Background
Methods
Participants
Data collection
Generating a consensus
Results
Participants’ characteristics from Delphi rounds
Country | Delphi I (n = 48), n (%) | Delphi II (n = 51), n (%) |
---|---|---|
UK | 19 (39%) | 20 (39%) |
USA | 7 (14%) | 6 (11%) |
Australia | 6 (12%) | 5 (9%) |
Canada | 2 (4%) | 4 (7%) |
France | 2 (4%) | 3 (5%) |
Brazil | 2 (4%) | 2 (4%) |
Netherlands | 2 (4%) | 2 (4%) |
Italy | 2 (4%) | 2 (4%) |
Belgium | 1 (2%) | 1 (2%) |
Japan | 1 (2%) | 1 (2%) |
Norway | 1 (2%) | 1 (2%) |
Spain | 1 (2%) | 1 (2%) |
Greece | 1 (2%) | 1 (2%) |
New Zealand | 1 (2%) | 1 (2%) |
Serbia | 0 (0%) | 1 (2%) |
Consensus on the Best Practice Guidelines
Best Practice Guidelines
E/Da
| Stage I. Define what you want to measure in terms of dietary intake: the key a priori considerations to guide your choice of the appropriate type of dietary assessment tool (DAT) | |
1
|
What? — Characteristics of the main dietary component of interest
| |
E | 1.1 | Clearly define what needs to be measured (e.g. intake of energy, food groups, specific or a range of macro- or micro-nutrients) |
E | 1.2 | Determine how the dietary data will be analysed and presented (e.g. total daily or meal level intakes, food groups or nutrients) |
2
|
Who? — Considerations around the characteristics of study participants
| |
E | 2.1 | Define the target sample in terms of characteristics (e.g. life stage, ethnicity, health status, body mass index (BMI), socio-economic level, country/region and setting — home, school, hospital) |
E | 2.2 | Identify other issues that could affect the choice of DAT (e.g. literacy, numeracy, language, cultural, disability, time or familiarity with technology) |
E | 2.3 | Consider the study sample size required in relation to the level of variation of your dietary component of interest and study power |
3
|
When? — Time frame considerations
| |
E | 3.1 | Are you interested in ‘actual’/short-term (hours or several days, up to one week) or ‘usual’/long-term intake (e.g. months or years)? Consider what reference period (e.g. daily, weekly, monthly, yearly) would be best suited to your dietary component of interest |
E | 3.2 | Will data collection in your study be retrospective or prospective? |
Stage II. Investigate the different types of DATs and their suitability for your research question | ||
4
|
Consider and appraise the different DAT types
| |
E | 4.1 | In relation to your research question, consider the suitability, strengths and weaknesses of different DAT typesb
|
E | 4.2 | Think about participant burden (e.g. study participants’ potential willingness, time, ability, ethical considerations, interest in using different tools and access issues associated with different DATs) |
E | 4.3 | Identify the availability of resources (e.g. staff skill, time, finances) |
Stage III. Evaluate existing tools to select the most appropriate DAT | ||
5
|
Research and evaluate available tools of interest
| |
E | 5.1 | Read any available published validation studies: |
• Has the DAT been evaluated to measure the dietary component you are interested in? | ||
• Has the DAT been evaluated in a population similar to your population of interest? | ||
• Is the nutrient database used appropriate? | ||
• Are the portion sizes used relevant? | ||
D | 5.2 | Assess the quality of validation in terms of: |
• Has the DAT been compared to an objective method (e.g. biomarkers)? | ||
• Has the DAT been compared to a subjective method (e.g. a different self-reported diet assessment)? | ||
• What were the limitations of the validation study? | ||
D | 5.3 | The strength of agreement between the two methods: |
• Is there any evidence of bias; do the methods agree on average? | ||
• Is there any evidence of imprecision; how closely do the methods agree for an individual? | ||
6
|
If, based on the validation studies, none of the existing DATs is entirely or wholly suitable, consider the need to modify or update an existing DAT, or create a new DAT and evaluate it
| |
E | 6.1 | Decide whether an existing tool can be improved. Investigate whether: |
• Foods and portion sizes included are characteristic of your target population, and frequency categories are appropriate | ||
• The time period that the questionnaire refers to could be modified to better suit your needs | ||
D | 6.2 | Consider the face validity of existing tools. Is there evidence the tool has been used to measure dietary intake in your population of interest? |
D | 6.3 | Updated or modified tools may require re-evaluation. Consider if validation can be integrated into your study |
Select your DAT | ||
Stage IV. Think through the implementation of your chosen DATs | ||
7
|
Consider issues relating to the chosen DAT and the measurement of your dietary component of interest
| |
E | 7.1 | Obtain information regarding DAT logistics (e.g. tool manual, relevant documents and other requirements from the DAT developer) |
E | 7.2 | Check that the chosen DAT has the most appropriate food/nutrient database and software |
E | 7.3 | Check the requirements for dietary data collection (e.g. entry, coding and software) |
D | 7.4 | Consider collecting additional related data (e.g. was intake typical, supplement use) |
8
|
Prepare an implementation plan to reduce potential biases when using your chosen DAT
| |
E | 8.1 | Consider potential sampling/selection bias and track non-participation/dropout/withdrawal at different stages |
E | 8.2 | Minimise interviewer bias (e.g. ensure staff qualifications and training are appropriate, develop standardised training protocols and monitoring procedures) |
E | 8.3 | Minimise respondent biases (e.g. use prompts, clear instructions) |
E | 8.4 | Quantify misreporting |
Pre-study: what is your research objective?
Stage I. Define what you want to measure in terms of dietary intake: the key a priori considerations to guide your choice of the appropriate type of dietary assessment tool (DAT)
1 What? — characteristics of the main dietary component of interest
2 Who? — considerations around the characteristics of study participants
3 When? — time frame considerations
Stage II. Investigate the different types of DATs and their suitability for your research question
4 Consider and appraise the different DAT types
Stage III. Evaluate existing tools to select the most appropriate DAT
5 Research and evaluate available tools of interest
-
Has the DAT been evaluated to measure the dietary component you are interested in?When possible, validated DATs should be used; however, validation should be relevant to the foods/nutrients of interest. It is important to check how well the DAT performed in the validation study for the food or nutrient of interest.The validity of a DAT will depend on accurate estimation of frequency and portion sizes, on the quality of the nutrient database and in the collection of data [43]. Measurement of absolute validity is difficult to establish, requiring the comparison method to be an objective measure such as recovery biomarkers, e.g. doubly labelled water. Relative validity (the comparison of two instruments of the same kind [30]), through use of multiple DATs, is more commonly used to detect bias [44].
-
Has the DAT been evaluated in a population similar to your population of interest?Determine whether validation studies support the use of the candidate DAT for your study population. Population characteristics/covariates to be considered are life stage, ethnicity, cultural differences in diets, geographical area, education/literacy, age range, sex, types of diets and relevance of foods consumed at the time the DAT was validated [45, 46].
-
Is the nutrient database used appropriate?The nutrient database used should be appropriate, comprehensive and up to date for the study population. Limited coverage of foods in the database, missing nutrient data, differences in software packages, incompatibility of databases [45], recipe, portion size allocations and bias in variability in recipes should be considered. This may be more difficult for processed foods due to the complexity of the food market and its rapid changes; most nutrient databases do not capture data on food reformulation. Composite dishes, either purchased or homemade, can vary due to differences in recipes. Weighing recipe ingredients is more practical than chemical analysis [47]. Standardised calculation procedures should take into account weight loss during cooking and nutrient losses into cooking water [48]. Nutrient retention factors may be applied to calculate the nutrient composition of a cooked food from the uncooked food [49]. Limitations and gaps in food composition tables need to be considered for coverage of nutrients. For example, total fibre is available in most food composition tables, but results differ according to the chemical analyses method used [50]. Sub-components of fibre, such as soluble and insoluble fibre, may not be available.
-
Are the portion sizes used relevant?Accurate estimation of food portion sizes is important; errors are often introduced due to incorrect portion size quantification or use of an ‘average’ portion size [51]. Food photographs or food models can be provided; however, they only provide a limited number of foods and food portion sizes [52]. Portion size measurements should be relevant to the study population, characteristics and life stage. The type of food will influence reliability of portion size estimation; pre-packaged foods will have a weight declared which could be recorded. Participants’ perception of portion sizes from photographs or ability to conceptualise amounts along with memory limitations will affect the precision of portion size recording [51].
-
Has the DAT been compared to an objective method (e.g. biomarkers)?Objective methods to assess nutrients include clinical indicators or biomarkers [53], which vary in response to intake [30]. Biomarkers can reflect intake over the short term (past hours/days), medium term (weeks/months) and long term (months/years), depending on the sample type, e.g. blood, hair [8]. Ideally all DATs should be validated against an objective measure of intake. Recovery biomarkers such as 24-hour urine nitrogen and potassium excretion and doubly labelled water reflect absolute nutrient intake over a short time [54]. These are the best approaches to use for absolute validity of the tool. Predictive biomarkers (e.g. urinary fructose, sucrose and dietary sugars) have a lower overall recovery, and concentration biomarkers (e.g. serum carotenoids) correlate with dietary intake [55, 56]. Predictive biomarkers may be useful for validation studies; however, since concentration biomarkers cannot be translated into absolute levels of intake, they are less reliable for validation studies. Concentration biomarkers may be used for estimation of diet-disease risk associations as a substitute for or as complementary to dietary assessments [57]. Recovery biomarkers provide an estimate of absolute intakes as they are based on the concept of the metabolic balance between intake and excretion over a period of time, but only a few are known [58].
-
Has the DAT been compared to a subjective method (e.g. a different self-reported diet assessment)?Although comparison with an objective method is preferable in terms of assessment of validity, this may not be available since these studies are costly and difficult to undertake. Comparison with an alternative form of dietary assessment is referred to as ‘relative validity’. However, comparison of one DAT against another risks correlated error between dietary assessment methods [30]. Any new dietary assessment should be compared against a more established method with greater face validity [59].It should be noted that the 7-day weighed food record was regarded as the ‘gold standard’ until studies that validated weighed food records with doubly labelled water found high levels of under-reporting [30]. Despite this, food records have been used as a standard to gain an insight into regular food intake [60], and they are often regarded as the most precise method for estimating food or nutrient intake [61].In addition to validity, test-retest reliability or reproducibility may also be relevant where diet is being measured at multiple time points.
-
What were the limitations of the validation study?The comparison DAT used in the validation study also needs to be assessed in terms of scope, the time frame/number of days, the main type of measurement error, memory requirements and also an assessment of cognitive difficulty. For an FFQ that is being validated, the agreement with an alternative method will be higher if multiple days of reference data have been collected. Furthermore, to measure within-individual variability, 2 or more days of dietary intake are required, from at least a sub-set of the population [30].When considering a validation study, it may be helpful to use a scoring system [14]. The authors in the study by Serra-Majem et al. [14] have developed a scoring system (0 = poorest quality to 7 = highest quality) for validation studies. This was based on the sample size, the statistics used, the data collected, seasonality and the inclusion of supplement measures. The authors identified issues relating to the poor quality of validation research: inadequate description of study details such as the respondent characteristics; design of the questionnaire; and adequacy of the reference data. Studies which reported relative validity, i.e. comparing two self-reported measures of diet, scored less than those which compared a self-report with a biomarker.
-
Is there any evidence of bias; do the methods agree on average?Consider the extent to which a DAT under- or over-estimates dietary intake compared to another, possibly better DAT. This can be described using the Bland-Altman technique [62] for method comparison. The mean difference of the two methods of measurement is plotted against the average. Errors associated with dietary intake may be correlated, and this can lead to overinflated agreement between methods. In general, the use of correlation as a method of comparison is not recommended, since it does not measure agreement between methods. Other statistical tests are also used in dietary assessment method validation [63], for example, the method of triads [64], which evaluates the association between three measurements: the test method, the reference method and a biomarker. This method calculates the validity coefficient between the observed and ‘true’ dietary intake and assumes a linear correlation between the three variables, for example, validating an FFQ (measuring carotenoid and vitamin E intake) using weighed food records and plasma biomarkers [65]. This method has limitations, whereby it is possible to generate validity coefficients greater than one [66].
-
Is there any evidence of imprecision; how closely do the methods agree for an individual?Precision provides a measure of the closeness of two methods for estimating diet for the individual [30], assessed over the whole sample. A DAT is considered precise if the estimated intake from the tool is close to the estimate from the reference tool, taking account of bias. The Bland-Altman technique also assesses precision with limits of agreement between the two DATs.
6 If, based on the validation studies, none of the existing DATs is entirely or wholly suitable, consider the need to modify or update an existing DAT, or create a new DAT and evaluate it
-
Foods and portion sizes included are characteristic of your target population, and frequency categories are appropriateFood consumption patterns change over time, influenced by income and socio-cultural preferences [67]. The DAT selected should be applicable to the population of the study. Investigate whether the food list and portion sizes used in the DAT are current.
-
The time period that the questionnaire refers to could be modified to better suit your needsAlteration of the time period the FFQ measures must be done with caution, as this could affect the validity of the tool and may require the FFQ to be revalidated. As one example, if an FFQ assesses the diet for 3 months, it could be converted to 12 months if the study was focusing on a nutrient that has seasonal variability.