Critically discuss the issues in assessing nutrient intake-the ability to accurately assess nutrient intakeallows identification of deficiencies and excesses of nutrients in bothindividuals and groups. Such information is important; under nutrition’s thought to contribute to 35% of childhood deaths each year (Alderman, Hodinott & Kinsey) and obesity is a major public health problem. Furthermore, it is well established thatdiet is associated withboth prevention and causation of many chronic diseases, such as cancer (Bingham, 2002). Therefore,accurate information regarding nutrient intake, and consequently nutrient status, is essentialinorder to form public health policies and promotion campaigns to both improve general health and preventdisease. There are several methods of assessing nutrient intakeand many issues associatedwith thesemethods. This essay will discussissues of biasassociated with self-report measures, including 24-hour recall and Food Frequency Questionnaires (FFQs), in addition to the cost of these methods.The difficulties that arise when comparing and validating these methods will then be considered, followed by the effect of introducing biomarkers as a method of assessing nutrient intake. The most common method of assessing nutrient intake isself-report methods. Two examples to beexplored in this essay include 24-hour recall and FFQs. 24-hour recall involves a trained interviewer asking respondents to recall their exact food intake during the previous 24-hour period -most often the previous day (Gibson, 2005). The foods listed are then entered into a specialised programme for analysis of nutrient intake. This method provides information-rich data, but only reflects short term intake. Thus, FFQsoffer a better assesmentof longerterm nutrient intake. FFQscollect information about an individual’s food consumption over a fixed period of time –usually months ora year, thus allowing for better assessment ofhabitual nutrient intake (Wrieden, Peace, Armstorng & Barton, 2003). Theseare usually self-administered questionnaires, requiringrespondents to record their consumption frequency of each food in a given list (Wiremen, Peace, Armstrong & Barton, 2003). The primary issue associated with these self-report methodsis the considerable amount of bias, which significantly impacts on their accuracy, reliability, and thus ability to assess nutrient intake. Whilst there are many different types of bias affecting these measures, including social desirability bias (Hebert et al., 2001), researcher biasand participant bias (in
that the individuals studied are likely to be those at-risk), the bias towards underestimation of food, and subsequentintake, has received the most research attention. This bias islikely heavily affected by social desirability bias. Underestimation of nutrient intakeis commonly reported in both 24-hour recall (Ferrari et al., 2002; Subar et al., 2000) and FFQs, where Carlsen et al. (2010) found a third of their sample to be ‘underreporters’ when using this method. This presents an issue when assessing nutrient intake, as it highlights issues of respondent accuracy and suggests collected information may not be reflective of actual intake. Such underreporting has been found to be related to individual factors, such as BMI, educational level and age (Freedman et al., 2014), further biasing responses, and subsequently questioning the ability of collected information to accurately assess nutrient intake and thus its usefulness as research data aimed at establishing causation and trends, improving health and preventing diseaseDespite this, there is evidence that recall accuracy may vary depending on the type of nutrientassessed. For example, although total energy and macronutrient intake appearparticularlysusceptible to underreporting (Ferrari et al., 2002; Subar et al., 2000; Carlsen et al., 2010), some research suggests that thismay be less prevalent when suchmethods are used to assess micronutrient intake. For example, FFQ results have been found to correlatehighly with both a 7-day food diary and vitamin D biomarker concentration (Wu et al., 2009). Molgaard, Sandstrom & Michaelson (1998) compared children’s calcium and phosphorus intake using FFQs and weighed food recordsand found no significant differences between the two methods. Furthermore, whilst Carlsen et al.’s (2010) study found a bias of respondents to underestimate their energy and macronutrient intake, they did not find significant differences between methods for reported intakes of antioxidant rich foods, suggesting that FFQsmay be an appropriate methodto assess micronutrient intakeand therefore,in such cases, reducing issuesin their use of nutrient intake assessment.Moreover,research has also shown that underreporting errors can be reduced. For example, previously underreported intake recall accuracy improves with multiple 24-hour recalls (Yunsheng et al., 2009). Similarly, reliability for total energy, fat, fruit and vegetable intake has been found to increase as a function of the number of recall days (St George, Van Horn, Lawman & Wilson,2016). Despite this, research suggests that the number of 24-hour recalls necessary for accurate recall may vary depending on the nutrients being assessed, and the individuals involved. Whilst Yunsheng et al. (2009) found three 24-hour recalls to be optimal
for accurate recall of overall energy intake, other research has suggested 21 to32 recalls to be necessary to achieve an acceptable level of reliability (St George et al., 2016). Furthermore,Ollberding et al.’s (2014) cross-sectional survey of the National Health and Nutrition Examination Survey (NHANES) found 6 to9 recalls to benecessary for children, whilst 3 to 6 recalls were optimum for adolescents. Therefore, whilst it seems evident that underreporting is an issue when assessing nutrient intake through self-report measures, it appears that underreporting can be minimised with appropriate use of measures -for example if multiple 24-hour recalls are used, or such methods are used to assess micronutrient rather than macronutrient intake. Althoughunderreporting may be improved through the collectionof multiple 24-hour recalls, this links directly to another issue in assessing nutrient intake: the substantial associated cost. By ‘cost’, we must consider not just financial cost but also cost in terms of researcher time and investment, in addition to participant burden. For example, whilst increasing the number of 24-hour recalls mayincrease accuracy (St George et al., 2016), participant burdenalso increasesparallel to this.At the extreme,the found necessity of21 to 32 recalls to ensure sufficient reliability(St George et al., 2016) causes such measurestobecome a decreasingly economical method of assessing nutrient intake. The burden on participants may deter certain individuals from completing nutrient intake assessments, further biasing results. This burden is alsoinflicted on the researchers; for example,24-hour recalls require a trained interviewer, and thus the burden and time needed from the participant is also needed from the researcher, in addition totime consumingdocumentation processes, coding and data entry. Finally, whilst FFQs have an estimated baseline cost of$1.2 million(Kristal, Peters & Potter, 2005), research suggests that they maybe poorer at assessing nutrient intake accurately (Subart et al., 2000) in comparison to 24-hour recalls. Whilst 24-hour recallsprovide more detailed information, they require multiple recalls, with threerecallsestimated to have a baseline cost of $25 million and thus a ‘bleak likelihood’offunding ( Kristal et al., 2005) when there is a cheaper alternative, highlightingthe issue of balancing cost with accuracy supporting the issue of nutrient intake assessment being a costlyprocess, not only financially but also in terms of individuals’time and resources.Whilst it is difficult to establish a method of nutrient intake without cost, new technology certainly reduces some of the costs associated with original versions of nutrient assessment. For example,internet versions of the 24-hour recall have been developed,including the
Automated Multiple Pass Method (AMPM), which has beenused in the US National Health and Nutrition Examination Survey and by the National Cancer Institute (Lombard, Steyn, Charlton & Senekal, 2015). AMPMreduces bothcostand burdenfor the researcher, who no longer needs to interview participantsnorcollectdata,which can be entered directly into programme software for analysis.Additionally, the participantis able to complete questionnaires in their own time, and from their own home, thus reducing the burden on them.However, the financial cost associated with thisnew technology iscomplex, as these programsare extremely expensiveto set up, but reduce costs in the long term(Lombard et al., 2015). Moreover, these technologies rely on respondents being familiar with the programme and whilst they don’t appear to increase underreporting,(Moshfegh et al., 2008) such adaptation does not overcome self-report issues, despite the additional information and graphical representations of portion size that could in theory, reduce an aspect of underestimation bias. Therefore,perhaps further research comparing online and original methods are warranted. However, the standardization of data collection with this method, thus reducing interview bias (Lombard et al., 2015). Overall, although cost is an issue to be considered, as technology progresses, cost should decrease alongside this, thus reducing this issue. Finally, amajor issue in assessing nutrient intake is the difficulty with validating assessment methods, as this requires being able to compare assessment methods to other already established and validated methods that are without error. Such method seems not toexist. Lombard et al. (2015) states that without direct observation, validation studies cannot compare methods of assessment with ‘absolute truth’, highlighting an important issue in assessing nutrient intake. This comparison is further complicated by the lack of standardisation of methods, with questionnaires varying between and within countries. The impact on results of this is demonstrated in a study by Wirfalt, Jeffery and Elmer (1998) who compared two widely used FFQs, and found one version (the Block questionnaire) to show underestimation bias, whilst the other version (Willet FFQ) did not show the same underestimate bias. Whilst the Block questionnaire wasbetter at accurately assessing relative fat and carbohydrate intakes, the Willet FFQ was better at accurately assessing Vitamin A and calcium intakes. Although this issuein comparing methodsremains,and makes it difficult to collate different data to form general theories and hypotheses,the introduction and development of
biochemical, or biomarkers, is beginning to somewhat reduce this issue. A nutritional biomarker is said to be ‘an indicator of nutritional status that can be measured in any biological specimen’ that can be interpreted ‘broadly as a physiological consequence of dietary intake’ (Celis-Moraleset al., 2015). There are 3 types of biomarkers; recovery biomarkers, predictive biomarkers and concentration biomarkers (Zamora-Ross et al., 2012). These biomarkers provide an objective, biological measure to compare self-report with, thus improving some of the difficulties in assessing the accuracy and validity of self-report measures; this is widely used(Bingham, 2002). Recovery biomarkers include the ‘gold-standard’doubly labelledwater method (DLW), which is used to compare individuals’ recall of their overall energy intake both in FFQs (e.g. Collins et al., 2013) and 24-hour recall (e.g. Yunsheng et al., 2009). Other recovery biomarkers alsoused include total urinary nitrogen and potassium. These recovery biomarkers provide an estimation of absolute intake level over a specific period of time (Zamora-Ross et al., 2012). Predictive markers are also used for comparison where appropriate –these highly correlate with intake (Zamora-Ross et al., 2012) and examples include urinary sucrose and fructose levels as biomarkers for reported sugar consumption (e.g. Tasevska et al., 2011). However, although biomarkers offer anobjective means of comparison and thus avoid the error that come with subjective methods, they are not without error. Individual differences, such as genetics, body weight and environmental factors, in addition to the fact that many nutritional indicators are homeostatically controlled, means the relationship between nutritional intake and biomarker value if often complex, and rarely does a direct relationship exist (Thompson, Subar, Loria, Reedy & Baranowski, 2010). Therefore, whilst biomarkers allow for an objective measure of nutrient intake which can be compared to other self-report measures, which we already know to be bias, it is still undergoing extensive research.Furthermore, this method is extremely expensive. Whilst it may be an appropriate way to validate and assess the accuracy of self-report measures using this method, in reality it is too expensive for routine use of dietary assessment amongst large groups of people. In conclusion,there are many different issues in the methods used to assessnutrient intake. Whilst self-report measures tend to have low reliability due to their susceptibility to bias, this is largely influenced by the nutrient being measured, and strategies to reduce this bias have been identified. It is also clear that there are issues with comparison studies and the validation of assessment methods. Although more recent use of biomarkers to assess nutrient intake has provided an objective measure, and allows assessment of the bioavailabilityof nutrients, this
method is not without error and there is complexity surrounding its use to validate self-report measures. In general, nutrient intake assessment methods carry high cost, be it financial, or resource and individual burden. The evidence seems to suggest that the issues associated with measuring nutrient intake can be minimised if the appropriate method for the nutrients of interest is selected(Potischman, 2003). There are also suggestions that different methods of nutrient intake assessment should be used simultaneously (Elmadja & Meyer, 2014) in order to reduce issues. However, as previously described in this essay, the need to balance accuracy with cost and burden becomes an unavoidable issue within itself.