1

OBSERVATIONAL ANALYTICAL
STUDY DESIGN

Abdul Gani Soulisa

Faculty of Dentistry
Trisakti University
 2

OVERVIEW

1. Study design in epidemiology

2. Observational analytical study
 3

SESSION OBJECTIVES
To enable students to:
- Understand the design principles of epidemiology study
- Understand the design of observational analytical studies
- Identify the most appropriate study design to answer a spesific
research question
 4

Study Design:
1. Observational (non-experimental) study
- The researcher observes what happens to people under exposure
conditions
- The researcher can choose what exposures to study, but does
not influence them

2. Experimental study (drg Armelia S)

11/06/2020
 5

 Observational Study:
1. Descriptive
- Descriptive studies describe how things are
- Oftenly use to establish disease prevalence
- Useful for public health and health care planning

11/06/2020
 6

2. Analytical
- Designed to test hypotesis (predicting an association between
variables)
- Analytical observational studies can be of three types:
a. Cohort
b. Case Control
c. Cross Sectional

11/06/2020
 7

OBSERVATIONAL
STUDIES

1. Cohort
2. Case-Control
3. Cross-Sectional
 8

1. COHORT
• Latin = cohors = enclosed yard or company of soldiers
• Concept: a groups of individuals that are all similar in same trait
and move forward together as a unit
• Epidemilogical definition is any group of people followed or
traced over time
• Typically examines multiple health effects of an exposure; subjects
are defined according to their exposure levels and followed for
disease occurrence
 9

ESSENCE OF COHORT STUDY

• Basis of all epidemiological studies
• A defined group iniatially free of the outcomes of interest (at
baseline), followed over time
• Rates/risks of disease are compared for exposed and not exposed
 10

THE NURSES HEALTH STUDY

- Established in 1976
- 121,700 nurses, age 30-55 yrs throughout USA
- Initial aim was to study the long term health effects of oral
contraceptives, especially the risk of breast cancer
- Nurses were chosen to improve accuracy of self-reported
information
- Initial questionnaire including reproductive risk factors, family
history of cancer, OC use, other lifestyle factors
- Questionnaires mailed every 2 years, additional questions added
over time
11
 12

BIAS IN COHORT
STUDIES
• Selection bias
- Inappropriate comparison groups
- Non-participation at baseline
- Loss to follow-up (major problem)
• Information bias
- Poor exposure measurement
- Poor outcome measurement
• Confounding
- Other risk factors
 13

CONTROLLING CONFOUNDING WHEN

DESIGNING COHORT STUDIES
• Restriction
- Restrict entry to people with same values for potential
confounding variables (eg women only)
• Matching
- Select unexposed subjects so that they have the same distribution
of confounding variables as the exposed
- Ensure that there is no relationship between exposure and
matching variable
 14

STRENGTHS OF COHORT
STUDIES
• Strong design for assessing causality as exposure occurs before
outcome
• Usually eliminates recall bias
• Good for rare exposure
• Good for fatal disease
• Multiple outcome can be studied
 15

WEAKNESSES OF COHORT
STUDIES
• Expensive
• Time consuming
• Ineffecient for rare disease
• Maintaining high levels of follow-up can be difficult
 16

2. CASE-CONTROL STUDY

- First step in identification of risk factors

- Method to measure exposure consistently both for cases and controls
17
 18

Strengths (relative to cohort studies)

- Cheaper
- Faster
- Useful for rare diseases

Weaknesses
- Usually cannot calculate incidence or absolute measures of association
- Can only study one outcome
- More susceptible to selection bias and information bias than cohort
studies
 19

3. CROSS-SECTIONAL STUDY

• Known as prevalence survey

• Identify risk factors for disease
• Measure prevalence of disease/risk factors
• Measure everything at the same time
• Generalisability is very important
20
 21

WHEN ARE CROSS-SECTIONAL

STUDIES USEFUL?

• For measuring prevalence of disease/risk factors

• For identifying risk factors when:
- Diseases are long duration or slow onset
- For exposures that do not change over time (eg blood group)
 22

Strengths
- Usually population-based, representative
- Study several outcomes
- Short, cheap

Weaknesses
- Can be difficult to determine temporal association between
exposure and disease
- Higher proportion of cases with longer duration are sampled
- Not good for rare exposure
 23

The Pyramic of evidence (Ho et al, 2008)

 24

RECOMMENDED TEXTBOOK
BR Kirkwood and JAC Sterne. Essential Medical Statistics. 2nd
edition. Blackwell Science, 2003
Chattopadhyay, A. Oral Health Epidemiology: Principles and
Practice. Boston: Jones and Bartlett Publishers, 2011
Elwood JM. Critical appraisal of epidemiological studies and
clinical trials. 3rd edition. Oxford: Oxford University Press, 2007
Hennekens CH, Buring JE. Epidemiology in medicine. Boston:
Little brown, 1987
Webb P, Brain C, & S Pirozzo. Essential Epidemiology.
Cambridge University Press, 2005
 25

REFERENCES
 Aschengrau, A & Seage, GR. Essentials of Epidemiology in Public Health. London: Jones
& Bartlett Publisher, 2003
 Elwood JM. Critical appraisal of epidemiological studies and clinical trials. 3rd edition.
Oxford: Oxford University Press, 2007
 English D. Epidemiology and Analytic Methods 1. Melbourne University, 2010
 Roeslan BO. Ikhtisar Metodologi Penelitian Bidang Kedokteran. Jakarta: Abadi Dhaya
Insani, 2003
 Russel M & Dharmage S. Study Design in Epidemiology. Melbourne University, 2010
 Slade, G. Study Designs Used in Clinical Research. Retrieved September 10, 2012, from
http://www.arcpoh.adelaide.edu.au/workshop/documents/ARCPOH_Slade.pdf
 Webb P, Brain C, & S Pirozzo. Essential Epidemiology. Cambridge University Press,
2005
 Woodward, M. Epidemiology: Study Design and Data Analysis. 2nd Edition. London:
Chapman & Hall/CRC, 2005