2.b Dietary assessment Flashcards
What are errors in dietary assessment?
Within persons: random and systematic errors
Between persons: random and systematic errors
What are Random within-person errors?
- natural, normal, day-to-day variation in intake
- reporting errors (mistakes, by accident ordeliberate) e.g. in portion size
- random = both over- and underestimation
What is a consequence of within person error?
Consequence:
- will not affect the mean intake of a population, but increases the SD of the population
- you need more (repeated) measures to get closer to the individual’s true value
What are Systematic within-person errors ?
= consistent over- or underestimation
- due to ‘technical’ issues: e.g. errors in food composition tables or food portion pictures
- Intake-related bias (proportional to intake)
- Specific for certain persons, e.g. men, obese, alcoholics etc.
What are Random between-persons errors?
- Caused by natural differences in intake (e.g. small/large portions)
- The more diverse the group, the larger the between-person SD
- Be selective in inclusion criteria to minimize between-person variation
- Becomes larger when random within-person variation is larger
Consequences of Random between-persons errors?
Consequence 1: % of population above or below cut-offs change
Consequence 2: Association with disease may disappear
What is Systematic between-persons error?
the observed population mean differs from the true mean
- due to ‘technical’ issues: e.g. errors in food composition tables or food portion pictures
- Intake-related bias (proportional to intake)
- Specific for certain persons, e.g. men, obese, alcoholics etc.
= consistent over- or underestimation
Consequences of Systematic between-persons error?
- the % of the population below or above a certain cut-off point is biased.
- correlation or regression coefficients are unaffected by systematic errors (ranking remains the same)
How to fix random vs systematic errors?
Systematic: correct population estimates accordingly
Not constant: regression equations to adjust population estimates
How are food assessment methods evaluated? (6 examples)
- Comparison with reference method
- Reproducibility
- Comparison with independent standard
- Correlation with physiologic response
- Ability to predict disease
- Goldberg cut-off (Only for determining energy underreporting)
What is a drawback for Comparison with reference method?
Drawback: errors may be similar or at least correlated
What does validity mean and is it the same as reproducability?
Not the same as validity! = measure true intake. Repeatability = get the same numbers each time.
What is an assumption for comparison with independent standard?
- Reference method has NO correlated errors (therefore ‘independent’ standard)
- Comparison of estimated intake with (Excellent) biomarker of intake, e.g. N excretion in urine/feces)
- Expensive; highly motivated compliant participates needed
What is meant with ‘correlation with physiologic response’?
Examples:
a. Decrease saturated fat intake should be reflected in a decrease in LDL-cholesterol
b. Increase folic acid: decrease homocysteine
c. Increase vit D, increase serum vit D
This is only rough estimate of quality assessment method.
What are 5 steps to choosing a method?
- What is the aim of the assessment?
- What type of information do you need?
- What is the target group?
- What is the reference period?
- How much time/finances/experience is available?
