Student Skills Training

Aims

• Know who to get advice from
• Be familiar with dietary data in the Unit
• Have practical info for carrying out analyses

Core Nutrition Group

Nita Forouhi

Fumiaki Imamura

Laura O'Connor

Ju-Sheng Zheng

Tammy Tong

Eirini Trichia

Nutrition & Aetiology

Zheng Ye

Sherly Li

Measurement Toolkit

Simon Wheeler

Interconnect

Silvia Pastorino

Measuring Diet

Buy?

Behaviour?

Untitled Slide

Dietary intake

• Foods
• Food groups
• Nutrients
• Bioactive dietary constituents

Dietary intake assessment tools

Choosing a tool

• Fit for purpose
• Error
• Validity

Untitled Slide

Diet questionnaire

Diet questionnaires

• Sabre Study
• Information on intake of some foods
• not validated
• not comprehensive
• can't compute energy/nutrient intake
• useful for adjustment when diet not main exposure
• Can be included in scores

Food frequency questionnaire

FFQs

FFQs

• Interact/EPIC, Fenland 1 & 2, Addition
• 131 food list
• ranks intakes - good at diet-disease ass.
• validated for food, energy & nutrient
• supplement info collected but not coded
• Not appropriate for micronutrient analyses
• Not appropriate to estimate abs. intakes

24 hr recall

24hr recalls

• Fenland 2
• self-reported (vs interview)
• estimates abs intakes foods, energy, nutrients
• not usual intake (can do multiple)
• good for long term exposure & variability
• good for assessing change

Food diaries

Food diaries

• EPIC Norfolk
• almost gold standard
• estimates absolute intake
• potential consumers/true non-consumers
• good for teasing apart diet-disease
• ideal for dose-response

All measurements of food intake are subject to error



Independent validation
Assess error structure - account for it in analysing & evaluating data

Sources of error #1

• messages from researcher
• attitudes about food
• socioeconomic status
• image management
• under/over reporting
• altered food choice

Sources of error #2

• weighing errors/ poor estimation
• Inadequate communication or descriptions
• poor memory
• false perception of own diet

Source of error #3

• poor quantification: questionnaire choices, subject perception
• Food table values differ from actual composition
• coding errors (multiple researchers)

Untitled Slide

Validity #1

• Gold standard: 14 day weighed food record

Validity #2

• 7 day diaries: validated against gold standard, considered valid for use in studies without further validation or references

Validity #3

• 4 day diaries & 24hr recalls: considered valid but need refs for e.g. how weekend days are treated & number of recalls

Validity #4

• FFQs need to have validation studies. You should always include the validation study r for your exposure in your manuscript.

Nutritional composition data

Untitled Slide

• Nutrients per 100g of food
• Food to nutrient conversion programs: FETA, EPIC Soft
• Composition of foods integrated dataset (CoFID)

Untitled Slide

• Food intake (g) as estimated by the measurement tool * nutrient per 100g of food = intake of nutrient from food

Example

• beefburger, once a week
• assign frequency per day = 0.14
• assign portion size of burger = 90g
• code as 19029 Beefburgers 98-99% beef, chilled/frozen, fried veg oil
• link to compositional data = 15.9g protein per 100g 19029
• (15.9 * (90/100))*0.14 = 2g protein per day from burger

Population & time appropriate

“The EPIC-Norfolk FFQ was developed in 1988. Its food list & portion sizes represent those of an adult population who follow a traditional UK diet”

FETA (FFQ EPIC Tool for Analysis)

• Open Source
• http://www.srl.cam.ac.uk/epic/epicffq/
• Methods paper: Mulligan, Luben, Bhaniani, Parry-Smith, O'Connor et al (2014) BMJ Open.

FETA (FFQ EPIC Tool for Analysis)

• Default: McCance & Widdowson 5th Ed
• Standard output: 41 nutrient & 14 food group intakes (g/d)
• Fully disaggregated output

FETA

• In house processing
• Fenland & Addition
• manipulate the data e.g. iron from meat
• add compositional data e.g. fatty acids
• add non-compositional, food related data e.g. GHG

36 flavonoids (not available in McCance)
Free sugar (no stand. calculation mthd)
DASH adherence score (study specific, includes researcher decisions)
Food price (method dev. by Pablo)

Merging nutritional data

• EPIC Europe: 10 European countries (Interact – 8 of these)
• Country specific dietary assessment tools
• Streamlined into compatible variables
• Merged into nutrient & food grp intakes

Pros:
1. much larger sample size
2. findings are more generalisable

Cons:
1. heterogeneous food groups mask associations
2. composition of the same foods may differ by region e.g. fish

Combining dietary assesment tools

all dietary assessment tools have limitations

Untitled Slide

• FFQs: habitual frequency of intake of commonly consumed foods
• 24hr recalls: portion size, detail on the variety of food consumed, capture intake of episodically consumed foods

Untitled Slide

• Multiple source models can be used to combine data to maximise the advantages of each tool
• NCI method
• MSM method
• Monte Carlo simulation
• Fenland 2: 1 FFQ & 1 24hr recall

Nutritional Biomarkers

intake

status

Intake

• objective measurement of intake
• Plasma vit C
• marker of citrus fruit intake in previous 2-3 wks
• easily measured, relatively low cost
• has been associated with better dietary quality
• may be used as a proxy

Status

• Circulating 25-hydroxyvitamin D
• reflects sun + intake
• Biomarkers of status don't reflect intake!

Distributions

Energy

normal distribution, expected

Iodine

ubiquitous in food supply

Kaempferol

distinct foods (tea, sprouts, witch hazel)

ergocalciferol

skewed: reflects inadequate or excessive nutrient intake

fruit juice

non-consumers

consumers & non-consuemrs

• zero-limited data
• true zeros?
• never consumers?
• Consider effect of dietary assessment method
• Comparison group?
• Consider the end message, role of food in diet (e.g. burgers vs all meat)

4d diary
23749 non-cons
1497 cons

What could this mean?

• only consumed by a sub-group of the population (alcohol)
• periodically consumed foods - certain times of the year (sprouts)
• consumed at weekends (roast dinners)

What can we do about it

• Use 2nd dietary assessment method to identify true non-consuemrs
• MSM not NCI
• Analysis options
• ntiles restricted to consumers only
• non-consumers, low-consumers, high consuemrs

Energy

“Failure to account for total energy intake can obscure associations between nutrient intakes and disease risk or even reverse the direction of association.” Walter Willet

Untitled Slide

• confounder, mediator, indep associated
• or all 3 e.g. SSB & T2D
• those that consume more energy tend to consume more total food & as such more of the food or nutrient exposure
• mediator - sugar intake associated with weight gain via contribution to energy intake

Common methods

• Regression method: regress nutrient/food on energy, include residuals as exposure
• Nutrient density method: divide nutrient/food by energy
• Partition method: adjust for energy from all other sources

Dietary quality

Suggestions for adjusting for overall DQ

• Simple, suitable for most situations: Energy, fibre (NSP), fruit, vegetable, fish intake
• Targeted, suitable for examining diabetes as an outcome: Energy, fibre (NSP), fruit, vegetable, processed meat, red meat, fish

Suggestions for adjusting for overall DQ

• Dietary energy density (kcal/weight of diet) simple calculation, data determined, proxy for the nutritional quality of the total diet
• Healthy Eating Index (HEI) complex calculation, researcher determined score for dietary quality of the diet

Available in the Fenland Study

• DASH adherence score classification according to intake of certain foods/nutrients
• Mediterranean Diet Pattern classification according to intake of certain foods (incl. olive oil)

Terminology

Sugars

• Free sugars: WHO definition – extracellular sugar
• Added sugars: American term introduced to help people make healthy dietary choices – no chemical definition , cannot be estimated
• Sugar: table sugar or sucrose ONLY

Fibre

• misnomer
• Much fibre is not fibrous
• Can mean dietary fibre or non-starch poly saccharides
• Usually NSP in UK, DF in US
• DF = NSP + resistant starches, lignans, waxes, gums and others

Energy

• Food energy = energy from carbs + protein + fats
• Total energy = food energy + energy from alcohol
• alcohol the nutrient, not alcoholic beverages!!

Miscellaneous

Atwater figures
Protein *4
Fat *9
Carbohydrate * 4

Sugars * 3.75
Alcohol * 7
Fibre ?

Conversion of kcal to kJ should always use 4.186 never 4.2

Intakes of supplements may or may not be included

Using FFQ to assess micronutrient intake can be inappropriate

Standards in reporting nutritional sciences:

http://journals.cambridge.org/download.php?file=%2FPNS%2FPNS31_02%2FS002966...

If using the residual method for energy adjustment be careful of non-consumers when making ntiles or scores

Untitled Slide