Skip to Main Content

An official website of the United States government

Feasibility of Short-term Individualized Dietary Recall Methodology Among U.S. Older Adults

Principal Investigator
Name
Alexandra Cowan
Degrees
M.S., Ph.D.
Institution
Texas A&M University AgriLife Research, Institute for Advancing Health Through Agriculture
Position Title
Postdoctoral Research Fellow
Email
About this CDAS Project
Study
IDATA (Learn more about this study)
Project ID
IDATA-62
Initial CDAS Request Approval
Feb 10, 2023
Title
Feasibility of Short-term Individualized Dietary Recall Methodology Among U.S. Older Adults
Summary
The dietary exposome serves as a significant determinant of several chronic diseases of public health relevance. However, the predominant methodologies utilized in dietary assessment research rely on self-report, subjecting the analysis to random and systematic measurement error that pervades self-report dietary assessment instruments. Both types of error affect the estimation of usual dietary intakes and can in turn, attenuate the relationship between diet and health outcomes and reduce the statistical significance of potential findings. A 24-hr dietary recall (24HR) possesses less systematic error but greater random error (i.e., within-person variation) when compared with other self-report methods; while the effects of systematic error largely cannot be abated, the effects of random error are well understood and can be mitigated via the use of statistical modeling techniques that adjust for random error in datasets that utilize a small number of repeat 24HRs. Previous research utilizing this approach has employed traditional, full-day 24HRs, however, in addition to within-person variation in intakes, variability in dietary intakes over the course of a single day is also an important consideration and type of pervasive bias that has yet to be adequately accounted for in existing studies. For this reason, there is a pressing need for an innovative, dietary assessment instrument that can account for both intra-individual and intra-day variations in dietary intake to enhance the accuracy of dietary exposure classification and improve our understanding of the diet-health relationship. Accordingly, in the present proposal, we intend to develop and assess the feasibility of a dietary assessment methodology designed to estimate long-term usual dietary intakes and account for within-day variation and temporal patterns of dietary intake, in addition to day-to-day variation in intake via the use of short-term, individualized dietary recalls. In Aim 1, we plan to leverage existing questionnaire data from the Interactive Diet and Activity Tracking in AARP (IDATA) study to develop an individualized, personal signature for each participant that provides details on each participant’s habitual eating patterns, as well as behavioral, lifestyle, and health-related characteristics. In Aim 2, we will employ personal signatures and novel statistical techniques to predict the optimal combinations of individualized, short-term (i.e., 4-hr) dietary recall collection intervals and evaluate the accuracy of energy reporting of this methodology in comparison to a traditional 24HR approach among the IDATA cohort. Lastly, in Aim 3, we will assess the feasibility and validity of the individualized, short-term recall approach via usual dietary intake estimation for energy, sodium, potassium, and protein relative to actual intake, estimated using recovery biomarkers. Thus, this proposed study has the potential to not only improve exposure classification, but also improve our ability to elucidate causal links to human health.
Aims

We propose to leverage existing data, and through the use of personal signatures and novel statistical techniques (e.g., machine learning), predict the optimal combinations of days and times (i.e., short-term, selected ranges of hours, rather than the complete 24hrs) across a large, diverse sample of older adults (n=1,110; ages 50-74 y; 92.5% non-Hispanic white, 6.5% African American, 3% Hispanic, 3% Asian) from the Interactive Diet and Activity Tracking in AARP (IDATA) validation study, via three specific aims. The IDATA study was developed and funded by the NIH National Cancer Institute to evaluate dietary assessment methods, including up to 6 24HRs and 2 food-frequency questionnaires, as well as recovery biomarkers (e.g., DLW, urinary nitrogen, potassium, and sodium). Evaluating the feasibility of this novel, dietary assessment instrument in a population subgroup in which many unique nutritional and health challenges exist (e.g., high prevalence of chronic disease) and in which innovative tools are greatly needed to aid in health promotion is of significant interest, especially given the increased potential for respondent burden (e.g., cognitive challenge) among the older adult population.
Aim 1. Utilize behavioral, lifestyle, and health-related information collected in IDATA to develop individualized, personal signatures among the IDATA cohort. Questionnaire and 24-hour physical activity data from IDATA will be employed to collect details on each participants’ habitual eating times, typical beverage consumption, supplement use, general sleep/wake schedule, and meal and snack patterns, and in turn, develop a personal signature for each participant to inform the optimal times for intervals of dietary data collection to match the participant’s typical patterns of eating, and to better approximate accuracy of reporting.
Aim 2. Employ individualized personal signatures, as well as novel, statistical techniques, to predict and augment optimal combinations of days and times for dietary data collection at an individual level among the IDATA cohort. 24HR data collected in IDATA will be utilized to simulate individualized, short-term (i.e., 4-hr) dietary recall collection schedules for each participant. A participant’s personal signature (Aim 1) will inform the simulated timing of data collection to increase the probability of capturing essential eating occasions for each participant. An additional approach will also be developed, in which novel statistical techniques (e.g., machine learning) are used to identify the optimal combinations of time intervals (and length of intervals) and compared for accuracy of energy reporting to the short-term recall with a personal signature method. Both approaches will also be evaluated for the accuracy of energy reporting in comparison to a traditional 24HR usual intake approach.
Aim 3. Examine the validity and feasibility of short-term, individualized dietary recall methodology for estimating long-term, usual dietary intakes among the IDATA cohort. Both short-term, individualized dietary recall approaches will be examined for validity and feasibility by estimating usual dietary intakes for energy, sodium, potassium, and protein for each participant, in comparison to the gold standard method of assessment (i.e., recovery biomarkers) for each respective nutrient, such as, DLW (i.e., energy), urinary nitrogen (i.e., protein), urinary potassium, and urinary sodium.

Collaborators

Regan L. Bailey, PhD, MPH, RD. Institute for Advancing Health Through Agriculture, AgriLife Research, Texas A&M University.
Diane Mitchell, MS, RD. Institute for Advancing Health Through Agriculture, AgriLife Research, Texas A&M University.
Terry Hartman, PhD, MPH, RD. Rollins School of Public Health, Emory University.
Janet A. Tooze, PhD, MPH. Wake Forest School of Medicine, Wake Forest University.