STUDY DESIGNS

Professor Dr. Syed Hatim Noor

Dr. Wan Arfah Nadiah Wan Abdul Jamil
Universiti Sultan Zainal Abidin

1 Syed Hatim Noor

Objectives of the session

 Recognize the most common ways causal

relationships or associations are explored
 Describe the advantages, disadvantages,
strengths and weaknesses of each method
for answering a particular type of clinical
question

2 Syed Hatim Noor

How do we begin to answer the
question?
 Start with the building blocks of any design
 Participants of investigation
 Outcomes of investigation
 Direction of inquiry (prospective or
retrospective)
 Other considerations (e.g. possibility and
resources)

3 Syed Hatim Noor

 About the investigation

 Presence of a comparison group

 Dependent on the objective of the study
 Generally increases the validity of an
observed association
 “Exposure” (or intervention) and outcome
 Must be measured with as little error as
possible

4 Syed Hatim Noor

5 Syed Hatim Noor
 Overview of study designs
 Systematic review and meta analysis
 Intervention [experimental] studies
 clinical trials & community trials
 Observational/ Analytical Studies:
 Cohort studies
• prospective, retrospective and historical
 Case-Control studies
• matched, unmatched and nested
 Cross Sectional studies
• classical, comparative, diagnostic test, questionnaire
 Observational descriptive studies:
 population : prevalence, incidence studies
 individuals: case reports, case series
6 Syed Hatim Noor
Randomised Controlled
Trials
 ‘Gold standard’ test of treatment
 Selection of groups entirely random
 Control group identical to treatment group at
start except for intervention
 Participants/investigators commonly ‘blind’
to group allocation to reduce bias
 May evaluate good and “bad” outcomes

7 Syed Hatim Noor

Example:

RD Randomized controlled trials

Effectiveness of new anti-

RQ
hypertensive drug

8 Syed Hatim Noor

 Parallel
Population

Eligible subjects

Randomisation

Pre-treatment assessement Pre-treatment assessement

test control

Post-treatment
9 assessment
Syed HatimPost-treatment
Noor assessment
Example

Hypertensive Patients

Eligible subjects

Randomization

Pre-treatment assessment Pre-treatment assessment

New drug Usual drug

Post-treatment assessment Post-treatment assessment

Syed Hatim Noor
10
 Cross-over

Population

Eligible subjects

Randomization

Test control

Outcome assessment

Control Test

Outcome assessment

11 Syed Hatim Noor

 Cross-over

Eligible subjects

Randomization

Pre-treatment assessment Pre-treatment assessment

New treatment Placebo

Post-treatment assessment Post-treatment assessment

Wast-out period Wash-out period

Pre-treatment assessment Pre-treatment assessment

Placebo New treatment

Post-treatment assessment Post-treatment assessment

12 Syed Hatim Noor

Prospective Cohort

13 Syed Hatim Noor

Cohort studies

 A group of people (cohort) is assembled,

none of whom has experienced the
outcome of interest
 On entry, people are classified according to
characteristics that might be related to
outcome
 Other names: longitudinal, prospective,
incidence studies

14 Syed Hatim Noor

 Design of a Prospective
Cohort Study

+ Disease
EXPOSED
No Disease

+ Disease
UNEXPOSED

- Disease

15 Direction of Inquiry
 Design of a Retrospective
Cohort Study (RRR)

+ Disease
EXPOSED
No Disease

+ Disease
UNEXPOSED

- Disease

16 Direction of Inquiry
RQ maternal & fetal outcome in
mothers with PIH

RD Prospective cohort

17 Syed Hatim Noor

 Risk estimates

 To estimates risk of event to occur when exposed to a

risk factor
 Relative risk (RR)
 Odds ratio (OR)
 OR/ RR = 1: No risk effect
 OR/ RR < 1: Protective effect
 OR/ RR > 1: Risk effect

18 Syed Hatim Noor

 Risk estimates (Relative risk)
Relative risk (RR)
 Used in cohort study

 The risk is the relative incidence in the exposed

and non exposed group
 RR = Proportion of disease in exposed group

Proportion of disease in unexposed group

= Incidence of disease in exposed
Incidence of disease in non-exposed

19 Syed Hatim Noor

 Risk estimates (Relative risk)
Disease Total
Yes No
Exposure Yes a b a+b
No c d c+d
Total a+c b+d N

RR = a/(a+b)
c/(c+d)

20 Syed Hatim Noor

Case-control

21 Syed Hatim Noor

Overview of case-control
study design
 Analytic study design

 looking back in nature

 we were not there to measure risk directly
 assoc. outcome (disease) with prior (?)
exposure

 Calculate indirect estimate of risk: odds ratio

22 Syed Hatim Noor

Case-control studies

 Compare the frequency of a risk factor in a

group of cases and a group of controls
 There must be a comparison group that
does not have the disease
 There must be enough people in the study
so that CHANCE does not play a large part
in the observed results
 Groups must be comparable except for the
factor of interest

23 Syed Hatim Noor

 Design of a Case-Control study

EXPOSED
CASES

UNEXPOSED

EXPOSED
Controls
UNEXPOSED

24
Direction of Inquiry
Risk estimates (Odds ratio)

Odds ratio
 Used in cross sectional, retrospective
studies, case control
 The odds of an event is the ratio of the
number of ways the event can occur to
the number of ways the event cannot
occur

25 Syed Hatim Noor

 Risk estimates (Odds ratio)
Disease Total
Yes No
Exposure Yes a b a+b
No c d c+d
Total a+c b+d N

OR = a/(a+c) ÷ b/(b+d) = a/c ÷ b/d = ad/bc

c/(a+c) d/(b+d)

26 Syed Hatim Noor

How might we design the
study?

Cross-Sectional Study
27 Syed Hatim Noor
 Observation

Population

Sample

Exposed Disease


Not exposed
No disease

28 Syed Hatim Noor

 Distinguishing Features
 Observe at one
particular point in time
or over a period
 exposures & outcomes
measured at the same
time
 information obtained
from subjects only
once

29 Syed Hatim Noor

Types of
cross-sectional
studies

30 Syed Hatim Noor

Types

 Descriptive - Prevalence studies (a

point prevalence or a period
prevalence)

 Analytical - Cross-sectional
- Comparative cross-
sectional

31 Syed Hatim Noor

Analytical

 analytical is valid only when the

current values of the exposure
variables are extremely stable over
time
 Classical cross-sectional and
comparative cross-sectional studies

32 Syed Hatim Noor

 Observation

Population

Sample

Exposed Disease


Not exposed
No disease

33 Syed Hatim Noor

Comparative cross-sectional

 A comparative way of conducting a

cross-sectional study
 Samples are drawn from two or more
defined different populations
 measure exposure and outcome
factors
 Investigate the association between
exposure and outcome
34 Syed Hatim Noor
 Observation
Population 1 Population 2

Sample 1 Sample 2

Exposed Disease


Not exposed
No disease

35 Syed Hatim Noor

Examples
 The prevalence of coronary heart
disease and its association with the
level of serum cholesterol
 The prevalence of hypertension in a
city
 Telephone interview on the respiratory
symptoms in an urban community
 The prevalence of smoking habit
amongst adolescents and factors
36
associated with it
Syed Hatim Noor
Examples

 A study on the relationship between

cataract and vitamin status
 A cross-sectional survey of asthma in an
occupational group of animal handlers
 A study on the association between milk
drinking and peptic ulcer
 A comparative cross-sectional study on
prevalence of stress among doctors and
nurses and factors associated with it

37 Syed Hatim Noor

 RR OR

IRR/ OR
 A Guide to Select an Appropriate
Observational Study Design

Case control
No constraint
Rare Prospective
of time and
cohort
budget

Outcome Determine risk

Constraint of Retrospective
time and cohort
Common budget

Do not Cross
determine risk sectional

39 Syed Hatim Noor