RESEARCH METHODOLGY 2

By
Dr. AHMED GHANDOUR
Cohort studies

Individuals in this study are defined on the basis of presence of

absence of exposure to suspected risk factor for a specific disease but
the group of individuals are free from the disease at the time their
exposure are defined.
Cohort study
Main features of cohort studies:
A - exposure to the risk factor → has occurred and known to the investigator but not the
disease.

B - The investigator proceeds from cause to effect

C - It uses a control group → to accept or reject the hypothesis.

D - The investigator can examine multiple effects of a single exposure

e.g., oral contraceptives may have multiple effects:

Breast cancer – Ovarian cancer – Cardiac effects – Melanoma – Thrombo-embolism.
Cohort studies

Develop
With disease
characteristic
Do not develop
Reference Sample disease
Population
Develop
disease
Without
characteristic Do not develop
disease
Cohort studies
e.g.:

Develop lung cancer

Smokers

Do not develop lung

Adolescent Sample cancer

Develop lung cancer

Non smokers
Do not develop lung
cancer
Cohort studies
Basic steps of cohort study:
1. Define your study population ( all population, sector, subsector)
2. Select the target population (the cohort)
3. Selection of cohort → based on exposure → exposed and non-
exposed cohorts.
4. Obtain data on exposure and ensure matching of both cohorts
5. Follow up the two cohorts
6. Obtain data on outcome
Types of cohort studies:

1. Prospective cohort study

2. Retrospective cohort study
3. Prognostic cohort (follow up study).
a. Prospective cohort:

It starts with a reference population or a representative sample →

some of whom → have certain characteristics or attributes relevant to
the study (exposed group) with other → do not have those
characteristics (unexposed group).

Both groups should be free from the condition under consideration at

the outset of the study.

Both groups are then observed over a specified period → to find out
the risk each group has at developing the conditions of interest
b- Retrospective cohort study: (historical cohort study):

This study depends upon the availability of data or records that allows
reconstruction of the exposure of cohorts to a suspected risk factor and
follow up of their mortality and morbidity overtime.

In other words although the investigator was not present when the
exposure was first identified → he reconstructs exposed and
unexposed populations from records and then proceeds as though he
has been present throughout the study.
c- Prognostic cohort: (follow up studies):

It is a special study used to identify factors that might influence the

prognosis after a diagnosis or treatment.

Features of follow up studies;

1. The cohort consists of cases diagnosed at fixed time or cases treated at
fixed time.
2. All cases have a specified disease but free of the outcome of interest.
3. The outcome of interest is usually survival, cure, improvement, disability,
recurrence, … etc.
• issues in analysis:
A. calculation of incidence rates:
among exposed (Ie) and non exposed (I0)

Outcome
Exposure Cases Controls Total
(develop a (do not
disease) develop a
disease)
Exposed a b a+b
Non c d c+d
exposed
• Ie = [a / (a + b)] x F
• I0 = [c /(c + d)] x F
• F = constant 100-1000……..

b. Ascertain whether there is a significant statistical association between

exposure and disease →
Z test or chi square
c. Quantify risk:
- Relative Risk
- Attributable risk
- Population Attributable risk
- Attributable Fraction (Ecologic Fraction or Attributable Risk Ratio)
Relative Risk:
relative risk = Ie / I0
Ie = incidence among exposed
I0 = incidence among non exposed

If above 1 → +ve association

• Relative risk → answer the question: how many times a person who is exposed →
is at risk compared to non-exposed ?
Attributable risk
= Ie – I0 ( risk difference)

Attributable risk → answer the question: how many (morbidity or

mortality) could be attributed to the risk factor?
Population Attributable risk:

=
Incidence rate among general population – incidence among unexposed x100
Incidence rate among general population

It represent the % of incidence in the total population attributable

to…..
Attributable Fraction (Ecologic Fraction or Attributable Risk Ratio):

= Attributable risk
Incidence among exposed group
Example:

Hypertension
Exposure Total
Yes No

Smoking 120 (a) 280 (b) 400 (a+b)

Non 30 (c) 570 (d) 600 (c+d)

smoking
• Incidence among population = 0.15

• Incidence among exposed = a / ( a+b )

= 120/400 = 0.30

• Incidence among unexposed = c / ( c+d )

= 30/600 = 0.05

• RR= 0.3/0.05 = 0.6 (> 1) → exposure to smoking → ↑ the chance of disease

occurrence 6 times.
• AR = 0.3 – 0.05 = 0.25 → 25% of new cases of HTN among exposed group can be attributed to
smoking.

• Population attributable risk = (0.15-0.05) / 0.15 x 100 = 66.6%

66.6 % of the HTN among population can be attributed to smoking

• Attributable fraction = (0.25/0.3) x 100 = 83.3 %

83.3 % of HTN among smoker can be attributed to smoking
Differences between cohort and case-control
studies:

Cohort study Case – control study

Advantage Disadvantages
1. minimal or no bias 1. liable to bias
selection bias
recall bias (poor memory)
information bias (poor designed tools
for data collection on disease or
exposure
2. can estimate the true risk of the 2. can not……….
disease
3. incidence rate of exposed or non 3. can not estimate incidence rate
exposed → can be determined
4. can calculate relative risk and 4. can not calculate RR – AR
attributable risk but can estimate Odds ratio
5. can determine the exposure / 5. Can not……
outcome relationship
6. possible to study several possible 6.can not ….
outcomes from exposure to same risk But allow study of several risk factors
factor
7. not liable to error 7. liable to error → depends on
NB. In retrospective → need accurate memory
records
8. valuable in rare exposure
(prospective)
Differences between cohort and case-control
studies:

Cohort study Case – control study

Disadvantage Advantages
1. long time 1. short time
2. expensive 2. economic
3. need ↑ manpower 3. need ↓ manpower
4. need large sample 4. does not require larger sample
5. complex statistical analysis 5. simple statistical analysis
6. loss of people (+ve attrition 6. no loss of people
phenomenon, ) - - in
prospective only
7. inefficient for evaluation of rare 7. used in study of rare disease and
disease disease of long latency
Clinical trial (randomized controlled trials)

- In which subjects in a population rare randomly divided into 2

groups → treatment (intervention) group [receive an experimental,
preventive or therapeutic regimen] and control group [ receive
placebo].

• The result are assessed by comparing the outcome of the 2 or

more groups
• single blind placebo trail: → patient is unaware of which ttt he is receiving
• double blind placebo trial: → clinicians and patients are unaware which the ttt or
placebo
• triple blind placebo trial → clinicians , patients, and person who reviewing the data
are unaware which the ttt or placebo
Intervention (experimental) studies
Experimental design characteristics:
a. Formulate hypothesis
b. Experimental group must represent reference population.
c. Informed consent
d. ttt and placebo
e. Randomization
f. Blinding (masking): single, double or triple
g. Appropriate sample size → for statistical power
Intervention (experimental) studies
Example:
Study population

Selection by defined criteria

Potential participants Non participants

Invitation to participate (do not meet selection criteria

Participants Non Participants

Randomization

ttt Control
NB. Randomization is the major difference between clinical trials and prospective trials
Randomization Methods:
1. Simple randomization: e.g.: tossing a coin, shuffling cards,……
2. Block randomization:
3. Stratified randomizations:
4. Unequal randomization:
 Meta analysis
Definition:
It is the systematic process of combining the results of different research
studies using statistical methods to obtain a numerical estimate of an
overall effect.

Its aim is to enhance the statistical power of research findings when number
in the available studies are too small.

• In public health → it is often applied by pooling results of small

randomized controlled trials when no single trial has enough cases to show
statistical significance
 Systemic Review
Systematic review is a review of a clearly formulated question that uses
systematic and reproducible methods to identify, select and critically
appraise all relevant research, and to collect and analyse data from the
studies that are included in the review.