dietary intake during the past 24 h. The 24-HDR will be performed through a telephone interview according to a standardized protocol applying the five-step-multiple-pass approach by trained dietitians [24]. Additionally, during the 24-HDR, current nutritional supplement use will be questioned as well. Three 24-HDR’s will be performed during the first two weeks of the study which will serve as baseline for dietary intake. Total intake of energy, nutrients and food groups (g/day) will be calculated using the Belgian Food Composition Database (NUBEL) [25].

Doubly labeled water method

To assess the validity of the newly developed ESDAM for estimating energy intake, the Doubly Labeled Water (DLW) method will be used. The DLW technique requires participants to drink water enriched with the stable isotopes deuterium (²H) and oxygen-18 (¹⁸O). Whereas ²H is eliminated from the body primarily as water, ¹⁸O is eliminated both as water and as carbon dioxide through respiration. Measuring the difference in elimination rate and using standard equations for indirect calorimetry allows to calculate energy expenditure [7]. Accuracy of the DLW method is 1% with a coefficient of variation of 2 to 12% [7]. According to the first law of thermodynamics, energy is conserved and energy expenditure can be considered equal to habitual energy intake if the subject is weight-stable [7]. The Maastricht protocol for DLW will be followed in this study which includes a baseline urine sample and six urine sample for isotope analysis on day 1, 7, 8 and 14 during the last two weeks of the study (Fig. 1) [8]. DLW will be given orally at a dose of approximately 2.5 g of ²H₂¹⁸O per kilogram of estimated total body water followed by a 50 mL water rinse of the dose bottle. The labeled water consists of 10 g of 10 atom percent H₂¹⁸O and 6.5 g 99.9 atom percent ²H₂O for 100 g ²H₂¹⁸O. Time of DLW ingestion will be registered together with timing of urine collections. Participants will be requested to store samples in the household freezer until the next study visit. Urine samples will be analysed for ²H and ¹⁸O abundance at the Laboratory for Digestion and Absorption. KU Leuven. Total daily Energy Expenditure (TEE) will be calculated by the multipoint method from the difference between elimination constants of hydrogen and oxygen, by linear regression, with the assumptions for fractionation [26]. Carbon dioxide production values will allow to calculate TEE with the classic equations of Coward and Cole [24].

Urinary nitrogen

Urinary nitrogen (N) is an established recovery biomarker of protein intake and serves as objective reference to assess validity of the ESDAM for estimating protein intake [27]. Urinary N will be measured in two 24-hour urine samples and analyzed according to the Dumas technique [28]. Participants will be requested to take two Para-Amino Benzoic Acid (PABA) pills on the days of the 24-hour urine collection to ensure completeness of sample collection [29]. Participants will be instructed to discard the first urine of the day of collection and collect all urine during the next 24 h until the first void of the next morning [30]. Participants will be requested to register time of collection (start, end), time of taking PABA pills, missed voids and medications taken on day of collection. The 24-hour urine samples will be handed to the study team the same day of finishing the 24-hour urine collection. Samples will be weighed, mixed and aliquoted into 1 mL cryovials and stored at −80 °C until analysis. Urinary protein content will be calculated according to the formula: 6.25 * (Urinary N/0.81) considering the assumption of 19% of fecal and skin loss [27]. Urine PABA concentrations will be measured in duplicate using the colorimetric technique [31]. PABA recovery of less than 70% will be considered incomplete while samples containing 70–85% PABA will be analyzed for nitrogen concentration however proportionally adjusted to 93% PABA recovery [32, 33]. Urinary nitrogen and PABA analysis will be performed at Laboratory for Digestion and Absorption KU Leuven.

Carotenoids and erythrocyte fatty acid composition

During the last study visit, following an overnight fast, three venous blood samples will be collected to assess serum carotenoids, and erythrocyte membrane fatty acid composition as biomarkers of fruit and vegetables, plant based foods and unsaturated fatty acids respectively. Venous puncture will be performed by a trained study nurse at the university hospital Leuven during the last study visit (Fig. 1). Aliquots for carotenoid measurement will be shipped at room temperature at the day of collection and will be analyzed by a certified clinical lab (Algemeen Medisch Labo, Antwerp, Belgium). Samples for quantification of erythrocyte membrane fatty acids will be centrifuged, aliquoted and stored at −80 °C. Erythrocyte membrane fatty acids aliquots will be shipped in batch and analyzed by OmegaQuant (South Dakota, U.S.A.). Erythrocyte membrane fatty acid composition analysis is a reflection of omega-3 and omega-6 fatty acid dietary intake of the past 90 to 12 days and, thus, habitual intake of the specified fatty acids. Fatty acid composition will be expressed as a percentage of total identified fatty acids. Omega-3 index (O3I) will be calculated as the sum of EPA and DHA content of red blood cells membranes, expressed as a percent of total identified fatty acids [34, 35].

Continuous glucose monitoring (CGM)

A blinded continuous Glucose Monitor (FreeStyle Libre Pro, Abbott, Diegem, Belgium) will be placed in the upper arm by a study nurse during the first study visit. Participants will be requested to wear the CGM during two weeks and will be blinded for the blood glucose measurements using a blinded reader. The CGM will serve as a measure of compliance for ESDAM. More specifically, date and timing of glucose peaks registered by the CGM will be used as reference for timing of dietary consumptions (assuming most dietary consumptions contain carbohydrates). Compliance of the ESDAM will be indicated by comparison of timing of dietary consumptions measured by the CGM with timing of registration of dietary intake in the ESDAM application by the participant.

Covariates

A socio-demographic questionnaire during the first study visit will collect participant data on age, gender and highest obtained educational level (high school level/college level/university level). All data of the study will be collated in an electronic participant record file (RedCap). Anthropometric measurements will be performed by a trained study employee following a standardized protocol at the first and last study visit (Fig. 1). Height of participants will be measured using a mounted stadiometer without shoes to the nearest 0.5 cm during the first study visit [35]. Weight of participants will be measured by a digital scale to the nearest 0.1 kg without shoes and heavy clothing during both the first and the last study visit to assess weight stability. A weight difference of ± 1 kg will be considered acceptable as a reflection of normal biological variations in body weight. Body composition (fat mass, fat free mass, body water) and Basal Metabolic Rate (BMR) will be measured by a calibrated impedance body composition analyzer. Body Mass Index (BMI) will be calculated as weight (kg) divided by the squared height (m²).

Analysis

Participants who completed at least 50% of ESDAM prompt questionnaires will be included for analysis based on previous research and ESM recommendations [36]. Compliance will be calculated as ratio of completed ESDAM prompts to eating episodes identified as blood glucose peak through the CGM during the corresponding time window.

Not normally distributed nutrients will be logarithmically transformed to achieve normal distributions. Means and standard deviations (SDs) of energy, macronutrients and food group intakes will be calculated for the ESDAM and reference methods (24-hour dietary recalls, urinary nitrogen and DLW). Mean differences between the ESDAM and reference methods will be calculated. Paired-difference t-test and Wilcoxon’s signed rank test will be calculated to assess significant differences between ESDAM and reference methods. Variance ratios of the repeated 24-hour dietary recalls and urinary nitrogen assessments will be calculated by dividing within-person variations by between-person variations. Pearson’s and Spearman’s correlation coefficients will be determined for crude and deattenuated nutrient intakes and for relative (energy-adjusted) crude and deattenuated nutrient intakes as well as for food group intakes in relation to the reference methods (DLW, urinary nitrogen, carotenoids, erythrocyte membrane fatty acids). Correlation coefficients will be interpreted according to Cohen’s cut-offs where r = ± 0.5 is considered to be strong, r = ± 0.30 moderate and r = ± 0.10 weak [37]. The formula of Beaton et al. will be applied to obtain deattenuation to correct for intraindividual variability [38]. Bland-Altman plots will be created to assess agreement and bias between ESDAM and the reference methods [39]. Cross-classification analysis will be performed by dividing study participants in quartile categories of energy, macronutrient and food group intakes for both the ESDAM and reference methods. To compare ESDAM with DLW, urinary nitrogen, carotenoids, study participants will be categorised into the same, adjacent or extreme quartile (misclassification). Weighed Kappa values will be calculated to assess agreement considering < 0.20 poor agreement, 0.21–0.40 fair agreement, 0.41–0.60 moderate agreement, 0.61–0.80 good agreement and >0.80 very good agreement [40].

Method of triads will be used to assess measurement error of ESDAM for estimating true energy and protein intake. Validity coefficients (VC) will be estimated between the three measurements (ESDAM, 24-hour dietary recalls and biomarkers) and the ‘unknown’ true intake [41]. VCs will be evaluated considering VC < 0.2 as low, 0.2 < VC < 0.6 as moderate or VC >0.6 as high [42]. The 95% CI for the validity coefficients will be estimated using the bootstrap method.

All statistical analyses will be two-sided with a significance level at α < 0.05. Statistical analyses will be performed using SPSS Inc. version 28.0 (SPSS Inc. and IBM Corp.) and R Studio version 4.3.2.

Discussion

This study includes an extensive validation protocol of a newly developed Experience Sampling-based Dietary Assessment Method against both self-reported and objective reference methods. The strength of this study includes the state-of-the-art validation techniques being used such as doubly labeled water, urinary nitrogen, carotenoids and erythrocyte membrane fatty acids as biomarkers for dietary intake. A limitation of the validation study is that an evaluation of reproducibility of the ESDAM is not yet included. Additionally, selection bias may be present as participants are recruited in the area of the university and university hospital, which may lead to a sample with a high prevalence of highly educated persons. Lastly, there is a potential bias due to the order effect since 24-hour dietary recalls are collected before the ESDAM.

Results from this research will be disseminated through presentations at public lectures, scientific institutions and meetings, web-articles and/or publications in scientific journals. Results will be reported according to STrengthening the Reporting of OBservational Epidemiology statement for nutrition research (STROBE-Nut) guidelines [41, 43]. International Committee of Medical Journal Editors recommendations will be followed regarding authorship to publications. 

Authors’ contributions

JV and CM conceptualized and designed the study. JV and CM acquired the funding. JV requested ethical approval and trial registration. JV wrote the manuscript. CM, KV, BVDS, and RV reviewed the manuscript. CM supervised the work.

Funding

This work was supported by a PhD fellowship Strategic Basic research grant (1S96721N) of.

Research Foundation Flanders (FWO) and KU Leuven Internal Funds grant number C3/22/050.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent participate

This study was approved by the Ethics Committee Research UZ/KU Leuven (Belgium) (S68363) and registered at ClinicalTrials.gov (NCT06485960) on July 2nd 2024. Any subsequent protocol amendments will be submitted to the appropriate Ethics Committees and national Regulatory Authorities for approval. All data will be pseudonymised and only the research team will have access to the data. All participants will be requested to consent prior to the start of the study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.