Caballero B. (ed.) Encyclopaedia of Food Science, Food Technology and Nutrition. Ten-Volume Set

Подождите немного. Документ загружается.

it is recognized that avoiding excess is also important.

Intakes of nutrients that are far in excess of the RDA

can disturb body functions and cause acute, progres-

sive, or permanent disability. Some examples include

fluorosis, hypervitaminosis A, and hypervitaminosis

D. The DRIs now also address risks of possible excess

of inadequate nutrient intakes. The tolerable upper

intake level (UL) of a nutrient is the highest level of

chronic, usual daily nutrient intake that is likely to

pose no risk of adverse health effects to almost every-

one in the population. Below the UL, individuals

should be able to biologically tolerate this amount

of the nutrient. Excesses of each nutrient may have

many possible adverse effects. The UL is set for the

effects that experts believe are most likely to be preju-

dicial for health and for which sufficient data are

available. The UL is set using qualitative and quanti-

tative evidence and inference-related judgments of

experts in a logical process that assesses the associ-

ations between nutrient ingestion and the probability

of adverse effects. The UL is set by determining the

levels at which no observed adverse effects are noted,

or the lowest level of intake associated with observed

adverse effects. An uncertainty factor is then applied

to insure that even very sensitive persons would not

experience adverse effects at the UL dose chosen. For

many nutrients, data on the adverse effects of large

amounts of nutrients are simply unavailable, or they

are so limited that a UL cannot be determined. The

lack of a UL does not mean that the risk of adverse

effects is nonexistent from high intakes; it simply

means that data are not yet available, and the verdict

is unknown. Therefore, caution is warranted in

consuming high amounts of nutrients. The UL is not

intended to be a recommended level of intake. There

is no established benefit for healthy individuals of

consuming nutrient levels above the RDA or AI. The

amounts of individual foods that most people eat rarely

reach levels that are likely to exceed the UL. In contrast,

nutrient supplements provide more concentrated

amounts of nutrients per dose, and thus the potential

risk of excess is somewhat greater with them.

Current Estimates of Nutrient

Requirements for Vitamins and Minerals

0020 The EAR, RDA, AI, and UL all have specific uses in

dietary assessment, for making nutritional recom-

mendations and for dietary planning for individuals.

The EARs for selected nutrients and age groups are

provided in Table 1. Note that, for all of these nutri-

ents, the estimated average requirements are based on

oral intakes of nutrients. The appropriate use of the

reference values is described in detail in the various

publications of the Food and Nutrition Board.

Water

0021For adults under usual conditions, 1–1.5 ml of water

per kcal energy expenditure is sufficient to allow for

normal variations in physical activity levels, sweating,

and solute loads of the diet. If renal function is normal,

and solute intakes are adequate, the kidneys adjust to

increased intakes of water by excreting the excess. If

water intakes are inadequate to cover obligatory

urine outputs, hydration status can be compromised.

Obligatory water losses include 50–100 ml per day in

the feces, 500–1000 ml by evaporation or exhalation,

and, depending on the renal solute load, 1000 ml or

more in the urine. If external losses increase, intakes

must increase accordingly to avoid dehydration.

0022The stage of life and health influence water require-

ments. Pregnancy increases water needs by perhaps

30 ml per day, but during lactation, milk production

increases requirements to at least 1 ml for each milli-

liter of milk produced or to 1 l per day or more,

depending on milk production. In addition, water

deficits may arise if disease causes increased losses,

damages the kidneys, requires the use of diuretics, or,

as in coma, makes it impossible for the individual to

communicate their thirst. Fever increases water losses

by approximately 200 ml per degree Celsius, and

diarrhea losses vary but may be as great as 5 l per

day in severe diarrhea. Vomiting and heavy sweating,

especially in hot, dry climates and at altitude, also

increase water losses substantially.

Energy

0023Recommendations on energy and the macronutrients

are expected from the DRI process in 2002. Energy

requirements or inputs must match energy outputs for

weight to remain stable. The major sources of energy

output are resting energy expenditure (REE) and

physical activity. Minor sources include the energy

cost of metabolizing food (thermic effect of food or

specific dynamic action) and shivering thermogenesis

(e.g., cold-induced thermogenesis). Because energy

needs are so rarely measured directly, they are esti-

mated in adults by assuming that if an individual is in

energy balance and their body composition is not

changing, they are in energy balance. To estimate the

energy needs of an individual whose weight is stable,

the formulas for estimating resting metabolism de-

vised by Owen and colleagues are useful. First, an

estimate of REE is obtained to provide a rough

approximation of energy needs at rest in health:

Male REE ¼900 þ10 (weight in kg) or female

REE ¼700 þ7 (weight in kg). The result is then

adjusted for physical activity level by multiplying it

by 1.2 for the very sedentary, 1.4 for moderately active,

1866 DIETARY REQUIREMENTS OF ADULTS

Table 1 Dietary reference intakes (DRIs): Estimated average requirements

Life stage group VitaminA

(mgper

day)

Vitamin C

(mgper

day)

Vitamin E

(mgper

day)

Thiamin

(mgper

day)

Riboflavin

(mgper

day)

Niacin

(mgper

day)

Vitamin

(mg

per day)

Folate

(mgper

day)

Vitamin

(mg

per

day)

Copper

(mgper

day)

Iodine

(mgper

day)

Iron

(mgper

day)

Magnesium

(mgper

day)

Molybdenum

(mgper

day)

Phosphorus

(mgper

day)

Selenuim

(mgper

day)

Zinc

(mgper

day)

Infants

7–12 months 6.9 2.5

Children

1–3 years 210 13 5 0.4 0.4 5 0.4 120 0.7 260 65 3.0 65 13 380 17 2.5

4–8 years 275 22 6 0.5 0.5 6 0.5 160 1.0 340 65 4.1 110 17 405 23 4.0

Males

9–13 years 445 39 9 0.7 0.8 9 0.8 250 1.5 540 73 5.9 200 26 1055 35 7.0

14–18 years 630 63 12 1.0 1.1 12 1.1 330 2.0 685 95 7.7 340 33 1055 45 8.5

19–30 years 625 75 12 1.0 1.1 12 1.1 320 2.0 700 95 6 330 34 580 45 9.4

31–50 years 625 75 12 1.0 1.1 12 1.1 320 2.0 700 95 6 350 34 580 45 9.4

51–70 years 625 75 12 1.0 1.1 12 1.4 320 2.0 700 95 6 350 34 580 45 9.4

>70 years 625 75 12 1.0 1.1 12 1.4 320 2.0 700 95 6 350 34 580 45 9.4

Females

9–13 years 420 39 9 0.7 0.8 9 0.8 250 1.5 540 73 5.7 200 26 1055 35 7.0

14–18 years 485 56 12 0.9 0.9 11 1.0 330 2.0 685 95 7.9 300 33 1055 45 7.3

19–30 years 500 60 12 0.9 0.9 11 1.1 320 2.0 700 95 8.1 255 34 580 45 6.8

31–50 years 500 60 12 0.9 0.9 11 1.1 320 2.0 700 95 8.1 265 34 580 45 6.8

51–70 years 500 60 12 0.9 0.9 11 1.3 320 2.0 700 95 5 265 34 580 45 6.8

>70 years 500 60 12 0.9 0.9 11 1.3 320 2.0 700 95 5 265 34 580 45 6.8

Pregnancy

18 years 530 66 12 1.2 1.2 14 1.6 520 2.2 785 160 23 335 40 1055 49 10.5

19–30 years 550 70 12 1.2 1.2 14 1.6 520 2.2 800 160 22 290 40 580 49 9.5

31–50 years 550 70 12 1.2 1.2 14 1.6 520 2.2 800 160 22 300 40 580 49 9.5

Lactation

18 years 880 96 16 1.2 1.3 13 1.7 450 2.4 985 209 7 300 35 1055 59 11.6

19–30 years 900 100 16 1.2 1.3 13 1.7 450 2.4 1,000 209 6.5 255 36 580 59 10.4

31–50 years 900 100 16 1.2 1.3 13 1.7 450 2.4 1,000 209 6.5 265 36 580 59 10.4

This table presents estimated average requirements (EARs), which serve two purposes: for assessing adequacy of population intakes and as the basis for calculating recommended dietary allowances (RDAs) for

individuals for those nutrients. EARs have not been established for vitamin D, vitamin K, pantothenic acid, biotin, choline, calcium, chromium, fluoride, manganese, or other nutrients not yet evaluated via the DRI

process.

As retinol activity equivalents (RAEs). 1 RAE ¼ 1 mg of retinol, 12 mgofy-carotene, 24 mg of I-carotene, or 24 mg y-cryptoxanthin. The RAE for dietary provitamin A carotenoids is twofold greater than retinol equivalents

(RE), whereas the RAE for preformed vitamin A is the same as RE.

As a-tocopherol. a-Tocopherol includes RRR-a-tocopherol, the only form of a-tocopherol that occurs naturally in foods, and the 2R-stereoisomeric forms of a-tocopherol (RRR-, RSR-, RRS-, and RSS-a-tocopherol) that

occur in fortified foods and supplements. It does not include the 2S-stereoisomeric forms of a-tocopherol (SRR-, SSR-, SRS-, and SSS-a-tocopherol), also found in fortified foods and supplements.

As niacin equivalents. 1 mg of niacin ¼ 60 mg of tryptophan.

As dietary folate equivalents (DFE). 1 DFE ¼ 1 m g of food folate ¼ 0.6 mg of folic acid from fortified food or as a supplement consumed with food ¼ 0.5 m g of a supplement taken on an empty stomach.

Data supplied by and used with permission of National Academy of Sciences Press.

tbl0001

or 1.8 for very active individuals. More precise esti-

mates of energy expended in physical activity may be

used if they are available. Taken together, the product

of REE times the activity factor provides an estimate of

total caloric needs in a state of energy balance.

0024 Energy needs decrease with declines in lean body

mass, the major determinant of resting energy

expenditure in health, and with physical activity. Be-

cause both actively metabolizing tissue and physical

activity usually decline with age, energy needs in older

persons, especially those over 70, are less than those of

younger persons; in contrast, energy needs rise with

pregnancy, because of the demands of fetal growth, in

lactation, because of the energy lost required for milk

production, and in growth and rehabilitation.

Macronutrients: Protein, Carbohydrate

(Including Fiber), and Fat

0025 Requirements for the macronutrients have been

reviewed most recently in publications of the Food

and Agricultural Organization/World Health Organ-

ization (FAO/WHO), and the update by the DRI

Committee is expected shortly.

Factors Affecting Nutrient Requirements

0026 Nutrient requirements are potentially affected by age,

sex, rate of growth, pregnancy, lactation, and physical

activity levels. They are also influenced by route of

administration (e.g., by mouth or vein), composition

of the diet with respect to nutrients and other constitu-

ents that may affect absorption, utilization or excre-

tion,coexistingdiseases,anddrugs. The specificfactors

vary, and every nutrient must be assessed individually.

Assessing Dietary Adequacy Using

Information on Nutrient Requirements

0027 Dietary assessment determines whether intake is ad-

equate while not being excessive. The first prerequis-

ite for doing this is to obtain a representative estimate

of usual dietary intake. Procedures for assessment of

individuals and groups vary; a report on assessing

dietary intakes is available. The Food and Nutrition

Board is also completing a report on the uses of the

DRI in planning for individuals and groups.

See also: Carbohydrates: Requirements and Dietary

Importance; Dietary Fiber: Properties and Sources;

Dietary Reference Values; Energy: Intake and Energy

Requirements; Fats: Requirements; Protein:

Requirements; Water: Physiology

Further Reading

Beaton GH (1999) Recommended dietary intakes: Individ-

uals and populations. In: Shils ME, Olson JA, Shike M

and Ross AC (eds) Modern Nutrition in Health and

Disease, 9th edn., pp. 1705–1726. Baltimore, MD:

Williams & Wilkins.

De Onis M, Clugston GA and Underwood B (1998) Pri-

mary malnutrition. In: Sadler MJ, Strain JJ and Cabel-

lero B (eds) Encyclopedia of Human Nutrition, pp.

1246–1254. San Diego, CA: Academic Press.

Dwyer JT (1999) Dietary assessment. In: Shils ME, Olson

JA, Shike M and Ross AC (eds) Modern Nutrition in

Health and Disease, 9th edn., pp. 937–962. Baltimore,

MD: Williams & Wilkins.

Food and Nutrition Board, Commission on Life Sciences,

National Research Council (1989) Recommended Diet-

ary Allowances, 10th edn. Washington, DC: National

Academy Press.

Harper AE (1999) Defining the essentiality of nutrients. In:

Shils ME, Olson JA, Shike M and Ross AC (eds) Modern

Nutrition in Health and Disease, 9th edn. Baltimore,

MD: Williams & Wilkins.

Owen OE, Holup JL, D’Alessio DA et al. (1987) A

reappraisal of the caloric requirements of men. Ameri-

can Journal of Clinical Nutrition 46: 875–885.

Owen OE, Kavle E, Owen RS et al. (1986) A reappraisal

of caloric requirements in healthy women. American

Journal of Clinical Nutrition 44: 1–19.

Panel on Dietary Antioxidants and Related Compound,

Subcommittees on Upper Reference Levels of Nutrients

and Interpretation and Uses of Dietary Reference

Intakes, and the Standing Committee on the Scientific

Evaluation of Dietary Reference Intakes (2000) Dietary

Reference Intakes for Vitamin C, Vitamin E, Selenium,

and Carotenoids. Washington, DC: National Academy

Press.

Panel on Micronutrients, Subcommittee on Upper Refer-

ence Levels of Nutrients and Interpretation and Uses of

Dietary Reference Intakes, and Standing Committee on

the Scientific Evaluation of Dietary Reference Intakes

(2001) Dietary Reference Intakes for Vitamins A, K,

Boron, Chromium, Copper, Iodine, Iron, Manganese,

Molybdenum, Nickel, Vanadium, and Zinc. Washing-

ton, DC: National Academy Press.

Standing Committee on the Scientific Evaluation of Dietary

Reference Intakes and its Panel on Folate, Other B Vita-

mins and Choline, Food and Nutrition Board, Institute

of Medicine (1998) Dietary Reference Intakes for Thia-

min, Ribroflavin, Niacin, Vitamin B

, Folate, Vitamin

, Pantothenic Acid, Biotin, and Choline. Washing-

ton, DC: National Academy Press.

Standing Committee on the Scientific Evaluation of

Dietary Reference Intakes, Food and Nutrition Board,

Institute of Medicine (1997) Dietary Reference In-

takes for Calcium, Phosphorus, Magnesium, Vitamin

D and Fluoride. Washington, DC: National Academy

Press.

Subcommittees on Interpretation and Uses of Dietary Ref-

erence Intakes and Upper Reference Levels of Nutrients,

and the Standing Committee on the Scientific Evaluation

of Dietary Reference Intakes (2000) Dietary Reference

Intakes: Applications in Dietary Assessment. Washing-

ton, DC: National Academy Press.

1868 DIETARY REQUIREMENTS OF ADULTS

DIETARY SURVEYS

Contents

Measurement of Food Intake

Surveys of National Food Intake

Surveys of Food Intakes in Groups and Individuals

Measurement of Food Intake

M Nelson, King’s College London, London, UK

Background

0001 Measuring food consumption is a complex and

difficult task. It may be carried out for a variety

of reasons. For example:

0002 to determine the adequacy of food supply available

to an entire country;

0003 to determine the adequacy of nutrient intake in a

selected group of individuals;

0004 to establish levels of food self-sufficiency in the

development of agricultural policy;

0005 to examine the relationship between diet and

health in groups or individuals;

0006 as part of fundamental research on nutrient

metabolism;

and for many other reasons.

0007 It is fair to say that there is no method available

that provides an entirely accurate measure of food

consumption or nutrient intake, at whatever level

the measurement is being made. It is impossible at

the national or regional level to measure all food

production completely, and to take into account all

food and production losses and wastage; at the house-

hold level, people alter their purchasing or food-use

habits while being observed; and at the individual

level, people change their eating habits during a

survey, or fail to describe their eating habits accur-

ately. Awareness of the likely sources of error, how-

ever, allows for a sensible interpretation of results

from studies of dietary assessment.

0008 The type of measurement made, whether at the

national, regional, household, or individual level,

will be dictated by the purpose of the study being

undertaken. This article identifies the points in the

food chain at which measurements of food consump-

tion can be usefully made, and highlights the

strengths and weaknesses of the techniques available.

It also considers the validity and reliability of these

techniques when trying to relate diet and disease.

Methods for Measuring Food

Consumption

0009Figure 1 shows the points in the food chain at which

food consumption is commonly measured. Point I is

at the level of domestic food production. In virtually

all countries, a Department or Ministry of Agricul-

ture requires food producers to report the amount of

food produced. This information is useful in estab-

lishing levels of food self-sufficiency, but cannot be

used alone to estimate the average consumption in the

population.

0010Measurements at Point II are used by the Food and

Agricultural Organization (FAO) of the United Nations

to establish food-balance sheets. These measurements

reflect domestic production, plus food imports and

food taken from stocks, minus food exports, food

moved into stocks, and food used for animal feed or

non-food purposes (e.g., sugar used in the brewing

industry). They can be used to show the average

amount of food available per person in the population.

0011The next point of food measurement is the house-

hold (Point III). Typically, the weights and costs (or

imputed value) of all food acquisitions (purchases,

gifts, food from gardens and allotments, payments

in kind, etc.) are recorded over a set period ranging

from 1 week to 1 year. In addition, an inventory of the

larder contents may be carried out at the beginning

and end of the survey period in order to determine

changes in food stocks, and allowances made for

estimated food wastage, to yield an estimate of

the amount of food available for consumption in the

household (Point IV). Alternatively, respondents may

be interviewed and asked to recall their purchases and

food use over the previous week or two weeks, usu-

ally having been forewarned and asked to keep re-

ceipts and other informal records of food use and

acquisition such as packaging and labels. Household

surveys may include records of food purchased and

eaten away from home, and household members’

consumption of items not directly under the purview

of the respondent, such as alcoholic beverages, sweets

and soft drinks.

0012The final point of measurement (Point V) is at

the level of the individual. Briefly, there are two

DIETARY SURVEYS/Measurement of Food Intake 1869

approaches, prospective and retrospective. Prospect-

ive methods include:

0013 Duplicate diet. The subject collects an exact dupli-

cate or fixed aliquot (e.g., 10% by weight) of all

food and drink actually consumed. The aliquot

technique is especially well suited to households

or institutions. A composite diet may be assembled

in which representative local foods are combined

into a single sample in proportion to the foods

consumed (assessed using other techniques such as

household food purchases or weighed inventories).

The sample is then analyzed chemically for its

nutrient content.

0014Weighed inventory or weighed record. The subject

records a detailed description and measured weight

of all food and drink consumed.

0015Household measures. The subject keeps a record

of all food and drink consumed. Amounts are

recorded in units of common household measures

such as cups, spoons, bowls, etc.

0016Food check list. The subject ticks a box on a pre-

printed list of food and drink to indicate each time

a named item is consumed. Foods or drinks that do

not appear on the list are recorded separately.

In general, the longer the period over which infor-

mation on diet is collected, the greater will be the

Production and distribution losses

Institutional and catering uses

Food taken from larder

Food from gardens, allotments, gifts, etc.

Food added to larder

Food wastage, pets, etc.

Food consumption by

individuals

Food available

in household

Household food

purchases

Total food available

Domestic food

production

Factors affecting food distribution

within the household

Imports

Food taken from stock

II.

III.

IV.

Exports

Food added to stock

Non-food uses

fig0001 Figure 1 Points in the food supply chain at which it is convenient to measure food availability or consumption. Reproduced from

Garrow J, James WPT and Ralph A (eds) Human Nutrition and Dietetics, 10th edn. London: Churchill Livingstone, wtih permission.

1870 DIETARY SURVEYS/Measurement of Food Intake

precision in the estimate of an individual’s food or

nutrient intake. However, the intake of some nutri-

ents (like vitamin C) varies more from day to day than

the intake of other nutrients (sugar, for example). The

greater the day-to-day variation in intake, the longer

the period of recording that will be needed to achieve

a given level of precision.

0017 Retrospective methods include:

0018 Diet history. The subject is asked by a trained

interviewer about usual consumption habits in an

extended interview lasting 1–2h.

0019 24-h recall. A trained interviewer asks quantita-

tively about consumption of food and drink over

the previous 24 h.

0020 Food-frequency questionnaires (FFQ). These are

completed either by the respondent or with the

help of an interviewer. FFQs typically ask about

the usual frequency and quantity of consump-

tion of a wide range of food and drink. Alterna-

tively, they may focus on the consumption of a

limited range of items that are the main sources

of intake of a particular nutrient. Some FFQs

ask about frequency alone without assessing

portion size.

Table 1 summarizes the principal applications of

these different techniques. Food-balance sheets are

appropriate for estimating national consumption

levels, and are therefore of use in epidemiological

studies comparing diets between countries. They

cannot be used for assessing consumption at the

regional, household, or individual level. Household

surveys lend themselves to a wide variety of uses. If,

for example, a nationally representative sample of

households is used, it will be possible to estimate

consumption levels nationally as well as in regional,

socio-economic, and other subgroups within the

population. Some household surveys include larder

inventories at the beginning and end of the survey

period, which allow estimates of consumption at the

level of single households. Techniques are available to

estimate the distribution of food and nutrient to indi-

viduals of similar age and sex within groups of house-

holds, and household surveys may therefore have

application in some epidemiological surveys where a

knowledge of the pattern of consumption to individ-

uals, but not individual intakes per se, is required.

0021Duplicate diet techniques are most suitable for

metabolic studies of individuals, where are a pre-

cise knowledge of nutrient intake is required. The

tbl0001 Table 1 Appropriate uses of dietary survey methods

Dietarysurveymethod

Food bal-

ance

sheets

Household

surveys

Surveys of individuals

Prospective Retrospective

Duplicate

diet

Weighed

inventory

Household

measures

Diethistory 24-hrecall Questionnaires

Level of dietary measurement

National þþþ þþþ þ

Regional – þþþ þ

Institution/group – þþþ þ

þþ

Household – þþ

þþ

d,e

þþ

Individual – þ

þþþ

Type of study

Epidemiological þþþ

þþþ

þþ

þþþ

Clinical – – þþ þþþ þþþ þþþ þþ

þþþ

Metabolic – – þþþ þ

–– – –

Includes other techniques of direct analysis (see text).

Requires sample representative of population, institution or group, or analysis weighted to reflect balance of subgroups.

Requires larder inventory. Short-term measures (e.g., 1 week) may not reflect usual diet in individual households.

The need for data from all household members may distort usual household food consumption patterns.

Semiweighed method (Nelson M and Nettleton PA (1980) Dietary survey methods. I. A semi-weighted technique for measuring dietary intake within

families. Journal of Human Nutrition 34: 325–348.) reduces total respondent burden within household.

Requires complex mathematical modeling of within-household food and nutrient distribution (Chesher A (1997) Diet revealed?: Semi-parametric

estimation of nutrient intake age relationships. Journal of the Royal Statistical Society, Series A ^ Statistics in Society 160(3): 389–420.).

Important to screen out individuals whose responses may not be valid (see text).

Appropriate for ecological studies (see Hiller JE and McMichael AJ (1997) Ecological studies. In: Margetts BM and Nelson M (eds) Design Concepts in

Nutritional Epidemiology, 2nd edn. Oxford: Oxford University Press.).

See Margetts B and Nelson M (1997) Design Concepts in Nutritional Epidemiology, 2nd edn. Oxford: Oxford University. For a detailed discussion of use of

dietary survey methods in epidemiological studies.

Requires repeat 24-h recalls for valid classification of subjects according to levels of intake (see text).

Useful only if range of foods is of limited variation in composition, allowing reliable use of food composition tables.

þþþ, very suitable; þþ, moderately suitable; þ, limited application; –, not suitable.

DIETARY SURVEYS/Measurement of Food Intake 1871

composite diet technique is particularly suitable for

estimating the nutrient content of foods eaten by a

group of people where food composition tables (q.v.)

are likely to be inadequate. The weighed inventory is

suitable for assessment of individual food consump-

tion or nutrient intake (using food-composition

tables), and thus lends itself to a wide variety of

uses, in that accurate estimates of dietary intake for

groups of people can be built up from individual data.

Where a limited range of foods of accurately known

composition is being consumed, the weighed inven-

tory may also be used in metabolic studies. Because

the household-measures technique, like the weighed

inventory, focuses on the individual as the unit of

assessment, it can also be used in a wide variety of

studies, but with some loss of precision in comparison

with weighed inventories.

0022The retrospective techniques, because they rely on

memory, are regarded as being less objective than the

prospective techniques, but they may free the

respondent of factors likely to distort the recording

of diet (see Table 2). The diet history, particularly,

may be used to assess the ‘usual’ diet of an individual

more effectively than other techniques, in that sea-

sonal variations in diet can be taken into account

more readily. A single 24-h recall from one individual

cannot be used to assess that individual’s diet because

of day-to-day variations in intake. Single 24-h recalls

collected from a large number of subjects, however,

may be useful for estimating the intake of a group.

tbl0002 Table 2 Main attributes of the different dietary survey methods for measuring food consumption or nutrient intake in human

populations

Typeof survey Strengths Weaknesses

Food-balance sheets Determine national food consumption National only

Compare diets between countries May overestimate food availability, especially

in more affluent countries

Household budget surveys Relatively cheap to obtain Home food only

Subgroupanalysis possible (by region,

income, household composition, etc.)

Usually excludes meals eaten away from home,

sweets, soft drinks, alcohol

Useful where individual data hard to obtain No individual data

Time trends if collected annually May be expenditure only, so weights of foods

must be imputed

Without larder inventory Lower respondent burden Bias towards overpurchasing, especially in low

income and elderly households

Food acquisition, not consumption

With larder inventory Consumption data Distorts normal food purchasing pattern

Higher respondent burden

Prospective surveys of individuals Current diet Labor-intensive

Direct observation of diet Requires literacy and numeracy skills

Vary duration of survey according to needs High motivation needed

Distortion and/or omission due to ‘health’

issues

Overweight subjects more likely to underreport

true consumption levels

Duplicate diet Direct analysis of nutrient content of diet (no

food tables needed)

Very expensive

Intensive supervision required

Likely to distort usual diet

Weighed inventory Widely used, facilitates comparisons between

studies or groups

Loss of precision of estimate of nutrient intake

compared with duplicate diet

Household measures No scales needed Loss of precision compared with weighed

inventory

Retrospective surveys of individuals Inexpensive Bias: memory, conceptualization, interviewer

Quick Daily variation not usually assessed

Low subject motivation needed Depends on regular eating habits

Good cooperation

Current or past diet

Diet history Assesses ‘usual’ diet Tends to overestimate nutrient intake

24-h recall Very quick Tends to underestimate nutrient intake

Repeat observations to obtain daily variation Does not reflect individual intake unless

repeated

Questionnaires Suitable for very large number of subjects Requires validation

Can be posted Literacy and numeracy skills needed if

self-completedFocus on particular nutrients

With the exception of the duplicate diet technique, all methods rely on food composition tables for estimating nutrient intake.

1872 DIETARY SURVEYS/Measurement of Food Intake

Repeat 24-h recalls collected from an individual can

provide an accurate assessment of that individual’s

diet. Questionnaires are of particular value in epi-

demiological studies, in that they can provide infor-

mation economically from a large number of subjects.

Strengths and Weaknesses

0023 Table 2 summarizes the main strengths and weak-

nesses of the methods available. When measuring

diet, it is important that the method chosen is accur-

ate enough to enable any hypothesis to be tested

adequately. It is likely that many studies that fail to

show a relationship between diet and disease do so,

not because no relationship exists, but because the

techniques used to assess diet were not appropriate,

and measurement error (under-reporting actual

intake, for example) was not taken into account.

Equally important, if the aim of a study is to relate

diet and health in groups of people, for example by

comparing diet and health status between countries

or regions, there is little point in spending resources

on expensive and labor-intensive surveys that charac-

terize the diets of individuals.

0024 It must be decided at the beginning of a study

whether measurements of food consumption, or

nutrient intake, or both are required. The required

precision (standard error) of the measurements will

determine the number of subjects and the number of

measurements per subject required, or the number

of regions or countries to be included in a regression

or correlation analysis. If ranking of subjects or

groups will suffice, then a technique that has a

systematic bias or nondifferential misclassification

may be acceptable, but if absolute measures of con-

sumption or intake are required, techniques with a

greater validity will be needed. Most prospective

methods (apart from duplicate or composite diet

techniques) and all retrospective methods rely on

food composition tables (q.v.) for estimating nutrient

intake, and the weaknesses associated with their use

should be considered in any evaluation of findings.

Measurements made will be affected by the time of

year (summer/winter, wet season/dry season, holy

days, festival days), the time of week (weekdays/

weekends), and even the time of day (interview re-

sponses may differ before and after meals). Also, the

precision and representativeness of the findings will

be influenced by whether it is present or past diet that

is being measured. Measurement of past diet is more

difficult but may be important in relation to under-

standing the dietary origin of chronic diseases such as

heart disease or some types of cancer.

0025In studies of individuals, different techniques re-

quire particular skills to complete records or recalls

of diet satisfactorily. For example, very young or very

old subjects may be unable to recall diet accurately;

people with poor literacy and numeracy skills will be

unable to keep adequate records.

Food-Balance Sheets (see ‘Food Supply’ database

at http://apps.fao.org/)

0026These are eminently suitable for comparing diets be-

tween countries, and are readily available from the

FAO, thereby reducing the cost of such studies. They

do, however, show significant bias, in that the esti-

mated availability of foods and nutrients may be

considerably greater than the actual levels of con-

sumption. In the UK, for example, food-balance

sheet estimates of energy intake are 25% greater

than the estimated energy requirement in the popula-

tion (although some allowance needs to be made for

waste and production losses). More important, this

bias is probably not constant between countries. More

affluent countries are more likely to overestimate con-

sumption than poorer countries. This may be due in

part to more efficient recording of food production

in affluent countries, and in part to the presence in

poorer countries of many small holdings and subsist-

ence farms whose production is underrecorded.

Household Budget Surveys

0027Household budget surveys of nationally representa-

tive samples are regularly conducted by many govern-

ments. The results provide a ready-made database for

analysis of food trends across time, and between

regions or subgroups within the population. Where

it is difficult to obtain individual data (such as in

cultures where families eat communally from large

serving vessels), household surveys may provide a

valuable way of characterizing diet.

0028Their principal disadvantage is that foods eaten away

from home are rarely recorded. These usually include a

high proportion of sweets, soft drinks, alcoholic bever-

ages, and snack foods. Also, individual intakes cannot be

estimated, which may be a particular disadvantage in

epidemiological studies, where it is important to relate

disease outcome in individuals to diet.

0029Some household surveys are conducted without a

larder inventory, so as to reduce the respondent

burden. The assumption is that over a sufficient

number of households (say 20), the differences within

individual households between purchases and actual

consumption will balance out, giving an overall

In nondifferential misclassification, measurement errors that are present

affect all subjects equally or proportionately. In differential misclassification,

errors in reporting are greater in some groups (for example, overweight or

elderly people) than in others. Consequently, comparison between groups is

difficult to interpret. Differential measurement errors may undermine

attempts to establish diet–disease relationships.

DIETARY SURVEYS/Measurement of Food Intake 1873

estimate of consumption close to the true level. In

practice, it has been observed that surveys of food

purchases in developed countries tend, on average,

to stimulate purchasing in excess of actual consump-

tion by as much as 15%. This bias is greater in house-

holds with respondents over 60 years of age and in

respondents from lower income households. As lower

income group consumption levels are generally lower,

comparisons of the food purchases or estimated nu-

trient intakes are likely to underestimate the true

differences between income groups except where

very major differences occur (in fruit consumption,

for example). Comparison of food or nutrient profiles

(per cent of income spent on certain foodstuffs, or

nutrient intakes per MJ) may be more reliable. This

problem is less pronounced in surveys where the var-

iety of food purchases is limited (as in many develop-

ing countries), or where purchases are reported

retrospectively rather than prospectively. Reported

household food purchases may be supplemented by

records from individual household members of con-

sumption of foods eaten outside the home, but are

often subject to errors of omission.

0030 Household surveys conducted with larder inventor-

ies give a more precise estimate of levels of food

consumption and nutrient intake, but they tend to

distort usual purchasing patterns when respondents’

attention may be drawn to the larder contents. They

also have a higher respondent burden. Some house-

hold budget surveys include expenditure only. It is

then necessary to impute purchase quantities from

expenditure. These are likely to be less reliable than

surveys that record amounts of food purchased as

well as expenditure.

Prospective Surveys of Individuals

0031 These have the advantage of direct observation of

current diet. They can be varied in duration to im-

proved the accuracy of classification of individuals’

nutrient intakes, i.e., the greater the number of days

of information recorded, the more likely it is to reflect

the true intake of the individual. For example, carbo-

hydrate intake may be estimated within a few days,

whereas energy intake may require a week, and fatty

acid intake 2 weeks. Where a large number of days is

needed, problems of respondent fatigue can be over-

come by having a series of shorter periods (four 4-day

periods, for example).

0032 The principal disadvantage of prospective tech-

niques is the level of literacy and numeracy skills

required to complete the surveys. This can be over-

come to some extent by having an interviewer do the

recording and measuring, but this is very invasive for

the subject and expensive in terms of staff time. Pro-

spective methods may also encourage distortion of

diet and avoidance of foods regarded as unhealthy,

particularly sweets and alcoholic beverages, or there

may be simple omission of items owing to poor

recording techniques. A particular problem that has

come to light with the advent of the (very accurate)

doubly labelled water method for estimating energy

expenditure is that overweight subjects are more

likely to underrecord their intake than normal weight

subjects. This has important implications when trying

to establish links between diet and disease where

being overweight may itself be a risk factor (heart

disease, for example).

0033The duplicate diet technique is appropriate mainly

for metabolic studies, as it is very expensive and time-

consuming. Its use in free-living populations requires

exceptionally close supervision to ensure that the

food samples collected are truly representative of the

foods that have been eaten. It has been used in epi-

demiological studies where a single day’s duplicate

has been collected from a number of subjects in

order to characterize the group diet rather than indi-

vidual diets. The composite diet technique is easier to

use for assessing group diet characteristics by chem-

ical analysis. It is very good for estimating average

levels of additives or contaminants present in the diet.

It has the disadvantage of relying on food intake or

purchasing records or dietary recall.

0034The weighed inventory has been used very widely,

and for this reason, there are many values in the litera-

ture against which new data can be compared. It is less

demanding of both respondent and interviewer than

the duplicate diet technique, but because it relies on

food composition tables, there is a loss of precision in

estimating nutrient intake. The household measures

technique is useful where subjects are unused to using

scales, or resources are limited, and requires a lower

level of numeracy than the weighed inventory, but

there is a loss of validity owing to a variation in

portion sizes not reflected in their descriptions.

Retrospective Surveys of Individuals

0035In comparison with prospective techniques, retro-

spective techniques are relatively inexpensive,

demanding less time of both respondent and inter-

viewer. They therefore tend to have slightly higher

cooperation rates. For foods that are consumed only

occasionally (like liver) or about which there may be

some stigma (like sweets or alcoholic beverages) in

some countries, and for assessing the intake of nutri-

ents (such as carotene or vitamin B

) that would

require many days of recording, there are advantages

in using retrospective techniques that ask about

the usual frequency of consumption of foods. Retro-

spective techniques also have the unique advantage

of allowing assessment of past diet, although, in

1874 DIETARY SURVEYS/Measurement of Food Intake

practice, estimates of past diet may be heavily influ-

enced by present diet. Most recalls of diet in the

distant past (2 or more years prior to the interview)

correlate as strongly with current diet as with the past

diet assessed at the time.

0036 Results from retrospective methods, like those from

prospective methods, are subject to differential mis-

classification, but for different reasons. Memory and

conceptualization skills (the ability to describe quan-

titatively what was actually eaten) differ in an unsys-

tematic way between individuals. The very young

(under 12) and the very old (over 70) are less likely

to be able to recall their diets correctly. The number of

food items recalled has been shown to correlate with

the estimated energy intake, independent of the true

level of intake. Apart from the use of repeat 24-h

recalls, retrospective methods provide little or no in-

formation on day-to-day variation in diet. Such meas-

ures of variation can be useful in determining the likely

accuracy of ranking of individuals according to nutri-

ent intake. For the diet history and questionnaires,

especially, subjects who do not have regular eating

habits will have difficulty in describing the ‘usual’

frequency of consumption or portion size.

0037 The diet history is especially valuable for assessing

the ‘usual’ diet, but the nature of the interview may

encourage overreporting, and the results from diet

history assessments usually exceed those from

weighed inventories, which, in normal-weight individ-

uals, are probably a good reflection of ‘usual’ intake.

0038 The 24-h recall is relatively quick, demanding little

of respondents’ time, but cannot be used to character-

ize individuals’ diets unless it is repeated over a

number of days. In general, because of the absence

of a ‘training effect,’ respondents are more likely to

omit items than in methods that extend over several

days. Estimates of nutrient intake using 24-h recalls

therefore tend to be low in comparison with other

techniques (whether prospective or retrospective). A

new ‘triple pass’ 24-h recall technique (in which

respondents are given two further opportunities in

the same interview to recall other items consumed)

has resulted in higher estimates of intake.

0039 Questionnaires are suitable for obtaining informa-

tion from very large numbers of subjects. Their ad-

ministration can be more readily standardized than

other types of dietary assessment (for example,

administered by computer or sent by post for self-

completion by respondents). Where the hypothesis to

be tested relates to specific foods (e.g., vegetables) or

nutrients (e.g., calcium), they can be of limited length

pertaining to relatively few foods. They can thus have

the advantage of brevity while still being able to rank

subjects with a known degree of validity. Because of

the uncertainty regarding the successful completion

and validity of assessment of newly designed question-

naires, it is essential that they are properly evaluated

against a dietary survey method in which the errors of

assessment can be estimated. If they are to be self-

completed after being posted to subjects, literacy and

numeracy skills will be needed, and differential mis-

classification may affect interpretation.

Use of Computers

0040Computers are used universally to calculate the nutri-

ent content of diets for which no direct chemical

analysis is made, but they can also be used directly

in the dietary assessment of individuals. Data-entry

programs such as UNIDAP (Unilever) have been

tailored to allow interviewers to enter patterned

food consumption data, thus effectively providing

the facility for an interactive diet history. Branching

interactive programmes (such as EPICSOFT) allow

for 24-h recall data to be entered directly at interview.

Short, structured questionnaires (e.g., CALQUEST

for measuring habitual calcium intake) have also

been devised for interactive use. In Integrated Dietary

Analysis (IDA), respondents use a hand-held com-

puter linked with digital electronic weighing scales

to collect weighed inventory data without the need

for written recording or subsequent computer entry

of data. Scales have also been linked to computers

with special keypads with symbols instead of letters

(FRED and FREDA), allowing respondents to use

pictures to indicate the food being consumed; again,

weights and codes are recorded automatically. A tape

recorder has also been linked to scales (PETRA),

allowing subjects to describe foods orally rather

than in writing, a weight being recorded automatic-

ally on the tape in association with each item de-

scribed. With the exception of interactive recalls, the

use of computers for data collection has been disap-

pointing because of high costs or laborious transcrip-

tion of data.

Use of Biological Markers

0041Given the errors that are inherent in any reporting of

diet, it is important to have objective measures that

reflect usual food consumption or nutrient intake.

Biochemical measurements of nutrients or related

metabolites in blood or urine have long been used

to identify nutrient deficiencies or irregularities.

Markers of diet that span the whole range of intake

are less common. Collection and storage of appropri-

ate sample material may be a problem in itself (e.g.,

ensuring completeness of collection of 24-h urine

samples). Biological markers do, however, provide

important tools for evaluation of the validity of

DIETARY SURVEYS/Measurement of Food Intake 1875