Transcript Max Feinberg

Metrological issues for food
consumption data.
Max Feinberg
INRA. Mét@risk
[email protected]
Max Feinberg
Using nutritional measurements to study
the potential link between nutrition and
disease and find methods to prevent the
disease.
The best the measurements, the best the
decisions…
Nutritional epidemiology
Max Feinberg
Foods
Exposures
A
c
ij


ai 
A
Foods

cij
ai
Contents qj
Consumers
Weights pi
=
Consumptions
qj
j
pi
a
i
i
N (i )
i : individual
(household)
j : food (group)
General nutritional measurement
Foods
Max Feinberg
Consumers
Weights pi
Consumptions
cij
Consumption measurement
Max Feinberg
24-hour Recall (24R)
Attempt to define and quantify food intake during a specific
day.
Dietary Records (DRs)
Detailed descriptions of types and amounts of foods and
beverages consumed, meal by meal, over a prescribed
period (3-7 days).
Food Frequency Questionnaires (FFQ)
Long-term diet over months or years, not just a few days.
Household Budget Survey (HBS)
Estimation of the average consumption of a population
based on point estimates of food consumptions.
Total Diet Study (TDS)
…
Estimation of consumer exposure to contaminants based on
the analysis of foods « as consumed ».
Consumption measurements (2)
Max Feinberg
cij
Weights pi
Individual
consumptions
Household
consumptions
Direct methods
mkj
Purchasing panels
i : individuals
k : households
j : foods
Survey reduction
Consumptions
hj
National budget
Uncertainty and variability
A proposal to evaluate
nutritional measurement
quality
Max Feinberg
Metrology: field of knowledge concerned with
measurement.
Metrology includes all theoretical and practical aspects
of measurement, whichever the measurement uncertainty
and field of application.
Measurand: quantity intended to be measured.
Uncertainty (of measurement): parameter that
characterizes the dispersion of the quantity values
that are being attributed to a measurand, based on
the information used.
International Vocabulary of Metrology
Max Feinberg
Reported
Value Y
y

y
n
True Value T
+
s
Y
Measurand
i
T
+
y
95% of replicates in the interval
2
(
y

y
)
 i
n 1
Does this
interval
represent
uncertainty?
y 1.96s
Analogy with chemical measurement
Max Feinberg
Milie
u
Environment
Personnel
Manpower
Method
Method
Calibration
Result +
Uncertainty
Traceability
Instrument
Equipment
Sampling
Samples handling
Material
Sources of uncertainty in analysis
Max Feinberg
Day 1
y1
+
yij
B1
Day 1
B2
Day 2
+
B1
y2
+
B2
B3
+
y3
Day 3
B3
Day 2
Standard deviation of:
sR 
reproducibility
sr2  s B2
Vary sources of uncertainty
Day 3
Max Feinberg
Reported
Value
Composed
standard uncertainty
True Value
+
Y
uc (Y )  sr2  sB2
T
Expanded uncertainty
U = kuc(Y)
p = 95% k = 2
p = 99% k = 3
Measurand
95% of the future
measurements
Standard and expanded uncertainty
Max Feinberg
Instrument
Milieu
Individual
Food
Material
Naming
Grouping
Grouping
Weight
Age,
Social category, …
Culture
?
Measurement
+
Uncertainty
Size
Competence
Location
Selection
Inquirer
Manpower
Panel
Duration
Study
Method
Uncertainty sources for nutritional
measurements
Max Feinberg
Consumption
pattern
=
Consumption
system
+
Variability
Individual
+
Uncertainty
Use multivariate techniques to identify
consumption systems
Consumption pattern
Max Feinberg
Ind 8
Ind 7
Ind 6
Ind 5
System 2
Fruits
Vegetables
Meat
Ind 4
Cereals
Dairy
Fish
Ind 3
Ind 2
Ind 1
System 1
0%
20%
40%
60%
80%
100%
Food groups (% Weight)
Simulated consumption systems
Max Feinberg
Needs for harmonisation.
Data collection procedures.
Data evaluation.
Identification of bias.
How far are methods complementary?
A proposal:
use uncertainty-variability approach to
organise harmonisation.
A new discipline: “Consumetrics” ?
Conclusions