Validating the 24-Hour Dietary Recall: A Research Guide for Accurate Energy Intake Estimation in Clinical and Population Studies

Abstract

Accurate estimation of energy intake is fundamental for nutritional research, clinical trials, and public health monitoring. This article provides a comprehensive guide for researchers and drug development professionals on the validation of the 24-hour dietary recall (24HR) method. We explore the foundational principles of validation, including the use of the doubly labeled water technique as a gold standard. The review details methodological best practices, from the Automated Multiple-Pass Method to modern web-based tools like ASA24. We address critical troubleshooting for common errors such as under-reporting and day-to-day variation, and provide evidence-based strategies for optimization. Finally, we present a comparative analysis of validation studies across diverse populations and settings, synthesizing key findings on the number of recalls required for precise energy intake estimation. This resource aims to empower scientists with the knowledge to collect, validate, and interpret dietary data with greater confidence and accuracy.

The Science of Validation: Core Principles and Gold Standards for Dietary Assessment

The Critical Role of Accurate Energy Intake Data in Clinical and Biomedical Research

Accurate measurement of energy intake (EI) is a cornerstone of nutritional epidemiology, clinical trials, and biomedical research. Flawed dietary data can ripple through research outcomes, leading to misclassification of participants, weakened exposure-outcome linkages, and ultimately, flawed public health policies and clinical interventions [1]. The 24-hour dietary recall (24HR) is one of the most widely used tools for assessing dietary intake in large-scale studies. This guide provides a comparative analysis of different 24HR methodologies, evaluating their validity and applicability across diverse populations and technological platforms.

Table 1. Validation of 24HR against Objective Energy Expenditure Measures

The following table summarizes key validation studies that compared self-reported energy intake from 24HRs to total energy expenditure (TEE) measured by the doubly labeled water (DLW) method, the gold standard for energy requirement in weight-stable individuals.

Population	Number of 24HRs	Key Finding (EI vs. TEE)	Reported Accuracy/Discrepancy	Reference
Middle-aged Women	7 recalls over 14 days	Significant underreporting on first recall; 3 recalls optimal for group mean	Call 1: ~1501 kcal (TEE: 2115 kcal); Calls 2-3: ~2246-2315 kcal [2]
Overweight/Obese Adults (Women & Men)	Observer-recorded records + 24-hr snack recalls over 14 days	No significant difference between EI and TEE	EI was 96.9% (women) and 103% (men) of TEE [3]
Women (In-person & Telephone 24HR)	4 recalls (2 in-person, 2 telephone) over 14 days	Significant underreporting with both methods	Telephone: 2253 kcal; In-person: 2173 kcal (TEE: 2644 kcal) [4]
Children (5-7 years)	One 24-hour multiple pass recall	Slight overestimation at group level; inaccurate at individual level	Overestimation of EI by 250 kJ/d; wide limits of agreement [5]

Detailed Experimental Protocols in Validation Research

To critically appraise 24HR validation studies, understanding the rigors of their experimental design is essential. The following are detailed methodologies from key studies.

Protocol 1: Validation Against Doubly Labeled Water in Adults

This protocol uses the doubly labeled water (DLW) method, the gold standard for measuring energy expenditure in free-living individuals, to validate self-reported energy intake [2] [4].

• Population & Setting: Typically involves weight-stable adults without major medical conditions, studied under free-living conditions [2].
• DLW Administration: At the start of the metabolic period (e.g., 14 days), subjects ingest a dose of water containing non-radioactive isotopes (^2H₂ and H₂^18O). Spot urine samples are collected pre-dose and at specific intervals post-dose (e.g., 1, 7, and 14 days). Isotope enrichment in the urine is analyzed by isotope ratio mass spectrometry to calculate carbon dioxide production and total energy expenditure [2].
• 24HR Collection: Multiple non-consecutive 24HRs are collected during the metabolic period, often using the multiple-pass method. This method involves:
  • Quick List: An uninterrupted listing of all foods/beverages consumed.
  • Forgotten Foods Probe: Queries about frequently forgotten food categories.
  • Time & Occasion: Collection of the time and eating occasion for each food.
  • Detail Cycle: Probing for detailed food descriptions and portion sizes, often with the aid of food models or images.
  • Final Review: A final pass to review and add any missing items [6] [2].
• Data Analysis: The mean energy intake from the 24HRs is compared to the TEE from DLW using paired t-tests. Underreporting is indicated if mean EI is significantly lower than TEE [2] [4].

Protocol 2: Validation Against Weighed Food Intake in Controlled Settings

This protocol compares recalled intake to directly measured weighed food intake, providing a precise measure of true consumption [7] [8].

• Population & Setting: Often conducted in controlled feeding studies or metabolic wards where researchers have complete control over the food supply. A study with older Korean adults used a controlled 5-day feeding protocol [8].
• Weighed Food Method: All ingredients are precisely weighed to the nearest 0.1g during meal preparation. The final served portions are also weighed before and after consumption to determine the exact amount eaten by each participant [8].
• 24HR Administration: An interviewer-administered 24HR is conducted on the day following the controlled intake, with participants unaware of the exact recall timing to mimic a free-living setting [8].
• Data Analysis: Accuracy is assessed by classifying reported foods as:
  • Matches: Correctly reported food items.
  • Omissions: Foods consumed but not reported.
  • Intrusions: Foods reported but not consumed. Portion size accuracy is also calculated, often defining a ≤10% error as a "corresponding" report [8]. Nutrient intakes from the recall are compared to those from the weighed data.

Diagram 1: Workflow for Validating 24-Hour Recalls against Doubly Labeled Water.

Comparative Analysis of 24HR Administration Modes

The method of 24HR administration—whether interviewer-led, self-administered via web, or by telephone—can significantly impact data quality, cost, and scalability.

Table 2. Comparison of 24HR Administration Modalities

This table compares the performance and characteristics of different modes of administering 24-hour dietary recalls.

Administration Mode	Reported Energy Intake (kcal)	Agreement with Criterion	Key Advantages	Key Limitations
Interviewer-Administered (In-person)	2,173 ± 656 (vs. TEE 2,644) [4]	Considered the traditional standard, but significant underreporting persists [4].	Interviewer can probe and clarify in real-time [9].	High cost, resource-intensive, requires trained staff [6].
Telephone-Administered	2,253 ± 688 (vs. TEE 2,644) [4]	No significant difference from in-person recalls; equally effective but also underreport [4].	Logistically simpler, broader reach, lower cost [4].	Lack of visual cues for portion sizing.
Web-Based Self-Administered (e.g., ASA24, Intake24)	Varies by platform and age group.	In children (8-13 yrs), 47.8% overall match rate with interviewer recall; lower in younger children [9].	Highly scalable, low cost, automated data processing, reduced social desirability bias [1] [6].	Requires computer literacy; can be challenging for children and older adults without support [9] [1].
Web-Based (FOODCONS - Adults)	No significant difference in energy/macronutrients vs. interviewer-led [6].	Good agreement for energy, carbs, fiber (Bland-Altman) [6].	Good concordance for food groups; less time-consuming for researchers [6].	Participants may find it less convenient or time-consuming [1].

The Scientist's Toolkit: Essential Reagents & Materials for 24HR Validation

Diagram 2: Essential tools for validating 24-hour dietary recalls.

• Doubly Labeled Water (DLW): A non-radioactive isotopic technique (²H₂ and H₂^18O) used to measure total energy expenditure in free-living individuals over 1-2 weeks. It is the criterion method for validating energy intake in weight-stable subjects [2] [5] [4].
• Portion Size Estimation Aids: Essential tools to improve the accuracy of reported food amounts. These include:
  • Two-Dimensional Food Models: Drawings or photographs of foods in various serving sizes [2] [4].
  • Food Picture Atlases: Comprehensive collections of food images with portion size variations, often integrated into software like FOODCONS [6].
  • Household Measures: Using cups, spoons, and rulers as visual references.
• Standardized Dietary Analysis Software: Computer-assisted systems for collecting, coding, and analyzing dietary data.
  • Nutrition Data System for Research (NDS-R): Commonly used for interviewer-administered recalls [2].
  • Automated Self-Administered 24-hr Recall (ASA24): A web-based system developed by the National Cancer Institute that automates the multiple-pass method for self-administration [9].
  • FOODCONS: An Italian web-based software used for both interviewer-led and self-administered 24HRs [6].
• Structured Interview Protocols: The Multiple-Pass Method is the standardized interview technique used in systems like ASA24 and the USDA's Automated Multiple-Pass Method (AMPM). Its structured passes (Quick List, Forgotten Foods, Time & Occasion, Detail Cycle, Final Review) are designed to minimize memory error and enhance completeness [9] [6].

Population-Specific Considerations for Accurate Data Collection

The validity of 24HR can vary significantly across different demographic groups, necessitating tailored approaches.

• Age:

  • Children: Accuracy is highly age-dependent. Children aged 8-9 years have substantially more omissions and require more assistance compared to 10-13 year olds when using self-administered tools [9]. In younger children (5-7 years), a single 24HR may be accurate for group-level energy intake but is unreliable for individual-level assessment [5].
  • Older Adults: While one study found older Korean women could recall >95% of food items [8], another found adults over 60 recalled about 71% of foods and overestimated portion sizes [7]. Cognitive decline and sensory impairments can affect reporting accuracy.

• Body Weight: Overweight and obese individuals have historically been considered more prone to underreporting. However, one study using observer-recorded food records in a cafeteria setting combined with 24-hour snack recalls showed no significant underreporting in overweight and obese participants [3], suggesting methodology can mitigate this bias.

• Cultural & Dietary Context: Validation studies in Western populations may not generalize to other diets. Korean diets with rice-based meals, multiple side dishes (banchan), and amorphous foods like kimchi present unique challenges. Studies show items like sauces and kimchi are frequently omitted or under-reported, and portion size estimation for these foods is difficult [7] [8]. This highlights the need for culturally specific tools and food databases.

The choice of a 24HR methodology involves a careful balance between precision, cost, and participant burden. Key takeaways for researchers include:

• Multiple Recalls Are Essential: A single 24HR is insufficient. For estimating usual energy intake at a group level, 3 non-consecutive recalls appear to be a robust minimum, as the first recall often shows significant underreporting [2].
• Leverage Technology for Scale: Web-based self-administered tools like ASA24, Intake24, and FOODCONS demonstrate good agreement with interviewer-led methods for adults and offer a scalable, cost-effective alternative for large studies [6].
• Account for Population Nuances: Research protocols must be adapted for specific populations, considering age, cultural diet, and literacy. Special attention is needed for children, older adults, and populations consuming non-Western diets [9] [7] [8].

Future research should focus on refining web-based tools for diverse age and cultural groups, improving portion size estimation for amorphous foods, and further integrating technology like artificial intelligence to enhance the accuracy and efficiency of dietary assessment.

In nutritional research, accurately measuring energy intake is fundamental to understanding energy balance, obesity development, and metabolic health. However, self-reported dietary assessment methods, including 24-hour recalls and food frequency questionnaires, are notoriously prone to systematic errors including underreporting, overreporting, and misrepresentation of actual consumption. The doubly labeled water (DLW) method has emerged as the unequivocal gold standard for validating these dietary assessment tools by providing an objective, precise measure of total energy expenditure in free-living individuals. This non-invasive, isotopic technique enables researchers to bypass the limitations of self-reporting by quantifying energy expenditure under real-world conditions, thereby serving as a reference against which all other dietary intake methodologies are calibrated [10] [11] [12].

The establishment of DLW as a validation tool represents a paradigm shift in nutritional science, allowing for the critical evaluation of dietary assessment methods that form the basis of public health recommendations and clinical interventions. By comparing reported energy intake against DLW-measured energy expenditure during weight stability, researchers can identify and quantify reporting biases that have historically compromised nutritional epidemiology. This article examines the methodological foundation of DLW, presents experimental evidence supporting its validation capabilities, and compares its performance against alternative dietary assessment technologies within the context of 24-hour recall validation research.

Methodological Foundation: The Scientific Principle of Doubly Labeled Water

Fundamental Biochemical Principles

The doubly labeled water method is founded on the differential elimination kinetics of two stable isotopes—deuterium (²H) and oxygen-18 (¹⁸O)—from the body water pool. After oral administration of water containing both isotopes, deuterium (²H) is eliminated from the body exclusively as water (through urine, sweat, respiration, and other water losses), while oxygen-18 (¹⁸O) is eliminated as both water and carbon dioxide (through the bicarbonate pool) [10] [12]. This fundamental difference provides the basis for calculating carbon dioxide production, as illustrated in Figure 1.

Figure 1: Principle of Doubly Labeled Water Method

The mathematical foundation for calculating carbon dioxide production from the differential elimination rates was established by Lifson and colleagues in the 1950s [12]. The core calculation accounts for isotope dilution spaces and fractionation effects during elimination, with the most widely adopted formula being:

rCO₂ (mol) = (N/2.078)(1.01K₁₈ - 1.04K₂) - 0.0246rGF

Where N represents the body water pool size, K₁₈ and K₂ are the elimination rates for ¹⁸O and ²H, respectively, and rGF is the rate of gaseous water loss [12]. This calculation provides a highly accurate measure of carbon dioxide production, which is then converted to total energy expenditure using established calorimetric equations based on the respiratory quotient.

Standard Experimental Protocol and Workflow

The DLW method follows a rigorous standardized protocol that has been refined through international consensus [10] [12]. The experimental workflow, depicted in Figure 2, involves precise dosing with stable isotopes and careful timing of biological sample collection.

Figure 2: DLW Experimental Workflow

The typical DLW protocol involves:

• Baseline Sample Collection: Participants provide a urine, saliva, or blood sample to determine background isotopic enrichment before dose administration [12] [13].

• Dose Administration: Precisely measured DLW is administered orally. The dose is calculated based on body mass to achieve target enrichments of approximately 180-200 ppm for ¹⁸O and 120-150 ppm for ²H above background levels [12].

• Equilibration Period: A 3-6 hour equilibration period allows for complete distribution of the isotopes throughout the body water pool [12].

• Post-Dose Sample Collection: The first post-dose sample is collected after equilibration to establish initial enrichment (Time 0).

• Free-Living Period: Participants resume normal activities for 1-3 weeks while the isotopes are eliminated from the body at differential rates.

• Final Sample Collection: Urine samples are collected at the end of the study period (typically 14 days) to determine final isotope enrichment levels.

• Isotopic Analysis: Samples are analyzed using isotope ratio mass spectrometry or laser-based absorption spectroscopy to determine isotopic enrichment [10] [13].

• Data Calculation: Isotopic elimination rates are calculated and applied to established equations to determine carbon dioxide production and total energy expenditure.

The minimal participant burden and non-restrictive nature of the protocol enables accurate measurement under free-living conditions, making it uniquely suited for validating self-reported dietary intake in real-world settings [13].

Experimental Validation: Evidence Supporting DLW as the Gold Standard

Accuracy and Precision Data from Controlled Studies

The validity of the DLW method has been extensively demonstrated through controlled studies comparing its measurements against direct and indirect calorimetry. Early validation studies by Schoeller and van Santen (1982) demonstrated that DLW-assessed energy expenditure differed from energy intake plus change in body composition by only 2 ± 6% during weight stability [12]. Subsequent comparisons with respiration chamber measurements have consistently confirmed the method's accuracy with a precision of 2-8% across diverse populations [12].

A critical demonstration of DLW's precision comes from the Comprehensive Assessment of Long-term Effects of Reducing Intake of Energy (CALERIE) study, which implemented rigorous protocols to evaluate longitudinal reproducibility [14]. As shown in Table 1, the DLW method demonstrated exceptional stability in key parameters over extended periods, confirming its reliability for long-term nutritional studies.

Table 1: Longitudinal Reproducibility of DLW Method in the CALERIE Study

Parameter	Time Frame	Reproducibility	Clinical Significance
Fractional Turnover Rate (²H)	4.5 years	Within 1%	Stable isotope kinetics
Fractional Turnover Rate (¹⁸O)	4.5 years	Within 1%	Stable isotope kinetics
Difference Between Fractional Turnover Rates	4.5 years	Within 5%	Precise CO₂ production measurement
Isotope Dilution Spaces	2.4 years	Highly reproducible	Accurate body water pool assessment
Total Energy Expenditure	2.4 years	Highly reproducible	Reliable energy requirement estimation

The CALERIE study implemented two validation protocols: a dose-dilution protocol assessing repeated analysis of dose dilutions over 4.5 years, and a test-retest protocol evaluating blinded analysis of randomly selected DLW studies over 2.4 years [14]. Both protocols confirmed that the DLW method produces highly reproducible results in longitudinal nutrition studies, establishing its validity for monitoring energy balance changes in humans over extended periods.

Direct Comparative Evidence: DLW Versus Alternative Methods

The superior accuracy of DLW becomes particularly evident when comparing its measurements against those obtained through self-reported dietary assessment methods. Multiple studies employing DLW as the reference standard have consistently revealed significant misreporting in traditional dietary recalls and records.

A recent controlled feeding study comparing four technology-assisted dietary assessment methods demonstrated substantial variability in accuracy when estimating energy intake (Table 2). The study found that even the most accurate automated self-administered tools showed significant deviations from true intake, with error rates increasing for specific nutrients [15].

Table 2: Accuracy of Technology-Assisted 24-Hour Dietary Recalls Compared to True Intake in Controlled Feeding Study

Assessment Method	Mean Difference in Energy Intake (% of True Intake)	Key Findings and Limitations
Image-Assisted Interviewer-Administered 24HR	+15.0% (95% CI: 11.6, 18.3%)	Significant overestimation; intake distributions inaccurate
Automated Self-Administered (ASA24)	+5.4% (95% CI: 0.6, 10.2%)	Moderate overestimation; intake distributions inaccurate
Intake24	+1.7% (95% CI: -2.9, 6.3%)	Reasonable average validity; accurate intake distributions
Mobile Food Record-Trained Analyst	+1.3% (95% CI: -1.1, 3.8%)	Reasonable average validity; intake distributions inaccurate

Similar disparities have been observed across diverse populations. In Canadian adolescents, a web-based self-administered 24-hour recall (R24W) demonstrated 8.8% higher mean energy intake compared to interviewer-administered recalls, with significant differences ranging from 6.5% for percentage of energy from fat to 25.2% for saturated fat [16]. In Japanese adults, a web-based 24-hour recall showed moderate correlations (median r = 0.51 for men, 0.38 for women) with weighed food records for most nutrients, but notable discrepancies for specific nutrients including iodine, retinol, and vitamin C [17].

The consistent demonstration of reporting errors across different methodologies and populations underscores the critical importance of using DLW as an objective validation standard. Without such reference methods, the systematic errors inherent in self-reported dietary data would remain undetected and uncorrected, compromising the scientific foundation of nutritional epidemiology and public health recommendations.

Comparative Analysis: DLW Validation of 24-Hour Dietary Recall

Performance Benchmarking Across Populations

The DLW method has been extensively employed to validate 24-hour dietary recalls across diverse demographic groups, revealing important patterns in reporting accuracy. Table 3 summarizes key findings from recent validation studies that utilized DLW as the reference standard.

Table 3: DLW Validation Studies of 24-Hour Dietary Recall Across Populations

Population	Study Reference	Key Findings	Reporting Bias Pattern
Canadian Adolescents	Drapeau et al. [16]	R24W overestimated energy by 8.8%; saturated fat intake overestimated by 25.2%	Systematic overestimation
Burkina Faso Adolescents	Arsenault et al. [18]	Energy intake equivalent within 15% bound for older adolescents; omissions of snacks, fruits, beverages	Underreporting specific food items
Japanese Adults	Nakadate et al. [17]	Moderate correlations for most nutrients; bias within ±10% for most nutrients	Variable accuracy by nutrient type
Multi-Center Clinical Trial	Wong et al. [14]	DLW method highly reproducible over 2.4-4.5 years; valid for monitoring adherence	Gold standard reliability established

These validation studies demonstrate that while 24-hour recalls can provide reasonable estimates of average intake at a group level for some nutrients, they frequently exhibit systematic biases and often fail to accurately capture intake distributions at the individual level. The Burkina Faso adolescent study revealed that approximately half of participants omitted foods in recalls, particularly sweet or savory snacks, fruits, and beverages [18]. This pattern of selective omission highlights the cognitive challenges and social desirability biases that compromise the accuracy of self-reported dietary data.

Advantages and Limitations in Validation Research

When evaluating DLW against alternative validation approaches, several distinct advantages emerge:

Advantages:

• Free-living assessment: Unlike calorimetry, DLW allows for measurement under completely free-living conditions without restricting participants' activities or diet [12] [13].
• Integrated measurement: DLW provides a cumulative measure of total energy expenditure over 1-3 weeks, smoothing day-to-day variability [10].
• Non-reactive: The method does not alter participants' behavior, avoiding the Hawthorne effect common in observed intake studies [12].
• Multicomponent data: DLW simultaneously assesses body composition through isotope dilution spaces [13].
• Broad applicability: Proven valid across all age groups, from newborns to elderly populations [13].

Limitations:

• Cost-prohibitive: The isotopes and analytical requirements make DLW expensive for large-scale studies [10] [19].
• Technical complexity: Requires specialized equipment and expertise in isotopic analysis [12].
• Group-level focus: While precise for groups, individual measurements have 2-8% error rates [12].
• Time-integrated: Provides energy expenditure over 1-3 weeks but cannot assess intake for specific days [10].

Despite these limitations, DLW remains the only method that provides objective validation of energy intake without disrupting normal living patterns, establishing it as an indispensable tool for evaluating and refining dietary assessment methodologies.

The Scientist's Toolkit: Essential Research Reagents and Materials

Table 4: Essential Research Reagents and Materials for DLW Studies

Reagent/Material	Function	Technical Specifications
Doubly Labeled Water (²H₂¹⁸O)	Isotopic tracer for assessing water turnover and CO₂ production	¹⁸O enrichment: 10-20%; ²H enrichment: 5-10%; Sterile, pyrogen-free
Isotope Ratio Mass Spectrometer	Precise measurement of isotopic enrichment in biological samples	Precision: ≤0.3‰ for δ¹⁸O; ≤1.0‰ for δ²H; Automated sample introduction
Laser-Based Absorption Spectrometer	Alternative to MS for isotopic analysis; lower cost and operational complexity	Precision: ≤0.1‰ for δ¹⁸O; ≤0.5‰ for δ²H; High-throughput capability
International Isotope Standards	Calibration and quality control of isotopic measurements	VSMOW2, SLAP2, GISP; Traceable to international reference materials
Sample Collection Kits	Standardized biological sample collection and storage	Urine/saliva containers; Parafilm seals; Freezer vials (-20°C storage)
Data Analysis Software	Calculation of elimination rates, CO₂ production, and TEE	Implementation of Schoeller, Speakman, or other validated equations

The DLW methodology requires careful attention to quality control throughout the experimental process. Recent advances in laser-based spectroscopy have created opportunities for more accessible and cost-effective isotopic analysis, with studies demonstrating that cavity ring-down spectroscopy and off-axis integrated cavity output spectroscopy can provide precision comparable to traditional mass spectrometry [10]. These technological developments promise to expand the application of DLW validation to broader research contexts.

The doubly labeled water method represents an indispensable gold standard for validating energy intake assessment in nutritional research. Its unique capacity to objectively measure total energy expenditure under free-living conditions, combined with its non-invasive nature and proven accuracy across diverse populations, establishes DLW as the reference method against which all other dietary assessment tools must be calibrated. The experimental evidence consistently demonstrates that self-reported methodologies, including 24-hour dietary recalls, exhibit significant and variable reporting biases that would remain undetected without objective validation using DLW.

For researchers focused on 24-hour recall validation, DLW provides the critical benchmark needed to quantify reporting errors, identify methodological limitations, and develop improved assessment strategies. The longitudinal reproducibility of DLW measurements further supports its application in studies monitoring dietary adherence and body composition changes over time. While cost and technical requirements remain challenging, ongoing methodological advances continue to enhance the accessibility and precision of this fundamental technique. As nutritional science continues to address complex questions regarding energy balance, metabolic health, and dietary interventions, the doubly labeled water method will remain essential for ensuring the validity of the dietary data underlying public health recommendations and clinical practice.

In nutritional epidemiology, the accurate measurement of energy intake (EI) is fundamental to understanding diet-disease relationships, validating dietary assessment instruments, and informing public health policy. The correlation between total energy expenditure (TEE) and reported energy intake (EI) serves as a critical validation criterion for dietary assessment methods, as TEE should theoretically equal EI in weight-stable individuals. However, extensive research reveals significant discrepancies between these metrics, largely attributable to methodological limitations and reporting errors inherent in self-reported dietary data.

The doubly labeled water (DLW) method has emerged as the gold standard for measuring TEE in free-living individuals, providing an objective benchmark against which self-reported EI can be validated. This comprehensive review examines the correlation between TEE and EI across different populations and assessment methodologies, synthesizing evidence from controlled feeding studies, biomarker-validated research, and large-scale epidemiological investigations to provide researchers with a rigorous evaluation of measurement validity in the context of 24-hour recall validation.

Methodological Foundations: Experimental Protocols for TEE and EI Assessment

Gold-Standard Protocol: Doubly Labeled Water (DLW) Method

The DLW method measures TEE by tracking the elimination of stable isotopes from the body after ingestion, providing an unobtrusive measure of carbon dioxide production over 1-2 weeks under free-living conditions.

• Isotope Administration: Participants ingest a measured dose of water containing stable isotopes (^2^H (deuterium) and ^18^O (oxygen-18)). Typical doses are 1.5 g/kg body weight of H218O (10 atom % excess) and 0.05 g/kg body weight of 2H2O (>99 atom % excess) followed by a 100-mL tap water rinse [2].
• Urine Sample Collection: Spot urine samples are collected at baseline (pre-dose), and at 1-, 7-, and 14-days post-dose. Samples are stored at -80°C until analysis [2].
• Isotope Ratio Mass Spectrometry: Samples are analyzed by isotope ratio mass spectrometry to determine elimination rates of both isotopes [2].
• TEE Calculation: Carbon dioxide production rate is calculated from the difference in elimination rates between ^2^H and ^18^O. Total EE is derived using an estimate of respiratory quotient [2].

Controlled Feeding Protocol for Validation Studies

Controlled feeding studies with unobtrusive weighing of foods provide the most rigorous validation of self-reported EI.

• Study Design: Randomized crossover designs where participants consume breakfast, lunch, and dinner under supervision [15].
• Food Weighing: All foods and beverages are weighed unobtrusively before and after consumption using digital scales accurate to 1 g [20].
• Recall Administration: Participants complete 24-hour dietary recalls the following day using the method under investigation [15].
• Data Analysis: True and estimated energy and nutrient intakes are compared using linear mixed models, equivalence testing, and Bland-Altman analysis [15] [17].

24-Hour Dietary Recall Protocol

The multiple-pass 24-hour recall method involves structured interviews to enhance completeness and accuracy.

• Quick List: Participants freely list all foods and beverages consumed the previous day without prompting [2] [20].
• Forgotten Foods Probe: Interviewer probes for commonly forgotten items (sweets, snacks, beverages) [20].
• Time and Occasion: Consumption time and eating occasion are collected for each item [20].
• Detail Cycle: Comprehensive details are collected including food description, preparation methods, and portion sizes using standardized aids [2] [20].
• Final Review: Participant reviews the complete dietary record for accuracy and completeness [2].

Table 1: Key Methodological Protocols in TEE and EI Correlation Research

Method Type	Primary Purpose	Key Procedural Steps	Duration	Output Metrics
Doubly Labeled Water	Objective TEE measurement	Isotope administration, urine collection, mass spectrometry	10-14 days	Total energy expenditure (kcal/day)
Controlled Feeding	Validation of dietary assessment	Unobtrusive food weighing, subsequent recall	1-3 days	Accuracy of estimated vs. true intake
24-Hour Recall	Self-reported EI assessment	Multiple-pass interview, portion size estimation	1 day	Estimated energy and nutrient intake
Weighed Food Record	Reference method in free-living	Participant-weighed all foods consumed	1-7 days	Detailed dietary intake data

Quantitative Synthesis: Correlation Data Across Populations

Pediatric Populations

Research in children reveals particular challenges in assessing EI, with studies showing considerable variability in the TEE-EI correlation.

In a study of 47 children aged 6-9 years, TEE was measured by DLW over 10 days and EI was assessed using 3-day food records. The mean values for EI (7514 ± 1260 kJ/d) and TEE (7396 ± 1281 kJ/d) were not significantly different at the group level, supporting the use of 3-day food records for population-level surveys. However, the lack of significant correlation between EI and TEE at the individual level, wide limits of agreement (118 ± 3345 kJ/d), and mean misreporting of 4 ± 23% highlight substantial measurement error in individual assessments [21] [22].

Adult Populations

Studies in adults demonstrate systematic underreporting in self-reported EI and illuminate factors influencing reporting accuracy.

Research with 79 middle-aged white women who completed seven 24-hour recalls over 14 days while TEE was measured by DLW found significant underreporting. The mean EE from DLW was 2115 kcal/day, while adjusted 24-hour recall-derived EI was substantially lower at calls 1 (1501 kcal/day), 2 (2246 kcal/day), and 3 (2315 kcal/day). Energy intake was significantly lower on Fridays compared to Sundays. Averaging multiple recalls significantly improved accuracy, with the first three recalls providing the optimal balance between participant burden and estimation precision [2].

Technology-Assisted Dietary Assessment

Recent research has evaluated the performance of technology-assisted 24-hour recall tools against objective measures.

A randomized crossover feeding study with 152 participants compared four technology-assisted dietary assessment methods: ASA24, Intake24, mobile Food Record-Trained Analyst (mFR-TA), and Image-Assisted Interviewer-Administered 24-hour recall (IA-24HR). The mean difference between true and estimated energy intake as a percentage of true intake was 5.4% for ASA24, 1.7% for Intake24, 1.3% for mFR-TA, and 15.0% for IA-24HR. While several methods estimated average intakes reasonably well, only Intake24 accurately captured intake distributions for both energy and protein [15].

Table 2: TEE and EI Correlation Across Populations and Assessment Methods

Population	Sample Size	TEE Method	EI Method	Mean TEE	Mean EI	Correlation/Agreement	Key Findings
Children (6-9y) [21]	47	DLW (10d)	3-d food record	7396 ± 1281 kJ/d	7514 ± 1260 kJ/d	No significant correlation; Wide LoA: 118 ± 3345 kJ/d	Valid for group surveys; Poor individual accuracy
Middle-aged women [2]	79	DLW (14d)	Seven 24HRs	2115 kcal/d	Call 1: 1501; Call 2: 2246; Call 3: 2315 kcal/d	3 recalls optimal for group mean	Significant under-reporting on first recall
Japanese adults [17]	228	-	Web24HR vs WFR	-	-	Moderate correlation (men: r=0.51; women: r=0.38)	Bias within ±10% for most nutrients
Adolescents (Burkina Faso) [20]	237	-	24HR vs OWR	-	-	Ratio 0.88-0.92; Equivalence within 15% bound	Acceptable underestimation for population assessment
Controlled feeding [15]	152	Weighed food	4 tech-assisted tools	-	-	Accuracy: mFR-TA (1.3%) > Intake24 (1.7%) > ASA24 (5.4%) > IA-24HR (15.0%)	Tech tools valid for group means

Visualizing the Validation Workflow: From Data Collection to Correlation Analysis

The following diagram illustrates the comprehensive workflow for validating self-reported Energy Intake against objectively measured Total Energy Expenditure, integrating the key methodological approaches discussed:

The Researcher's Toolkit: Essential Reagents and Materials

Table 3: Essential Research Materials for TEE and EI Correlation Studies

Category	Specific Items	Research Function	Application Notes
Stable Isotopes	^2^H₂O (Deuterium oxide), H₂^18^O (Oxygen-18 water)	DLW method for TEE measurement	Require isotope ratio mass spectrometry for analysis [2]
Dietary Assessment Software	ASA24, Intake24, NDS-R, mFR	Standardized nutrient analysis	Must use country-specific food composition databases [15] [17]
Portion Size Estimation Aids	2D food models, household measures, playdough models, digital images	Enhanced portion size reporting	Culture-specific aids improve accuracy [2] [20]
Weighing Equipment	Digital scales (7-kg capacity, ±1g accuracy)	Gold-standard food intake measurement	Required for controlled feeding and weighed record studies [20]
Biological Sample Collection	Urine containers, -80°C freezer, cryovials	Sample preservation for DLW analysis	Strict chain-of-custody protocols required [2]
Quality Control Instruments	Social desirability scales, approval motivation scales	Assessment of reporting bias	Marlowe-Crowne Social Desirability Scale commonly used [2]

Implications for Research and Practice

The body of evidence examining the correlation between TEE and EI reveals several critical considerations for researchers designing studies involving dietary assessment:

• Multiple Recalls Are Essential: For population-level estimates, three non-consecutive 24-hour recalls appear optimal for balancing participant burden with statistical precision, as single recalls demonstrate significant underreporting and day-to-day variability [2].

• Technology-Assisted Tools Show Promise: Automated systems like Intake24 and ASA24 demonstrate reasonable validity for group means, with accuracy comparable to interviewer-administered recalls but with substantially reduced resource requirements [15].

• Population-Specific Validation Needed: The accuracy of self-reported EI varies substantially across age groups, cultural contexts, and physiological status, necessitating population-specific validation studies rather than extrapolation from other groups [21] [20].

• Objective Biomarkers Strengthen Design: Incorporating DLW or other objective measures of TEE in validation substudies significantly enhances the credibility of dietary assessment in large-scale epidemiological research [2] [23].

These findings underscore that while self-reported EI methods have significant limitations for estimating absolute intake at the individual level, they remain valuable for assessing group-level means, ranking individuals by intake, and evaluating diet-disease relationships when implemented with appropriate methodological rigor and statistical adjustment for measurement error.

In scientific research, particularly in studies validating 24-hour dietary recalls for energy intake estimation, understanding measurement error is fundamental to interpreting data accurately. Measurement error refers to the difference between an observed value and the true value of something [24]. These errors are typically categorized into two main types: random error and systematic error [24] [25]. In the context of dietary assessment, self-reported data from tools like 24-hour recalls are inherently susceptible to both types of error, which can have serious consequences for study findings and their interpretation [26]. The distinction between these errors is critical for researchers, scientists, and drug development professionals who rely on accurate energy intake data to investigate diet-health relations, inform public health policy, and assess the efficacy of nutritional interventions [27] [28].

The following diagram illustrates the core concepts and distinct impacts of random and systematic error on a set of measurements, where the bullseye represents the true value.

Theoretical Foundations of Random and Systematic Error

Characteristics of Random Error

Random error is a chance difference between the observed and true values of a measurement [24]. It affects measurements in unpredictable ways, making them equally likely to be higher or lower than the true values [25]. This type of error mainly affects precision, which is the degree to which repeated measurements of the same thing under equivalent circumstances produce the same result [24] [25]. In dietary assessment, random errors can arise from day-to-day variation in food intake, imprecise estimation of portion sizes, or fluctuations in the assessment environment [27] [28]. Because random errors occur in different directions, they tend to cancel each other out when multiple measurements are averaged, bringing the mean closer to the true value [25].

Characteristics of Systematic Error

Systematic error, also referred to as bias, is a consistent or proportional difference between the observed and true values of something [24]. Unlike random error, it skews measurements in a specific direction, meaning every measurement will differ from the true measurement in the same way [25]. Systematic error primarily affects accuracy, which is how close the observed value is to the true value [24]. In dietary recalls, systematic errors can include energy underreporting, social desirability bias (where participants report what they believe the researcher wants to hear), and reactivity (where the act of recording alters normal eating behavior) [27] [26]. Systematic error does not average out with repeated measurements; instead, it introduces a consistent distortion that requires specific mitigation strategies [25].

Table 1: Core Characteristics of Random and Systematic Error

Characteristic	Random Error	Systematic Error
Definition	Unpredictable, chance variations [24]	Consistent, directional bias [24]
Impact on Data	Reduces precision and reliability [25]	Reduces accuracy and validity [25]
Directionality	Occurs in both directions (high and low) [25]	Occurs consistently in one direction [25]
Effect of Averaging	Diminishes with repeated measurements [25]	Not diminished by repeated measurements [25]
Common Causes in 24HR	Day-to-day intake variation, rounding of portion sizes [27] [26]	Under-reporting, social desirability, instrument miscalibration [27] [26]

Experimental Protocols for Error Validation in 24-Hour Recalls

Validating 24-hour dietary recalls (24HR) requires robust experimental designs that can isolate and quantify both random and systematic errors. The following methodologies are considered gold standards in the field.

Controlled Feeding Studies with Observed Intake

Controlled feeding studies serve as a critical protocol for directly comparing reported intake against a known, true intake. In this design, participants consume meals prepared and weighed by research staff in a controlled setting, providing a definitive baseline for comparison [15]. A recent randomized crossover feeding study utilized this method to evaluate the accuracy of four technology-assisted dietary assessment methods, including the Automated Self-Administered Dietary Assessment Tool (ASA24) and Intake24 [15]. Participants were randomized to consume breakfast, lunch, and dinner on separate feeding days, with all foods and beverages unobtrusively weighed. The following day, participants completed a 24HR using one of the assigned methods. The direct comparison of true (weighed) intake versus estimated intake from the 24HR allows researchers to calculate the total measurement error and, with careful design, parse out its systematic and random components [15].

Biomarker-Based Validation Studies

The use of recovery biomarkers represents the most robust method for detecting systematic error in energy and specific nutrient intake. A recovery biomarker is a biological product that is directly related to intake and not subject to homeostasis or substantial inter-individual differences in metabolism [28]. The primary biomarkers used are:

• Doubly Labeled Water (DLW) for total energy expenditure (a proxy for energy intake) [27] [28].
• Urinary Nitrogen for protein intake [27] [28].
• Urinary Potassium and Sodium for potassium and sodium intakes, respectively [28].

In studies such as the Observing Protein and Energy Nutrition (OPEN) study and the Women's Health Initiative Nutrition Biomarkers Study (NBS), participants complete 24HRs while their physiological levels of these biomarkers are measured [28]. Because the biomarkers provide an objective measure of intake independent of self-report, the difference between the 24HR estimate and the biomarker-derived estimate quantifies the systematic error (bias) [27] [28].

Relative Validity Studies with Weighed Food Records

When controlled feeding or biomarker studies are not feasible, relative validity studies using weighed food records (WFR) as a reference method are commonly employed. In this protocol, participants maintain a detailed WFR, where all consumed foods and beverages are weighed and recorded, typically over multiple days [17]. Subsequently, participants complete the 24HR tool under investigation. For example, a 2024 study validating a web-based 24HR for Japanese adults collected 12 days of WFR over a year and compared them to three non-consecutive Web24HR administrations [17]. While WFRs are themselves subject to error (e.g., reactivity), they are often considered a superior method to recalls and thus a practical reference for estimating validity in free-living populations [17].

The workflow below summarizes the key experimental pathways for validating 24-hour dietary recalls.

Comparative Performance Data of Dietary Assessment Methods

The following tables synthesize quantitative data from recent validation studies, providing a clear comparison of how different dietary assessment methods perform with respect to measurement error.

Table 2: Accuracy of Energy Intake Estimation from a Controlled Feeding Study (n=152) [15]

Dietary Assessment Method	Mean Difference (True vs. Estimated) (% of True Intake)	95% Confidence Interval	Key Conclusion
ASA24 (Australia)	+5.4%	(+0.6%, +10.2%)	Moderate overestimation
Intake24 (Australia)	+1.7%	(-2.9%, +6.3%)	Accurate at the group level
mFR-Trained Analyst	+1.3%	(-1.1%, +3.8%)	Accurate at the group level
IA-24HR	+15.0%	(+11.6%, +18.3%)	Significant overestimation

Table 3: Relative Validity of Web-Based 24HR in Different Populations

Study Population & Tool	Reference Method	Correlation for Energy (r)	Mean Bias for Energy	Key Findings
Canadian Adolescents (R24W) [16]	Interviewer 24HR	Not specified	+8.8% (p < 0.05)	Significant overestimation; acceptable relative validity for most nutrients.
Japanese Adults (Web24HR) [17]	Weighed Food Record (WFR)	Median r = 0.51 (Men), 0.38 (Women)	Within ±10% for most nutrients	Moderate correlations; bias acceptable for most nutrients.

Table 4: Systematic Error (Under-Reporting) Identified via Recovery Biomarkers [28]

Dietary Assessment Method	Nutrient	Range of Mean Under-Reporting	Implication
24-Hour Recalls (24HR)	Energy	6% to 26%	Substantial systematic error exists
Food Frequency Questionnaires (FFQ)	Energy	24% to 33%	Greater systematic error than 24HR
Food Records	Energy	~20%	Systematic error similar to 24HR

The Scientist's Toolkit: Key Reagents and Materials

Table 5: Essential Research Reagents and Tools for Dietary Validation Studies

Item	Function/Application	Key Features
Doubly Labeled Water (DLW) [27] [28]	The gold-standard recovery biomarker for validating total energy intake.	Measures carbon dioxide production to calculate total energy expenditure; not subject to self-report bias.
Automated Self-Administered 24HR (ASA24) [26] [15]	A web-based, self-administered 24-hour dietary recall system.	Uses a multiple-pass method to enhance memory recall; eliminates interviewer bias; adaptable for different populations.
Weighed Food Records (WFR) [17]	A reference method in relative validity studies.	Involves weighing all food and drink before consumption; provides highly quantitative intake data.
Automated Multiple-Pass Method (AMPM) [27] [26]	A structured interview protocol for 24-hour recalls.	Developed by the USDA; uses five passes to minimize memory lapse and improve portion size estimation.
GloboDiet (formerly EPIC-SOFT) [27] [26]	A computerized 24-hour recall interview software.	Standardized across countries; includes culture-specific probing questions to enhance detail and accuracy.

The validation of 24-hour dietary recalls for energy intake estimation hinges on a clear understanding of random and systematic errors. The body of evidence demonstrates that while all self-report methods contain error, 24-hour recalls generally exhibit less systematic error than food frequency questionnaires, though they are still prone to significant energy under-reporting [28]. Random error, largely driven by day-to-day variation, can be mitigated by repeating dietary assessments [27] [25]. In contrast, addressing systematic error, such as the consistent under-reporting of energy, requires more sophisticated methods like the use of recovery biomarkers (e.g., Doubly Labeled Water) for detection and calibration [27] [28].

The choice of assessment method, as shown in comparative studies, directly impacts the error structure of the resulting data. Tools like ASA24 and Intake24 can provide reasonable estimates of average group intake, but their accuracy varies [16] [15] [17]. For researchers and drug development professionals, this implies that study design must account for these errors. Utilizing multiple 24-hour recalls per participant and incorporating statistical adjustments that account for the error structure, potentially informed by biomarker sub-studies, are essential strategies for producing reliable data that can accurately inform public health policy and clinical practice [27] [28].

From Theory to Practice: Implementing Robust 24-Hour Recall Protocols

The Automated Multiple-Pass Method (AMPM) is a computerized, interviewer-administered method for collecting 24-hour dietary recalls. Developed by the USDA, it employs a research-based, multiple-pass approach with five distinct steps designed to enhance the completeness and accuracy of food recall while reducing the burden on respondents [29]. It serves as the instrument for collecting 24-hour dietary recalls in major surveys like the National Health and Nutrition Examination Survey (NHANES), forming the backbone of national nutrient intake data [30].

This guide objectively compares the AMPM against other technology-assisted dietary assessment methods, framing the evaluation within research on validating 24-hour recalls for energy intake estimation.

Experimental Comparison of Dietary Assessment Methods

A 2024 controlled crossover feeding study compared the accuracy of four technology-assisted dietary assessment methods against objectively measured true intake [15].

• Study Design: The study involved 152 participants (55% women, mean age 32, mean BMI 26 kg/m²) who were randomized to consume breakfast, lunch, and dinner on separate feeding days, with all foods and beverages unobtrusively weighed [15].
• Method Comparison: The following day, participants completed a 24-hour recall using one of four methods:
  • ASA24: The Automated Self-Administered Dietary Assessment Tool.
  • Intake24: An online 24-hour dietary recall tool.
  • mFR-TA: The mobile Food Record, analyzed by a trained analyst.
  • IA-24HR: An Image-Assisted Interviewer-Administered 24-hour recall, where participants used images captured with the mFR app [15].
• Outcome Measures: The primary outcomes were the accuracy of energy and nutrient intake estimation, calculated as the mean percentage difference between estimated and true intake [15].

The table below summarizes the core findings on energy estimation accuracy from this controlled study.

Method Name	Method Type	Mean Difference in Energy Estimation (% of True Intake)
ASA24	Automated Self-Administered	5.4% overestimation [15]
Intake24	Automated Self-Administered	1.7% overestimation [15]
mFR-TA	Image-Based, Analyst Reviewed	1.3% overestimation [15]
IA-24HR	Image-Assisted Interviewer	15.0% overestimation [15]
AMPM (for context)	Interviewer-Administered	Data from other studies deemed plausible [30]

The study concluded that under controlled conditions, Intake24, ASA24, and mFR-TA estimated average energy and nutrient intakes with reasonable validity. However, only Intake24 accurately estimated the distribution of energy and protein intakes, not just the group mean [15].

Detailed Experimental Protocols

To ensure reproducibility and critical appraisal, this section details the methodologies of key experiments cited.

This protocol describes the rigorous validation study comparing the four dietary assessment methods.

1. Participant Recruitment & Randomization

• Participants: 152 adults (55% women) with a mean age of 32 years and a mean BMI of 26 kg/m² [15].
• Randomization: Participants were randomized to one of three separate feeding days [15].

2. Controlled Feeding & True Intake Measurement

• Procedure: Participants consumed breakfast, lunch, and dinner in a controlled setting.
• True Intake Measurement: All foods and beverages were unobtrusively weighed before and after consumption to establish a "true intake" baseline [15].

3. Dietary Assessment Method Application

• 24-Hour Recall: The day after the feeding session, participants undertook a 24-hour dietary recall using one of the four assigned methods (ASA24, Intake24, mFR-TA, or IA-24HR) [15].

4. Data & Statistical Analysis

• Comparison: Researchers compared the estimated energy and nutrient intakes from each method to the true intake values [15].
• Analysis: Differences among methods were assessed using linear mixed models [15].

This study assessed the practicality of conducting AMPM interviews in participants' homes, a critical step for large-scale surveys facing declining response rates and rising costs.

1. Study Design and Sampling

• Design: A feasibility study conducted in central North Carolina between November 2019 and February 2020 [30].
• Sampling: A two-stage design was used, purposively selecting census block groups for demographic variability, followed by random selection of residential addresses [30].
• Participants: 133 completed interviews with English-speaking adults. Pregnant women and individuals who were fasting were excluded [30].

2. Experimental Interventions & Randomization

• Randomization: Participants were randomly assigned to one of six study arms, representing combinations of two variables [30]:
  • Interviewer Type: Nutritionist (standard) vs. Field interviewer (no specialized nutrition knowledge).
  • Portion Estimator: 3D food models (standard), 2D food model booklet, or a tablet with augmented reality images.
• Training: All interviewers underwent a standardized "train-the-trainer" certification process, including didactic sessions, role-playing, and a passing score on a take-home exam [30].

3. Interview Execution

• Setting: Interviews took place in participants' homes, most frequently in living rooms (52%) or kitchens (22%) [30].
• Tool: The USDA's 2016 AMPM computer-assisted personal interviewing (CAPI) software was used on a password-protected laptop [30].

4. Outcome Measures & Analysis

• Interview Time: Mean interview time was 40 minutes (range 13-90), with no significant difference by interviewer type or portion estimator [30].
• Participant Behavior: 45% of participants referenced items from their own homes (e.g., bowls, cups) to aid portion estimation and memory [30].
• Data Quality: Collected data for energy and key nutrients were deemed plausible and within expected ranges (e.g., average 3,011 kcal for men and 2,105 kcal for women) and met NHANES consistency requirements, regardless of interviewer type or portion estimator used [30].

The Scientist's Toolkit: Research Reagent Solutions

The table below details essential materials and tools used in AMPM and comparative dietary assessment research, with a brief explanation of each item's function.

Item/Category	Function in Dietary Assessment Research
AMPM CAPI Software	The core computerized system that structures the 5-pass interview, standardizes questioning, and facilitates data entry [30].
Portion Estimation Aids	Assist respondents in converting the food they recall into quantifiable amounts. Critical for accurate nutrient calculation [30].
3D Food Models	Three-dimensional, life-size models of common foods and dishes. The standard tool used in NHANES mobile examination centers [30].
2D Food Model Booklet (FMB)	A portable booklet containing life-size, two-dimensional photographs of food models. Used for telephone and home interviews [30].
Augmented Reality (AR) Tablet	An emerging technology that renders life-size, 3D images of food models into the user's real-world environment via a tablet screen [30].
Standardized Recipe Database	A comprehensive database of food composition and, crucially, recipes for mixed dishes. Essential for accurately breaking down meals into ingredients [17].
Doubly Labeled Water (DLW)	An objective, biomarker-based method for measuring total energy expenditure in free-living individuals. Considered a gold standard for validating self-reported energy intake [31].
Activity Monitors (e.g., Fitbit)	Consumer-grade wearable devices that estimate total energy expenditure using accelerometry and heart rate. Offer a lower-cost alternative to DLW for large-scale studies [32].

Discussion and Research Context

The data demonstrates a trade-off between the high feasibility and standardization of the AMPM and the variable accuracy of different technology-assisted methods. While the AMPM produces plausible and consistent data used for national policy [30], the controlled feeding study reveals that some automated methods (Intake24, mFR-TA) can achieve high accuracy for average energy intake estimation, with the added benefit of lower cost and reduced interviewer burden [15].

A critical challenge in this field, relevant to validating any 24-hour recall method, is the widespread underreporting of energy intake in self-reported data. A 2024 study comparing an online tool (Intake24) with energy expenditure from an activity tracker found energy intake was underreported by an average of 33% [32]. This systematic error varies by demographic, with greater underreporting observed in men, younger individuals, and those with higher BMI [32]. This underscores the importance of using objective validation measures, like controlled feeding or DLW, and of complementing dietary survey data with other population-level energy intake proxies [31].

For researchers and drug development professionals, the choice of dietary assessment method should be guided by the study's primary objective. The AMPM remains the benchmark for structured, interviewer-administered recalls, particularly when high-quality control and detailed food-based data are required for regulatory or national surveillance purposes. However, for large-scale epidemiological studies where cost and participant burden are primary concerns, validated automated tools like Intake24 and ASA24 present a viable and accurate alternative for estimating group-level energy and nutrient intakes.

The accurate measurement of dietary intake is a cornerstone of nutritional epidemiology, chronic disease research, and public health monitoring. For decades, researcher-administered 24-hour dietary recalls (24HRs) served as the standard method, despite limitations in cost, scalability, and potential interviewer bias. The digital transformation has introduced web-based, self-administered tools like the Automated Self-Administered 24-hour Dietary Assessment Tool (ASA24), developed by the National Cancer Institute (NCI), and the R24W, a similar tool developed for French-speaking populations. These tools automate the dietary recall process, offering a promising solution for large-scale studies. Framed within the broader thesis of validating 24-hour recalls for energy intake estimation, this guide objectively compares the performance of these digital tools against traditional methods and recovery biomarkers, providing researchers with the experimental data necessary for tool selection.

This section provides a detailed comparison of the core features and technical foundations of ASA24 and the R24W.

Table 1: Core Tool Specifications

Feature	ASA24	R24W
Developer	National Cancer Institute (NCI), USA	Research team at Université Laval, Canada
Primary Language(s)	English, Spanish [33]	French [34] [35]
Underlying Methodology	Adapted from USDA's Automated Multiple-Pass Method (AMPM) [36]	Inspired by USDA's AMPM, uses a meal-based approach [34]
Key Database (US Versions)	Food and Nutrient Database for Dietary Studies (FNDDS) [33]	Canadian Nutrient File (CNF) [34]
Cost for Researchers	Free [33]	Information not specified in search results
Mobile Responsiveness	Yes (HTML5) [33]	Information not specified in search results

ASA24 is updated approximately every two years to incorporate updated food composition databases. As of November 2025, the ASA24-2024 version is the most recent, with underlying data from FNDDS 2019-2020. The previous version, ASA24-2022, is scheduled for retirement in April 2025 [33]. This continual update process ensures nutrient intake data remains current.

Performance Comparison: Validation Against Objective Measures

A critical step in validating any dietary assessment tool is comparing its estimates against objective, unbiased measures known as recovery biomarkers. These biomarkers, including doubly labeled water for energy expenditure and urinary nitrogen for protein intake, provide a reference point for assessing the accuracy of self-reported data.

Criterion Validity in Feeding Studies

Feeding studies, where researchers provide all meals and know the precise composition and weight of consumed foods, offer the strongest design for criterion validity testing.

• ASA24 Performance: One feeding study with 81 participants found that ASA24 respondents reported 80% of items truly consumed, which was slightly lower than the 83% reported with an interviewer-administered AMPM recall, though the difference was not statistically significant (P=0.07) [37]. The study found little evidence of differences between ASA24 and the interviewer-led method in the gap between true and reported energy, nutrient, and food group intakes [37].
• R24W Performance: In a controlled feeding study (n=62), the R24W demonstrated a high rate of item reporting, with participants recording 89.3% of the food items they received [34]. The correlation between offered and self-reported portion sizes was strong (r=0.80, P<0.001), though accuracy varied by portion size. Small portions (under 100g) were overestimated by 17.1%, while larger portions (100g and above) were slightly underestimated by 2.4% [34].

Comparison with Recovery Biomarkers in Free-Living Populations

Studies in free-living populations that compare tool outputs to recovery biomarkers reveal how tools perform under real-world conditions.

• ASA24 and Energy Reporting: Data from the Interactive Diet and Activity Tracking in AARP (IDATA) study (n=1,077) confirmed that energy intake estimates from ASA24-2011 were lower than energy expenditure measured by doubly labeled water, a common finding known as energy underreporting [38]. However, the reported intakes for protein, potassium, and sodium were closer to their respective urinary biomarkers for women than for men [38].
• Relative Validity of ASA24: The 2017 Women's Lifestyle Validation Study concluded that, in general, averaged ASA24s (using the beta version) had lower validity relative to biomarkers than a paper-based food frequency questionnaire (SFFQ) or 7-day dietary records (7DDRs). The study further noted that "an average of 3 days of [ASA24] measurement will not be sufficient for some important nutrients" [39].

Table 2: Summary of Key Validation Study Results

Study (Tool)	Design	Key Findings on Validity
Kirkpatrick et al. [37] (ASA24)	Feeding study (n=81) vs. true intake	Reported 80% of consumed items; energy/nutrient estimates comparable to interviewer-administered recall.
Brassard et al. [35] (R24W)	Population survey vs. interviewer recall	R24W produced 18% higher energy intakes in women and 15% higher in men than traditional recall.
IDATA Study [38] (ASA24)	Free-living adults (n=1,077) vs. biomarkers	Energy underreporting observed; protein/sodium/potassium reports were closer to biomarkers for women.
Women's Lifestyle Study [39] (ASA24)	Free-living women (n=627) vs. biomarkers	Averaged ASA24s had lower validity than SFFQs or diet records for many nutrients.
Drapeau et al. [16] (R24W)	Adolescents (n=272) vs. interviewer recall	Energy intake was 8.8% higher with R24W; "acceptable relative validity" for most nutrients.

Methodological Insights from Experimental Protocols

The validation of these tools relies on rigorous and varied experimental protocols. Understanding these methodologies is crucial for interpreting results and designing future studies.

The Feeding Study Protocol

The feeding study design, as employed in validation research for both ASA24 and R24W, provides a high level of control [34] [37].

Diagram 1: Feeding Study Validation Workflow

Key steps include:

• Precise Weighing: All food and beverage items are inconspicuously weighed before being provided to participants [37].
• Waste Measurement: After consumption, any leftover food or plate waste is weighed to determine the exact amount consumed [37].
• Recall Administration: The following day, participants complete the self-administered recall (e.g., ASA24 or R24W), reporting their intake from the previous day without knowledge that their actual intake is known to researchers.
• Data Comparison: The reported intake from the web-based tool is directly compared to the "true" intake calculated from the weighing data, allowing for analysis of reporting accuracy for foods, portion sizes, energy, and nutrients [34] [37].

The Biomarker Validation Protocol

An alternative and highly robust protocol involves comparing dietary intake from the tool against objective recovery biomarkers in free-living participants [39] [38].

Diagram 2: Biomarker Comparison Study Design

This design often includes:

• Multiple Dietary Methods: Participants typically complete the tool under investigation (e.g., ASA24), alongside other traditional methods like FFQs and diet records, to allow for cross-comparison [39].
• Repeated Biomarker Measures: To account for day-to-day variation, biomarkers are collected multiple times over an extended period (e.g., over 15 months). This includes:
  • Doubly Labeled Water (DLW): To measure total energy expenditure and validate reported energy intake [39] [38].
  • 24-Hour Urine Collections: To measure urinary nitrogen (for protein intake), sodium, and potassium [39] [38].
• Statistical Modeling: Correlation and regression analyses are used to assess the strength of the relationship between reported intakes and biomarker values, providing a measure of the tool's validity [39].

The Scientist's Toolkit: Key Reagents & Materials

Table 3: Essential Research Reagents and Materials for Validation Studies

Item	Function in Validation Research
Doubly Labeled Water (DLW)	A gold-standard recovery biomarker for measuring total energy expenditure in free-living individuals, used to validate self-reported energy intake [39] [38].
Urinary Nitrogen Analysis	The analysis of nitrogen in 24-hour urine collections serves as a recovery biomarker for validating protein intake [39] [38].
24-Hour Urine Collection Kit	Standardized containers and instructions for participants to collect all urine output over a 24-hour period for the analysis of nitrogen, sodium, and potassium [39].
Metabolic Kitchen	A controlled facility for the preparation, precise weighing, and provision of all meals and snacks in a feeding study [34].
Digital Food Scales	High-precision scales used in a metabolic kitchen to weigh food items before and after consumption to determine "true" intake [34] [37].
Standardized Recipe Database	A comprehensive database of food items and mixed dishes with defined nutritional composition, which forms the backbone of tools like ASA24 and R24W for converting reported foods into nutrient intakes [33] [17].

The digital transformation of dietary assessment through tools like ASA24 and R24W offers undeniable advantages in scalability, cost-effectiveness, and automated data processing. Validation studies indicate that these tools perform reasonably well for assessing population-level intakes of many nutrients and food groups.

However, evidence from recovery biomarker studies underscores that they are not without measurement error. ASA24 demonstrates performance similar to interviewer-administered recalls in feeding studies but shows systematic underreporting of energy in free-living populations. The R24W has shown a tendency to yield higher intake estimates than traditional recalls. For researchers, the choice of tool must be guided by the study's specific objectives, population, and required precision. The consensus is that multiple administrations (more than three) are necessary to estimate usual intake for many nutrients, and the integration of biomarker subsudies in large cohorts remains a critical strategy for correcting for measurement error and strengthening diet-disease findings.

Accurate estimation of habitual energy intake is a cornerstone of nutritional epidemiology, yet it is perpetually challenged by methodological biases. A critical, though often underexplored, source of this bias is the temporal framework of data collection. This guide examines how structured protocols for day-of-week and seasonal coverage in 24-hour dietary recall (24HR) studies serve as a fundamental strategy to mitigate these biases, ensuring data more accurately reflects true long-term consumption patterns. Evidence from recent validation studies demonstrates that the failure to account for temporal variations can compromise the validity of intake estimates for energy and key nutrients.

The Impact of Temporal Coverage on Dietary Assessment Validity

Dietary intake is not uniform across days of the week or throughout the year. Weekdays often differ from weekends in dietary patterns, and seasonal availability of food can significantly alter consumption. Ignoring these variations during study design introduces selection bias and representation bias, systematically skewing intake estimates [40]. The following data from recent validation studies highlight how different dietary assessment methods perform when evaluated against objective measures.

Table 1: Comparative Accuracy of Technology-Assisted 24HR Methods in a Controlled Feeding Study cited from [15]

Method	Description	Mean Difference in Energy Intake vs. True Intake (%)	Key Findings
ASA24	Automated Self-Administered Dietary Assessment Tool	5.4%	Overestimated energy intake on average.
Intake24	Online self-administered 24-hour recall	1.7%	Most accurate for estimating average intake and intake distributions.
mFR-TA	Mobile Food Record analyzed by a trained analyst	1.3%	High accuracy for average energy intake estimation.
IA-24HR	Image-Assisted Interviewer-Administered Recall	15.0%	Significantly overestimated energy intake.

Table 2: Performance of a Web-Based 24HR Tool in a Multi-Season Japanese Cohort cited from [17]

Metric	Result	Interpretation
Study Design	Web24HR administered 3 times vs. 12-day Weighed Food Records (WFR) over a year.	Reference data collected over four seasons to capture habitual intake.
Correlation (Energy)	Men: median r = 0.51; Women: median r = 0.38	Moderate validity for estimating energy intake.
Bland-Altman Analysis	Bias within ±10% for most nutrients.	Good agreement for most nutrients, though with underestimation for some.

Detailed Experimental Protocols for Validation

The credibility of a dietary assessment tool rests on the rigor of its validation protocol. The studies cited above provide exemplary methodologies for controlling bias through comprehensive temporal coverage.

Protocol 1: The Controlled Crossover Feeding Study

This design, as executed in [15], provides a high level of control for evaluating accuracy.

• Objective: To compare the accuracy of four technology-assisted 24HR methods against objectively measured true intake.
• Participants: 152 adults (55% women, mean age 32).
• Feeding Phase: Participants were randomized to consume breakfast, lunch, and dinner on one of three separate feeding days. All foods and beverages were unobtrusively weighed to establish "true intake."
• Recall Phase: The following day, participants completed a 24HR using one of the four assigned methods (ASA24, Intake24, mFR-TA, or IA-24HR).
• Data Analysis: True and estimated energy and nutrient intakes were compared using linear mixed models to assess differences and variances.

Protocol 2: The Longitudinal Multi-Season Validation Study

This protocol, detailed in [17], is designed to validate a tool's ability to capture habitual intake over time.

• Objective: To validate a web-based 24HR (Web24HR) developed for the Japanese population against the reference method of Weighed Food Records (WFR).
• Participants: 228 Japanese adults aged ≥20 years.
• Reference Data Collection (WFR): Participants completed 3-day WFRs at 3-month intervals over one year, covering all four seasons. This resulted in 12 days of reference data per participant, capturing seasonal variation.
• Test Data Collection (Web24HR): Participants completed the Web24HR three times: twice non-consecutively within 3 weeks of a WFR, and once 3 months later.
• Data Analysis: Validity was assessed using Pearson’s correlation coefficients and the Bland–Altman method to evaluate agreement and systematic bias between the Web24HR and WFR for energy and multiple nutrients.

Research Workflow: From Protocol Design to Validated Tool

The following diagram illustrates the logical pathway for designing a validation study that effectively mitigates temporal bias.

The Scientist's Toolkit: Essential Research Reagents & Materials

Table 3: Key Reagents and Tools for Dietary Validation Research Compiled from [15] [41] [17]

Item	Function in Research	Example from Literature
Weighed Food Records (WFR)	Serves as a high-fidelity reference method; participants weigh all consumed food and beverages.	Used as the reference standard in the Japanese validation study [17].
Controlled Feeding Study Setup	Provides the "ground truth" of intake by precisely preparing and measuring food given to participants.	Used to obtain "true intake" for comparing 24HR methods [15].
Standardized Recipe Database	Essential for accurately converting reported food consumption into nutrient intake, especially for mixed dishes.	The AWARDJP system used a database of typical Japanese mixed dishes [17].
Web-Based 24HR Platform	Automated self-administered tool for scalable, repeated dietary data collection in large cohorts.	Examples include ASA24, Intake24, and the custom AWARDJP system [15] [17].
Image-Assisted Recall Tools	Uses photos of meals captured by participants to improve the accuracy of portion size estimation and food identification.	Used in the mFR-TA and IA-24HR methods to assist recall [15].

Accurate dietary assessment is a cornerstone of nutritional epidemiology, public health monitoring, and clinical research. The 24-hour dietary recall (24HR) stands as a predominant method for capturing detailed dietary intake data in population studies [42]. However, the accuracy of this method fundamentally depends on the precise estimation of portion sizes, a well-documented source of measurement error [27] [43]. Portion size misestimation can lead to systematic biases, potentially distorting relationships between diet and health outcomes in scientific studies. Traditional methods often rely on household measures (e.g., cups, spoons) and food models to assist participant memory [42]. Recently, image-assisted tools have emerged, leveraging digital technology to potentially enhance accuracy. This guide objectively compares the performance of these methodological approaches, providing researchers with experimental data to inform their selection of dietary assessment tools within the critical context of validating 24-hour recall for energy intake estimation.

Comparative Performance Data: Image-Assisted Tools vs. Established Methods

Evaluations under controlled feeding studies, where true intake is known, provide the most robust data for comparing the accuracy of different dietary assessment methods. The table below synthesizes key findings from recent validation studies on energy and portion size estimation.

Table 1: Comparison of Dietary Assessment Method Accuracy in Controlled Studies

Assessment Method	Type	Mean Difference in Energy vs. Observed Intake	Key Findings on Portion Size/Other Nutrients
Image-Assisted Interviewer-Administered 24HR (IA-24HR) [15]	Interviewer-administered with participant-captured images	+15.0% (95% CI: 11.6, 18.3%)	Not specifically reported for portion sizes in the results.
Automated Self-Administered 24HR (ASA24) [15] [43]	Self-administered web tool with digital images	+5.4% (95% CI: 0.6, 10.2%)	Mean portion size difference: +3.7g; 16.2% of estimates within 10% of truth [43].
Intake24 [15]	Self-administered web tool	+1.7% (95% CI: -2.9, 6.3%)	Accurately estimated intake distributions for energy and protein [15].
Mobile Food Record-Trained Analyst (mFR-TA) [15]	Image analysis by trained analyst	+1.3% (95% CI: -1.1, 3.8%)	Not specifically reported for portion sizes in the results.
Interviewer-Administered AMPM [43]	Interviewer-administered with food model booklet	Not reported for energy in these results.	Mean portion size difference: +11.8g; 14.9% of estimates within 10% of truth [43].
24hR-camera (with Food Atlas) [44]	Interviewer-administered with photos and a food atlas	Not reported for energy in these results.	High correlation for most food groups; lower accuracy for oils, fats, and condiments [44].

The data reveals a clear performance gradient. Fully automated image-based analysis (mFR-TA) and streamlined web systems (Intake24) demonstrated the highest accuracy for energy intake at the group level, with mean differences closest to true intake [15]. Interestingly, simply adding participant-captured images to a traditional interviewer-administered recall (IA-24HR) resulted in the highest level of over-estimation in one study, suggesting the method of image integration and analysis is critical [15]. For portion size specifically, a different study found that a self-administered tool using digital images (ASA24) yielded a smaller average discrepancy from true portion size compared to an interviewer-led method using a food model booklet (AMPM) [43].

Experimental Protocols: Validating Portion Size Estimation Tools

Understanding the experimental designs that generate comparative data is crucial for interpreting results and planning future research. The following are detailed methodologies from key cited studies.

Protocol 1: Randomized Crossover Feeding Study for 24HR Method Comparison

This protocol, designed to compare multiple technology-assisted methods, provides a robust model for validation [15] [45].

• Objective: To compare the accuracy, acceptability, and cost-effectiveness of three technology-assisted 24HR methods (ASA24, Intake24, mFR24) against observed intake.
• Population: 150 healthy adults aged 18-70 years.
• Feeding Procedure: Participants attend a study center on three separate days to consume breakfast, lunch, and dinner. All foods and beverages are inconspicuously weighed before and after consumption to determine "true" intake.
• Recall Procedure: The day after each feeding day, participants complete a 24HR using one of the three methods. The sequence of methods is randomized for each participant in a crossover design.
• Data Analysis: Estimated energy, nutrient, and food group intakes from each 24HR method are compared to observed intake using linear mixed models. Omission (forgotten foods) and intrusion (incorrectly reported foods) rates are also calculated.

Protocol 2: Validation of Image-Assisted 24HR with Food Atlas

This study validates a specific hybrid approach combining participant-taken photos with a food atlas [44].

• Objective: To validate the 24hR-camera method against the gold standard of Weighed Food Records (WFR).
• Population: 30 Japanese males who rarely cook.
• Procedure:
  • WFR: A registered dietitian weighs all food and drink items before serving and weighs leftovers to calculate actual intake.
  • 24hR-camera: On the same day, participants photograph every food and drink item before and after consumption, using a reference object for scale.
  • Interview: The following day, a dietitian conducts a 24-hour recall, using the participant's photographs and a Japanese food atlas (containing photos of common foods in various portion sizes) to estimate intake.
• Data Analysis: Spearman's correlation coefficients and Bland-Altman plots are used to assess agreement between the 24hR-camera and WFR methods for food groups and nutrients.

Protocol 3: Assessing Portion Size Accuracy in ASA24 vs. AMPM

This protocol focuses specifically on the accuracy of portion size estimation, a key component of overall dietary assessment [43].

• Objective: To assess the accuracy of portion size reporting in the ASA24 compared to interviewer-administered AMPM recalls.
• Population: 81 adults from the Washington, DC area.
• Feeding Procedure: Participants select and consume foods from a buffet for three meals. All foods are weighed before serving and after waste is collected to determine "true" intake. The buffet includes a variety of food types (liquids, amorphous foods, single units, spreads).
• Recall Procedure: Participants are randomly assigned to complete either an unannounced ASA24 recall or an interviewer-administered AMPM recall the following day.
• Data Analysis: An adapted Bland-Altman approach is used to assess agreement between true and reported portion sizes. The proportion of portion sizes reported within 10% and 25% of the true value is calculated.

The workflow for the validation of image-assisted tools in a controlled feeding study is summarized in the diagram below.

The Researcher's Toolkit: Essential Reagents and Materials

Successful implementation of portion size estimation methods, particularly in validation studies, requires specific tools and reagents. The following table details key items and their functions.

Table 2: Essential Research Reagents and Materials for Dietary Assessment Validation

Tool/Reagent	Function & Application in Research
Digital Food Scales	Precisely measure the weight of foods and beverages served and leftover in controlled feeding studies to establish "true" intake [43] [44].
Doubly Labeled Water (DLW)	A biomarker used as an objective reference method to validate total energy intake estimates from self-report tools under free-living conditions [46] [47].
Standardized Food Atlas	A photographic guide with images of common foods in multiple portion sizes; used by researchers or participants to improve visual estimation of amounts consumed [44].
3D Food Models & Household Measures	Physical aids (e.g., measuring cups, spoons, shape models) used during interviews to help participants conceptualize and report portion sizes [43] [42].
Fiducial Marker	An object of known size, shape, and color included in photographs of food. It provides a scale reference for software and analysts to estimate portion size from 2D images [45].
Automated Dietary Assessment Platforms (e.g., ASA24, Intake24)	Web-based or mobile systems that automate the 24HR process, incorporating digital portion size images and standardized probes to reduce interviewer burden and cost [15] [42] [48].
Nutrient Composition Database	A standardized database (e.g., Food Patterns Equivalents Database, Canadian Nutrient File) that converts reported foods and portion sizes into estimated nutrient intakes [42] [48].

The empirical data demonstrate that the choice of portion size estimation method significantly impacts the accuracy of energy and nutrient intake assessment. Fully automated image analysis (mFR-TA) and efficient web-based systems (Intake24) show particular promise for achieving high accuracy while managing resource constraints [15]. The integration of digital images, when implemented effectively, can provide a superior alternative to traditional household measures and food models alone [43]. For researchers designing studies on energy intake estimation, the selection of a dietary assessment tool must balance accuracy, participant burden, and cost-effectiveness. The continued development and validation of image-assisted and automated tools are critical for advancing the precision of nutritional epidemiology and strengthening the evidence base linking diet to health and disease.

Identifying and Correcting for Systematic Errors and Biases

The Pervasive Challenge of Under-Reporting and Its Determinants

Accurate dietary assessment is a cornerstone of nutrition research, public health policy, and clinical practice. Among various dietary assessment tools, the 24-hour dietary recall (24HR) has emerged as a predominant method in large-scale nutrition surveillance and epidemiological studies due to its relatively lower bias compared to food frequency questionnaires and its feasibility for population-level administration [49] [27]. However, the validity of 24HR data is fundamentally compromised by systematic measurement errors, with under-reporting of energy intake representing the most pervasive and challenging issue [50]. This systematic review synthesizes current evidence on the extent and determinants of under-reporting in 24-hour dietary recalls, providing researchers with a comprehensive analysis of methodological considerations and potential solutions.

The problem of under-reporting is not merely a statistical concern but has profound implications for interpreting diet-disease relationships and formulating evidence-based dietary guidelines. When energy intake is underestimated, it is probable that intakes of other nutrients are also underestimated, potentially leading to erroneous conclusions about nutritional adequacy and diet-health relationships [50]. Understanding the magnitude, patterns, and determinants of under-reporting is therefore essential for advancing nutritional science and developing more robust dietary assessment methodologies.

Quantitative Evidence of Under-Reporting Across Methods and Populations

Comparative Accuracy of Dietary Assessment Methods

Table 1: Comparison of Energy Intake Under-Reporting Across Dietary Assessment Methods

Assessment Method	Study Population	Reference Standard	Magnitude of Under-Reporting	Citation
ASA24 (multiple recalls)	Adults aged 50-74 years	Doubly Labeled Water	15-17% lower than energy expenditure	[51]
4-Day Food Records	Adults aged 50-74 years	Doubly Labeled Water	18-21% lower than energy expenditure	[51]
Food Frequency Questionnaires	Adults aged 50-74 years	Doubly Labeled Water	29-34% lower than energy expenditure	[51]
Interviewer-Administered 24HR (Korean adults)	Adults aged 20-49 years	Doubly Labeled Water	12.0% lower than energy expenditure (307.5 ± 629.3 kcal/day)	[52]
Web-Based R24W	French-Canadian adults	Estimated Energy Requirement	10% lower prevalence of under-reporting vs. interviewer-administered	[35]
Image-Assisted 24HR (IA-24HR)	Australian adults (controlled feeding)	Weighed Food Intake	15.0% over true intake	[49]
Intake24	Australian adults (controlled feeding)	Weighed Food Intake	1.7% over true intake	[49]

Controlled feeding studies, which directly compare reported intake to actual consumption under observation, provide particularly insightful evidence on the accuracy of different dietary assessment methods. A 2024 Australian study with a crossover design compared four technology-assisted dietary assessment methods under controlled conditions and found notable differences in accuracy [49]. The Image-Assisted Interviewer-Administered 24-hour recall (IA-24HR) overestimated energy intake by 15.0%, while the Automated Self-Administered Dietary Assessment Tool (ASA24) overestimated by 5.4%. In contrast, Intake24 and the mobile Food Record-Trained Analyst (mFR-TA) showed the closest agreement with true intake, with overestimations of just 1.7% and 1.3% respectively [49]. These findings demonstrate that method-specific differences significantly impact reporting accuracy, with self-administered web-based tools generally outperforming interviewer-administered approaches in controlled settings.

Biomarker-based validation studies using doubly labeled water (DLW) have consistently revealed substantial under-reporting across various self-assessment methods. In a comprehensive study comparing multiple dietary assessment tools against recovery biomarkers, all self-reported instruments systematically underestimated absolute intakes of energy, protein, potassium, and sodium [51]. The under-reporting was most pronounced for energy intake, with food frequency questionnaires performing substantially worse (29-34% under-reporting) than multiple ASA24 recalls (15-17%) or 4-day food records (18-21%) [51]. This pattern of under-reporting has been observed across diverse populations, with a study of Korean adults finding that energy intake estimated by 24HR was 12.0% lower than total energy expenditure measured by DLW, equating to an average under-reporting of 307.5 ± 629.3 kcal/day [52].

Population-Specific Patterns in Under-Reporting

Table 2: Determinants and Patterns of Under-Reporting in Different Populations

Determinant	Population	Effect on Under-Reporting	Citation
Obesity	Korean NHANES participants	Higher prevalence among obese individuals	[53]
Gender	Korean adults	Men: 12.2% under-reporting; Women: 11.8% under-reporting	[52]
Age	Korean NHANES	Highest under-reporting in 30-49 age group	[53]
Education Level	Korean NHANES	Higher under-reporting in women with elementary education or less	[53]
Household Status	Korean NHANES	Higher under-reporting in women living alone	[53]
Self-Rated Health	Korean NHANES	Higher under-reporting in those with poor self-rated health	[53]

Analysis of data from the Korean National Health and Nutrition Examination Survey (KNHANES) revealed that under-reporters accounted for 14.4% of men and 23.0% of women, with the highest under-reporting rates observed in the 30-49 age group for both genders [53]. Socioeconomic characteristics also influenced reporting accuracy, with higher under-reporting observed in women living alone and those with only elementary school education or no formal education [53]. Health-specific characteristics showed that a larger proportion of under-reporters had poor self-rated health or were obese compared to non-under-reporters [53]. These findings highlight how sociodemographic and health-related factors systematically influence reporting accuracy in dietary assessments.

A validation study specifically focused on older Korean adults found that age-related factors influenced reporting accuracy differently than in younger populations. While participants aged 60 years and older recalled only 71.4% of the foods they actually consumed and tended to overestimate portion sizes, their reported energy and macronutrient intakes did not statistically differ from weighed intakes [7]. Interestingly, women in this older cohort demonstrated significantly better recall accuracy than men, reporting 75.6% of foods consumed compared to 65.2% in men [7]. This gender difference in reporting accuracy among older adults merits particular attention in study design and data interpretation.

Methodological Considerations in Validation Research

Validation Study Designs and Protocols

Diagram 1: Validation methodologies for 24HR. This workflow illustrates the three primary approaches used to validate 24-hour dietary recalls.

Research on under-reporting employs various validation methodologies, each with distinct strengths and limitations. The doubly labeled water (DLW) technique represents the gold standard for validating energy intake assessment under conditions of weight stability [52] [50]. This method involves administering doses of stable isotopes (²H₂O and H₂¹⁸O) and tracking their elimination rates through serial urine samples over 1-2 weeks to calculate carbon dioxide production and total energy expenditure [52]. The fundamental principle is that in weight-stable individuals, energy intake should equal total energy expenditure, allowing identification of under-reporting when reported energy intake falls significantly below measured expenditure [50].

Controlled feeding studies provide an alternative validation approach by directly measuring true consumption. In these protocols, participants consume meals under supervision with unobtrusive weighing of foods and beverages consumed [49]. The following day, participants complete the dietary assessment method being validated, allowing direct comparison between reported and actual intake. This design eliminates the day-to-day variability in consumption and provides a precise measure of reporting accuracy for specific eating occasions [49].

The weighed food record method serves as a practical validation tool, particularly in field settings. Participants weigh and record all foods and beverages consumed on the same day as the 24HR is conducted [27]. While still subject to some measurement error, this approach provides a more objective reference than the recall alone and has been utilized successfully in various populations, including older adults [7].

Emerging Technologies and Methodological Innovations

Diagram 2: Evolution of dietary assessment methods. This diagram shows the transition from traditional approaches to increasingly sophisticated technology-assisted methods.

Recent methodological advances focus on technology-assisted approaches to mitigate systematic errors in dietary reporting. Web-based self-administered 24HR systems such as ASA24 (Automated Self-Administered Dietary Assessment Tool), Intake24, and R24W standardize the recall process and eliminate interviewer effects [49] [35]. These platforms typically employ an automated multiple-pass method with standardized probes to enhance memory and reduce forgetting [49]. Comparative studies indicate these web-based systems may reduce under-reporting compared to traditional interviewer-administered recalls, with one study finding the R24W produced 18% higher energy intake estimates in women and 15% higher estimates in men compared to traditional interviewer-administered recalls [35].

Image-assisted methodologies represent another innovative approach to improving accuracy. These include image-assisted recalls (where participants capture images of their meals to aid later recall) and image-based records (where images serve as the primary data source) [49]. The mobile Food Record (mFR) app, for instance, incorporates a fiducial marker of known size and shape to provide reference for portion size estimation when images are analyzed by trained raters [49]. Emerging evidence suggests that incorporating digital images may attenuate portion size misestimation compared to traditional food booklet aids [35].

The most technologically advanced approaches involve fully automated image-based assessment systems utilizing computer vision and artificial intelligence. The SNAQ app, for example, employs depth-sensing hardware and computer vision to automatically recognize foods and estimate volume and nutrient content from photographs [54]. Validation studies comparing this approach to doubly labeled water have shown closer agreement with energy expenditure than traditional 24HR, though significant challenges remain in real-world implementation [54].

Table 3: Research Reagent Solutions for Dietary Validation Studies

Tool/Technique	Primary Function	Key Features	Application Context
Doubly Labeled Water	Measure total energy expenditure	Gold standard biomarker; requires mass spectrometry	Criterion validation for energy intake reporting
Stable Isotopes (²H₂O, H₂¹⁸O)	DLW administration	Precisely measured doses based on body weight	Energy expenditure measurement over 1-2 weeks
Computerized 24HR Systems (ASA24, Intake24)	Self-administered dietary recall	Automated multiple-pass method; standardized probes	Large-scale studies; reduced interviewer burden
Image-Assisted Tools (mFR app)	Food capture and identification	Fiducial marker for portion size estimation	Enhanced portion size estimation in recalls
Food-Recognition Apps (SNAQ)	Automated food identification	Computer vision; depth-sensing; nutrient database	Real-time dietary assessment; reduced user burden
Body Composition Analyzers	Measure fat and fat-free mass	Bioelectrical impedance; DEXA alternatives	Energy requirement estimation; under-reporting detection

The Goldberg cutoff method provides a practical statistical approach for identifying under-reporters in large datasets where direct biomarker validation is not feasible [50]. This method calculates the ratio of reported energy intake to estimated basal metabolic rate (EI:BMR) and compares it to expected values based on population physical activity levels. Values falling below the 95% confidence limit of agreement between reported intake and energy expenditure signify under-reporting [50]. This approach has been widely applied in epidemiological studies to identify characteristic patterns associated with misreporting and to adjust analyses for this systematic bias.

Standardized dietary assessment platforms like GloboDiet (formerly EPIC-Soft) have been developed to enhance comparability across studies and populations. This computerized 24-h recall instrument has been adapted and validated in multiple countries and provides a standardized framework for collecting, coding, and processing dietary data [27]. Such standardization is particularly valuable in multi-center studies and for international comparisons, as it reduces methodological variability that can confound cross-population analyses.

The pervasive challenge of under-reporting in 24-hour dietary recalls remains a significant methodological concern in nutritional research. The evidence consistently demonstrates systematic under-reporting of energy intake across diverse populations and assessment methods, with magnitude varying by participant characteristics, assessment tool, and study context. Technology-assisted methods such as web-based platforms and image-assisted tools show promise for mitigating some sources of error, particularly through standardized administration and enhanced portion size estimation.

Future directions for advancing the field include further development and validation of automated dietary assessment technologies, improved statistical methods for identifying and adjusting for misreporting, and enhanced understanding of the cognitive and behavioral factors underlying reporting errors. Researchers should carefully consider the potential for under-reporting in their study designs, incorporate appropriate validation methodologies when feasible, and transparently address the limitations imposed by this systematic bias in their interpretations and conclusions. Only through rigorous attention to these methodological challenges can the field advance toward more accurate dietary assessment and more valid understanding of diet-health relationships.

The Impact of Social Desirability and Participant Characteristics on Reporting Accuracy

The 24-hour dietary recall (24HR) is a cornerstone method for assessing energy and nutrient intake in nutritional epidemiology and clinical research [27]. However, its reliance on self-report makes it susceptible to measurement errors that can compromise data validity and lead to erroneous conclusions in diet-disease association studies [55] [27]. These errors are not random; they are systematically influenced by participant characteristics, with social desirability bias emerging as a critical factor. This bias describes the tendency of individuals to alter their reported intake to align with perceived social norms, often leading to the underreporting of "bad" foods and overreporting of "good" ones [56]. Accurate data is paramount for researchers and drug development professionals relying on dietary intake to understand disease risk factors or intervention outcomes. This guide provides a comparative analysis of how social desirability and other participant characteristics impact the reporting accuracy of 24HRs, detailing key experimental protocols and offering tools to mitigate these biases.

Quantitative Comparison of Reporting Biases

The table below summarizes the empirical evidence on the direction and magnitude of reporting errors associated with various participant characteristics.

Table 1: Impact of Participant Characteristics on Self-Reported Dietary & Physical Activity Data

Participant Characteristic	Impact on Reporting	Magnitude of Effect	Key Evidence
Social Desirability	Underreporting of total energy and "unhealthy" foods; Overreporting of "healthy" foods like fruits and vegetables [57] [56].	In children, higher SDB associated with significantly fewer calories from snack foods [57]. In young adult women, a positive correlation was found between social desirability and diet quality scores [56].	Observational studies using validated scales (e.g., Marlowe-Crowne) compared with objective intake measures [57] [56].
Gender/Sex	Females are more likely to underreport energy intake and under-report 'socially undesirable' foods [56].	Women recalled 75.6% of foods consumed vs. 65.2% in men [7]. Social approval scores are significantly higher in females than males [56].	Validation studies comparing self-report with weighed food records or Doubly Labeled Water (DLW) [7] [56].
Body Mass Index (BMI)	Often associated with underreporting of energy intake, though not universally [55].	One study on postmenopausal women found no association between BMI and underreporting (P = 0.95) [55].	Epidemiological studies correlating BMI with the difference between reported energy intake and TEE from DLW [55].
Age Group	Younger age trended towards greater underreporting in one study; older adults may face memory-related errors [55] [58].	Postmenopausal women underreported energy by 20.8%, with error trending upward with younger age (P = 0.07) [55]. Older adults overestimated portion sizes by a mean ratio of 1.34 [7].	Cross-sectional studies across different age cohorts [55] [7].
Social Approval	Can lead to underestimation of physical activity levels and influence reporting of specific food groups [59] [56].	Weak association with underestimation of physical activity energy expenditure (-0.15 kcal/kg/day) on a 24-hour physical activity recall [59]. In females, negative correlation with reported vegetable intake; positive correlation with dairy [56].	Studies using the Martin-Larson Approval Motivation Scale alongside objective measures like DLW or accelerometers [59] [56].

Experimental Protocols for Validation

A critical step in assessing the validity of 24HRs is the use of rigorous experimental protocols that compare self-reported data with objective, high-quality reference measures. The following methodologies are central to this validation process.

Doubly Labeled Water (DLW) Protocol

The DLW method is considered the gold standard for objectively estimating total energy expenditure (TEE) in free-living individuals, providing a benchmark for validating self-reported energy intake [27] [59].

• Objective: To validate self-reported energy intake by comparing it against objectively measured total energy expenditure.
• Procedure:
  • Baseline Sample Collection: A fasting urine sample is obtained from participants to establish baseline levels of hydrogen and oxygen isotopes [59].
  • Isotope Administration: Participants orally ingest a carefully weighed dose of water containing stable, non-radioactive isotopes of deuterium (²H) and oxygen-18 (¹⁸O) [59].
  • Post-Dose Sample Collection: Further urine samples are collected over a specific period (e.g., 7 and 14 days post-dose) to track the elimination rates of the two isotopes from the body. The differential elimination rate of ²H versus ¹⁸O is proportional to carbon dioxide production, which is used to calculate TEE [59].
  • Energy Intake Calculation: Assuming energy balance (weight stability), TEE is equivalent to energy intake. Physical activity energy expenditure (PAEE) can be derived by subtracting resting metabolic rate (estimated using predictive equations based on fat-free mass) from TEE [59].
• Application: This protocol was used to reveal a 20.8% underreporting of energy intake on a Food Frequency Questionnaire (FFQ) in postmenopausal women and to link social desirability with overreporting of physical activity [55] [59].

Weighed Food Record Protocol

This method involves the direct weighing of all foods and beverages consumed, providing a highly accurate measure of actual intake against which 24HRs can be compared [7].

• Objective: To obtain a precise measure of actual food and nutrient intake for validating the accuracy of a subsequent 24-hour recall.
• Procedure:
  • Controlled Feeding Study: In a study with older Korean adults, participants consumed three self-served meals where all food items were discreetly weighed before and after consumption without the participants' knowledge, thereby minimizing behavioral changes [7].
  • 24-Hour Recall Interview: On the day following the feeding study, a trained interviewer conducted a detailed 24HR with the participant, either in person or via an online video call. The interview used a multiple-pass method to minimize forgotten foods [7].
  • Data Comparison: The researchers then calculated the proportion of food items correctly reported (matches), not reported (omissions), and incorrectly reported (intrusions). They also compared the reported and weighed portion sizes, energy, and nutrient intakes [7].
• Application: This protocol demonstrated that older adults accurately recalled 71.4% of consumed foods but systematically overestimated portion sizes by 34%, and revealed that women had significantly higher recall accuracy than men [7].

Visualizing the Validation Workflow

The following diagram illustrates the typical workflow and key decision points in a validation study that investigates the impact of participant characteristics on 24HR accuracy.

The Researcher's Toolkit

To conduct rigorous studies on dietary reporting accuracy, specific tools and reagents are essential. The table below lists key materials and their functions.

Table 2: Essential Reagents and Tools for Dietary Validation Research

Tool / Reagent	Function / Purpose	Application Example
Doubly Labeled Water (²H₂¹⁸O)	Gold standard for measuring total energy expenditure (TEE) in free-living individuals to validate self-reported energy intake [27] [59].	Served as the objective criterion to quantify a 20.8% underreporting of energy on an FFQ [55].
Stable Isotope Ratio Mass Spectrometer	Analyzes the ratio of deuterium and oxygen-18 isotopes in urine samples to calculate the rate of CO2 production and TEE [59].	Used in the DLW protocol to process post-dose urine samples and compute energy expenditure [59].
Social Desirability Scales (e.g., Marlowe-Crowne)	Quantifies a participant's tendency to respond in a culturally normative fashion. A 33-item scale is common [59] [56].	Regressed against the difference between reported energy intake and TEE to identify systematic underreporting [55] [59].
Social Approval Scales (e.g., Martin-Larson)	Measures the need to obtain positive responses in a testing situation. A 20-item scale is used [59] [56].	Used to identify participants whose reporting of physical activity or specific food groups may be biased by a need for approval [59] [56].
Multi-Pass 24-Hour Recall Protocol	A structured interview technique with multiple passes (quick list, detailed description, review) to enhance completeness and reduce memory lapse [27].	Served as the self-report method in validation studies against weighed records, helping to quantify food item omission rates [7].
ActiGraph Accelerometer	An objective, wearable device that measures body movement and intensity, used to validate self-reported physical activity [59].	Provided criterion measures of activity duration and intensity, against which self-reported physical activity recalls were compared [59].

The evidence clearly demonstrates that self-reported dietary data is systematically biased rather than randomly erroneous. Social desirability and participant characteristics such as gender, age, and BMI significantly influence reporting accuracy, threatening the validity of research findings [55] [7] [56]. To enhance the reliability of dietary assessment, researchers should integrate objective measures like DLW or weighed records into their study designs, even if only for a subset of participants [55] [27]. The data from these validation studies can then be used to calibrate self-reported intake and correct for person-specific bias, ultimately strengthening the conclusions drawn from nutritional epidemiology and clinical trials [55].

The accurate assessment of habitual energy intake is a cornerstone of nutritional epidemiology, clinical research, and public health monitoring. The 24-hour dietary recall (24HR) stands as one of the most widely used methods for capturing dietary data in large-scale studies due to its relatively low participant burden and potential for standardization. However, this method faces a fundamental challenge: day-to-day variability in individual food consumption means that a single day of data provides a poor estimate of a person's usual intake. This limitation necessitates the collection of multiple recalls, yet the optimal number balancing precision with practical constraints has been a long-standing question in nutritional science.

This guide synthesizes current evidence on optimizing the number of 24-hour dietary recalls, with a specific focus on the requirement for multiple, non-consecutive days to obtain reliable estimates of energy intake. We examine foundational and emerging research that employs robust validation methodologies, including comparison with doubly labeled water measurements and weighed food records, to provide evidence-based recommendations for researchers designing dietary assessment protocols.

Quantitative Evidence: How Many Recalls Are Needed?

The number of recall days required depends significantly on the nutrient of interest and the desired level of reliability. Recent large-scale studies provide specific guidance on minimum day requirements for energy and key nutrients.

Table 1: Minimum Days Required for Reliable Dietary Intake Assessment

Nutrient/Food Group	Minimum Days for Reliability (r > 0.8)	Key Findings
Total Energy	2-3 days	Achieves good reliability for estimating usual intake [60]
Macronutrients (Carbohydrates, Protein, Fat)	2-3 days	Most macronutrients reach good reliability within this timeframe [60]
Micronutrients	3-4 days	Generally require more days than macronutrients [60]
Water & Coffee	1-2 days	Can be reliably estimated with the fewest days [60]
Food Groups (Meat, Vegetables)	3-4 days	Specific food groups require more days for reliable estimation [60]

A 2025 analysis of over 315,000 meals from 958 participants in the "Food & You" digital cohort demonstrated that three to four days of dietary data collection are sufficient for reliable estimation of most nutrients [60]. The study further emphasized that including both weekdays and weekends significantly increases reliability, with specific day combinations outperforming others.

Foundational Evidence from Doubly Labeled Water Validation

Groundbreaking research utilizing doubly labeled water (DLW) as an objective biomarker has been instrumental in validating 24HR protocols for energy intake assessment.

Table 2: Key Validation Studies Using Doubly Labeled Water

Study Population	Methodology	Key Finding on Recall Numbers
79 Middle-Aged Women [2]	7 x 24HR vs. DLW over 14 days	Three 24HRs appeared optimal for estimating energy intake; first recall significantly underreported [2]
Adults with Overweight/Obesity (NY-TREAT) [61]	3-6 non-consecutive 24HR vs. DLW over 2 weeks	Underreporting was identified in 50% of dietary recalls, highlighting need for multiple days to identify misreporting [61]

The seminal study with middle-aged women demonstrated a critical learning effect: energy intake reported on the first recall was significantly lower (1501 kcal/day) than on subsequent calls (2246 and 2315 kcal/day for calls 2 and 3) [2]. Averaging the first two calls better approximated true energy expenditure than the first call alone, and averaging the first three calls further improved the estimate. This finding underscores why single 24-hour recalls are particularly vulnerable to misreporting and cannot provide valid estimates of usual intake.

Experimental Protocols and Methodologies

DLW-Validated Protocol for Energy Intake Assessment

The following workflow visualizes the key experimental design from foundational research that established the optimal number of recalls using doubly labeled water validation:

This protocol established that the first recall consistently underreports energy intake, and that averaging across multiple days corrects this systematic bias [2]. The finding that three recalls provided optimal estimation while additional calls did not significantly improve accuracy has profound implications for efficient study design.

Digital Cohort Methodology for Minimum Days Estimation

Recent technological advances have enabled more precise determination of minimum day requirements through large-scale digital tracking:

This digital methodology confirmed that including both weekdays and weekends significantly increases reliability, with specific day combinations outperforming others [60]. The analysis also revealed significant day-of-week effects, with higher energy, carbohydrate, and alcohol intake on weekends—particularly among younger participants and those with higher BMI.

Population-Specific Considerations in Recall Accuracy

The validity of 24-hour recalls and the optimal number of assessment days can vary substantially across different population groups.

Age-Specific Variations

Infants and Toddlers: A validation study comparing 24HR with 3-day weighed food records found that a single telephone-administered recall overestimated energy intake by 13% among infants and 29% among toddlers [62]. This overestimation was primarily attributed to portion size estimation errors, with dairy and grains accounting for most of the excess.

Older Adults: Research with free-living Korean adults aged 60+ demonstrated they recalled approximately 71.4% of foods consumed but tended to overestimate portion sizes (mean ratio: 1.34) [7]. Women showed significantly better recall accuracy than men (75.6% vs. 65.2% of foods), suggesting potential need for sex-specific protocols in older populations.

Cultural and Ethnic Adaptations

The expansion and validation of Foodbook24 for diverse populations in Ireland highlights the importance of culturally adapted dietary assessment tools. The inclusion of 546 additional foods commonly consumed by Brazilian and Polish populations, along with translation into Portuguese and Polish, resulted in 86.5% of participant-listed foods being available in the updated database [63]. However, Brazilian participants still omitted a higher percentage of foods in self-administered recalls (24%) compared to Irish participants (13%), indicating that cultural and linguistic adaptation, while beneficial, does not fully eliminate reporting disparities.

Similarly, the development of AWARD-J, a web-based 24HR system for Japanese adults, addressed the unique challenge of assessing intake in populations consuming predominantly mixed dishes by creating a standardized recipe database for Japanese cuisine [17].

The Scientist's Toolkit: Essential Research Reagents

Table 3: Key Reagents and Tools for 24HR Validation Research

Tool/Reagent	Function in Research	Application Example
Doubly Labeled Water (DLW)	Gold standard for measuring total energy expenditure in free-living individuals [2] [61]	Validation of self-reported energy intake against objective energy expenditure [2]
Multiple-Pass 24HR Interview	Structured interview technique to enhance completeness of food recall [2]	Reducing omission of foods and beverages in dietary reporting [2]
Web-Based 24HR Platforms	Automated self-administered dietary recall systems for large-scale data collection [63] [17]	Foodbook24, ASA24, INTAKE24, AWARD-J for efficient multi-day assessment [63] [17]
Nutrition Analysis Software	Conversion of food intake data to nutrient composition using standardized databases [2]	Nutrition Data System for Research (NDS) software for nutrient analysis [2]
Weighed Food Records	Detailed prospective method serving as reference for validation studies [17]	3-day weighed records as criterion method in Japanese validation study [17]
Social Desirability Scales	Assessment of psychological trait influencing dietary misreporting [2]	Marlowe-Crowne Social Desirability Scale to identify reporting bias [2]

Implications for Research and Practice

The consistent evidence across multiple studies indicates that three to four non-consecutive 24-hour recalls, strategically including both weekdays and weekend days, provide the optimal balance between scientific rigor and practical feasibility for estimating usual energy intake in adult populations. This approach mitigates the profound underreporting typically observed in first recalls and accounts for day-to-day variability in eating patterns.

For research targeting specific nutrients or food groups, the minimum number of required days may increase, with micronutrients and specific food groups generally needing more assessment days than total energy or macronutrients [60]. Additionally, special populations including children, older adults, and culturally diverse groups may require adapted protocols to address population-specific reporting challenges.

Future directions in this field should focus on refining day combination strategies, developing more sophisticated correction factors for single-day assessments, and leveraging technology to reduce participant burden while maintaining data quality in multi-day dietary assessment protocols.

Statistical Adjustments for Within-Person Variation to Estimate Usual Intake

Accurate estimation of usual nutrient intake is fundamental to nutritional epidemiology, public health policy, and clinical drug development. The 24-hour dietary recall (24HR) stands as a predominant method in large-scale surveillance due to its lower bias compared to food frequency questionnaires [64] [49]. However, a single 24HR captures only short-term intake, which varies day-to-day within an individual. This within-person variation obscures the true, habitual "usual intake" distribution, potentially leading to misclassification of nutrient status and biased diet-disease associations [65]. Consequently, statistical adjustments are imperative to distinguish this non-systematic day-to-day variation from the between-person variation that reflects true, long-term differences in dietary habits.

Within the broader thesis of 24HR validation research, these adjustment models are not merely statistical corrections but are essential for aligning self-reported data with biological reality. Studies using recovery biomarkers like doubly labeled water (DLW) for energy have consistently revealed significant underestimation in unadjusted 24HR data [64] [31]. This guide objectively compares the prominent statistical methodologies developed to address this challenge, detailing their protocols, performance, and appropriate applications for researchers and drug development professionals.

Comparative Analysis of Major Adjustment Methods

The following table summarizes the core statistical approaches for estimating usual intake, highlighting their key methodologies, outputs, and validation contexts.

Table 1: Comparison of Major Statistical Methods for Usual Intake Estimation

Method Name	Core Statistical Approach	Primary Output	Data Requirements	Key Advantages / Disadvantages
NRC / Iowa State University (ISU) Method	- Separates within- & between-person variance.- Shrinks individual's mean intake toward the group mean.	Usual intake distribution for a population or sub-population.	At least two non-consecutive 24HRs from a representative sample.	Advantage: Established, robust method for population distributions.Disadvantage: Does not correct for systematic bias (e.g., under-reporting) [64].
National Cancer Institute (NCI) Method	- Sophisticated mixed-effects models.- Correlates intake of different dietary components.- Can incorporate covariates (e.g., age, sex).	Usual intake distributions, prevalence of inadequacy/excess, and relationships between foods and nutrients.	Two or more 24HRs for at least a portion of the sample; can use one 24HR with an external variance ratio.	Advantage: Highly flexible and powerful; models complex dietary data.Disadvantage: Complex to implement; requires advanced statistical expertise [65].
Statistical Program to Assess Usual Dietary Exposure (SPADE)	- Resamples intake data to model a usual intake distribution.- Available as an R package.	Usual intake distribution and prevalence of inadequacy based on the Estimated Average Requirement (EAR).	Multiple 24HRs per person or a single 24HR with a variance ratio.	Advantage: User-friendly interface for a specific analytical task.Disadvantage: May be less flexible than the NCI method for complex modeling [66].
nutriR Package	- Fits a "best-fit" distribution to the observed intake data (e.g., normal, log-normal, gamma).	A modeled usual intake distribution intended to improve prevalence estimates.	Can be applied to either single 24HR data or usual intake data from other methods (e.g., SPADE).	Advantage: Simplifies analysis by automating distribution fitting.Disadvantage: Recent study found it did not significantly alter prevalence estimates compared to standard methods applied to usual intake data [66].

Quantitative Performance Data from Validation Studies

The true test of any adjustment method is its performance against objective validation criteria. The following table synthesizes key quantitative findings from controlled feeding studies and biomarker-based research, providing a critical lens for comparison.

Table 2: Performance Data of 24HR Methods and Adjustments from Validation Studies

Study Context / Method	Key Performance Metric (vs. True Intake or Biomarker)	Findings on Usual Intake Distribution	Implication for Adjustment
OPEN Study (Biomarker Reference)Traditional 24HR Analysis [64]	- Energy: 10-15% underestimation of mean.- Protein: 6-7% underestimation of mean.	Overestimation of the standard deviation (SD).	The NRC/ISU methods improved distribution shape but did not correct the mean bias, highlighting that adjustments cannot fix systematic reporting error.
Controlled Feeding StudyFour Technology-Assisted 24HRs [15] [49]	- Mean Energy Error: ASA24 (5.4%), Intake24 (1.7%), mFR-TA (1.3%), IA-24HR (15.0%).	Variance: Only Intake24 produced a variance in energy and protein intake that was not significantly different from the true variance.	Methods with accurate mean and variance (e.g., Intake24) provide a superior foundation for subsequent adjustment modeling.
Variance Ratio (WIV:Total) AnalysisCompilation of 40 Publications [65]	- Observed WIV:Total ratios ranged widely from 0.02 to 1.00.	Using an incorrect external WIV:Total ratio in models can lead to inaccurate prevalence estimates of inadequacy.	Emphasizes the need for population- and setting-specific variance ratios. Collecting ≥2 days of data from a subsample is strongly recommended.
nutriR Package ValidationIvorian School-Age Children [66]	- Compared nutriR to EAR cut-point and Probability of Adequacy (PA) methods.	Using "best-fit" distribution shapes with nutriR did not significantly affect prevalence estimates of inadequacy.	For usual intake data, the choice of distribution shape may be less critical than the initial correction for within-person variation.

Detailed Experimental Protocols from Cited Studies

To ensure methodological reproducibility, this section outlines the core experimental designs that generated the comparative data.

3.1.1 The OPEN Study Protocol [64]

• Objective: To quantify measurement error in dietary self-reports using recovery biomarkers.
• Design: Observational study in 484 healthy volunteers.
• Reference Biomarkers: Doubly labeled water (DLW) for total energy expenditure (proxy for energy intake) and 24-hour urinary nitrogen (UN) for protein intake.
• Dietary Instruments: Multiple FFQs and 24HRs administered to the same participants.
• Validation Analysis: Compared the self-reported intake distributions from the 24HRs (analyzed with traditional, NRC, and ISU methods) to the biomarker-estimated intake distributions.

3.1.2 Controlled Feeding Study for Technology-Assisted 24HRs [15] [49]

• Objective: To compare the accuracy of four technology-assisted 24HR methods against true intake under controlled conditions.
• Design: Randomized crossover feeding study.
• Participants: 152 adults (55% women, mean age 32, mean BMI 26).
• Feeding Protocol: Participants attended three separate feeding days, consuming breakfast, lunch, and dinner with all items unobtrusively weighed.
• 24HR Methods Tested: On the day after each feeding day, participants completed one of four 24HRs in a randomized order:
  • ASA24-Australia: A self-administered, web-based tool based on the Automated Multiple-Pass Method.
  • Intake24-Australia: A self-administered, web-based 24HR system.
  • mFR-TA (mobile Food Record-Trained Analyst): An image-based method where a dietitian analyzed images of meals taken by participants.
  • IA-24HR (Image-Assisted Interviewer-Administered 24HR): An interviewer-led recall where participants used their meal images as a memory aid.
• Primary Outcome: The difference between the estimated energy/nutrient intake from the 24HR and the true intake from weighed food.

Conceptual Workflow for Usual Intake Estimation

The following diagram illustrates the logical sequence and decision points involved in estimating a population's usual nutrient intake distribution, integrating the methods and challenges discussed.

This table details key methodological "reagents" — software, biomarkers, and databases — essential for conducting research in this field.

Table 3: Key Research Reagent Solutions for Usual Intake Estimation

Tool / Reagent	Type	Primary Function in Research	Key Considerations
ASA24 (Automated Self-Administered 24HR)	Software / Dietary Instrument	Self-administered web-based 24HR system that automates the Multiple-Pass Method. Allows for large-scale, cost-effective data collection [15] [49].	Requires adaptation to local food databases. Validation studies show variable performance in estimating energy variance [15].
Intake24	Software / Dietary Instrument	Open-source, self-administered web-based 24HR system. Designed to be easily adaptable for different countries and languages [49].	In a controlled study, it showed low mean bias and accurate estimation of intake distribution for energy and protein [15].
Doubly Labeled Water (DLW)	Biomarker	The gold-standard method for measuring total energy expenditure in free-living individuals, serving as a reference for validating reported energy intake [64] [31].	Very high cost and technical complexity limit its use to validation subsamples rather than entire studies.
24-hour Urinary Nitrogen (UN)	Biomarker	A recovery biomarker for protein intake, used to validate self-reported protein consumption and correct for reporting bias [64].	Like DLW, it is costly and burdensome, but incorporating it into surveys can significantly improve accuracy of protein intake estimation [64].
NCI Method Macros/SAS Programs	Statistical Software	A set of freely available, well-documented statistical macros (primarily in SAS) for implementing the complex NCI method for usual intake estimation [65].	Considered a gold-standard but has a steep learning curve. Requires advanced statistical knowledge for proper implementation and interpretation.
SPADE (R Package)	Statistical Software	An R package that provides a user-friendly interface for modeling usual intake distributions and calculating the prevalence of inadequacy [66].	Simplifies the analysis process for a specific set of tasks compared to the more generalizable NCI method.
nutriR (R Package)	Statistical Software	An R package designed to fit the "best-fit" statistical distribution to observed nutrient intake data to improve prevalence estimates [66].	Recent evidence suggests that for usual intake data, its impact on prevalence estimates may be minimal compared to standard methods [66].

Evidence-Based Performance: Analyzing Validation Studies Across Populations

How Many Recalls Are Enough? A Synthesis of Findings from Key Validation Studies

Accurate estimation of habitual energy intake is a cornerstone of nutritional epidemiology, intervention studies, and public health monitoring. The 24-hour dietary recall (24HR) has become a standard method for collecting dietary data in many large-scale studies due to its ability to capture detailed intake without the high participant burden of food records. However, a single day of intake is a poor proxy for an individual's usual consumption due to considerable day-to-day variability in eating patterns. Consequently, a critical methodological question persists: how many recalls are required to reliably estimate habitual energy intake?

This guide synthesizes evidence from key validation studies to provide evidence-based recommendations on the optimal number of 24-hour recalls. We compare the performance of various recall protocols against objective biomarkers and reference methods, presenting quantitative data to help researchers design efficient and accurate dietary assessment protocols.

Comparative Performance of Dietary Assessment Methods

Validation studies typically compare self-reported dietary intake against objective biomarkers such as doubly labeled water (DLW) for energy expenditure and urinary nitrogen for protein intake. The table below summarizes the relative validity of different dietary assessment methods based on such biomarker comparisons.

Table 1: Relative Validity of Dietary Assessment Methods Against Biomarkers

Assessment Method	Comparison Biomarker	Key Findings	Correlation with Biomarker	Underreporting Magnitude
Web-based 24HR (myfood24) [67]	Urinary biomarkers (Protein, K, Na)	Attenuation similar to interviewer-based recall	Partial r ≈ 0.3-0.4 for protein, potassium	Not specified
Interviewer-based 24HR (MPR) [67]	Urinary biomarkers (Protein, K, Na)	Considered a standard but costly	Partial r ≈ 0.3-0.4 for protein, potassium	Not specified
ASA24 (beta version) [39]	Recovery biomarkers (Protein, Na, K)	Lower validity than SFFQ/records for some nutrients	Deattenuated r = 0.46 (protein vs. biomarker)	Not specified
2×24h Recalls (AMPM) [68]	Doubly Labeled Water (TEE)	Minimal overall underreporting	Not specified	None on average (reported EI = TEE)
7-day Food Diary [68]	Doubly Labeled Water (TEE)	Significant underreporting	Not specified	-22% on average
Food Frequency Questionnaire (EPIC) [69]	Doubly Labeled Water (TEE)	Moderate agreement at group level	r = 0.48 (EI vs. TEE)	-22% on average

Determining the Optimal Number of Recall Days

Evidence from Biomarker Validation Studies

The Women's Lifestyle Validation Study provided high-quality evidence by comparing multiple dietary assessment methods against biomarkers over a 15-month period. This study found that the average of multiple ASA24 recalls (beta version) generally showed lower validity compared to food frequency questionnaires and 7-day dietary records for most nutrients when compared against recovery biomarkers [39]. This suggests that for the ASA24 system in particular, averaging three days of measurement may not be sufficient for capturing usual intake of some important nutrients [39].

A Danish validation study demonstrated superior performance of the 2×24h recall method using the Automated Multiple-Pass Method (AMPM) compared to a 7-day food diary. The 2×24h recall showed no significant underreporting on average (reported energy intake matched total energy expenditure from DLW), while the 7-day food diary underestimated energy intake by 22% [68]. The proportion of under-reporters was substantially lower with the 2×24h recall (4%) compared to the 7-day diary (34%) [68].

Evidence from Digital Cohort Studies

Recent large-scale digital studies provide more granular insights into day-to-day variability and minimum days required for reliable assessment. Analysis of the "Food & You" cohort (n=958), which collected detailed dietary data using the AI-assisted MyFoodRepo app, revealed differential requirements across nutrient types [60]:

Table 2: Minimum Days for Reliable Estimation by Nutrient Type

Nutrient/Food Category	Minimum Days for Reliability (r > 0.8)	Special Considerations
Water, Coffee, Total Food Quantity	1-2 days	Highest reliability with minimal data
Macronutrients (Carbohydrates, Protein, Fat)	2-3 days	Consistent across most studies
Micronutrients	3-4 days	Higher variability requires more days
Food Groups (Meat, Vegetables)	3-4 days	Depends on consumption frequency

This research identified significant day-of-week effects, with higher energy, carbohydrate, and alcohol intake on weekends—particularly among younger participants and those with higher BMI [60]. The study concluded that 3-4 days of dietary data collection, ideally non-consecutive and including at least one weekend day, are sufficient for reliable estimation of most nutrients [60].

Experimental Protocols in Key Validation Studies

The Women's Lifestyle Validation Study Protocol

This comprehensive study employed a rigorous design to evaluate multiple dietary assessment methods [39]:

• Duration: 15-month study period divided into 5 phases representing 3-month intervals
• Participants: 627 women from the Nurses' Health Study cohorts
• Dietary Methods: Two paper SFFQs, one Web-based SFFQ, four ASA24 recalls (beta version), two 7-day dietary records
• Biomarkers: Four 24-hour urine samples, one doubly labeled water measurement (repeated in a subset), two fasting blood samples
• Design: Measurements spread over 15 months to represent a typical year, with altered order of measurements into 4 randomized groups to avoid artificially high correlations

The Danish Validation Study Protocol

This study directly compared the 2×24h recall method with a 7-day food diary [68]:

• Design: Cross-over study with participants randomly assigned to start with either method
• Participants: 120 adults (52 men, 68 women) aged 18-60 years
• Reference Method: Total energy expenditure measured by doubly labeled water technique
• 24HR Protocol: Two detailed 24-hour recalls (one in-person, one by telephone) using the Automated Multiple-Pass Method
• Additional Measures: Pedometer-determined step counts, body composition measurements

Digital Cohort Methodology ("Food & You")

This study leveraged modern technology to collect extensive dietary data [60]:

• Participants: 958 adults who tracked meals for 2-4 weeks
• Tool: MyFoodRepo app with three logging methods: image capture (76.1% of entries), barcode scanning (13.3%), and manual entry (10.6%)
• Data Processing: Machine learning algorithm for automatic food segmentation and classification, with verification by trained annotators
• Analysis Methods: Coefficient of variation and intraclass correlation coefficient analyses across all possible day combinations to determine minimum days required

Visualizing the Validation Workflow

The following diagram illustrates the typical experimental design for validating 24-hour recall methods against objective biomarkers:

Diagram Title: Biomarker Validation Study Workflow

The Researcher's Toolkit: Essential Methodological Components

Table 3: Key Research Reagents and Tools for Dietary Validation Studies

Tool/Component	Function in Validation Research	Examples/Specifications
Doubly Labeled Water (DLW)	Objective measure of total energy expenditure for validating energy intake reports	Considered gold standard; requires specialized laboratory analysis
24-hour Urine Collection	Recovery biomarkers for protein (nitrogen), sodium, and potassium intake	Requires complete collection; 4 samples over year recommended [39]
Blood Biomarkers	Concentration biomarkers for specific nutrient intakes	Fatty acids, carotenoids, α-tocopherol, retinol, folate [39]
Automated Multiple-Pass Method (AMPM)	Standardized interview protocol for 24-hour recalls	Developed by USDA; used in ASA24 and interviewer-administered recalls [68]
Food Composition Databases	Convert reported food consumption to nutrient intakes	Must be comprehensive and updated; country-specific versions needed
Portion Size Estimation Aids	Help participants quantify food amounts consumed	Photographic atlases, household measures, digital images [17]
Digital Dietary Assessment Platforms	Self-administered 24-hour recall systems	ASA24 [36], myfood24 [67], INTAKE24, Web24HR [17]

The synthesis of validation evidence indicates that 3-4 non-consecutive days of 24-hour recalls, strategically including weekend days, provide a reasonable balance between reliability and participant burden for estimating habitual energy and nutrient intakes in most research contexts. The 2×24h recall method demonstrates particularly favorable performance against objective biomarkers with minimal underreporting [68]. However, researchers should consider that specific nutrients and food groups with higher day-to-day variability may require additional days of assessment [60]. Methodological choices should be guided by study objectives, population characteristics, and available resources, with careful consideration of the trade-offs between different dietary assessment methods.

The accurate measurement of energy and nutrient intake is a cornerstone of nutritional epidemiology, clinical research, and public health monitoring. The validity of dietary assessment methods directly impacts the quality of data used to establish diet-disease relationships, evaluate nutritional status, and inform dietary guidelines. Among the various tools available, the 24-hour dietary recall (24HR) and food diaries (including weighed and estimated food records) are widely used, yet each presents distinct advantages and limitations. This comparison guide objectively evaluates the relative validity of these methods and other alternatives, framing the analysis within the broader context of scientific validation research, particularly against the doubly labeled water (DLW) technique, the reference standard for estimating energy expenditure in free-living individuals.

Comparative Performance Data

The table below summarizes key validity findings from studies comparing self-reported energy intake (EI) from various dietary assessment methods to total energy expenditure (TEE) measured by doubly labeled water.

Table 1: Validity of Dietary Assessment Methods Compared to Doubly Labeled Water

Method	Study Population	Under-Reporting Rate (vs. TEE)	Correlation with TEE	Key Findings & Context
24-Hour Recall (Online - Intake24)	98 UK adults (40-65 years) [70]	25% (single recall) [70]	0.31 (single recall) [70]	Under-reporting comparable to interviewer-led recalls. Correlation improved to 0.47 with two recalls [70].
24-Hour Recall (Interviewer-led)	Adults (Systematic Review) [71]	Variable, but common [71]	Not specified	Shows less variation and degree of under-reporting compared to other methods like FFQs [71].
Food Diary (Weighed)	Young & Older Women [72]	~2.0 MJ/day (significant) [72]	Not significant [72]	Significantly lower than TEE in both age groups and did not correlate significantly with individual TEE values [72].
Food Frequency Questionnaire (FFQ)	Young & Older Women [72]	Closer to TEE in older women [72]	Significant in young women only [72]	Mean intakes were closest to TEE in older women; the only method correlating with TEE in young women [72].
Web-Based Tool (Nutrition Data)	42 Swedish adults with Type 1 Diabetes [73]	No significant difference [73]	0.79 for energy [73]	Good validity for energy and macronutrients (e.g., carbohydrate correlation: 0.94) compared to 24HRs [73].

For assessing habitual intake, the number of required days of assessment varies significantly. The table below provides replication recommendations for 24-hour recalls and food records.

Table 2: Number of Replications Needed to Estimate Usual Intake

Nutrient / Outcome	Required Replications (24HR or Food Record)	Population Context
Energy Intake	14-23 days [74]	To achieve 90% precision for estimation in adults and adolescents [74].
Energy Intake	4-7 days [74]	To classify individuals' intake with a correlation of 0.9 [74].
Macronutrients	Fewer days than for energy [75]	Major macronutrients may require ~3 days [75].
Micronutrients	More days than for energy [75]	Can require weeks of assessment due to high day-to-day variability [75].

Detailed Experimental Protocols

To ensure the reliability of the data presented, it is crucial to understand the experimental protocols used in the key validation studies cited.

The Doubly Labeled Water (DLW) Reference Method

The DLW technique is the gold standard for validating self-reported energy intake in free-living, weight-stable individuals, under the principle that in energy balance, energy intake equals total energy expenditure [71] [76].

• Protocol: Participants ingest a dose of water containing stable, non-radioactive isotopes of hydrogen (²H) and oxygen (¹⁸O). Subsequently, urine samples are collected over 7-14 days. The differential elimination rates of these isotopes from the body are used to calculate carbon dioxide production, which is then converted to total energy expenditure [70] [71].
• Advantage: This method is objective and independent of the memory, literacy, or portion-size estimation skills required by self-report methods [71].

The Multiple-Pass 24-Hour Recall Protocol

This structured interview approach is designed to enhance memory and reduce omission of foods [75]. Systems like Intake24 and ASA24 automate this protocol for online self-administration [70] [77].

• Quick List: The respondent freely recalls all foods and beverages consumed in the previous 24 hours.
• Forgotten Foods: The interviewer or system probes for commonly missed items (e.g., snacks, condiments, sugary drinks).
• Time and Occasion: The time and eating occasion for each food item are recorded.
• Detail Cycle: A detailed description of each food is collected, including portion size (aided by photographs or models), cooking method, and brand names.
• Final Probe: A last opportunity is provided for recalling any additional items [75].

Weighed Food Diary Protocol

This is a prospective method where participants record all food and drink consumed as it occurs.

• Protocol: Participants are provided with a digital scale and a logbook. They are instructed to weigh and record every item before consumption, including any leftovers. Details on cooking methods, recipes for mixed dishes, and brand names are also noted [72]. Recording periods typically range from 3 to 9 days [76].

The following diagram illustrates the typical workflow for validating a dietary assessment method against the DLW reference standard.

The Scientist's Toolkit: Key Research Reagents & Materials

The following table details essential components and tools used in dietary validation research.

Table 3: Essential Research Reagents and Tools for Dietary Validation Studies

Tool / Reagent	Function & Application	Examples & Specifications
Doubly Labeled Water (DLW)	Gold-standard measure of Total Energy Expenditure (TEE) for criterion validation [71].	Stable isotopes ²H₂O and H₂¹⁸O; dose calibrated to body weight; analysis via isotope ratio mass spectrometry [70] [71].
Standardized Food Composition Database	Converts reported food consumption into energy and nutrient intakes [70] [63].	UK CoFID, USDA FoodData Central, Swedish Food Database. Must be compatible with the dietary assessment tool and updated regularly [70] [73].
Portion Size Estimation Aids	Visual aids to improve the accuracy of reported food amounts [75].	Standardized food photograph atlases (e.g., with portion sizes from small to large), household measure guides, 2D grids, or physical food models [70] [63] [75].
Online 24HR Platforms	Automated, self-administered dietary recall systems that reduce cost and interviewer bias [70] [77].	Intake24, ASA24, MyFood24, Foodbook24. Feature integrated food lists, portion images, and automated nutrient analysis [70] [63] [77].
Structured Interview Protocols	Guide for interviewer-led recalls to standardize data collection and minimize omission [75].	The 5-step Multiple-Pass Recall Method [75].

Critical Analysis & Research Implications

The data demonstrates that no self-reported dietary assessment method is perfectly accurate at the individual level. Under-reporting of energy intake is a pervasive issue across all methods [71] [72]. However, the choice of method depends heavily on the research question, target population, and resources.

• 24HR demonstrates utility for estimating group-level mean intakes, especially when multiple non-consecutive recalls are collected to account for day-to-day variation [75] [74]. Its validity is bolstered by structured protocols like the multiple-pass method. A significant advantage is that it does not alter intake behavior, as it is retrospective [75].

• Food Diaries, while prospective and potentially more detailed, often show significant under-reporting and can be highly burdensome, leading to participant fatigue and reactivity (changing diet because it is being recorded) [72]. Their strength lies in detailed food description but may not be practical for large-scale studies.

• Emerging Technologies like online 24HR systems (e.g., Intake24, Nutrition Data) and AI-based tools show promise. They offer a favorable balance between cost, participant burden, and validity, producing data not largely different from traditional methods [70] [63] [78]. Their adaptability for different languages and cuisines makes them particularly valuable for diverse populations [63].

The relationship between different dietary assessment methods and their validation pathways can be conceptualized as follows.

The 24-hour dietary recall (24HR) is a foundational tool in nutritional epidemiology, used to assess individual food and nutrient intake. Its validation is crucial for ensuring the accuracy of data informing public health policy, clinical research, and our understanding of diet-disease relationships. While the method is used globally, its application in low- and middle-income countries (LMICs) presents unique challenges and opportunities. This guide objectively compares the performance of the 24-hour recall for estimating energy intake in LMIC settings against alternative methods and gold-standard measures, providing a synthesis of current experimental data to inform researchers and health professionals.

Comparative Performance Data

The table below summarizes key performance metrics of the 24-hour dietary recall method from validation studies conducted in various LMIC settings, using reference standards such as the Doubly Labeled Water (DLW) method and Observed Weighed Records (OWR).

Table 1: Performance of 24-Hour Dietary Recall for Energy Intake Estimation in LMIC Settings

Study Population & Location	Reference Method	Mean Energy Intake (24HR)	Mean Energy Expenditure/Intake (Reference)	Mean Difference (Bias)	Under-reporting Rate	Key Findings
Korean Adults (20-49 years) [52]	Doubly Labeled Water (DLW)	2,084.3 ± 684.2 kcal/day	2,401.7 ± 480.3 kcal/day	-317.4 kcal/day (Under)	60.5% (All)	Significant under-reporting (p<0.001); under-reporting was 12.0% on average [52].
Adolescents, Burkina Faso (10-14 years) [18]	Observed Weighed Record (OWR)	88-92% of OWR	(Reference = 100%)	-8% to -12% (Under)	Not Specified	Energy intakes were equivalent within a 15% bound, indicating acceptable underestimation [18].
South Asian Adults (via Intake24) [79]	Not directly validated	Median reported intake varied by age/sex	Not Available	Not Available	Not Available	Demonstrated feasibility with median recall completion time of 13 minutes and good food list coverage [79].
Sport Science Students, Ethiopia [80]	Estimated Requirements	Suboptimal intake prevalent	Not Available	Not Available	Not Available	High prevalence of inadequate energy intake, highlighting context-specific dietary challenges [80].

Detailed Experimental Protocols

Understanding the methodology of key validation studies is critical for interpreting their findings. The following are detailed protocols from seminal research.

This study provides a high-quality benchmark for validating self-reported energy intake against total energy expenditure.

• Objective: To compare the accuracy of total energy intakes (TEI) estimated by the 24-hour diet recall method with total energy expenditure (TEE) obtained using the doubly labeled water (DLW) method.
• Subjects: 71 Korean adults (35 men, 36 women) aged 20-49 years with a BMI between 18.5 and 25.0 kg/m².
• 24HR Protocol: Trained investigators conducted three multiple-pass 24-hour recalls per subject (two weekdays and one weekend day) over a 14-day period. To reduce recall bias, subjects were provided with and used digital cameras to photograph all foods consumed.
• DLW Protocol:
  • Dosing: Subjects ingested a dose of DLW (1.1g per kg body weight) prepared from H₂¹⁸O and ²H₂O.
  • Urine Collection: Urine samples were collected at five time points: baseline and on days 1, 2, 13, and 14 after ingestion.
  • Analysis: Isotopic enrichment of ²H and ¹⁸O in urine was analyzed using an isotopic mass analyzer. Carbon dioxide production rate was calculated from the elimination rates of the two isotopes.
• Data Analysis: The paired t-test was used to determine the significance of differences between TEI and TEE. Accuracy was further determined by calculating the accuracy predictions percentage, root mean square error, and bias.

This study represents a robust validation in a classic LMIC context.

• Objective: To validate the 24-hour recall method against observed weighed records in adolescents in Burkina Faso.
• Subjects: 132 younger adolescents (10-11 years) and 105 older adolescents (12-14 years).
• Study Design: Dietary data were collected for the same day by both methods. The observed weighed record (OWR) was conducted first, where all foods consumed by adolescents from the first to the last meal of the day were weighed. The following day, a 24-hour recall was conducted where adolescents reported foods consumed using portion aids like photographs and household utensils.
• Data Analysis: Nutrient intakes from both methods were converted and tested for equivalence by comparing the ratios (24HR/OWR) with pre-defined equivalence margins of within ±10%, 15%, and 20%.

Visualization of 24HR Validation Workflow

The following diagram illustrates the standard workflow for validating a 24-hour dietary recall against a reference method, synthesizing the protocols from the cited studies.

Comparison with Alternative Dietary Assessment Methods

The performance of the 24-hour recall must also be understood in relation to other common dietary assessment tools, particularly in the context of large-scale surveys in LMICs.

Table 2: Comparison of Dietary Assessment Methods in LMIC Contexts

Method	Key Features	Advantages in LMICs	Limitations in LMICs	Comparative Performance Data
24-Hour Dietary Recall (24HR)	Relies on memory to recall past 24h intake; often interviewer-administered.	Lower respondent burden than records; can capture detail; suitable for low-literacy populations with interviewer.	Prone to memory lapses and under-reporting; requires trained interviewers; translation/adaptation of tools needed [52] [18].	Underestimates energy by 8-12% vs. OWR [18] and ~13% vs. DLW [52].
Household Consumption & Expenditure Surveys (HCES)	Assesses household-level food acquisition over 4-7 days.	Low cost; nationally representative; often existing data.	Does not measure individual intake or away-from-home consumption; assumptions on food distribution introduce error [81].	HCES showed substantially lower energy intakes vs. 24HR, with 42% difference in large households [81].
Food Frequency Questionnaire (FFQ)	Assesses long-term habitual intake via frequency of food groups.	Captures usual intake; low cost for large samples once developed.	Difficult to quantify portions; limited by food list; requires validation for each population [52].	Not directly compared in results, but noted as less quantitative than 24HR [52].
Technology-Assisted 24HR (e.g., Intake24)	Digital, often self-administered 24HR with automated coding.	Reduces interviewer burden; faster; potentially more accurate portion sizing.	Requires digital literacy and access; needs extensive local food database [15] [79].	Intake24 showed low mean bias (1.7%) for energy vs. true intake in a controlled study [15].

The diagram below outlines a decision framework for selecting an appropriate dietary assessment method in LMICs based on research objectives and constraints.

The Scientist's Toolkit: Key Research Reagents & Materials

The following table details essential materials and tools required for conducting and validating 24-hour dietary recall studies, particularly in LMIC settings.

Table 3: Essential Research Reagents and Solutions for 24HR Validation

Item	Function/Description	Example Use in Context
Doubly Labeled Water (DLW)	A gold-standard method for measuring total energy expenditure in free-living individuals. It involves orally administering a dose of water enriched with the stable isotopes ²H (Deuterium) and ¹⁸O (Oxygen-18) and tracking their elimination rates in urine over 1-2 weeks [52].	Used as a non-invasive, objective reference method to validate the accuracy of energy intake reported in 24-hour recalls [52].
Stable Isotope Mass Spectrometer	An analytical instrument used to measure the precise ratio of stable isotopes (e.g., ²H/¹H and ¹⁸O/¹⁶O) in biological samples like urine. This is essential for DLW analysis [52].	Used to analyze urine samples collected from subjects in a DLW protocol to calculate the rate of carbon dioxide production and, consequently, total energy expenditure [52].
Local Food Composition Table (FCDB)	A database detailing the nutritional content of foods commonly consumed in a specific country or region. It is the foundation for converting reported food consumption into nutrient intakes [80] [79].	Integrated into software (e.g., CAN-Pro, NutriSurvey) to calculate energy and nutrient intake from the foods and portions reported during a 24-hour recall interview [52] [80].
Portion Size Estimation Aids	Visual tools to help respondents estimate and report the volume or weight of consumed foods more accurately. These can include photographic atlases, food models, standard household utensils, or even real food [18] [81].	Presented to respondents during the 24-hour recall interview to improve the accuracy of portion size reporting, thereby reducing a major source of measurement error [18].
Digital Dietary Assessment Platform	Software or web-based applications designed to administer 24-hour recalls, often with automated food coding and nutrient analysis. Examples include Intake24, ASA24, and OpenDRS [15] [79] [81].	Used to streamline data collection, reduce interviewer burden, improve data quality, and standardize the interview process across a large or multi-site study [79].

The 24-hour dietary recall (24HR) serves as a cornerstone method for collecting dietary intake data in nutritional surveillance, epidemiological research, and clinical trials [82]. The validation of these tools—assessing their accuracy against a measure of true intake—is paramount for generating reliable data. However, a one-size-fits-all approach to validation is inadequate. A method demonstrating excellent validity in one group may perform poorly in another due to differences in cognitive function, cultural familiarity with foods, or health status. This guide examines the critical considerations for validating 24HRs across special populations, providing a structured comparison of methodological approaches and their performance data to inform researchers and professionals in drug development and public health.

Key Considerations for Special Populations

The accuracy of 24HR is influenced by a complex interplay of population-specific characteristics. The following table summarizes the primary considerations for different groups.

Table 1: Key Considerations for Validating 24HRs in Special Populations

Population	Primary Challenges	Key Validation Findings	Recommended Methodological Adjustments
Children & Adolescents	Developing cognitive skills (memory, attention), susceptibility to omitting foods (especially snacks) [18].	In adolescents (10-14 years), 24HR underestimated energy intake by 8-12% vs. observed intake; omission of snacks, fruits, and beverages was common [18]. A web-based tool (R24W) in Canadian adolescents overestimated energy by 8.8% vs. interviewer-administered recall [16].	Use of age-appropriate probes and portion-size aids; integration of caregiver reporting for younger children; multiple recalls to improve precision [18] [16].
Elderly	Potential age-related cognitive decline affecting memory retrieval and executive function.	Neurocognitive processes, specifically visual attention and executive function (measured by Trail Making Test), are significantly associated with greater error in energy estimation in self-administered 24HRs [82].	Interviewer-administered recalls may mitigate cognitive load [82]; further research is needed on cognitive screening within validation studies.
Low- and Middle-Income Countries (LMIC)	Diverse and often non-standardized food compositions, low literacy levels.	In adolescents in Burkina Faso, 24HR underestimated energy intake, but the degree was deemed acceptable for 12-14-year-olds within a 15% equivalence bound [18].	Development of localized food composition databases; use of image-assisted methods and portion aids relevant to the local context.
Populations with Unique Diets	Assessment of complex mixed dishes, unfamiliar food items.	A Japanese Web24HR, developed with a recipe database for mixed dishes, showed moderate correlations (median r=0.51 men, 0.38 women) with weighed food records for most nutrients [17].	Creation of specialized food databases (e.g., for mixed dishes); validation must be repeated when a tool is adapted for a new cuisine [17].

Experimental Protocols for Validation

To generate the data required for comparisons, robust and controlled experimental protocols are essential. The following are detailed methodologies commonly employed in 24HR validation studies.

The Controlled Feeding Study

Purpose: To measure the absolute validity of a 24HR method by comparing reported intake to a known, true intake under controlled conditions [15].

Workflow:

• Participant Recruitment: Recruit a sample representative of the target population, applying specific inclusion/exclusion criteria (e.g., no serious illnesses, special diets) [82] [15].
• Controlled Feeding: Provide participants with all meals and beverages for one or more days. All items are unobtrusively weighed before and after consumption to establish "true" intake [15].
• Dietary Recall: On the day following the feeding day, participants complete the 24HR method being validated (e.g., ASA24, Intake24, interviewer-administered). In crossover designs, participants test multiple methods on separate occasions [82] [15].
• Data Analysis: Calculate the difference between reported energy/nutrient intake and true intake. Statistical analyses (e.g., paired t-tests, linear mixed models) are used to assess the significance and magnitude of the error [15].

The Comparison of Methods Experiment

Purpose: To assess the relative validity or systematic error of a new (test) 24HR method against a comparative method, which could be another 24HR, a food record, or a biomarker [83].

Workflow:

• Specimen/Sample Selection: A minimum of 40 participant specimens (or dietary records) should be selected to cover the entire working range of nutrient intakes [83].
• Paired Measurements: Each participant completes both the test method (e.g., a web-based 24HR) and the comparative method (e.g., an interviewer-administered 24HR or a weighed food record) [83] [16] [17].
• Data Graphing and Analysis: Data are graphed using difference plots (Bland-Altman) or comparison plots for visual inspection. Statistical calculations (e.g., linear regression, paired t-tests, correlation coefficients) are performed to estimate systematic error (bias) [83] [16] [17].

Comparative Performance Data

The following tables synthesize quantitative data from recent validation studies, highlighting how the performance of different 24HR methods varies.

Table 2: Accuracy of Technology-Assisted 24HR Methods in a Controlled Feeding Study (Adults) [15]

Dietary Assessment Method	Mean Difference in Energy Intake (vs. True Intake)	Statistical Significance (p-value)
Image-Assisted Interviewer-Administered 24HR (IA-24HR)	+15.0% (Overestimation)	p < 0.001
Automated Self-Administered 24HR (ASA24)	+5.4% (Overestimation)	p < 0.05
Intake24	+1.7% (Overestimation)	Not Significant
mobile Food Record-Trained Analyst (mFR-TA)	+1.3% (Overestimation)	Not Significant

Table 3: Relative Validity of a Web-Based 24HR in Adolescents and Japanese Adults [16] [17]

Population & Comparison	Key Metric	Result
Canadian Adolescents (n=272) [16]	Mean Energy Intake (Web vs. Interview)	Web-based R24W was 8.8% higher
	Correlation for Nutrients	Significant for most nutrients (range: r=0.24 to 0.52)
	Misclassification Rate	5.7% were misclassified (extreme quartiles)
Japanese Adults (n=228) [17]	Correlation with Weighed Food Record	Median r = 0.51 (men), 0.38 (women)
	Bias for Most Nutrients	Within ±10% of reference method

The Scientist's Toolkit: Research Reagent Solutions

This section details essential materials and tools used in the validation of dietary assessment methods.

Table 4: Essential Reagents and Tools for 24HR Validation Studies

Item	Function in Validation	Examples / Specifications
Doubly Labeled Water (DLW)	Gold standard for measuring total energy expenditure, which equals energy intake in weight-stable individuals [2].	Provides an objective criterion to validate the accuracy of energy intake reporting [2].
Cognitive Assessment Tasks	Quantifies neurocognitive abilities (e.g., memory, attention) that may contribute to measurement error in 24HR [82].	Trail Making Test (visual attention), Wisconsin Card Sorting Test (cognitive flexibility), Visual Digit Span (working memory) [82].
Standardized Food Composition Database	Converts reported food consumption into estimated nutrient intakes; critical for accuracy.	Nutrition Data System for Research (NDSR), country-specific databases (e.g., Japanese food composition tables) [2] [17].
Portion-Size Estimation Aids	Helps participants describe the quantity of food consumed, reducing one source of measurement error.	Two-dimensional food models, photographs, household measures, or digital portion-size guides [2] [16].
Statistical Analysis Software	Performs complex data analysis to compare methods, estimate bias, and model usual intake.	Software supporting linear mixed models, Bland-Altman analysis, and the NCI method for usual intake (e.g., SAS, R, Stata) [2] [84].
The NCI Method (Software/Macro)	A statistical method to estimate the distribution of "usual intake" for a population by accounting for day-to-day variation from 24HRs [84].	Corrects for within-person variation and can incorporate FFQ data; superior to simple within-person mean calculations [84].

Conclusion

The validation of 24-hour dietary recalls is not a one-size-fits-all endeavor but a rigorous process essential for generating reliable data. The synthesis of evidence confirms that a single 24HR is insufficient for estimating usual energy intake due to significant day-to-day variability and pervasive under-reporting, particularly on the first recall. The consensus from multiple validation studies indicates that three non-consecutive 24-hour recalls, utilizing standardized methods like the AMPM and leveraging modern, web-based tools, provide a robust balance between accuracy and participant burden. For the biomedical research community, this underscores the necessity of integrating validation protocols into study design, whether using objective biomarkers like DLW or statistical adjustments for within-person variation. Future efforts should focus on enhancing the accessibility and adaptability of validated digital tools, refining methods for diverse global populations, and further integrating biomarker-based validation to strengthen the evidence base linking diet to health and disease outcomes.

References

• 1. Comparing the web-based and traditional self-reported 24- ... [https://www.sciencedirect.com/science/article/abs/pii/S0169260723003474]
• 2. Number of 24-Hour Diet Recalls Needed to Estimate ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC2803049/]
• 3. Validation of energy intake measurements determined from ... [https://pubmed.ncbi.nlm.nih.gov/11815316/]
• 4. In-person vs telephone-administered multiple-pass 24-hour recalls in women: validation with doubly labeled water - PubMed [https://pubmed.ncbi.nlm.nih.gov/10916515/]
• 5. Validation of energy intake by 24-hour multiple pass recall: comparison with total energy expenditure in children aged 5-7 years - PubMed [https://pubmed.ncbi.nlm.nih.gov/15975166/]
• 6. Comparing Self-Administered Web-Based to Interviewer- ... [https://www.mdpi.com/2674-0311/4/2/17]
• 7. Accuracy of 24-hour dietary recalls among free-living older ... [https://www.sciencedirect.com/science/article/abs/pii/S0022316625006868]
• 8. Validity of Interviewer-Administered 24-h Dietary Recalls in ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC10096839/]
• 9. Comparison of a web-based versus traditional diet recall ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC3379547/]
• 10. Doubly Labeled Water Is a Validated and Verified Reference ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC3985818/]
• 11. Recent Advances from Application of Doubly Labeled ... [https://www.sciencedirect.com/science/article/pii/S0022316623021624]
• 12. Doubly labelled water assessment of energy expenditure [https://pmc.ncbi.nlm.nih.gov/articles/PMC5486561/]
• 13. The Doubly Labeled Water (DLW) Method [https://sites.baylor.edu/samuel_urlacher/the-doubly-labeled-water-method/]
• 14. The Doubly Labeled Water Method Produces Highly ... [https://www.sciencedirect.com/science/article/pii/S0022316622008914]
• 15. Accuracy of energy and nutrient intake estimation versus ... [https://www.sciencedirect.com/science/article/pii/S0002916524004568]
• 16. Assessing the relative validity of a web-based self ... [https://pubmed.ncbi.nlm.nih.gov/38907225/]
• 17. Validity of a Web-Based 24-Hour Dietary Recall of Energy ... [https://www.mdpi.com/2072-6643/16/23/4140]
• 18. Validation of 24-h dietary recall for estimating nutrient ... [https://pubmed.ncbi.nlm.nih.gov/32337835/]
• 19. Validity of predictive equations for total energy expenditure ... [https://www.nature.com/articles/s41598-024-66767-7]
• 20. Validation of 24‐h dietary recall for estimating nutrient intakes ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC7503205/]
• 21. Comparison of total energy expenditure and energy intake ... [https://www.sciencedirect.com/science/article/pii/S0002916523063839]
• 22. Comparison of total energy expenditure and energy intake ... [https://pubmed.ncbi.nlm.nih.gov/11684533/]
• 23. Total Energy Expenditure, Energy Intake, and Body ... [https://sportsmedicine-open.springeropen.com/articles/10.1186/s40798-017-0076-1]
• 24. Random vs. Systematic Error | Definition & Examples [https://www.scribbr.com/methodology/random-vs-systematic-error/]
• 25. Random Error vs Systematic Error [https://statisticsbyjim.com/basics/random-error-vs-systematic-error/]
• 26. Measurement Error in Dietary assessment [https://nutritionalassessment.org/errors/]
• 27. Measurement Errors in Dietary Assessment Using Self- ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC5683000/]
• 28. Evaluating the Measurement Error Structure of Self-Report ... [https://dietassessmentprimer.cancer.gov/concepts/error/error-affecting.html]
• 29. AMPM - USDA Automated Multiple-Pass Method [https://www.ars.usda.gov/northeast-area/beltsville-md-bhnrc/beltsville-human-nutrition-research-center/food-surveys-research-group/docs/ampm-usda-automated-multiple-pass-method/]
• 30. Field Interviewer–Administered Dietary Recalls in Participants ... [https://www.ncbi.nlm.nih.gov/books/NBK577065/]
• 31. Estimates of energy intake, requirements and imbalances ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC12458847/]
• 32. Comparing self-reported energy intake using an online ... [https://www.sciencedirect.com/science/article/abs/pii/S0899900723002861]
• 33. of Comparison ® Versions | EGRP/DCCPS/NCI/NIH ASA 24 [https://epi.grants.cancer.gov/asa24/comparison.html]
• 34. Validation of a newly automated web-based 24-hour dietary ... [https://bmcnutr.biomedcentral.com/articles/10.1186/s40795-017-0153-3]
• 35. Frontiers | Differences in Population-Based Dietary Intake Estimates... [https://www.frontiersin.org/journals/nutrition/articles/10.3389/fnut.2020.00137/full]
• 36. ASA24® Dietary Assessment Tool | EGRP/DCCPS/NCI/NIH [https://epi.grants.cancer.gov/asa24/]
• 37. Performance of the Automated Self-Administered 24-hour ... [https://www.sciencedirect.com/science/article/pii/S0002916523046919]
• 38. and feasibility of recalls completed using the Automated... Performance [https://pmc.ncbi.nlm.nih.gov/articles/PMC7606702/]
• 39. Relative Validity of Nutrient Intakes Assessed by ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC5928456/]
• 40. Bias recognition and mitigation strategies in artificial intelligence healthcare applications - PMC [https://pmc.ncbi.nlm.nih.gov/articles/PMC11897215/]
• 41. Validation and evaluation of a tablet-based dietary record ... [https://pubmed.ncbi.nlm.nih.gov/41289268/]
• 42. 24-hour Dietary Recall (24HR) At a Glance [https://dietassessmentprimer.cancer.gov/profiles/recall/]
• 43. The Use of Digital Images in 24-Hour Recalls May Lead to ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC5118765/]
• 44. Validation of improved 24-hour dietary recall using a ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC8283912/]
• 45. Accuracy and Cost-effectiveness of Technology-Assisted ... [https://www.researchprotocols.org/2021/12/e32891/]
• 46. Estimation of Energy Intake by a Food Frequency ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC6683088/]
• 47. Validation of an inexpensive and accurate mathematical ... [https://www.sciencedirect.com/science/article/pii/S0002916523125056]
• 48. Assessing the relative validity of a web-based self ... [https://nutritionj.biomedcentral.com/articles/10.1186/s12937-024-00954-0]
• 49. Accuracy of energy and nutrient intake estimation versus ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC11347807/]
• 50. Markers of the Validity of Reported Energy Intake [https://www.sciencedirect.com/science/article/pii/S0022316622157056]
• 51. Comparison of self-reported dietary intakes from the ... [https://www.sciencedirect.com/science/article/pii/S0002916522027459]
• 52. Comparison of total energy intakes estimated by 24-hour ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC9523208/]
• 53. Under-reporting of Energy Intake from 24-hour Dietary ... [https://www.sciencedirect.com/science/article/pii/S2210909914000095]
• 54. Assessing daily energy intake in adult women: validity of a ... [https://www.frontiersin.org/journals/nutrition/articles/10.3389/fnut.2023.1255499/full]
• 55. Participant characteristics associated with errors in self ... [https://www.sciencedirect.com/science/article/pii/S0002916523059695]
• 56. Gender differences in social desirability and approval ... [https://www.sciencedirect.com/science/article/abs/pii/S019566632200126X]
• 57. Social Desirability Bias is Related to Children's Energy ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC10922645/]
• 58. Protocol for a Mixed Methods Study [https://www.researchprotocols.org/2025/1/e70194]
• 59. The Effect of Social Desirability and Social Approval on Self ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC2958515/]
• 60. Minimum days estimation for reliable dietary intake ... [https://www.nature.com/articles/s41430-025-01644-8]
• 61. Dietary misreporting: a comparative study of recalls vs ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC12034172/]
• 62. Overestimation of infant and toddler energy intake by 24-h ... [https://www.sciencedirect.com/science/article/pii/S0002916523241180]
• 63. Expansion and Assessment of a Web-Based 24-Hour Dietary ... [https://ojphi.jmir.org/2025/1/e52380/]
• 64. Adjustments to Improve the Estimation of Usual Dietary ... [https://www.sciencedirect.com/science/article/pii/S0022316622131536]
• 65. Within-Person Variation in Nutrient Intakes across ... [https://www.sciencedirect.com/science/article/pii/S2161831322001776]
• 66. Assuming the "Best-Fit" Shape of the Usual Nutrient Intake ... [https://pubmed.ncbi.nlm.nih.gov/40414299/]
• 67. Validity of an online 24-h recall tool (myfood24) for dietary ... [https://bmcmedicine.biomedcentral.com/articles/10.1186/s12916-018-1113-8]
• 68. Validation of the 2 × 24 h recall method and a 7-d web- ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC10511678/]
• 69. Validation of a self-administered food-frequency ... [https://www.sciencedirect.com/science/article/pii/S0002916522041259]
• 70. 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 | Cambridge Core [https://www.cambridge.org/core/journals/journal-of-nutritional-science/article/validity-and-reliability-of-an-online-selfreport-24h-dietary-recall-method-intake24-a-doubly-labelled-water-study-and-repeatedmeasures-analysis/3CB8BB3B415198218F7E6B9D5AB3DA84]
• 71. Validity of Dietary Assessment Methods When Compared ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC6928130/]
• 72. Evaluation of four methods for determining energy intake in ... [https://www.sciencedirect.com/science/article/pii/S0002916523192434]
• 73. Validation of a web-based dietary assessment program ... [https://www.frontiersin.org/journals/nutrition/articles/10.3389/fnut.2024.1395252/full]
• 74. How many 24-hour recalls or food records are required to estimate usual energy and nutrient intake? - PubMed [https://pubmed.ncbi.nlm.nih.gov/21180983/]
• 75. Measurement Toolkit - 24-hour dietary recalls [https://www.measurement-toolkit.org/diet/subjective-methods/24-hour-dietary-recall]
• 76. The validity of dietary assessment methods to accurately ... [https://www.nature.com/articles/s41430-017-0029-2]
• 77. a review and evaluation of online 24-h dietary recall tools [https://pmc.ncbi.nlm.nih.gov/articles/PMC11705012/]
• 78. Validity and accuracy of artificial intelligence-based dietary ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC12229984/]
• 79. The Adaptation, Implementation, and Performance ... [https://www.sciencedirect.com/science/article/pii/S2475299125000022]
• 80. Inadequate intake of energy and nutrients: A comparative ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC10804097/]
• 81. A Comparative Analysis of Food Consumption Data From ... [https://www.medrxiv.org/content/10.1101/2025.07.03.25330715v1.full-text]
• 82. The association of cognitive task scores with energy intake ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC12169953/]
• 83. The Comparison of Methods Experiment [https://www.westgard.com/lessons/basic-method-validation/lesson23.html]
• 84. Usual Dietary Intakes: The NCI Method | EGRP/DCCPS/NCI/NIH [https://epi.grants.cancer.gov/diet/usualintakes/method.html]