Introduction to Human Nutrition

318 Introduction to Human Nutrition

Table 13.7 Design options in nutritional epidemiology

Design Setting Uses Advantages Limitations

Clinical trial Secondary prevention
(diseased participants)

Treatment–outcome

association

Strongest evidence for

causality

Highest internal validity

Very low potential for bias

Low external validity

Ethical problems

High cost

Field trial Primary prevention (healthy
participants)

Exposure–onset of

disease association

Strong evidence for causality

High internal validity

Low potential for bias

Very large sample and
long follow-up
Low external validity
Can assess only single-
nutrient effects
Highest costs
Community
trial

Group randomization (towns,

work-sites, schools)

Evaluation of

community

interventions or

educational activities

If multiple, and small groups

are randomized, it has

the advantages of an

experimental design

Low internal validity
if the number of
randomized units
is low
Quasi-experiment Intervention study (not
randomized)

Evaluation of

community

interventions or

educational activities

High feasibility

More applicable

Investigator controls exposure

Diffi culties in fi nding
comparable groups
High potential for bias
Underlying trends may
alter results
Cohort Participants are initially
classifi ed as exposed or
nonexposed and followed
up in time to monitor the
incidence of the outcome.
Retrospective or historical
cohort studies are
conducted using previously
collected information (fi les)

The most powerful

observational tool

in nutritional

epidemiology to

study diet–health

associations

Very low potential for bias

Ability to study rare exposures,

complex dietary patterns

and multiple outcomes of a

single exposure

Allows direct estimation of

risks and rates

Minimal ethical problems

Large sample and very
long follow-up
No ability to study rare
outcomes
Bias by low follow-up
(attrition)
Requires collaborative
participants
High costs
Case–control Exposure is compared between
subjects with and without
the outcome. Nested
case–control studies are
conducted within an
ongoing cohort using the
data of cohort members
who develop the disease
(cases) and a sample of
nondiseased members
(controls)

Practical analytical

tool in nutritional

epidemiology to

study diet–health

associations

Ability to study rare outcomes

Ability to study multiple

potential causes of a single

outcome

No problems with losses to

follow-up

Minimal ethical problems

Low cost

Potential for biased recall

of exposure and biased

participation of

controls

Inability to study rare

exposures and multiple

outcomes of a single

exposure

Inability to estimate risks

and rates

Cross-sectional Past exposure and outcome
are simultaneously assessed
in a representative sample
of the population

Estimation of the

prevalence of a

disease or an

exposure

Population assessment

in health planning

Monitoring trends if it

is periodically

repeated

Highest external validity

Relatively low costs

Minimal ethical problems

A wide spectrum of

information about diet and

health can be collected

Diffi cult to assess the

temporal sequence:

very low ability for

causal inference

Potential for biased

participation and

response bias

Ecological The unit of analysis is not the
individual but a community.
Exposure and/or disease are
not measured at the
individual level

Generation of new

hypothesis and

contextual or

multilevel analysis

Ability for assessing exposures

at the community level

Relatively low costs

Minimal ethical problems

Very low internal validity

(“ecological fallacy”)