Currently, a range of different tools are used in nutrition research with relevance to childhood obesity. This variability, combined with the use of tools that may not be optimal for the purpose and population, hinders the development of a cohesive evidence base to inform policies and programs for reducing childhood obesity. To further this area of research, it is critical to use the best possible methods, along with appropriate analysis, interpretation, and reporting. This User Guide, along with other resources highlighted, is intended to help with this goal. Selection of measures will involve weighing various considerations, including logistical issues that may limit the feasibility of some measures in particular environments or populations, with the overall aim of arriving at the best measure for a given research study. The use of the Measures Registry along with this User Guide can also contribute to a greater standardization of measures, with positive implications for comparability of studies, potential for pooling of data to address questions that might not be feasible within any one study, and syntheses of evidence to inform interventions.

To assess dietary behaviors, we typically rely on self-report measures, which are known to be affected by error.⁸⁰ Sources of error in studies of children can include reliance on proxy reporters (who may have incomplete knowledge of true consumption). Further, tools may not be appropriate for use with particular age groups due to varying attention span, literacy, and numeracy. Factors that affect self-report data more generally, such as social desirability biases, also are sources of error. These errors can result in under- or over-estimation of intake of different dietary components. It has been well established that energy is reported with substantial error and for this reason, it is recommended that self-report data not be used to estimate absolute energy intake.⁶⁷ Misreporting has been shown to be associated with body weight status, with greater under-reporting of energy with higher body mass index.^65,112 For this reason, it is challenging to assess associations between intake and body weight. Cautions of this nature must be extended to variables that may be correlated with body weight, including race/ethnicity, education, and other dietary behaviors, such as restricted eating behaviors or body image. Estimates of other dietary components, such as protein and potassium, based upon self-report appear less biased compared to energy,^65,66 which supports the overall value of self-report for understanding dietary patterns. Less is known about misreporting of particular foods and beverages, and this is an area in which additional research is needed.

Study design must also be considered in selecting measures and analyzing and interpreting data. For example, caution is needed in approaching evaluations of the impact of interventions on intake, particularly in cases in which individuals have been counseled or otherwise encouraged to alter their eating patterns.¹ Differential exposure to the intervention can result in differences in error in reporting before and after or between intervention and control groups, with implications for the validity of comparisons.

Despite these caveats and challenges, measuring dietary intake in children is well worth the effort given the invaluable data yielded to inform policies and programs to prevent obesity.

Evolution in the Field of Dietary Assessment

The long-standing recognition of the limitations of self-report measures has fueled considerable efforts to improve them, for example, using technology to reduce researcher and respondent burden. The explosion of technology extends to children, for example, with the development of mobile device food records that can be used by adolescents to capture food intake in various settings. Portion size estimation is another area of ongoing inquiry, with links to technology through image-based assessment that may reduce error associated with traditional portion size aids. Work to better understand how children interact with dietary measures, such as the age at which they can independently respond to self-report measures, is also ongoing. However, it must be borne in mind that technology cannot address all limitations of self-report methods, and indeed, may introduce new challenges, for example, related to computer literacy.

Complementing efforts to improve methods and how they are used are significant efforts to enhance our understanding of error and approaches to mitigate it, for example, through combining measures to exploit the strengths of each as well as analytic techniques. The discovery of concentration and predictive biomarkers and other innovative methods that may help offset the effects of error in self-reported data in analysis is an ongoing area of research. Approaches to combine self-report and biomarker data have been developed for both epidemiological and intervention research.

Despite these innovations, much work remains. Many studies to assess the psychometric properties of self-report methods with children have used other self-report tools as references, with limitations in interpretation due to the lack of an unbiased reference such as a recovery biomarker. Additional biomarker-based studies in children could help to further this evidence, though the number of dietary components for which unbiased references are available remains limited. To complement improvements in methods, additional research is needed to better understand misreporting of a broad range of dietary components, including foods and food groups, as well as to inform the evaluation of interventions by better understanding intervention-related biases and the sensitivity of measures to change.