A Decade of Dietary Assessment Methodology Research at the National Institutes of Health, 2012–2021

Abstract

Background

Assessment of individual and population-level dietary intake is critical for public health surveillance, epidemiology, and dietary intervention research. In recognition of that need, the National Insitutes of Health (NIH) has a history of funding research projects designed to support the development, implementation, and refinement of tools to assess dietary intake in humans.

Objectives

This report provides data and information on NIH-funded dietary intake assessment methodological research over the period of 2012–2021.

Methods

Data were extracted from an internal NIH data system using the Research, Condition, and Disease Categorization (RCDC) spending category for Nutrition. Data were then examined to identify research focused on dietary assessment tools or methods to capture or analyze dietary intake.

Results

Over the decade of 2012–2021, NIH supported 46 grants and 2 large contracts specific to dietary assessment methods development. The top 6 Institutes and Offices funding dietary assessment methods research were identified. Most projects were limited to adults. Projects ranged from novel methods to capture dietary intake, and refinement of analytical methods, to biomarkers of dietary intake. One key contract supported the automated self-administered 24-h dietary assessment tool (ASA24), a widely used, free tool available to the research community for assessing dietary intake.

Conclusions

NIH’s support for dietary assessment methods development over this 10-y period was small but grew over time with an expanding number and variety of methods, data sources, and technological advancements in the assessment of dietary intake. NIH remains committed to supporting research seeking to advance the field of dietary assessment methods research.

Introduction

Accurate assessment of dietary intake remains critical for public health surveillance, epidemiology, nutrition, and dietary intervention research; however, measuring diet without overburdening individuals has been a persistent challenge for scientists [1,2,3]. Diet is also not static for an individual and varies across the day, week, month, and lifetime [4]. Moreover, factors such as health status; disease status; developmental stage(s); food access and affordability; a myriad of sociocultural, behavioral, and individual-level characteristics; and norms or dietary preferences influence dietary intake in a highly variable manner over time and across individuals and populations [5].

For decades, assessment of dietary intake and composition relied primarily on paper-based self-reported intake, such as 24-h dietary recalls, food diaries, dietary screeners, and food frequency questionnaires. Although still valuable, self-reported dietary intake is subject to measurement errors, such as those from inaccurate memory, the variable composition of mixed meals, and difficulty estimating portion sizes/quantity of foods. Finally, another type of diet assessment is weighed food records. Weighed food records have high accuracy; however, this method is labor-intensive and only truly possible in highly controlled feeding studies where clinical research dietitians and staff are available.

More recently, novel technological advances, such as image and video capture and other devices capable of measuring dietary intake and eating behaviors, have been used in research studies, though this field is in its infancy [6,7,8]. Methods to assess biomarkers of nutrients and dietary patterns in the blood, skin, urine, hair, and/or stool are also being developed; however, to date, these approaches remain unproven, aside from a limited number of foods and in small to moderate-sized populations. Further, even as biomarkers develop, identifying and capturing food intake will be necessary to examine the total dietary pattern—including the timing of food intake and other contextual factors surrounding a meal and eating behaviors—to develop tailored or targeted strategies for behavior change to promote health. Multiple research purposes also need to be considered, including methods for surveillance, epidemiology, and evaluation of behavior change in response to interventions. The need to improve dietary assessment methodologies has been highlighted by leading nutritional scientists and communities [9,10] as well as the most recent Dietary Guidelines Advisory Committee report [11].

To address this research need, the United States NIH developed the first ever 2020–2030 Strategic Plan for NIH Nutrition Research [12], which includes a focus on improving dietary assessment. Specifically, the Strategic Plan calls for developing inexpensive and customizable dietary intake data-capture tools, including those that meet the needs of vulnerable populations, such as older adults with cognitive challenges, acutely ill patients, and infants and young children. In addition, the plan calls for integrated strategies to employ software and statistical models to combine data from multiple, distinct methods of dietary intake assessment, integrate datasets with nutrient-composition databases, and develop and disseminate statistical approaches to mitigate the effect of measurement error and thus improve accuracy.

As NIH begins to implement the Strategic Plan, the first step is to assess what research has recently been supported over the period of 2012–2021 to improve dietary assessment methodologies. To this end, the purpose of this manuscript is to characterize the NIH-funded dietary assessment portfolio over the last decade and highlight gaps and research opportunities. Herein, we highlight the NIH’s history of supporting dietary assessment research and aim to spur future research to continue addressing this great need.

Methods

Identifying nutrition research awards

NIH maintains an internal, confidential database with information on all NIH applications and awards. This portfolio analysis searched the database using a tool called NIH Query View and Report (QVR), which includes a label for which awards involve nutrition research. This label is derived from the NIH’s Research, Condition, and Disease Categorization (RCDC) System [13], which uses sophisticated text data mining in conjunction with NIH-wide definitions to create a “fingerprint” to identify NIH awards involving nutrition research. For this manuscript, we used QVR to identify all new, competing nutrition research awards (including grants, cooperative agreements, NIH intramural research, administrative supplements, and contracts), as labeled by RCDC, during fiscal years (FYs) 2012–2021. Each FY spans from October 1 through September 31. FY22 data were not available at the time of analysis. The same projects may also be categorized as 100% in other overlapping categories (that is, prevention, obesity, etc.). As this analysis required the use of data from an internal NIH database, the data does not align with what is publicly available from RePORT/RePORTER; therefore, these analyses cannot be wholly replicated using the publicly available data.

Identifying dietary assessment research awards

Within the set of nutrition research awards, subject matter experts (ME, JR, and KR) identified awards as focusing on dietary assessment, first by narrowing the list down using keywords (Biomarker of Dietary Intake, Diet Monitoring, Diet Records, Diet Surveys, Dietary Assessment, Dietary Monitoring, Dietary Questionnaires, or Dietary Records) found within the award title, abstract, or specific aims. Awards were then manually reviewed to determine those that supported dietary assessment research. An analysis of submitted compared with funded dietary assessment research awards was not possible. NIH publishes funding success rates for research projects by several variables in NIH RePORT/RePORTER, but success rates by research topic are not available.

Awards associated with applications submitted to 5 unique funding opportunity announcements specific to dietary assessment methods research were also categorized as dietary assessment research while avoiding duplication. The funding opportunities announcements (FOAs) included:

Diet and Physical Activity Assessment Methodology (R21) (example: PAR-18-857) and Diet and Physical Activity Assessment Methodology (R01) (example: PA-18-856). Both FOAs were active through the period of FY2012–2021.

Food-Specific Molecular Profiles and Biomarkers of Food and Nutrient Intake, and Dietary Exposure (R01 Clinical Trial Optional): PAR-18-727. FOA has been active from FY2015 to 2020.

Development of Biomarkers of Dietary Intake and Exposure consortium: RFA-DK-20-005 (U2C) and RFA-DK-20-007 (U24 Coordinating Center) issued for FY21. These FOA invited applications to form a consortium of multidisciplinary teams to establish Dietary Biomarkers Development Centers that explore and develop metabolomics-based dietary intake biomarkers.

Awards were further categorized as research project grants (R01, R21, R03), small business grants (R43/R44), awards submitted/funded through an NIH Institute or Center-specific initiative [that is, the NIDDK-Food-specific Biomarker Program Announcement with Special Review (PAR) or Request for Applications (RFAs)], or “Other” (K99/R00 or administrative supplement). Applications submitted as R01s and converted to a different mechanism (that is, cooperative agreement, R56) were treated as R01-equivalent awards.

Type of research topics

Awards were further reviewed by subject matter experts and assigned to 1 of 3 types of research, as follows:

• 1)New/refined method- or device-based approaches for capturing dietary intake, such as automated image capture via mobile phone or video camera; methods to determine specific patterns of intake; electrochemical sensors; ecological momentary assessment (EMA) or other methods to incorporate contextual factors, such as location, timing of dietary intake, or environmental factors; speech recognition technologies; methods to incorporate other dietary constituents such as dietary supplements; and methods to improve automated identification of food items and portion sizes.
• 2)Biomarkers of food and nutrient intake, such as stable isotopes for determining energy intake, or novel markers associated with intake of specific food types or nutrients, such as sugar-sweetened beverages, fruits and vegetables, meat, or fish.
• 3)Analytical/Statistical Methods, including applications that seek to develop and test analytical techniques, including machine learning or artificial intelligence approaches, to improve or correct measurement error in self-reported dietary intake methods.

Award abstracts and/or specific aims were also manually evaluated to determine the population of interest, that is, adults, children, and/or specific populations, including those with chronic or genetic diseases and conditions and individuals with disabilities.

Other NIH funding for dietary assessment

NIH can also fund research using a contract mechanism. Contracts supporting dietary assessment methodologies were identified using the same method as grant awards. NIH contracts are managed differently than grant awards; however, like grant applications, the contracts identified in this analysis were invited and reviewed through an open, competitive process and awarded based on merit and availability of funds. Contracts are described separately as a group from the other grant awards identified in this analysis. Projects and publications associated with these contracts were also evaluated.

Results

Total number of dietary assessment awards from FYs 2012 through 2021 compared with the total number of nutrition research awards

Over the 10-y period, 46 new competing awards for dietary assessment were issued out of a total of 13,185 new competing nutrition-related awards, or ∼0.35% of total nutrition research awards (Table 1). The number of dietary assessment awards increased modestly over time but remained a very small proportion of all nutrition-related awards. A list of grant awards as appearing in NIH RePORT/RePORTER is available in Supplemental Table 1.

TABLE 1.

Number and percentage of total funded competing dietary assessment awards compared to the total number of funded competing awards in nutrition research by fiscal year.

Fiscal year	Number of funded competing awards in dietary assessment	Total number of funded competing awards in nutrition research	Percentage of dietary assessment awards per total nutrition research awards (%)
2012	2	1176	0.17
2013	2	1167	0.17
2014	4	1150	0.35
2015	3	1276	0.24
2016	2	1386	0.14
2017	2	1279	0.16
2018	4	1486	0.27
2019	7	1318	0.53
2020	7	1468	0.48
2021	13	1479	0.88
Total	46	13,185	0.35

Dietary assessment awards by NIH institute/center

Awards were issued across 6 NIH Institutes, Centers, and Offices (Table 2), with most awards issued by NIDDK (37%) and the NCI (35%), followed by the NHLBI (17%), the Eunice Kennedy Shriver NICHD (7%), the NIMHD (2%), and the Office of Dietary Supplements (2%).

TABLE 2.

Number of funded dietary assessment research awards in fiscal years 2012–2021 by NIH institute or office.

NIH institute/office	Number of dietary assessment awards	Percentage of all dietary assessment awards (%)
NIDDK	17	37
NCI	16	35
NHLBI	8	17
NICHD	3	7
NIMHD	1	2
ODS1	1	2
Total	46	100

Data include all types of grant awards except for contracts. ODS, Office of Dietary Supplements.

¹ODS does not have direct grant funding authority but may provide funding through administrative supplements to the institute or center managing the parent grant award.

Funding opportunities and mechanisms of dietary assessment awards

The majority (54%) of the dietary assessment awards were submitted to dietary assessment-specific funding opportunities, with 11% submitted as small business applications and another 11% submitted to NIH parent R01 or R21 announcements (Figure 1A). Another 11% were submitted to the dietary biomarker consortium RFA, and the remainder (13%) were submitted to other NIH institutes, centers, or office-specific opportunities.

FIGURE 1.

Percentage of all dietary assessment awards funded in fiscal years 2012–2021 by type of NIH Funding Opportunity Announcement (Panel A) and type of award by activity code (Panel B). Funding Opportunity Announcements include the following: Methods Program Announcements (PAs): PA-18-856 and PAR-18-857 (several renewals); Biomarker PAs PAR-18-727 (several renewals); Small Business Innovation Research Program Announcements; Parent PAs: NIH Parent Program Announcements or NIH Institute-Specific Program Announcements; Biomarker RFA: Biomarker Request for Application U2C or U24 awards submitted to RFA-DK-20-005 or RFA-DK-20-007; or Other: 1 Office of Dietary Supplements administrative supplement, 3 R03 applications, and 1 R00 application. The Office of Dietary Supplements does not have direct grant funding authority but may provide funding through administrative supplements to the Institute or Center managing the parent grant award. RFA, request for applications.

Approximately 70% of the awards were submitted to program announcements as research project grants, specifically R01 applications (∼41%), R21 applications (∼24%), and R03 applications (∼4%). Another 11% were submitted as small business grants (R43/R44) (Figure 1B). Approximately 11% of awards were submitted to RFAs for the dietary biomarker consortium (U2C or U24). The remainder (∼9%) were R56 awards, R00 awards, or administrative supplements.

Types of dietary assessment awards

Awards were distributed across 3 types of research approaches (Figure 2). Most awards included projects developing or refining new technologies and procedures to detect dietary intake (46%), followed closely by projects proposing to develop biomarkers of intake (41%). Smaller numbers of awards (13%) proposed to conduct statistical analyses generally to address measurement error within existing methods.

FIGURE 2.

Percentage of all dietary assessment awards funded in fiscal years 2012–2021, by research focus area. Methods: development or refinement of methods/devices to capture dietary intake. Biomarkers: novel biomarker considered to reflect dietary intake (assessed via skin, urine, or blood). Statistical Analysis: analytical method (statistical methods or artificial intelligence/machine learning).

Over time, biomarker research has increased slightly, with awards specific for biomarker analysis issued in 2014 (Figure 3). The substantial increase observed in FY21 corresponded both with unsolicited awards submitted to the general dietary assessment methods PAs or other nonspecific funding opportunities as well as a consortium of 5 awards (highlighted in gray) designed to develop metabolomic biomarkers of dietary intake [submitted in response to the Development of Biomarkers of Dietary Intake and Exposure consortium: RFA-DK-20-005 (U2C) and RFA-DK-20-007 (U24 Coordinating Center)].

FIGURE 3.

Percentage of dietary assessment awards focused on biomarker development by fiscal year. Black bars: awards made to applications with a biomarker focus across all Program Announcements and Program Announcements with Special Review. Gray bars: applications submitted to the Biomarker Request for Applications: RFA-DK-20-005 and RFA-DK-20-007. No biomarker grant awards were funded in fiscal years 2012, 2013, and 2015.

Special populations and diseases and conditions and research type for dietary assessment awards (data not shown)

Most awards (91%) involved research in adults. A limited number of projects proposed to assess dietary intake in children over the age of 6 (3 awards), with 1 additional award specific to preschool age children. Other populations of interest included non–English speakers, populations with low literacy/numeracy, or methods specific for populations with chronic diseases or conditions, such as cancer, diabetes, kidney disease, or inborn errors of metabolism.

Other NIH funding for dietary assessment

Two large projects to advance diet assessment were funded through contracts during this time, including the automated self-administered 24-h dietary assessment tool (ASA24) [14] and the interactive diet and activity tracking in AARP (formerly the American Association of Retired Persons) (IDATA) [15]. The ASA24 is a free, web-based tool that enables researchers or users to collect multiple, automatically coded, self-administered 24-h diet recalls or single or multiday food records. Nutrient and food group information is immediately available to the researcher once the respondent completes the 24-h diet recall or record. Development of ASA24 began in 2006 through a contract directed by the NCI with contributions from 8 NIH Institutes and Centers to the research firm Westat (Rockville, MD). ASA24-2011 was the first version released after the original beta version. ASA24 was migrated to a mobile platform (both Android and iPhone versions) in 2016. Between 2012 and 2021, over 8,000 new studies have been registered for the United States version (independent efforts have supported 2 Canadian versions and an Australian version over the years). For the same time period, the United States version has collected over half a million (524,598) dietary recalls or records. Figure 4 shows the number of publications extracted from administrative data that used ASA24 to collect data until 2020, when publications jumped to 166, over 3 times (54) the number of publications the previous year.

FIGURE 4.

Number of peer-reviewed publications that used the automated self-administered 24-h dietary assessment tool (ASA24) to collect data by calendar year.

The IDATA project has been described elsewhere [15], but in brief, the project was designed to evaluate and compare the measurement error structure of diet and physical activity assessment tools against reference biomarkers. The IDATA study recruited participants from a list of AARP members aged 50–74 y residing in and around Pittsburgh, Pennsylvania. Data collected included physical activity self-report (4 instruments), physical activity objectives monitors (3 monitors), diet (4 instruments), biomarkers (from urine and saliva), and participant characteristics. IDATA data and/or biospecimens are available as public use data; researchers must submit a project proposal, which is reviewed by NCI [16]. Since 2016, IDATA has logged 46 approved projects [17] and 11 publications [18]. The IDATA Biospecimen Round 10 is currently open for applications, and they will be accepted through February 2023.

Discussion

This analysis of NIH-funded dietary assessment methods research from 2012 through 2021 indicates that the portfolio of funded grants and contracts has increased over time but remains a small proportion of NIH’s overall nutrition research portfolio. Although small in numbers, this portfolio encompasses a wide and growing variety of novel tools and analytical methods assessing food and nutrient intake. Across all of NIH, the majority of NIH funding supports untargeted research, that is, investigator-initiated applications submitted to funding opportunity announcements without set-aside funding. To continue increasing NIH support for dietary assessment methods research, combined efforts will be necessary, including more high quality, investigator-initiated applications in particular, as well as new NIH-supported initiatives for dietary assessment methods research as specific research gaps are identified.

Over the 10-y period in this analysis, much of the funded dietary assessment research has focused on developing or refining accurate, user-friendly, and customizable dietary intake data-capture tools, ranging from self-reported questionnaires or recalls to mobile image-based approaches. More recently, NIH has supported an expanding number of grants and programs designed to develop biomarkers of nutrient intake. Support for novel analytical methods has also grown with the emergence of artificial intelligence/machine learning methods in recent years. Despite modest increases in the number of awards in dietary assessment methods, more research and improvement in dietary assessment methodologies is needed. As mentioned above, the NIH has added a focus on dietary assessment in the first ever NIH Nutrition Research Strategic Plan.

Since 2006, NIH has sponsored the ASA24 through a contract managed by the NCI. ASA-24 is a free, widely used research grade dietary intake assessment tool, which has made the collection of research quality dietary data cost effective and timely. Because ASA24 is self-administered, there is no reliance on trained interviewers to conduct dietary interviews. Moreover, ASA24 is automatically coded and does not rely on individual coders, nutritionists, or dieticians to link foods, beverages, and dietary supplements reported to the associated nutrients and food groups. NIH has also invested in the USDA-managed Food and Nutrition Database of Dietary Studies [19], which codes foods and beverages based on amounts consumed to calculate nutrient composition. The value of the ASA24 is reflected in the growing number of users and an increasing number of publications presenting ASA24 data over the years.

NIH has also supported a range of methodological advances over the past decade. These methods include research using image capture coupled with image analysis, machine learning, and artificial intelligence methodology to improve interpretation, detection, and processing of data as well as methods to mitigate measurement error. Back-end image detection, linkages, and context detection have also been expanded and refined with associated privacy protections, funded, in part, with NIH contributions. Dietary biomarker research grew during this period through an NIH and USDA-supported partnership funding investigator-initiated awards. This area of research has also grown more recently through an NIH and USDA-supported consortium of 5 awards and a Data Coordinating Center bringing together multidisciplinary teams to explore and develop metabolomics-based dietary intake biomarkers within urine and blood samples. More specifically, this consortium will employ untargeted metabolomics analyses designed to identify metabolite signatures for major food groups or types within small, controlled feeding studies.

Although the field of dietary assessment has grown with the use of a variety of methods, gaps, and challenges remain. A well-known, critical challenge is the capture of energy intake, particularly underreporting [20]. Underreporting by self-report and/or failure to capture foods using digital methods, as well as difficulty in evaluating mixed meals and food/beverage composition (that is, fat in dairy products, sweetened compared with artificially sweetened or unsweetened beverages, etc.) diminishes the accuracy of estimation of energy intake. Some technologies, such as automated densitometry and volumetrics, may help to mitigate this limitation; however, some user reporting will likely be necessary [21]. The addition of automated, personalized reminders to capture food intake, integration of dietary biomarkers, and the development of other tools and analytical procedures, among others, may be promising approaches to improve the precision of assessment [22,23]. Additionally, NIH funding for assessment of dietary intake in special populations, including infants and young children, was limited; almost all the funded awards identified in this analysis were focused on adults, with a small number of awards focused on children over 6 y of age and only 1 award in children between 3–5 y of age. No awards had a focus on individuals with disabilities; moreover, current digital dietary assessment technologies are also unable to accommodate individuals with limited digital proficiency.

Absolute quantification of dietary intake also requires true measures of the content and amounts of foods and beverages consumed by an individual. The nutrient and other content (water, ash, fiber, etc.) of foods can only be absolutely determined by testing the food itself, which is usually not possible in most studies; therefore, studies must rely on food databases that provide average estimates of food content. A food’s nutrient content, particularly for produce, varies by where and when the ingredients were grown/made, how they were produced or processed, and how long the items remained on the shelf, among many other factors [24,25]. Therefore, there is a known error in relying on averages from food databases to estimate food and beverage content. Though this error can be overcome by measuring large amounts of participants, that is often not feasible

Dietary patterns, including timing [26] and contextual factors such as the location where foods are consumed (that is, at home, school or work, restaurants, in the car, etc.) [27], consumption of foods in social settings [28], and whether foods were consumed while using a screen [29], are also known to influence dietary intake. Newer tools are beginning to incorporate reminders, EMA, or specific questions to capture these personal and contextual factors [22,23]. Additional priorities include strategies to combine data from multiple methods of dietary intake assessment; integrating datasets with nutrient-composition databases, and continuing to support the development and dissemination of statistical approaches to mitigate the effect of measurement error. Further research also needs to accommodate a more holistic view that considers multidimensionality (that is, analyzes dietary patterns as opposed to individual nutrients) and timing of consumption along with other behaviors.

Although not included in this analysis, in FY 2022, NIH funded the Nutrition for Precision Health (NPH), powered by the All of Us Research Program [30]. The NPH program will conduct controlled community-dwelling and domiciled feeding studies with a goal of developing algorithms that predict individual responses to foods and dietary patterns in adults. The program also includes a component intended to advance dietary assessment methodologies through validation, evaluation, and modeling efforts using existing and emerging technologies.

Limitations for this analysis include reliance on information contained within grant applications, including abstracts and specific aims, as well as NIH’s system for categorizing research topics; therefore, it is possible that awards may have been missed. Although contracts were included in this analysis, contract abstracts often provide limited insight into the nature of the work, so some relevant contracts may also have been missed. Finally, the authors note that many organizations or groups, including nonprofit organizations, for-profit entities, other Federal agencies (that is, the National Science Foundation), and other larger research foundations support the development of dietary assessment methods; therefore, this analysis does not provide a comprehensive assessment of ongoing efforts to improve dietary assessment. Nevertheless, NIH plays an important and substantial role in advancing dietary assessment methodologies available for research use.

In conclusion, this analysis of NIH support for dietary assessment methodologies between 2012 and 2021 revealed a small but growing number of grants and contracts designed to improve dietary assessment. Technologies ranged from self-reported dietary intake through the ASA24, a freely available, automated method for collecting dietary recalls and records, to image-based capture of foods and beverages through mobile phone cameras or video capture, to biomarkers of nutrient intake. Support for novel, artificial intelligence/machine learning-based methods to improve the analysis and overcome methodological limitations is also expanding. As the recent increase in awards demonstrates, assessment of dietary intake and efforts to improve dietary intake, including precision nutrition for optimal health, represent important priorities for NIH. NIH continues to encourage the research community to pursue innovative methods to assess or improve dietary assessment methodologies, particularly across diverse populations and settings.

Author disclosures

The authors report no conflicts of interest.

Funding

All authors were employees of the NIH when they were actively engaged in work related to this article. NIH is the sole source of support for the work reported.

Data availability

Publicly available data on funded NIH grants and contracts can be found on NIH RePORT/RePORTER (https://reporter.nih.gov/). NIH RCDC data is confidential and may not be shared.

Appendix A. Supplementary data

The following is the Supplementary data to this article: