Professional Documents
Culture Documents
Choice of Methods To Assess Dietary Intakes
Choice of Methods To Assess Dietary Intakes
Dunedin, NZ
Choice of methods to assess dietary What will you learn from this session?
intakes of individuals or populations
• Factors influencing choice of dietary methods
• How to achieve four levels of study objectives
• What objectives can be met at each level of
objectives?
Rosalind S Gibson, • Sources of uncertainty in true usual intakes
Emeritus Professor, o measurement errors
Department of Human Nutrition, o true variation in usual intakes
University of Otago, Dunedin, New Zealand • Other factors to consider
RSG RSG
Level 1: Average intake of a population group Level 2: Proportion of the population ‘at risk’
Single 24-hr recall/record per person Measure 24-hr recalls/records on two non-
Ensure all days of the week proportionately consecutive days or three consecutive days
represented
Size of study group: depends on the precision Measure 24-hr recalls/records for two days on at
required and the between-subject variation least a subsample (40-50 per stratum) of the
participants
RSG RSG
1
Choice of methods to assess dietary intakes Rosalind S Gibson, University of Otago,
Dunedin, NZ
RSG RSG
What objectives can be met with Level 1 data? Median (IQR) nutrient intakes of Ethiopian
children from complementary foods
Describe usual mean or median nutrient intake for a
group with a certain precision Nutrient Age 9-11 months Age 12–23 months
Demonstrate a significant difference in mean or median Energy (kcal) 169 (80, 291) 292 (193, 402)
intakes between two groups Protein (g) 3.6 (1.6, 7.3) 7.3 (4.8, 9.7)
Demonstrate a significant change in group mean Calcium (mg) 7.6 (3.6,14.2) 22.4 (10.7,112.6)
intakes based on paired measurements (eg: before
Iron (mg) 2.1 (0.9, 3.9) 4.5 (3.0,6.6)
and after intervention)
Zinc (mg) 0.9 (0.3, 1.9) 1.8 (1.4, 2.3)
Demonstrate a significant change in group mean
Niacin (mg) 0.64 (0.26, 1.35) 1.37 (0.90, 2.15)
intakes based on unpaired measurements
RSG RSG
From Abebe et al. (2008)
Median (IQR) nutrient intakes of two groups of Change in mean intakes (95th CI) over time in NZ
Malawian children: post-intervention toddlers based on paired measurements
Utilizable protein (g) 26 (20, 32) 204, 28)** Energy (kJ) 4126 (3917, 4334) 4093 (3864, 4322)
Animal protein (g) 8.9 (5.1, 12.7) 5.1 (14. 28)**
Protein (g) 41.1 (38.4, 44.1) 40.5 (37.9, 43.3)
Animal source (% energy) 5 3**
Fat (% energy) 11 (8. 15) 9 (2.9, 9.0)** Calcium (mg) 898 (821, 975) 782 (703, 861)
Heme iron (mg) 0.51 ( 0.33** Vitamin C (mg) 33.7 (27.9, 40.7) 39.0 (32.8, 46.5)
[Phy]:[Zn]* molar ratio 16 (12, 20) 20 (15, 23)**
Iron (mg) 4.7 (4.1, 5.3) 5.4 (4.7, 6.1)
[Phy]: [Zn]* < 15 (%) 23 12**
Meat / fish /
Analysis via Mann-Whitney-U-test; ** p<0.05 4.1 ( 3.1, 5.3) 10.0 (8.1, 12.4)
poultry (g)
From Yeudall et al. (2005) RSG From Szymlek Gay et al.(2009) RSG
2
Choice of methods to assess dietary intakes Rosalind S Gibson, University of Otago,
Dunedin, NZ
What objectives can be met with Level 2 data? Assessing prevalence of inadequate intakes after
adjusting distribution of observed to usual intakes
Determine distribution of intakes within the group to
assess proportion at risk of inadequate intakes Adjust 1-d intakes to usual intakes
to remove effects of within-
Determine a significant change in proportion at risk of
subject day-to-day variation
inadequate intakes before and after an intervention
Select appropriate EAR
Determine a significant difference in proportion at risk Use cutpoint method for Zn to
to inadequate intakes between two groups asses percent with inadequate
intakes
Assess risk of inadequate intakes of a nutrient in
Results in a reduction in likelihood
specific sub-groups defined by variables such as
of over- or under-estimating
age, sex, geographic region (urban vs. rural), prevalence of inadequate intakes
household composition, SES, etc
RSG RSG
See IZiNCG Brief No. 3
What objectives can be met with Level 3 data? Median daily intake of dairy products by decile
versus mean (SEM) PDA in adipose tissue
Assess relationship between frequency of daily
• Pentadecanoic acid (PDA) level
intake of food groups (e.g. deciles (times/d) of in adipose tissue correlates with
dairy products) versus mean level of a intake of dairy products via decile
biomarker (eg. adipose pentadecanoic acid (r= 0.31; p<0.01) based on a FFQ
(PDA)) in adults (n=503) in Costa Rica
where intake of dairy products is
rather low
Divide individuals into terciles of nutrient intake
(eg. cholesterol) and calculate the • Age, sex, BMI, smoking status
corresponding average biomarker level (eg. taken into account in analysis
plasma cholesterol) for each tercile
RSG From Baylin et al. (2002) RSG
3
Choice of methods to assess dietary intakes Rosalind S Gibson, University of Otago,
Dunedin, NZ
What objectives can be met with Level 4 data? IYCF & sentinel food indicators at 9 mos in relation to
LAZ scores of Sumedang, W Java infants at 12 mos
Assess inter-relationships between nutrient intakes of
individuals to other indices of nutritional status in same persons IYCF indicators
ᵝ 95%Confidence I p-value
Sources of true variation in usual intakes Other factors to consider when selecting a
method for measuring dietary intakes
Between-subject variation: Depends on study group & nutrient. Characteristics of the study group and setting: Level of
Reduce by increasing number of subjects literacy and numeracy; Age group:(elderly problems with
Within-subject variation: Reduce with multiple days of intake per memory)
subject
Respondent burden: Weighed intakes in home: high respondent
Day-of-the-week effects: weekends: market days. Avoid by burden
proportionately including these days in study design
Available resources: Experience of research assistants?
Seasonal effects: Greater for intakes of foods than nutrients
Availability of dietary scales; Availability of local food
(except Vit A). Accounted for by using random days
representative of all seasons composition data
Training effect: May be severe if recalls/records on consecutive Validity and reproducibility of the proposed method: Must be
days. Avoid by completing on randomly selected non- confirmed for population group and study setting before
consecutive days adopting any method
RSG RSG
4
Choice of methods to assess dietary intakes Rosalind S Gibson, University of Otago,
Dunedin, NZ
Key References