02 Study Designs - Research Methodology Workshop - Aug 2011.ppt

Study Designs in Clinical Research
an overview..
an overview..
Workshop on Research Methodology
Workshop on Research Methodology
NIRT-NIE
NIRT-NIE
Chennai
Chennai
3-4 August 2011
3-4 August 2011
Dr. M S Jawahar
Dr. M S Jawahar
Deputy Director (Sr. Gr.)
Deputy Director (Sr. Gr.)
National Institute for Research in Tuberculosis
National Institute for Research in Tuberculosis
Chennai
Chennai

Research question
Research question
 Is it relevant?
Is it relevant?
 Will it fill a gap in knowledge?
Will it fill a gap in knowledge?
 Has it been done before?
Has it been done before?
 Is it feasible?
Is it feasible?

To minimise
To minimise
 Random error
Random error
 Wrong result due to
Wrong result due to chance
chance
 Systemic error
Systemic error
 Wrong result due to
Wrong result due to bias
bias
The goal of clinical research
The goal of clinical research

Clinical research methods…
Clinical research methods…
 Case reports, case series
Case reports, case series
 Cross sectional studies (surveys)
Cross sectional studies (surveys)
 Cohort studies
Cohort studies
 Case-control studies
Case-control studies
 Clinical trials
Clinical trials
 Qualitative research methods
Qualitative research methods
 Meta analysis
Meta analysis
 Systematic reviews
Systematic reviews

Type of studies – hierarchy of efficiency
Type of studies – hierarchy of efficiency
B
I
A
S Case series
Cohort Studies
RCT

Case reports, Case series
Case reports, Case series
 Collections of reports of individual patients with
Collections of reports of individual patients with
the same condition, or of a single patient.
the same condition, or of a single patient.
 Usually of an unusual problem
Usually of an unusual problem
 Used to illustrate an aspect of a condition, the
Used to illustrate an aspect of a condition, the
treatment or the adverse reaction to treatment.
treatment or the adverse reaction to treatment.
Eg: A patient presents with an unfamiliar illness.
Eg: A patient presents with an unfamiliar illness.
You would search for case reports that could
You would search for case reports that could
help you decide on a direction of treatment or to
help you decide on a direction of treatment or to
assist on a diagnosis
assist on a diagnosis

Research Design
Descriptive Analytical
Observational Experimental
Survey
Cohort Case-Control RCT Non-RCT

Research Design
Survey

Research Design
Cross-
sectional

Research Design
Observational Interventional
Cross-
sectional

Temporal profile in observational studies
Temporal profile in observational studies

Cross-Sectional Study
 Conducted at a single point in time or over a
Conducted at a single point in time or over a
short period of time. No Follow-up.
short period of time. No Follow-up.
 Exposure status and disease status are
Exposure status and disease status are
measured at one point in time or over a period.
measured at one point in time or over a period.
 Prevalence studies. Comparison of prevalence
Prevalence studies. Comparison of prevalence
among exposed and non-exp.
among exposed and non-exp.

 Diseases of slow on-set and long duration.
Diseases of slow on-set and long duration.
 Conducted over short period of time
Conducted over short period of time
 Many outcomes can be assessed
Many outcomes can be assessed
 Relatively inexpensive
Relatively inexpensive
 Hypothesis generating
Hypothesis generating
 Very useful for public health planning
Very useful for public health planning
(number of beds in a hospital).
(number of beds in a hospital).

Cross-sectional Study
Cross-sectional Study
Disadvantages
Disadvantages
 Difficult to separate cause from effect, because
Difficult to separate cause from effect, because
measurement of exposure and disease done at the
measurement of exposure and disease done at the
same time.
same time.
 A person’s exposure status at time of study may
A person’s exposure status at time of study may
have little to do with their exposure status at the
have little to do with their exposure status at the
time disease began.
time disease began.

Cohort Studies
Cohort Studies
 Can either start with group that is exposed or
Can either start with group that is exposed or
with a defined population and wait for exposure
with a defined population and wait for exposure
to occur
to occur
 Exposure determined prior to outcome
Exposure determined prior to outcome

Become Diseased
Remain non-diseased
Become Diseased
Remain non-diseased
Exposed
Non-
Exposed
Incidence
Incidence
Relative
Risk
Cause Effect
Time
Population
Cohort Study sequence

Disease No disease Total
Exposed a b a+b
Non exposed c d c+d
Total a+c b+d a+b+c+d
Cohort study in a 2 x 2 table
Known at the
beginning of the study

Types of cohort studies
Types of cohort studies
Lancet 2002; 359: 341-45

Cohort Studies - advantages
Cohort Studies - advantages
 Able to determine time sequence between
Able to determine time sequence between
exposure & disease
exposure & disease
 Avoids bias in measuring exposure
Avoids bias in measuring exposure
 Able to study multiple exposures & multiple
Able to study multiple exposures & multiple
outcomes
outcomes
 Able to calculate incidence of disease in
Able to calculate incidence of disease in
exposed and unexposed
exposed and unexposed
 Good for rare exposures
Good for rare exposures

Cohort Studies - disadvantages
Cohort Studies - disadvantages
 Requires long follow up period
Requires long follow up period
 Potential for loss to follow up
Potential for loss to follow up
 Require large sample size
Require large sample size
 Inefficient for rare diseases
Inefficient for rare diseases
 Expensive
Expensive

Case-Control Studies
Case-Control Studies
 Cases: persons with a given disease
Cases: persons with a given disease
 Controls: persons without the given disease
Controls: persons without the given disease
 Ascertain exposure or background of the two
Ascertain exposure or background of the two
groups and compare the proportion
groups and compare the proportion
 Best suited for study of diseases where
Best suited for study of diseases where
medical care usually sought, (hip fracture,
medical care usually sought, (hip fracture,
cancer) because this makes it easier to identify
cancer) because this makes it easier to identify
cases
cases

Design of case-control study
Design of case-control study
Cases
Outcome
Exposure
Controls
Time
Exposed
Un-exposed
Odds of
smoking
Odds of
smoking
Odds
Ratio

Cases Controls Total
Exposed a b -
Non exposed c d -
Total a+c b+d -
Case control study in a 2 x 2 table
Case control study in a 2 x 2 table
Fixed at the
beginning of the study

Selection of Cases
Selection of Cases
 Ideally, investigator enrolls all incident cases in
Ideally, investigator enrolls all incident cases in
a defined population in a specified period
a defined population in a specified period
 Select cases from registries, hospitals, clinics
Select cases from registries, hospitals, clinics
 When all incident cases in a population are
When all incident cases in a population are
included, the study is representative; otherwise
included, the study is representative; otherwise
there is potential for bias (e.g. referral bias)
there is potential for bias (e.g. referral bias)

Selection of controls
Selection of controls
 Critical that exposure in controls is
Critical that exposure in controls is
representative of exposure in the population
representative of exposure in the population
 Ideal controls would have same/similar
Ideal controls would have same/similar
characteristics as the cases
characteristics as the cases
 Matching cases to controls
Matching cases to controls

Population-Based Controls
Population-Based Controls
 Best control group is a random sample of
Best control group is a random sample of
individuals from same source population (as
individuals from same source population (as
the cases) who have not developed the disease
the cases) who have not developed the disease
 Population-based controls are the best way to
Population-based controls are the best way to
ensure that the distribution of exposure among
ensure that the distribution of exposure among
the controls is representative
the controls is representative

Hospital Controls
Hospital Controls
 Most frequently used source
Most frequently used source
 May not be representative of exposure rates in
May not be representative of exposure rates in
target population
target population
 Use of other ill persons as controls will
Use of other ill persons as controls will
provide valid result only if their illness is
provide valid result only if their illness is
unrelated to the exposure in question
unrelated to the exposure in question

Advantages of Hospital Controls
Advantages of Hospital Controls
 Convenient
Convenient
 Cheap
Cheap
 Numerous
Numerous
 Avoids non-response
Avoids non-response

Use of Multiple Controls
Use of Multiple Controls
 Case to control ratio used is usually 1:1
Case to control ratio used is usually 1:1
 If a study has a small number of cases,
If a study has a small number of cases,
increasing the number of controls
increasing the number of controls
(maximum 4) increases power of study
(maximum 4) increases power of study

Advantages of Case Control Design
Advantages of Case Control Design
 Relatively inexpensive
Relatively inexpensive
 Good for diseases with long latency
Good for diseases with long latency
 Optimal for rare diseases
Optimal for rare diseases
 Multiple etiologic factors can be evaluated
Multiple etiologic factors can be evaluated
 Shorter time
Shorter time
 Smaller sample
Smaller sample

Limitations of Case Control Design
Limitations of Case Control Design
 Identifying controls may be difficult
Identifying controls may be difficult
 Difficult to determine representativeness of
Difficult to determine representativeness of
cases & controls
cases & controls
 Temporal relationship between exposure &
Temporal relationship between exposure &
disease difficult to establish
disease difficult to establish
 Prone to recall bias
Prone to recall bias

A Case–Control Study of HIV Seroconversion in
A Case–Control Study of HIV Seroconversion in
HCW after Percutaneous Exposure
HCW after Percutaneous Exposure
NEJM 1997; 337 (21): 1485-1490
NEJM 1997; 337 (21): 1485-1490
 Risk of HIV infection after percutaneous exposure to HIV-infected blood is 0.3
Risk of HIV infection after percutaneous exposure to HIV-infected blood is 0.3
percent, but the factors that influence this risk are not well understood.
percent, but the factors that influence this risk are not well understood.
 Methods
Methods: case–control study of HCWs with occupational, percutaneous exposure to
: case–control study of HCWs with occupational, percutaneous exposure to
HIV-infected blood. The cases - those who became seropositive after exposure to
HIV-infected blood. The cases - those who became seropositive after exposure to
HIV. The controls were those exposed to HIV but did not seroconvert.
HIV. The controls were those exposed to HIV but did not seroconvert.
 Results
Results Logistic-regression analysis based on 33 case patients and 665 controls showed
Logistic-regression analysis based on 33 case patients and 665 controls showed
that significant risk factors for seroconversion were
that significant risk factors for seroconversion were
 deep injury (odds ratio = 15; 95 percent confidence interval, 6.0 to 41), injury with a device
deep injury (odds ratio = 15; 95 percent confidence interval, 6.0 to 41), injury with a device
that was visibly contaminated with the source patient's blood (odds ratio = 6.2; 95 percent
that was visibly contaminated with the source patient's blood (odds ratio = 6.2; 95 percent
confidence interval, 2.2 to 21),
confidence interval, 2.2 to 21),
 a procedure involving a needle placed in the source patient's artery or vein (odds ratio = 4.3;
a procedure involving a needle placed in the source patient's artery or vein (odds ratio = 4.3;
95 percent confidence interval, 1.7 to 12),
95 percent confidence interval, 1.7 to 12),
 exposure to a source patient who died AIDS within two months (odds ratio = 5.6; 95
exposure to a source patient who died AIDS within two months (odds ratio = 5.6; 95
percent confidence interval, 2.0 to 16).
percent confidence interval, 2.0 to 16).
 The case patients were significantly less likely than the controls to have taken zidovudine
The case patients were significantly less likely than the controls to have taken zidovudine
after the exposure (odds ratio = 0.19; 95 percent confidence interval, 0.06 to 0.52).
after the exposure (odds ratio = 0.19; 95 percent confidence interval, 0.06 to 0.52).
 Conclusions
Conclusions The risk of HIV infection after percutaneous exposure increases with a
The risk of HIV infection after percutaneous exposure increases with a
larger volume of blood and, probably, a higher titer of HIV in the source patient's
larger volume of blood and, probably, a higher titer of HIV in the source patient's
blood. Postexposure prophylaxis with zidovudine appears to be protective.
blood. Postexposure prophylaxis with zidovudine appears to be protective.

Clinical Trials
Clinical Trials

Types of Clinical Trials
Types of Clinical Trials
 Phase I – assesses safety, in volunteers
Phase I – assesses safety, in volunteers
 Phase II – efficacy and safety, in patients
Phase II – efficacy and safety, in patients
 Phase III – detailed investigation of efficacy and
Phase III – detailed investigation of efficacy and
safety in large number of patients
safety in large number of patients
 Phase IV – assessment of post-marketing
Phase IV – assessment of post-marketing
surveillance
surveillance

The experimental study design, or
The experimental study design, or
the clinical trial, is a procedure that
the clinical trial, is a procedure that
can provide data of such quality
can provide data of such quality
that it most closely resembles the
that it most closely resembles the
controlled experiment done by
controlled experiment done by
basic science researchers
basic science researchers

Ecological studies
Ecological studies
 Epidemiological studies in which the unit of analysis is a
Epidemiological studies in which the unit of analysis is a
population rather than an individual (smoking and lung
population rather than an individual (smoking and lung
cancer deaths in different countries). Inferior to cohort
cancer deaths in different countries). Inferior to cohort
and case-control studies because it is susceptible to
and case-control studies because it is susceptible to
ecological fallacy.
ecological fallacy.
 Can be confused with cohort studies. The difference is
Can be confused with cohort studies. The difference is
that in ecological studies there is no information on the
that in ecological studies there is no information on the
individual members of the populations compared (e.g.
individual members of the populations compared (e.g.
comparing several states based on state-wide average air
comparing several states based on state-wide average air
pollution and state-wide average prevalence of respiratory
pollution and state-wide average prevalence of respiratory
diseases); whereas in a cohort study the data pair
diseases); whereas in a cohort study the data pair
exposure/outcome is known for each individual
exposure/outcome is known for each individual
 Can be carried out easily, quickly and inexpensively using
Can be carried out easily, quickly and inexpensively using
data that are already available. If interesting and strong
data that are already available. If interesting and strong
associations are observed, the results can provide the
associations are observed, the results can provide the
opportunity for later, more carefully designed studies
opportunity for later, more carefully designed studies

Ecological fallacy
Ecological fallacy
 Fallacy in interpretation of statistical data in an
Fallacy in interpretation of statistical data in an
ecological study, where inferences about nature of
ecological study, where inferences about nature of
specific individuals are based upon aggregate statistics
specific individuals are based upon aggregate statistics
for group. This fallacy assumes that
for group. This fallacy assumes that individual
individual members
members
have the
have the average
average characteristics of the group at large.
characteristics of the group at large.
However, statistics that accurately describe
However, statistics that accurately describe group
group
characteristics do not necessarily apply to
characteristics do not necessarily apply to individuals
individuals
within that group.
within that group.
 Stereotypes, which assume that groups are
Stereotypes, which assume that groups are
homogeneous, are one form of ecological fallacy. Eg, if
homogeneous, are one form of ecological fallacy. Eg, if
a particular group of people are measured to have a
a particular group of people are measured to have a
lower average IQ than the general population, it is an
lower average IQ than the general population, it is an
error to assume that
error to assume that all
all members have a lower IQ than
members have a lower IQ than
the general population.
the general population.

Meta-Analyses
Meta-Analyses
 A systematic, objective way to combine data from
A systematic, objective way to combine data from
many studies, usually from RCTs, and arrive at a
many studies, usually from RCTs, and arrive at a
pooled estimate of Rx effectiveness and statistical
pooled estimate of Rx effectiveness and statistical
significance.
significance.
 Can also combine data from case/control and
Can also combine data from case/control and
cohort studies. The advantage is that it increases
cohort studies. The advantage is that it increases
sample size and allows for analyses that would not
sample size and allows for analyses that would not
otherwise be possible.
otherwise be possible.
 Two problems - publication bias (negative studies
Two problems - publication bias (negative studies
often not published) and quality of the design of
often not published) and quality of the design of
the studies from which data is pulled
the studies from which data is pulled

Systematic Reviews
Systematic Reviews
 A comprehensive survey of a topic that takes
A comprehensive survey of a topic that takes
great care to find all relevant studies of the
great care to find all relevant studies of the
highest level of evidence, published and
highest level of evidence, published and
unpublished, assess each study, synthesize
unpublished, assess each study, synthesize
the findings from individual studies in an
the findings from individual studies in an
unbiased, explicit and reproducible way and
unbiased, explicit and reproducible way and
present a balanced and impartial summary
present a balanced and impartial summary
of the findings with due consideration of
of the findings with due consideration of
any flaws in the evidence.
any flaws in the evidence.

More rigorous than a traditional literature review
More rigorous than a traditional literature review
 Clearly formulated research question
Clearly formulated research question
 Published & unpublished literature searched
Published & unpublished literature searched
 Identified research is assessed according to an explicit
Identified research is assessed according to an explicit
methodology
methodology
 Results of critical assessment of studies combined
Results of critical assessment of studies combined
 Final results placed in context, addressing issues such
Final results placed in context, addressing issues such
as quality of studies, impact of bias & applicability of
as quality of studies, impact of bias & applicability of
findings
findings
The difference between a systematic review and meta-
The difference between a systematic review and meta-
analysis is that a systematic review looks at the whole
analysis is that a systematic review looks at the whole
picture (qualitative view), while a meta-analysis looks
picture (qualitative view), while a meta-analysis looks
for specific statistical picture (quantitative view).
for specific statistical picture (quantitative view).
Systematic Reviews
Systematic Reviews

Summary
Summary
 Clinical research maybe observational or
Clinical research maybe observational or
experimental
experimental
 Experimental trials maybe randomised or non-
Experimental trials maybe randomised or non-
randomised
randomised
 Observational studies maybe descriptive (no
Observational studies maybe descriptive (no
comparison group) or analytical (with comparative
comparison group) or analytical (with comparative
group)
group)
 Analytical studies may be prospective (cohort) or
Analytical studies may be prospective (cohort) or
retrospective (case-control)
retrospective (case-control)
 Descriptive studies cannot investigate associations as
Descriptive studies cannot investigate associations as
exposure and outcome are measured at the same
exposure and outcome are measured at the same
time
time

02 Study Designs - Research Methodology Workshop - Aug 2011.ppt

More Related Content

Similar to 02 Study Designs - Research Methodology Workshop - Aug 2011.ppt

More from ParameshwariPrahalad

Recently uploaded

02 Study Designs - Research Methodology Workshop - Aug 2011.ppt

Editor's Notes