262 Introduction to Human Nutrition
● factors that infl uence retention of dietary informa-
tion over time
● the ways in which individuals conceptualize foods
and food quantities.
Coding
Coding refers to the allocation of a specifi c code
to each food item. Since the nutritional content of a
food varies with different processing and preparation
methods, it is vital that the correct codes be assigned
to each food item. Coding errors arise when the
food that has been consumed is not described in suf-
fi cient detail to enable unambiguous allocation, by
the investigator, to a food category in a food composi-
tion table or database. Food frequency questionnaires
are often precoded to reduce the time needed for
coding and the possibility of coding errors (see Table
10.3). Making it easy for respondents to describe
foods with the level of detail required is therefore
an important consideration in study design. This is
increasingly diffi cult, particularly in industrialized
countries where the food supply now consists of
thousands of different manufactured foods, the names
of which are often no longer a good guide to their
nutrient content.
Coding errors are also likely to arise when more
than one person is involved in coding and there is
no agreed procedure and/or comprehensive coding
manual. Coding errors arising exclusively from
inadequate description of foods have resulted in
coeffi cients of variation ranging from 3% to 17% for
different nutrients. Note that a standard procedure
for coding foods, while minimizing differences
between coders (random error), can also introduce
bias if the coding decisions that are made are not
based on up-to-date knowledge of the local food
supply and food preparation methods. Gross errors
associated with weights of foods can be checked,
before analysis, by means of computer routines that
identify values outside a prescribed range and by
using data-checking techniques such as duplicate data
entry.
Use of food composition tables
Most dietary studies use food composition tables or
databases rather than chemical analysis to derive the
nutrient content of the foods consumed. Chapter 2
describes in detail the way in which data on food
composition are derived and compiled. The purpose
of this section is simply to review briefl y the kinds of
error that can arise as a consequence of using food
composition tables to calculate nutrient intake, com-
pared with chemical analysis of the diet, and which
can lead to both random and systematic errors.
Systematic error can result from:● the way in which results are calculated or
expressed
● the analytical method used
● the processing and preparation methods in common
use.Food composition tables for different countries often
use different ways of expressing results and different
analytical methods. The ways in which food items are
processed or prepared are also likely to differ and for
these reasons different sources will not necessarily
provide comparable data for the same foods.
Systematic differences, which may not necessarily be
errors (e.g., when foods are prepared differently in
different countries), often only become evident when
different food composition tables are used to evaluate
the same diets.
Random error arises from the fact that most foods
vary in their composition as a result of changes in
composition associated with the conditions of pro-
duction, processing, storage preparation, and con-
sumption. The random error associated with the use
of food composition databases generally decreases as
the size of the sample group increases. This may not
be true, however, in institutional settings where every-
one is likely to be consuming food from the same
source.
To compare calculated and analyzed data without
the complication of other sources of error it is neces-
sary that the diets are analyzed by collecting a dupli-
cate of what has been eaten at the same time as the
diet record. At group level it has been observed that
mean intakes calculated from the food tables are gen-
erally within approximately 10% of the mean ana-
lyzed value for energy and macronutrients, but not
for micronutrients. However, a large proportion of
individuals have values that fall outside this range.
In general, calculated and analyzed values for nutri-
ents agree more closely:● for groups than for individuals
● for macronutrients than for micronutrients
● when data for locally analyzed foods are used.