2. Module 3- Study types/designs
Learning Objectives
At the end of this section, you will be able to:
Describe the common types of study designs used in
HSR.
Mention the advantages and limitations of each type of
study design
Identify the most appropriate study design for the
research proposal you are developing.
2
3. Study types or study designs
The type of study chosen depends on:
The type of problem;
The knowledge already available about the problem;
and
The resources available for the study.
Study designs broadly can be classified as
interventional or non interventional studies.
3
4. Study design….
Table: Classification of research study designs
4
I. Non-interventional (observational) studies
Exploratory Qualitative
Ecological (correlational) population as
study unit
Descriptive
Studies
Epidemiologicalstudy
designs(Quantitative)
Case reports
Case series
Cross-sectional surveys
individualas
studyunit
Cross-sectional comparative study
Case control
Cohort
Analytical
Studies
II. Interventional studies
Experimental studies (Randomized)
Quasi-experimental studies (Not Randomized)
5. Study designs…
An EXPLORATORY STUDY is
a small-scale study of relatively short duration,
carried out when little is known about a situation or a
problem.
It may include description as well as comparison.
It may include description as well as comparison.
For example:
Description: To explore needs of HIV positive and AIDS
patients, a number of in-depth interviews can be held with
various categories of patients (males, females, married, single)
and with some counselors working on a program that is already
under way.
5
6. Exploratory study…
Comparison: To identify variables that help to
explain why one group of persons or objects
differs from another.
To explain the differences we observe (e.g., in the
needs of male and female AIDS patients) or to
identify causes of problems.
Note: If the problem and its contributing factors
are not well defined, it is always advisable to do
an exploratory study before embarking on a
large-scale descriptive or comparative study.
6
7. 7
Study designs….
EPIDEMIOLOGICAL STUDIES: Purpose
Descriptive studies Analytic studies
Characterize disease
occurrence by time, place
and person.
Generate testable
hypothesis as to the cause
of disease
Concerned with the
search for causes and effects.
Test hypothesis about
association between
exposure and outcome.
10. Study designs…
10
Descriptive studies
A. Dealing with individuals
- Case report/case study
- Case series
- Cross sectional surveys (usually)
B. Dealing with population
- Correlational or ecological(some times)
11. Case Report/case study
11
Careful and detailed report of the profile of a single
patient by one or more clinicians
• Document unusual medical occurrences
• Can generate hypothesis, provide clues in
identification of a new disease or adverse effects of
exposures
(E.g. It was a single case report that formulated the
hypothesis of oral contraceptive use increases venous
thromboembolism)
It is made using
Simple history, physical examination and Lab./ radiologic
12. Case Series Studies
12
Description of clinical/epidemiologic characteristics of
a number of patients (usually 5-12) with a given
disease having similar diagnosis
• Collection of individual case reports occurring within
a fairly short period of time
Used as an early means to identify the beginning or
presence of an epidemic, generate hypothesis and
gives information about natural history of disease
Can suggest the emergence of a new disease (i.e.
PCP …. AIDS)
13. Example of case-series studies
Five young, previously healthy homosexual men
were diagnosed as having PCP at Los Angeles
hospital during a six month period from 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
13
14. Uses of case series studies
Can be valuable early evidence for associations
between exposures and diseases which can be
studied in more detail
Useful for the recognition of new diseases,
Useful for constructing of the natural history of a
disease,
Use to formulate a hypothesis and to detect an
epidemic
14
15. Cont….
15
Limitations of case report and series studies :
• No appropriate comparison group
• Based on the experience of one person so
Can’t be used to test for presence of a valid
statistical association …prone to atomistic
fallacy
not a true epidemiologic design
16. Cross-sectional study
16
Measure disease and exposure status simultaneously
among individuals in a well-defined population at a
point in time …..also called a “prevalence
study/survey”)
Snapshot of the health status of populations at a
certain point in time
17. Cont…
17
Can have descriptive or analytic purposes
The descriptive type is carried out to study prevalence
of health related events at a point in time/snapshot
Diseases, risk factors, coverage of interventions,
health service utilization, knowledge, attitude and
practice
The analytic type is carried out to assess association
between exposure and outcome
Exposure and disease status are assessed
simultaneously among individuals at the same point in
time
Compare prevalence of disease in persons with and
without the exposure of interest
18. 18
Cross-Sectional Studies….
Steps in the conduct of cross-sectional studies:
1. Define a population of interest (reference or source
population)
2. Recruiting a representative sample (adequate size,
random selection)
3. Measure the variables of interest (disease &or exposure)
at the same point in time
4. Analyze the data
20. 20
Cross-Sectional Studies….
Example:
Respiratory problems
Yes no total
Smoking Yes 70 50 120
no 30 70 100
Total 100 120 220
Prevalence of smokers (among respiratory problems)
=70 x100=70%
100
Prevalence of respiratory problems (among smokers)
=70 x100=58.3%
120
21. 21
Cross-Sectional Studies….
Examples…
General household surveys
National Health and Nutrition Examination
Survey
International surveys (International Study of
Asthma and Allergies in Childhood (ISAAC)
22. Cross-sectional Study….
22
Advantage:
helps to determine prevalence …disease burden
Fast/Inexpensive - no waiting!
No loss to follow up
multiple factors and outcomes at same point in time
can be studied
Helps to generate hypotheses
24. Ecologic Studies
24
A study in which one or more exposures or disease
is measured at the population level rather than the
individual level
Uses data from entire population to compare
disease frequencies (average values)
- between different groups during the same
period of time, or
- in the same population at different points in time.
Correlation coefficient (r) is the measure of
association
25. Examples of correlational studies
1. Trend of HIV in Ethiopia
HIV prevalence of Ethiopia at different years or points in
time
2. Geographic distribution of HIV in the regions of Ethiopia
HIV prevalence of different regions of Ethiopia at the
same year or point in time
3. Fluoride content of water and dental caries (correlation)
Proportion of people with dental caries in villages
Vs
Fluoride content of water in villages during the same
period or point in time
25
26. Breast Cancer Mortality and Dietary Fat
Intake
26
Ecologic correlation of breast cancer mortality
and dietary fat intake
27. Ecologic Studies…
27
Limitation
Lack of ability to control for effects of potential
confounding factors.
Inability to link exposure with disease at
individual level
association found with aggregate data (average
values) may not apply to individuals (Prone to
ecological fallacy)
Measurement limitation (Ecological conditions are
difficult to measure at individual level)
E.g environmental contact, fluoride content of water
28. Ecological fallacy: example
28
Imagine a study of the rate of coronary heart disease in
the capital cities of the world relating the rate to average
income.
Within the cities studied, coronary heart disease is
higher in the richer cities than in the poorer ones.
We might predict from such a finding that being rich
increases your risk of heart disease, but
In the industrialised world the opposite is the case -
within cities such as London, Washington and
Stockholm, poor people have higher CHD rates than rich
29. Atomistic fallacy
29
Studies of individuals, case report and case series study,
are prone to the opposite of the ecological fallacy, the so-
called atomistic fallacy.
Wrongly assuming from observations on the causes of
disease in individuals that the same forces apply to whole
populations.
For example, at an individual level a high income or a marker
of material success such as employment, car access etc., is
associated with a lower rate of suicide. But,
Does not mean that populations or societies which are rich
have a lower rate of suicide or better mental health, rather
the opposite seems to be true.
30. 30
Study designs…
Assignment
of exposure by researcher
yes no
Random allocation
to comparison gps
yes No
Experiment e.g
Randomized
Clinical trial
Quasi- experiment
yes
no
Comparison
Descriptive Case-control cohort
ObservationalInterventional
31. ANALYTIC STUDIES
31
Focuses on identifying determinants of a disease by
testing the hypothesis formulated from descriptive
studies
the ultimate goal is judging whether a particular
exposure causes or prevents disease (unwanted health
related event)
Analytic studies are broadly classified into two -
observational and interventional studies.
Both types use "control group", the use of control group
(comparison grp) is the main distinguishing feature of
analytic studies.
In Observational, information is obtained by observation
of events.
No intervention is done, no deliberate interference with
natural course of disease. (cross-sectional, case control,
32. Cont…
32
In Interventional study, the researcher does something
about the exposure and observes the changes on the
outcome or disease.
Investigator has control over who gets exposure and
who don't.
The key is that the investigator assign study participants
into either group, whether it is done randomly(RCT,
Experimental) or not randomly (quasi-experimental).
33. 33
Cohort Studies
Cohort
a group of people who share a common experience or
condition
E.g. Birth cohorts, cohort of smokers, occupational exposures
Cohort studies
The observation of a cohort over time to measure outcome(s)
Because the data on exposure and disease refer to different
points in time, cohort studies are longitudinal
Longitudinal, follow-up or incidence studies
34. 34
Cont…
They have 2 primary purposes:
Descriptive: to describe the incidence rates of
an outcome
Analytic: to analyze associations between the
outcomes and risk factors (Usual type)
35. 35
Cont…
begin with a group of people free of disease
who are classified into subgroups according to
exposure to a potential cause of outcome
and the whole cohort is followed up to see how
the subsequent development of outcome
differs between the groups with and without
exposure (Figure below)
37. 37
Types of Cohort Studies
Closed vs. Open
Closed cohort: exposure groups are defined at the
start of follow-up and no new members are added
during the follow-up
Open/dynamic cohort: people move in and out the
study
38. 38
Cont…
Incidence cohort vs. Prognostic (clinical)
Incidence Cohort Study
To assess incidence of disease
To identify risk factors for disease onset
Incidence greater in exposed than non-exposed?
39. Cont…
39
Prognostic Cohort Study
Follow diseased cohort to assess factors
associated with outcome (recovery or death)
Goal is to identify explanatory/prognostic
factors/ factors helped to the dev’t of the out
come of the disease.
40. 40
Cont…
Prospective vs. Retrospective (Concurrent vs.
Non-concurrent)
Depending on temporal relationship between
initiation of the study and time of collection of
exposure and outcome data from the study
subjects or participants
Cohort studies have been called prospective
studies, but this terminology is confusing and should
be avoided
the term “prospective” refers to the timing of data collection
and not to the relationship between exposure and effect
Thus, there can be both prospective and
41. 41
Prospective cohort studies
Exposure and outcome data is collected
after start of the study
cohorts Identified in the present
exposure status or possible
explanatory/prognostic factors determined in
the present
Cohorts followed-up to identify outcome
Ascertainment of outcome done in future
43. 43
Cont…
Advantages
Exposure precedes outcome
Outcome unknown when exposure determined
Can examine many outcomes of the exposure
Disadvantages
Cost
Time delays
Loss to followup
44. 44
Retrospective cohort studies
all the exposure and effect data have been
collected before the actual study begins
This type of investigation is called a historical
cohort study
Conduct
Identify cohort in the past using
records/databases
Determine exposure or prognostic factors in
the past using again records or databases
then
Identify outcome in past or present or future
(in case of mixed cohort)
45. 45
Cont…
Costs can occasionally be reduced by using a historical
cohort (identified on the basis of records of previous
exposure)
This sort of design is relatively common for studies of
cancer related to occupational exposures
For example, records of military personnel exposure to
radioactive fall-out at nuclear bomb testing sites have
been used to examine the possible causal role of fall-out
in the development of cancer over the past 30 years
47. 47
Cont…
Advantages:
Do not require a very long time (exposures and/or
outcomes have already occurred)
Cheap, if used record linked for outcomes
Disadvantages:
Feasible only when a list of exposed individuals is
available
Exposure data often of poor quality
Usually unable to measure confounders
48. 48
Design and data collection of
cohort study (1)
1. Define and identify cohorts
1.1. Identify population at risk
Selection of Exposed Population
Depends on research question
Depends on frequency of exposure
Common exposures: general population
Rare exposures: selected groups
49. 49
Design and data collection (2)
Outcome must not be rare in exposed
Attributable risk must be high
Accessible and compliant subjects
E.g., Nurse’s Health Study, Physicians
Selection of Non-exposed Group
Similar to exposed
Control for confounding factors
50. 50
Design and data collection (3)
1.2. Screen identified subjects for the disease and
omit the prevalent cases
2. Define, assess, identify and classify exposure
3. Follow-up and ascertain outcome
Timing of outcome events-case definition
51. 51
Sampling
Sample size - for test of significant difference between two
proportions, the following formula can be used:
Parameters:
n - size of sample in each group
P1 ,P2–estimated population prevalence in the comparison
groups
β = 1- Power (the probability that if the two proportions
differ the test will produce a significant difference)
Usually a power of 80% is used
2
21
2211
2
2 11
pp
ppppZZ
n
52. 52
Analysis
(statistically prospective cohorts are summarized using RR but retrospective
by OR)
RR=incidence exp/inc non exp
Presence of association
Population
RR=1 – no association; RR<1 – negative association;
RR>1 – positive association
Sample
P-value<0.05 – statistically significant association
RR≠1
Statistical methods – survival analysis
Strength of association
Weak – RR close to 1; Strong – RR far from 1
53. Case control study
Design concept
Starts with cases and comparative group(control)
We determine what proportion of cases were exposed
and what proportion were not
We also determine what proportion of controls were
exposed and what proportion were not
Also called case-referent or retrospective
53
55. Designing case control studies
I. Selection of cases (definition, eligibility criteria)
Hospitals, other medical care facilities/general population
II. Selection of controls (definition, eligibility criteria)
General population, neighborhood, friends/relatives,
hospital or clinic-based
***The benefit of increased sample size is not as relevant
past the 1:4 ratio (e.g. increase in statistical power).
III. Ascertaining Exposure
Sources of exposure data (cases and controls)
***The measure of association in case control study is Odds
Ratio(OR)
55
56. Advantages of Case-Control Studies
Quick and easy to complete, cost
effective
Most efficient design for rare diseases
Usually requires a smaller study
population than a cohort study
56
57. Uncertainty of exposure-disease time
relationship
Inability to provide a direct estimate of risk
Not efficient for studying rare exposures
Subject to biases (recall & selection bias)
Disadvantages of Case-Control
Studies57
58. INTERVENTION STUDIES
58
Investigator determines who is exposed, ideally
using random methods
Investigator allocates the exposure and follows
for an outcome
Types of interventional studies include
Randomized Clinical Trials
Field Trials
Community Intervention Trials
Quasi-experimental Studies
60. What is an experimental study?
60
Randomized controlled trials are sub-types of
cohort studies in which exposure (i.e.,
treatment) is randomly assigned by the
investigator (or by some other, observable
phenomenon)
Have a long history in clinical medicine
Although experimental studies come in many
types, principles are the same and clinical
trials dominate the field
61. What is a clinical trial?
61
A clinical trial is a prospective study evaluating
the effect and value of intervention(s) in
human beings under pre-specified conditions.
A controlled clinical trial is a prospective study
comparing the effect and value of
intervention(s) against a control in human
beings.
62. 62
The clinical trial is the most definitive tool for
evaluation of the applicability of clinical
research.
It represents a key research activity with the
potential to improve the quality of health care
and control costs through careful comparison
of alternative treatments.
63. When might a RCT be
indicated?63
Exposure is a modifiable factor which persons will let you
modify, e.g., taking a pill, trying a different diet
When there is ethical equipoise, that is when we really do
not know whether a particular exposure is associated with
benefit or with harm- Imperative that informed
consent be obtained
A particular exposure may have an influence on multiple
outcomes of tremendous importance
In experimental trials, in contrast with other epidemiologic
study designs we have discussed, we are doing something
to participants so we have to be certain that, first, we do no
harm
65. Phases…65
Phase I: clinical pharmacology and
toxicity
1st experiment in human for new drug, schedule,
or combination
Primary concern: Safety
Goal: define the maximum tolerated dose (MTD)
in a dose-escalation study
Typically required 15-30 patients
66. Phases…
66
Phase II
Small randomized, controlled, blinded
Tests tolerability and different doses
E.g., optimal dosage without side effects
Applied to patients with relevant illness
Goal - Identify suitable formulation of drug
67. Intervention Trials
67
Phase III
Referred to as clinical trial
Evaluation of efficacy of drug
Usually randomized, blinded, controlled trial
If successful, licensed and marketed
Phase IV
Large studies after approval of drug
Often observational, study long-term effects
Long term efficacy, rate of serious side effects
Evaluate drug in “real life”, additional uses
68. Conducting Trials
68
1. Selecting participants
2. Measure baseline characteristics and describe
sample
3. Randomizing
4. Apply intervention
5. Follow-up and adherence to protocol
6. Measuring outcome
7. Analysis
69. 1. Selection of Participants
69
Terminology
Target population
People to which findings will be generalized
Study population
Subset of target population available/accessible to
study
Selecting subjects
Establish inclusion/exclusion criteria
Sample size
72. Sample size
72
Calculating sample sizes for trials with
dichotomous outcomes (eg, sick vs well)
requires four components:
type I error (α),
power,
event rate in the treatment group(p1),
RR or event rate in the control group(p2)
,RR=P1/p2
74. Participants
74
Inclusion criteria
Define main characteristics of target population that
are relevant to research question
Demographic characteristics
E.g., adults, aged 20-69
Clinical characteristics
E.g., in good health
Geographic characteristics
E.g., living in northern Ethiopia
Temporal characteristics
E.g., inception period Jan 1, 2003 to Dec 31, 2003
75. Participants
75
Exclusion criteria
Subsets of people meeting inclusion criteria (potentially suitable
for research question) except for characteristics that might
interfere with quality of data etc.
High likelihood of being lost to follow-up
E.g., transients
Inability to provide good data
E.g., language barrier or cognitive incapacity
High risk for side effects
E.g., pregnant, lactating
Unethical to withhold treatment
E.g., severe depression
76. Exclusion Criteria
76
Five main reasons for exclusion from clinical
trial
Safety concerns (susceptibility to adverse effects of
active treatment)
Unethical to withhold treatment (tx so beneficial for
some not acceptable to assign placebo
Active treatment unlikely to be effective
Unlikely to adhere to treatment
Unlikely to provide outcome information (e.g., die or
move before study completion)
77. Conducting Trials
77
1. Selecting participants
2. Measure baseline characteristics and describe
sample
3. Randomizing
4. Apply intervention
5. Follow-up and adherence to protocol
6. Measuring outcome
7. Analysis
78. 2. Baseline Characteristics
78
Need enough information to track subjects
Contact persons; address, etc.
Description of participants
Aid in assessing generalizability (e.g., gender, age,
disease severity, etc.)
Risk factors for outcome or to define subgroups
E.g., smoking status, smoking status of spouse
Measure of “outcome” variable
E.g., if pain is “outcome”, need baseline pain
79. Conducting Trials
79
1. Selecting participants
2. Measure baseline characteristics and describe
sample
3. Randomizing
4. Apply intervention
5. Follow-up and adherence to protocol
6. Measuring outcome
7. Analysis
80. 3. Randomization
80
Should be done at the last possible moment,
after eligibility criteria has been determined
and informed consent has been obtained
Important to conceal randomization scheme
from attending health care providers
81. Advantages of Randomization
81
Removes the potential of bias in the allocation
of participants.
Prevents confounding
produce comparison groups
83. Stratified Randomization
83
Select factor(s) of interest
Stratify (divide) group by that factor
Randomize the appropriate proportion of each
group into your treatment groups
Increases the likelihood that your treatment
groups will be comparable on that factor
84. Stratified Randomization
84
E.g., age is important in treatment response
100 in sample want 50 per group
Say 20% of sample are >60 (high risk group)
Randomly select 10 people over 60 for each
group
Then randomly select 40 people under 60 for
each group
85. Conducting Trials
85
1. Selecting participants
2. Measure baseline characteristics and describe
sample
3. Randomizing
4. Apply intervention
5. Follow-up and adherence to protocol
6. Measuring outcome
7. Analysis
86. 4. Applying Intervention
86
Intervention strategy compared with
Placebo
Standard treatment
May have more than one comparison group
Advantages, disadvantages of “placebo”?
Ethical issues
Advantages, disadvantages of “standard treatment”
control?
Interpretation of findings?
87. Blinding
87
Randomization
Control for confounding bias at baseline
Does not control for confounding during follow-up
E.g., differential attention to subjects in treatment arm
Does not control for information bias
Blinding
Controls for
Information bias (e.g., observer bias)
Reduce loss to follow up (reduce selection bias)
88. 88
Blind studies
Single blind The patients do not know
which treatment they receive
Double blind The patient and the observer or
the physician do not know
Triple blind The patient, the observer and
the analyst do not know
89. Conducting Trials
89
1. Selecting participants
2. Measure baseline characteristics and describe
sample
3. Randomizing
4. Apply intervention
5. Follow-up and adherence to protocol
6. Measuring outcome
7. Analysis
90. 5. Follow-up and Adherence
90
Ideally
All subjects adhere to treatment regimens; do not
seek additional treatment; do not drop out, die,
move or have to be withdrawn from study; attend
follow-up sessions and provide outcome data
91. Unplanned Crossovers
91
Unplanned crossover is said when subjects
switch to either treatments
When subjects choose the alternative treatment
Subjects in experimental group start using
“control” treatment or vice versa
Usually have selective cross-over (more subject
from one group cross over)
A large proportion of crossovers may invalidate
study
92. Conducting Trials
92
1. Selecting participants
2. Measure baseline characteristics and describe
sample
3. Randomizing
4. Apply intervention
5. Follow-up and adherence to protocol
6. Measuring outcome
7. Analysis
93. 6. Measuring Outcome
93
Outcome should be
Clinically relevant
Feasible
E.g., choose outcome that is sufficiently common for time and
number of subjects
Able to be measured accurately and precisely
Not too costly
Above all, valid, reliable and sensitive to change
94. Conducting Trials
94
1. Selecting participants
2. Measure baseline characteristics and describe
sample
3. Randomizing
4. Apply intervention
5. Follow-up and adherence to protocol
6. Measuring outcome
7. Analysis
95. Analysis
95
Analyze according to which treatment the
patient was randomized to (which treatment was
intended?)
“intention to treat” analysis
Or according to which treatment they actually
received?
“per protocol” analysis
96. Intention to treat vs. per protocol
analysis96
Intention to treat analysis (management trial):
Includes all randomized patients in the
groups to which they were randomly
assigned, regardless of their adherence with
the entry criteria, regardless of the treatment
they actually received, and regardless of
subsequent withdrawal from treatment or
deviation from the protocol
97. ITT---cont
97
Key points
Use every subject who was randomized
according to randomized treatment
assignment
Ignore noncompliance, protocol deviations,
withdrawal, and anything that happens after
randomization
As randomized, so analyzed
99. Per protocol---cont
99
Per protocol analysis:
Patients who deviate from the protocol are
excluded from the analysis
Advantage: determine efficacy of intervention
Disadvantage: vulnerable to all source of bias
100. QUASIEXPERMENTAL
STUDIES100
In these studies, one characteristic of true
experiment (i.e randomization ) is missing.
But, they always include intervention or
manipulation of the independent variable.
The common quasi-experimental studies are
described below.
a)Non-equivalent control group design
b) Before-After Study design
101. Non-equivalent control group design
101
Uses two or more groups (one serves as a control
group)
The subjects in study (intervention) group and control
group are not randomly assigned.
Figure: Diagram of a quasi-experimental design with two groups
102. Before-After Study design
102
Uses only one group in which an intervention is carried
out. The situation is analyzed before and after the
intervention to test if there is any difference in the
observed problem.
Figure: Diagram of a before-after study