What are common dietary questionnaires?
- 24-hour recall
- Food record or food diary
- FFQ - Food frequency questionnaire
- Brief screener
What are the typical effects of systematic error (bias)?
Think of this by type
All –> Inaccuracy: biased estimates of mean and distribution
Intake-related –> exaggerated relationships
Person-specific* –>
a) Attenuated relationships
b) Loss of power
*Due to characteristics of the person – ex: scientists have found that people with higher BMI tend to underestimate systematically what they had to eat.
What are the types of effects associated with random error (within-person)?
E.g. difference between one-day reported intake and long-term average reported intake
Not biased but may be imprecise
- Imprecision: estimates scattered around true value
- Distributions too wide, leading to overestimations of tail probabilities
- Attenuated relationships
- Loss of power
What is within-person variability and what is the impact on nutritional studies?
Within-person variability is the natural variation in what a person eats from day to day, which can reduce the accuracy of dietary intake measurements in nutrition studies.
Ideally we want to reduce within-person variability in order to improve accuracy.
How to reduce?
Validate using other measurement instruments, including imperfect ones.
- “Data from multiple administrations (at least two) of each instrument is needed to assess within-person variability
- Caution! Validation might be overoptimistic, as the same type of measurement error is likely to affect both questionnaires (i.e. FFQ and 24-h recall)”
[Michaelopoulou lecture]
How can we address measurement error in the design phase?
“Do a validation study
- The only way to know how good your measurement tool is – quantifies the accuracy and precision
- Modify the tool after validation, if needed
Include bigger sample
- Measurement error leads to lower power
Repeat measurement of diet
- In full cohort or in a sub-sample
- Less random error with more measurements
- Diet changes over time”
[Michaelopoulou lecture]
How can we address measurement error in the analysis phase?
“Averaging repeat observations
If not possible to have repeat measurements, use statistical modelling to adjust for the effects of the day-to-day variation (given at least two measurements on at least a portion of the sample)
Exclusion of participants with unreliable results
- E.g. energy intake outside of a pre-defined range
Energy adjustment
- Adjustment may cancel out some of the measurement error, allowing for the fact that some participants just report a lot/a little of everything
Use information from a validation study
- E.g. attenuation factor to correct measures of association
Use a second measure of diet and combine with main instrument data
- E.g. estimates from second measure as covariates”
[Michaelopoulou lecture]
What are macronutrients?
They contribute energy:
1. Fats
2. Carbs
3. Protein
4. Alcohol
What are micronutrients?
They are essential for body function:
- Vitamins (A, B, C, D, E, K)
- Minerals (Calcium, Iron, Magnesium, Sodium, Potassium)
What are DRVs?
Dietary reference values:
- Estimated Average Requirement (EAR)
- Reference Nutrient Intake (RNI)
- Lower Reference Nutrient Intake (LNRI)
- Safe intake
- Population average
What is EAR?
Estimated Average Requirement (EAR)
- Half of people need more, half need less
- For energy, protein, vitamins, minerals
What is RNI?
Reference Nutrient Intake (RNI)
- Amount which is enough for 97% of people
- Calculated as EAR + 2xSD
- For protein, vitamins, minerals
What is LNRI?
Lower Reference Nutrient Intake (LNRI)
- Intakes below this are almost certainly inadequate for most people
- Calculated as EAR – 2xSD
What is safe intake?
“A level of intake that is judged to be adequate and safe for most people, when there isn’t enough data to calculate an EAR, RNI, or LRNI.”
[ChatGPT based on lecture]
What is population average?
Consistent with good health, mostly used for total fat, SFA, total carbs, sugar, fibre
Expanded by ChatGPT:
An observed average intake in the population that is consistent with good health across a group, not based on physiological requirements.
Used for:
- Macronutrients (e.g. total fat, saturated fat (SFA), carbohydrate, sugar, fibre)
- Nutrients where there isn’t a clearly defined individual “requirement,” but we can still observe what’s associated with health at a population level
⚠️ Important:
- It’s not meant for assessing adequacy at the individual level.
- It’s used in dietary guidelines and public health targets, e.g., “average saturated fat intake should be no more than 10% of total energy.”
What is change in energy stores?
Change in energy stores= Energy intake (EI) − Energy expenditure (EE)
What is total energy expenditure (EE)?
Total EE=Basal Metabolic Rate (BMR)+Physical Activity+Thermal effect of food
What is one signal that someone might be under-reporting their energy balance?
If their
Energy intake (EI) / Basal metabolic rate (BMR) < Physical activity level (PAL)
then they might be under-reporting their energy intake.
A person’s reported energy intake (EI) is too low compared to how much energy their body is expected to need based on their basal metabolism (BMR) and activity level (PAL).
