MEASURES REGISTRY USER GUIDE
Key Considerations in Measuring Dietary Behavior Among Children
In most research relevant to dietary behaviors, self-report measures are used. This is because it is generally not possible to objectively assess usual intake in community-dwelling individuals. Various considerations come into play in assessing dietary intake using self-report (Box 1). These considerations may have particular salience depending on the age of children and their respective cognitive skills, as well as literacy and numeracy. Errors in the capture of children’s intake may include under- and over-reporting of foods and beverages, incorrect identification of foods, and portion size misestimation.31 Commonly consumed foods and beverages may be reported, even if they weren’t consumed during the time period of interest.31 In addition, measurement of children’s dietary behavior may be affected by motivation to provide accurate information. In sum, intake may be either under- or over-estimated and this may vary by age, dietary component, and in relation to the tool used for assessment.31
In assessing diet among infants and toddlers, measures are typically completed using proxy-reporting by parents or caregivers, except perhaps in the case of observation. Parents or other caregivers may not accurately report their children’s intake, with potential misreporting particularly for eating occasions for which they were not present.32 This becomes a more significant issue as children get older and spend more time away from parents, for example, in early care and education settings. Additionally, caregivers who report their children’s food intake are susceptible to many of the sources of error that affect reporting of their own diet, including social desirability biases,33 because of concerns that their child’s eating patterns and weight might be perceived negatively. For example, Börnhorst et al. used the Goldberg equation to classify parent proxy reports as plausible, under-reported, or over-reported, based on energy estimates from 24-hour recalls completed for children aged 2 to 9 years.34 Under-reporting was positively associated and over-reporting was negatively associated with z-score for body mass index. Further, proxies who perceived their children as overweight were more likely to be classified as under-reporters.
Box 1. Factors That Can Influence the Quality of Data on Dietary Intake Behavior
Cognitive abilities: The capacity to learn, remember, and pay attention. Immature cognitive skills in young children necessitate proxy reporting, and still-developing cognitive skills in older children may limit options for independent self-reporting.
Literacy: Ability to read and write. Essential for completion of most self-administered tools.
Numeracy: Ability to understand and manipulate numbers. Critical to self-administration of most dietary assessment tools, particularly when calculations to average frequency of intake and typical portion sizes over time are required.
Proxy reporting: Provision of data on dietary behavior by someone other than the person of interest. The extent to which the proxy has first-hand knowledge of the person of interest’s diet can impact the accuracy of reporting, as can other sources of bias such as those related to body weight or portion size estimation.
Recall bias: Lapses in memory, either short-term in reporting of intake for a recent period (e.g., yesterday) or long-term in reporting of intake for a longer period (e.g., the past year).
Reactivity: Tendency to change one’s behavior in response to monitoring or the expectation of being measured. When the intention is to capture usual intake, data collected using food records can suffer from reactivity bias. When the intention is to support self-monitoring for the purpose of enabling behavior change, reactivity is the desired effect.
Retention interval/recency: The length of time between the dietary behavior of interest and reporting of that behavior. Longer retention intervals (i.e., lower recency) may reduce children’s ability to accurately report.
Social desirability bias: Tendency to respond in a way perceived to be socially desirable. “Social desirable responding is presumed when an individual reports never performing a behavior that most everyone performs at least occasionally or reports always performing a behavior that most people usually perform but omit occasionally.”35 For example, consumption of foods and beverages perceived as less healthy, or “bad”, may be under-reported whereas consumption of foods and beverages perceived as “good” may be over-reported.
School-age children may be enthusiastic about reporting their dietary intake.36 However, depending on the method used, their developmental stage may hinder their ability to accurately report food intake and related behaviors. Potential factors affecting reporting include variable levels of literacy,32,37,38 limited attention span,32 and inadequate concepts of time and memory.39 It has been suggested that children begin to be able to conceptualize time at the age of around 7 or 8 years.31 However, at this age, the time periods for which children can report their intake are likely to be limited.31 Limited knowledge of food preparation methods and ingredients may also pose a barrier to accurate reporting of details of foods and beverages consumed.
Adolescents may be less interested in reporting their intake, increasing the risk of reporting error.36 Further, their diets may be less structured and more variable than those of younger children,32,37 and this complexity may be difficult to capture, for example, with tools administered a single time or that do not consider complexities such as snacking between meals. Additionally, weight management and restrained eating efforts, which are much more frequent among adolescents than younger children, as well as weight status, may bias reported intake.32,36 Technology-enabled tools, such as food records on mobile devices, developed for use with this population may help lessen some of these barriers to the capture of accurate dietary data, for example, by increasing engagement and motivation.40
Children of all ages struggle with portion size estimation32,38 (as do adults). The literature on the utility of training or the use of portion size aids that reflect typical portion sizes consumed by children is mixed, with the overall conclusion that this aspect of dietary assessment poses an ongoing challenge.31,41
Social desirability bias is an additional issue that cuts across population groups. It is possible that this source of bias is becoming more of an issue due to increasing rates of, and stigma related to, overweight and obesity,42 as well as constant media (including social media) attention to food and diets, as well as body shapes and sizes.43 A study conducted using data from the mid-1990s and early 2000s found higher under-reporting of protein in the later time period among adults, which the authors suggested might reflect growing misreporting of other macronutrients, potentially due to heightened awareness of dietary intakes due to public health campaigns.44 Further research is needed to understand secular trends in misreporting among children, particularly in the age of social media.
Research in children suggests social desirability is an issue affecting this population. For example, social desirability (measured using a questionnaire) has been associated with lower accuracy of reporting (potentially under- or over-estimation) of dietary intake on 24-hour recalls among fourth-grade children.45,46 Scales to assess social desirability bias can be administered in nutrition research as a means of better understanding this possible source of error.35,47–49 Also possible is social desirability-related misreporting of other variables of interest, such as weight,37 with implications for study findings. Weight perceptions and bias may interact with social desirability,46 leading children and youth to report lower energy intakes if they perceive themselves as overweight.50 More research is needed to investigate this complex interaction and its implications for dietary data.35 However, researchers can aim to reduce this source of under-reporting by maintaining neutrality in describing the study and not promoting perceived desirable responding,37 as well as by measuring and adjusting for social desirability.35,47–49
With technological innovation in the assessment of diet,51 such as web-based 24-hour recalls52,53 and mobile food records,54,55 it is becoming feasible to collect dietary data in a broader range of studies and settings than previously possible. Technological developments have been targeted to engage children in particular in more accurately self-reporting their diets.40 For example, recent innovations have incorporated animated avatars or cartoon characters, interactive questionnaires, and the addition of narratives to engage children during the completion of assessments.40 However, technology does not mitigate all of the challenges in accurately assessing diet, and it remains critical to ensure that the measure chosen is well-suited to the research question, the study population, and the setting.40,56,57 Although the Measures Registry does not categorize methods and tools that incorporate technology, the developments in this area suggest this as a potential focus of the Registry and similar tools in the future.
With these various considerations in mind, Box 2 outlines potential strategies to enhance the accuracy of data captured in studies of children’s dietary behaviors. Within Section 5, we briefly highlight unique considerations related to specific types of measures for use with children. Section 12 provides a list of selected resources that can be used in combination with the Measures Registry and this User Guide to help inform measurement of dietary behaviors in children. This list includes numerous reviews related to assessment in children specifically.
Box 2. Potential Strategies to Enhance Accuracy of Dietary Behavior Data Among Children
- Ensure tools are tailored to the target group in terms of demands related to attention span, literacy, numeracy, and other factors influenced by developmental stage and corresponding cognitive skills.
- Take advantage of technology (e.g., web-based recalls, mobile food records, skin scanner technology, ‘gamified’ programs) if applicable to the target group to reduce burden and improve motivation and engagement. However, new challenges that can be introduced by technology (e.g., related to Internet connectivity or computer literacy) also need to be considered.
- Provide clear instructions, complemented with hands-on training if necessary. This extends to proxy reporters in studies in which children cannot report independently.
- Use portion size aids appropriate to the population. Training in portion size estimation may be helpful to reduce error associated with this component of assessment.
- Allow adequate time for completion, recognizing that children may need more time to complete assessments than adults due to less mature cognitive skills.
- Use techniques to reduce social desirability bias, such as using neutral probing and querying an array of food and drink items rather than focusing narrowly on specific items of interest, particularly when these may be perceived as unhealthy (e.g., sugar-sweetened beverages).
- Be aware of potential sources of error and interpret and report the data accordingly.