Using Short-Term Dietary Intake Data to Address Research Questions Related to Usual Dietary Intake among Populations and Subpopulations: Assumptions, Statistical Techniques, and Considerations

Sharon I. Kirkpatrick, Patricia M. Guenther, Amy F. Subar, Susan M. Krebs-Smith, Kirsten A. Herrick, Laurence S. Freedman, Kevin W. Dodd

Research output: Contribution to journalArticlepeer-review

24 Scopus citations

Abstract

Many research questions focused on characterizing usual, or long-term average, dietary intake of populations and subpopulations rely on short-term intake data. The objective of this paper is to review key assumptions, statistical techniques, and considerations underpinning the use of short-term dietary intake data to make inference about usual dietary intake. The focus is on measurement error and strategies to mitigate its effects on estimated characteristics of population-level usual intake, with attention to relevant analytic issues such as accounting for survey design. Key assumptions are that short-term assessments are subject to random error only (i.e., unbiased for individual usual intake) and that some aspects of the error structure apply to all respondents, allowing estimation of this error structure in data sets with only a few repeat measures per person. Under these assumptions, a single 24-hour dietary recall per person can be used to estimate group mean intake; and with as little as one repeat on a subsample and with more complex statistical techniques, other characteristics of distributions of usual intake, such as percentiles, can be estimated. Related considerations include the number of days of data available, skewness of intake distributions, whether the dietary components of interest are consumed nearly daily by nearly everyone or episodically, the number of correlated dietary components of interest, time-varying nuisance effects related to day of week and season, and variance estimation and inference. Appropriate application of assumptions and recommended statistical techniques allows researchers to address a range of research questions, though it is imperative to acknowledge systematic error (bias) in short-term data and its implications for conclusions.

Original languageEnglish
Pages (from-to)1246-1262
Number of pages17
JournalJournal of the Academy of Nutrition and Dietetics
Volume122
Issue number7
DOIs
StatePublished - Jul 2022
Externally publishedYes

Bibliographical note

Publisher Copyright:
© 2022 Academy of Nutrition and Dietetics

Keywords

  • 24-hour dietary recall
  • Dietary Assessment
  • Food record
  • Measurement error
  • Usual dietary intake

Fingerprint

Dive into the research topics of 'Using Short-Term Dietary Intake Data to Address Research Questions Related to Usual Dietary Intake among Populations and Subpopulations: Assumptions, Statistical Techniques, and Considerations'. Together they form a unique fingerprint.

Cite this