Chapter 4

Epidemiological Study Designs

1
 learning Objectives
After completing this chapter, the student will
be able to
• Understand and differentiate descriptive and
analytical study designs
• Select a study design according to the study
objectives

2
 Definition of design
Design
- is an arrangement of conditions for the collection &
analysis of data that leads to the most accurate answer
to the research question and in the most economical
way.
-is overall structure of the study
Selection of study design
Selection of a design depends on:
1)State of knowledge on the problem
2)research questions

3
Classification of epidemiologic
Study Designs

Ecological

4
5
6
 Selection of study design
• Depends on
1.Hypothesis being tested

2.Resources available

3.Current state of knowledge

7
State of knowledge of the problem Types of research questions Study design

Knowledge that problem -Who is affected? -Qualitative(e.g.

exists, but knowing little -How do the affected people FGD)
about its characteristics behave? Or
of possible causes -what do they know, believe, Quantitative
think about the problem? (Descriptive)
-What is the magnitude of the
problem?

Suspecting that certain -Are certain factors indeed Analytic

factors contribute to the associated with the problem? (observational)
problem

-Having sufficient -What is the effect of a Intervention

knowledge about the particular intervention? (experimental)
cause -Which of the alternative
-to develop & assess an strategies gives better result?
intervention that would -Are the results in proportion to
prevent, control, or solve time/ money spent
the problem

8
Types of descriptive study designs
1. Case report:
 consists of a careful, detailed report by one or
more clinicians of the profile of a single patient
 more emphasis is given for unusual findings
Example: Case report in 1961
A 40-year old pre-menopausal woman developed
pulmonary embolism 5 weeks after beginning to
use an oral contraceptive preparation to treat
endometriosis.
9
 2. Case series
 describes the characteristics of a number of patients
with a given disease (same diagnosis)
Example: Five young, previously healthy homosexual
men were diagnosed as having pneumocystis carinii
pneumonia at 3 Los Angeles hospitals during a 6
month period in 1980 to 1981.
 This form of pneumonia had been seen almost
exclusively among older men and women whose
immune systems were suppressed
 This unusual circumstance suggested that these
individuals were actually suffering with a previously
unknown disease, subsequently it was called AIDS 10
11
12
 3. Correlational (Ecological) studies
• measures that represent characteristics of entire population
are used to describe disease in relation to some factor of
interest
• The units of analysis are populations or groups of people
rather than individuals
• compare populations in different countries at the same time
or the same population of a country at different times
• Incidence and prevalence rates are commonly used to
quantify disease occurrence in groups.
• Example: Incidence of hypertension and average per capita
salt consumption compared between two communities.

13
14
15
16
17
18
19
20
21
22
 4. Cross sectional surveys
• often called prevalence studies
• Exposure and disease status are assessed
simultaneously
• help in assessing the health status and health
care needs of a population
• Several countries conduct regular cross-
sectional surveys on representative samples
of their population.

23
24
25
26
27
28
 Characteristics of cross-sectional
studies:

• Assess both exposure and outcome simultaneously

• Are based on point prevalence rates with only few

exceptions

• Are frequently made on total population samples

• Subdivision of the total population takes place after data

collection unlike in cohort & case- control studies

• Are cheaper, easier

• Measures of association is made using odds ratio

29
 Advantages of cross sectional studies:

• one-stop, one-time collection of data

• less expensive & easier to conduct
• provide much information useful for planning health
services and medical programs
• show relative distribution of conditions, disease,
injury and disability in groups and populations
• studies are based on a sample - do not rely on
individuals that present themselves for medical
treatment

30
 Disadvantages of cross-sectional studies

1) chicken or egg dilemma -which occurred first? the

exposure or the outcome ?
Example: study conducted to explore relationship
b/n physical activity & CHD
• Cross sectional studies can reveal valid
association only when the current values of the
exposure variables are unalterable over time
• Such variables include factors present at birth,
such as blood group.
2) It may not show strong cause-effect relationships
if sample size is small.
31
3. Potential bias in measuring exposure

4. Not feasible for rare diseases

5. Does not yield incidence or relative risk.

6. the exposure could be associated with survival after

dx occurrence, rather than development of the disease

32
 Analytic study designs
Analytic studies focus on the determinants
(causes) of diseases.
They are used to test hypothesis with the
ultimate goal of judging whether a particular
exposure causes or prevents disease.
One major distinguishing feature of analytic
studies is the use of controls.

33
 Types of Analytic study designs
 There are two major categories of analytic study
designs, observational and experimental.

 In observational studies the investigator can not

take an active role in allocating people into
groups and administering an exposure to one of
the groups. He/she simply observes what is
happening or what has happened to the groups
under study.

 In experimental studies the investigators

themselves allocate the exposure.
34
 Observational analytic studies
The commonest types of observational analytic
studies
are:-
1. Case control
2. Cohort
3. Cross sectional

35
 A) Case control studies
 Subjects are selected with respect to presence
or absence of disease (outcome), and then
inquiries are made about past exposure to the
factor of interest.

36
 Case-Control Study Design
Exposed
Cases
Not-exposed

Exposed
Controls
Not-exposed

37
Example: Is cigarette smoking a cause of lung
cancer?
Case control design:
 Identify people with lung cancer (cases) and
people without lung cancer (controls), then ask
both groups whether they are/were smokers.
 If cigarette smoking is a cause of lung cancer,
large proportion of lung cancer cases will give
history of cigarette smoking compared to the
normal individuals (controls)

38
 Steps in conducting case control
study
Step 1: Define cases
 One of the first issues to be considered in the design of
case control study is the definition of the disease or
outcome of interest.
Step 2: Select cases
The investigator should select cases on which he/she can
get complete and reliable information
Places where we can get cases:
1-Hospitals (health institutions)
easy & inexpensive
selection bias is one of the major problems
2-Population (community)
expensive
avoids selection bias

39
 Step cont……
Step 3: Select controls
The controls should be similar with the cases except
that the cases have the disease or other outcome of
interest.
Sources of controls
1-Hospital controls
Advantage:
 easily identified & readily available in sufficient number
 less cost
 more likely to be aware of antecedent exposures or
events than healthy individuals . This decreases recall
bias
 They are more likely to be cooperative
40
 Disadvantages
They are different from healthy individuals in
many ways
If the controls are patients with diseases known to
be associated with the exposure of interest (either
positively or negatively), there will be danger of
altering the direction of association or masking a
true association between the exposure and
outcome.

41
 2-General population
controls
Advantages:
 Generalization is possible
 If cases are selected from the population, it is
good to select controls from the population
too.
Disadvantages
 Costly & time consuming
 recall bias (may not be concerned about past
exposure since they are healthy)
 people might be less motivated to participate
42
 Step cont…
Step 4:Check the exposure status of individuals
both in the cases and controls
 Information regarding the exposure status can be
obtained by interview or from different records.
Step 5: Analysis
 Prepare 2X2 table
 calculate Odds Ratio (OR)
 Perform statistical tests to check whether there is
significant association

43
Reading assignment

Nested case control

Matched case control
Case –cross over
 Cohort studies
 Subjects are selected by exposure, or
determinant of interest, and followed to see
the development of the disease or other
outcome of interest
 Other terms used instead of cohort study are:
– Follow up study
– Incidence study
– Longitudinal study

50
Example: Is cigarette smoking a cause of lung
cancer?
 Cohort design: Identify smokers (exposed) and
non smokers (not exposed) then follow both
groups over time (e.g. 10 years) and check for
development of lung cancer in both groups.
 If cigarette smoking is a cause of lung cancer,
large proportion of smokers will develop lung
cancer compared to the non-smokers

51
 Types of cohort studies

 Classification is based on the temporal

relationship between the initiation of the
study and the occurrence of the disease.
1- prospective cohort study
2- retrospective cohort study

52
 A) Prospective cohort study
 at the beginning of the study the outcome has
not yet occurred
 is the commonest type (compared to the
retrospective cohort)
 unless specified cohort study refers to the
prospective type of cohort
 is regarded more reliable than the
retrospective cohort

53
 Prospective vs. Retrospective

Selection of
Exposed & Unexposed Follow - Up
Participants
2010 PROSPECTIVE
2005 COHORT
STUDY
End of Follow
Investigator Up
begins the study

2000 2005
RETROSPECTIVE
1990 COHORT
STUDY
End of Follow
Up
Investigator
begins the study

54
 B) Retrospective (Historical) cohort study

 The investigation is initiated at a point in time

after both the exposure and disease have
already occurred
 less costly and less time consuming
 Often uses data collected for other purposes,
hence information obtained might be
incomplete and non-comparable for all
subjects

55
 Steps in conducting
cohort study
Step 1: Define exposure
Step 2: Select exposed group
During selection consider:
 the frequency of the exposure in the population
 the need to obtain accurate information
Exposure/outcome)
 the ease to obtain relevant information and to
follow up
Step 3: Select non-exposed group
 Non exposed groups should be comparable to the
exposed group
56
 Steps cont….
Step 4: Identify sources of data for
exposure and outcome
Possible sources of exposure data:
• pre-existing records
• conducting interview
Possible sources of outcome data:
– routine surveillance
– death certificate
– periodic health examination
– hospital records etc..

57
 Steps cont….
Step 5: collect data
Step 6: Analyze data
 prepare 2X2 table
 calculate Relative Risk (RR)
 perform statistical tests to check whether there is
statistical significant association

58
Case control Cohort
Advantages: valuable when exposure is rare
optimal for evaluation of rare can examine multiple effects
disease of a single exposure
can examine multiple factors temporal relationship is known
for a single disease allows direct measurement of
Quick & inexpensive risk
relatively simple to carry out minimize bias in ascertainment
of exposure

Limitations: inefficient in evaluation of

inefficient in evaluation of rare rare diseases
exposure expensive
can not directly compute risk time consuming
difficult to establish temporal loss to follow up create
relationship problem
determining exposure will
often rely on memory
59
 3. Cross sectional study
 Even though the main purpose of cross sectional
study is for describing occurrence of disease by time,
place and person, it can be considered as both
descriptive and analytic study design.
 The data collected by cross-sectional study design
can be analyzed in two ways, either by comparing
the prevalence rate of the outcome in exposed
versus non-exposed people, or by comparing the
prevalence rate of the exposure in those with and
with out the outcome.

60
 One feature which distinguishes cross
sectional studies from other types of
observational analytic studies is the timing of
the subdivision of the study population into
comparison groups.
 In cohort and case control studies, this takes
place prior to the data collection process.
 In a cross sectional study, this takes place
after the information has been collected.
61
 II- Experimental/ Intervention
studies
 Individuals are allocated into experiment or control
group by the investigator.
 If done properly, experimental studies can produce
high quality data.
 Experimental is the gold standard study design
compared to other designs.
Key Features of Experimental Design
1)Investigator manipulates the condition under study
2)Always prospective

62
 Classification of
Intervention Studies
 Intervention studies can generally be considered
either therapeutic or preventive.
a. Therapeutic (or secondary
prevention) trials:
 are conducted among patients with a particular
disease to determine the ability of an agent or
procedure to diminish symptoms, prevent
recurrence, or decrease risk of death from that
disease.

63
 Therapeutic Trial

Improved
outcome
New
Treatment No improvement

Sick B
R
Individuals Improved
outcome
Existing
Treatment
No improvment

R=randomization B=double blinding

64
b. Preventive (or primary
prevention) trials: involves the
evaluation of whether an agent or procedure
reduces the risk of developing disease among
those free from that condition at enrollment.
While therapeutic trials are virtually always
conducted among individuals, primary prevention
measures can be studied among either individuals
(e.g. vaccine trial), or entire population
(Community trial).
65
 Prevention Trial
With outcome

Vaccinated
Without outcome

Healthy B
R
Individuals With outcome
Not-
Vaccinated
Without outcome

R=randomization B=double blinding

66
 A. Classification based on the
population studied
Clinical trial
usually performed in clinical settings and the
subjects are patients
Field trial
used in testing medicine for preventive purpose
subjects are healthy people e.g. vaccine trial
Community trial
unit of the study is group of people/community
e.g. fluoridation of water to prevent dental caries
67
B. Classification based on design
Uncontrolled trial
no control group
control will be past experience (history)
Non-randomized controlled
there is control group
allocation to either group is not randomized
Randomized controlled
there is control group
there is random allocation of subjects to either
group

68
 C. Classification based on
objective
Phase I -clinical pharmacologic study
 trial on small subjects to test a new drug with small
dosage to determine the toxic effect /to provide
preliminary information on drug safety.
 20-80 healthy volunteers needed.
Phase II-efficacy studies
 trial on small group to determine the therapeutic effect
and additional information on safety including side
effects.
 100-200 ill volunteers needed.
 Approximately 70% drug trials proceed from phase I to
phase II
69
Phase III
 study on large population
 to test effectiveness
 usually randomized controlled trial
 show an advantage of an experimental therapy/new
drug over a standard one using an objective outcome
such as improved survival
Phase IV
 Trials are about “post-marketing surveillance” to
determine long- term safety and efficacy of the drug
 A phase 4 trial may be needed because rare and slowly
developing adverse events may not become evident
during the phase 3 trial
70
 Problems related to experimental studies
1. Ethical considerations prevent evaluation of many
treatments or procedures using intervention studies
Some of the ethical issues:
Practices or substances already known to be harmful
should not be used in this study.
Therapies known to be beneficial should not be
withheld from any affected individuals in the study
population.
Investigators have to have a complete knowledge of
the subject under study.
The researcher must have at least informed consent
from each study participant and subjects should be
left free to withdraw from the study at any time.
A written research protocol is a must

71
2. Feasibility / practical issues
– subject recruitment, getting adequate
individuals to enroll into the study is not
easy.
– difficult to achieve satisfactory
compliance.
3. Cost
Experimental studies are very expensive

72
 Steps in conducting experimental
studies

Step 1. Identify new drug/intervention/prevention

Step 2. Identify comparison - e.g. standard
treatment or placebo
Step 3. Define eligible population/ exclusions
Step 4. Define the outcomes and how to assess
them
Step 5. Write the protocol
Step 6. Obtain research ethics committee
approval
Step 7. Recruit and consent required sample of
patients or subjects
Step 8. Allocate individuals into: 73
A. Experimental (study group)
 group that will receive a drug, vaccine or other
procedure
B. Control group
 group that will receive no treatment, a placebo, or
standard form of therapy
Random allocation of the subjects in to experimental
and control group is important
Step 9. Collect data
 Collect all relevant information including the outcome
 The knowledge of participant's treatment status
might influence the identification or reporting of
relevant events. This can be overcome by the use of
placebo.

74
 Experimental con’t
• The Gold standard is achieved through
– Randomization
– Blindness
– Use of placebo

75
 Randomization
Random Allocation
Every one with a known chance receiving a
treatment
Maximize the chance of comparability of study
groups at baseline:
Ideally, both groups are at the same stage of the
natural history of disease
Ideally, only one factor affecting the outcome of
interest would vary between the study groups
76
 Advantages of randomization
 The potential for bias in allocation to study
groups is removed, it eliminates selection bias
 On average the study groups will be
comparable (confounders will be equally
distributed controls confounding effect ),
study subjects will tend to be comparable with
respect to all variables except for the
interventions being studied.

77
 Techniques of Randomization
 Standard ways:
– Random number tables
– Computer programs
 NOT legitimate
-Birth date(before 15th /after 15th)
-Last digit of the medical record number(1-5/6-10)
-Odd/even room number
-Arrival time(morning/afternoon)
-Alphabetical order of names (A-M=Rx, N-Z=control)
78
Placebo - is an inert agent indistinguishable
from the active treatment
Placebo effect - is the tendency of individuals to
report favorable response to any therapy
regardless of the physiologic efficacy of what
they received

79
 Blinding
 Blinding is when the observers and/or
subjects are kept ignorant as to the group to
which the subjects are assigned
 Subjects who can be masked/blinded
- Study participants
-Caregivers/treaters
-Data collectors/assessors of outcome
-Data analysts
- Investigators 80
 Types of blinding/ masking
1. Single blinding - the observer is aware but the
subject is not aware of treatment assignment

2. Double blinding - Neither the observer nor the

subject is aware of treatment assignment

3. Triple blinding - The observer, subject, and data

analyst are not aware of treatment assignment

81
 Why Not Blind?

Impossible
- Surgery
-Exercise
-Diet
-Education
 Possible, but
-Dangerous
-Painful
-Cumbers 82
Step 10. Analyze the results
Analysis is similar to cohort study

Step 11. Publish/ disseminate

findings

83
 Experimental designs are chosen when

The research question cannot be answered by

observational studies
Earlier observational studies have not
answered the research question
Existing knowledge is not sufficient to
determine clinical or public health policy
An experiment is likely to provide an
important extension of this knowledge

84