SAIF ALGHAMDI

‫توكلت على اهلل‬

4th YEAR
2017-2018

BY:. Hasan Daghriri & Abdullelah Arishi

 Study Designs

Definition:
Study design is an organized process used to evaluate a hypothesis to test wither
its accepted to rejected or not.

Overview of the design tree

The Study design can be divided into two main types :

Descriptive study and Analytical study , with their subtypes.
 Descriptive study:
We are trying to describe the outcome of something among a specific sample or
group of people (only describing what is happening) like describing its number
or other things by time ,place and person.
Establish the prevalence.
No control group.
* (PO) = (P : the population or the group sample) , (O : the outcome)
Example : The prevalence of insomnia among medical student
P= Medical student , O = insomnia
*So we are trying just to Describe how is common (prevalent) is insomnia among medical
student
* We are just Observing the outcome !
Analytical study (PECO) :
We are trying to Analyze or study the relationship between an Exposure and an
Outcome.
So, there must be an exposure or intervention and an outcome that we are looking to
see if the study sample is exposing to it.
* (PECO) = (P : population or group sample), ( E: Exposure or intervention), (C:
comparison group), (O: outcome)
Example : The prevalence of insomnia among medical student and its effect on
academic performance Or The effect of insomnia on academic performance among
medical student
P = medical student, E= insomnia, C= group with no insomnia, O= academic performance
So here we are trying to analyze the relationship between insomnia and academic
performance , is there an effect or no ? is it a harmful effect or protective effect ?
** Note : The exposure can be a drug , risk factor, education or other things that we
can study its effect on something (outcome).

Types of Descriptive study (PO) :

Case study
Case series
Cross-sectional study
Incidence and prevalence
Types Of Analytical study (PECO) :
It can be divided into two types :-
 1. Observational analytical study : here we observe the relation between
exposure and outcome , just observation without changing the amount of
exposure or its dose.
It include : Cross-sectional study , Case-Control, Cohort study
2. Experimental analytical study : here we manipulate exposure, that mean
we can change the exposure dose or amount to all the study groups not
just observing.
It include : Randomized control trials
Example for analytical study : (The effect of caffeine consumption on blood sugar)

** If we want to conduct an observational analytical study : we will choose a groups of

people who consume Caffeine(E) regard its amount and another group who do not take
caffeine and we will see the blood sugar level(O) in both group does caffeine increase or
decrease blood sugar ??

** if we want to conduct an experimental analytical study : we will choose 2 groups of

people , the study group will be given caffeine of 50mg daily (same amount of exposure
in all individual), and the control group (with no exposure) will be given another drink
which has no caffeine , and we will see the blood sugar in both group.

**Note : cross-sectional study can be descriptive or analytical , it depend on your

research :
- The prevalence of insomnia among medical student using cross-sectional method
(here we detect the number of student having insomnia only)-(PO)
- The prevalence of insomnia among medical student and its effect on academic
performance (here we detect the number of insomniac student and we have a relation
that we would like to study which is the insomnia(E) and academic performance(O)-
(PECO)
Cross sectional: Prevalence ‫تحديد نسبة مثالً السرطان في مناطق مختلفة‬

This Graph represent the hierarchy of study design , which one is stronger and more
evident .
The RCT (randomized control trial) is the gold stander study and the most evident one .
 Based on outcomes
Case-Control Study
Type of study :
Observational analytical study (PECO).
A
In this study we are trying to see a relation between an exposure and an
B outcome in which the outcome is already present and we try to see the past
history of the individual, did he expose to our exposure or no ?
Based on the outcome ( Study begin from the outcome ).
Exposure is a RISK, not a cause ( If the exposure associated with outcome, we
can say that the exposure X is risk for the disease Y " NOT a cause of disease Y " )
It is retrospective study , which mean it goes back in time of the already present
outcome .
There are two groups , one with outcome and the other without outcome
(control).
No randomization. No intervention.

How is it conducted ?
1st : Select 2 group of people : A case group and A control group
The case group is the study group which has the outcome
The control group is the group without the outcome
Note : control group must be equivalent in most aspect of case group
like age and o her fac or b he don ha e he disease of co rse
Note : the case group must have the best confirmative test to say that
they have the outcome
nd
2 : Collect data :
The data can be collected by using a standardized question used in both
groups , you have to ask both groups whether they exposed to the
desired exposure in the past or no , You should ask the same questions in
both with no bias to one group , you can use also other method then
equations like detecting the exposure level in blood (biomarker) .
rd
3 : Measure the association between exposure and outcome and interpret the
result.
The used measure in Case-Control study is Odds ratio

Interpretation of the Odds Ratio :

OR = 1 Exposure NOT related to disease(outcome)
OR >1 Exposure POSITIVELY related to disease(outcome)
OR <1 Exposure NEGATIVELY related to disease(outcome)
* Note : as odds ratio increase , as the strength of relation between E+O is
more .
 A 1200 4
s
exposure 300
Noexposure 100
Example: B 1400 s 0,07
The effect of caffeine(E) on insomnia(O) :
4
0,07 57 I
o Case Group(a+b): Group with insomnia
o Control Group(c+d) : Group with normal sleep pattern
We will ask both groups , did you consume caffeine before ? how much and for
how long?
Collect data from both group , and calculate odds ratio
If result were as follow :

Diseased(Case) Non-diseased (Control)

Exposure 1200 (a) 100 (c)
Non-exposure 300 (b) 1400 (d)
Total 1500 (a+b) 1500 (c+d)

- (of both Groups)

- Case group: , Control Group:

- Then we divide the case group ratio by control group ratio to calculate odds ratio :

- , so there is a true relation between caffeine consumption and

insomnia , caffeine is a risk factor for insomnia
Interpret : People who consume caffeine are 57 times more likely to develop
insomnia than non-caffeine consumer.

Advantage Disadvantage
Easy and quick Cannot generate incident data
Cheap and has good rank Bias (specially recall) is common
Good for rare disease Control selection can be difficult
 Based on exposure
Cohort Study
Type of the study:
Observational analytical study
Aim to see the relation between the exposure and outcome among two groups
without previous outcome.
There are two groups completely healthy or without outcome, one is exposed
and the other not exposed ( control ) , then follow the two groups in the future
to see the outcome.
Based on exposure ( Study begin from the exposure ).
No randomization. No intervention
Time of the study depends on the feature of the outcome ( How long does the
outcome take to appear ? ).
Exposure is a RISK, not a cause ( If the exposure associated with outcome, we
can say that the exposure X is risk for the disease Y " NOT a cause of disease Y " ).

Two designs:
1. Prospective cohort design.
2. Retrospective cohort design.

1) Prospective cohort design:

o Exposure & non-exposure occur during the beginning of the study, then
followed up into the future to see the outcome. Prospective = Less bias
o Requiring the collection of new data Exposure: Beginning of the study
Outcome: From past
o Can measure the incidence.
o Less liable for confounding or bias. Retrospective = More bias
Exposure: From past
How is it conducted ? Outcome: Beginning of the study

1st : Select 2 group of people completely free of outcomes :

(A) Exposed group.
(B) Control group without exposure.
nd
2 :Follow the two groups in the future then collect data about the outcome.
3rd : Measure the association between exposure and outcome and interpret the result:
The used measure in Cohort study is Relative Risk (RR).
2) Retrospective cohort design:
o Exposure(including zero exposure or no exposure) established from past
records, and outcome determined at the beginning of the study.
o Looking back in time, thus using existing data such as medical records or claims
database
o Used to minimize the time spend in the study and decrease the cost of the study.
o More liable for confounding or bias.

How is it conducted ?
1st : Select 2 large groups of exposed and not exposed from a point in the past.
2nd : See the records and determine the incidence of outcome among both.
3th: Measure the association between exposure and outcome and interpret the
result:
The used measure in Cohort study is Relative Risk (RR).

3) Retro-Prospective cohort study:

o Combine between the two designs.
 Interpretation of the Relative Risk :
RR = 1 Exposure NOT related to disease(outcome)
RR >1 Exposure POSITIVELY related to disease(outcome)
RR <1 Exposure NEGATIVELY related to disease(outcome)
* Note : as relative risk increase , as the strength of relation between E+O is
more .
A 1400 s 0,9
exposure 1500
Noexposure 300
B s 012
Example: 1500
The effect of caffeine(E) on insomnia(O): 0,9 4,5
o Exposed group(b+d): Group drinking caffeine . 0,2
o Non-exposed (Control) group(a+c) : Group not drinking caffeine.
Then follow them in the future to see insomnia development in both groups
If result were as follow :

Non-exposed (Control) Exposed

of Healthy 1200 (a) 100 (b)
2
if Diseased 300 (c) 1400 (d)
Total 1500 (a+c) 1500 (b+d)

- RR=

- RR= , so there is a positive relation between caffeine

consumption and insomnia , caffeine is a risk factor for insomnia.
Interpret : The people who consume caffeine are 4.66 times more likely to
develop insomnia than non-caffeine consumer.

Advantage Disadvantage
Easy to understand Cannot determine the causal conclusion
Can estimate both incidence rate & Costly
incidence rate ratio
Suitable for addressing risk factors Require long time
 Experimental Study
Type of the study:
Analytic study
Aim to establish the causal relationship between exposure and outcome.
It is a prospective study.
Exposure is a CAUSE, ( If the exposure X associated with outcome Y, we can say
that the exposure X is a cause for the outcome Y ).

Two types:
1. Randomized Controlled Trials
2. Quasi-experimental

1) Randomized Controlled Trials (RCTs):

o Characters:
Manipulation (Trial) : Researcher does something ( Intervention ).
Control group : Researcher introduces one or more control groups to compare
with
Randomization ( everyone has an equal chance to be selected ) : The researcher
takes care to randomly assign subjects to the control and experimental groups
o Use Blinding process to avoid bias:
The process by which the researcher tries to make sure that as few as much people
know about which one belongs to which group , treatment or placebo, including: The
patients, the nurses and the doctors.

How is it conducted ?
1st : From a pool of study, subjects/participants randomly select into two groups:
(A) Group exposed to intervention.
(B) Control group without intervention, or with placebo, or standard
treatment.
2nd : Give the first group the intervention (as drug) , and leave the control group
without intervention or with placebo, or standard treatment.
3rd : Follow them in the future to see the outcome.
 Phases of RCT:
Phase I:
Healthy volunteers, answers the question whether the intervention is
compatible with life.
Phase II:
Uncomplicated cases ,
Does the intervention have an effect on the outcome?
Phase III:
The Proper RCT
Phase IV:
Post-marketing. In case of drug intervention, you can ask in hospitals or
clinics.

Example:
The effect of caffeine(E) on insomnia(O):
o Exposed group(b+d): Group drinking caffeine (Intervention) .
o Control group(a+c) : Group not drinking caffeine (Non intervention).
o (Both group are selected randomly).
Then follow them in the future to see insomnia development in both groups

Pre-test/Post-test Control Group:

2) Quasi-Experimental Study:
o One characteristic of a true experiment is missing, either randomization
or the use of separate control group.

o Always includes the manipulation of an independent variable which

serves as the intervention.
 Example:
The effect of caffeine(E) on insomnia(O):
o Exposed group(b+d): Group drinking caffeine (Intervention) .
o (No control group)
Then follow them in the future to see insomnia development.
Or
o Exposed group(b+d): Group drinking caffeine (Intervention) .
o Control group(a+c) : Group not drinking caffeine (Non intervention).
o (Both group are not selected randomly).
o Then follow them in the future to see insomnia development in both groups

One-shot Case Study :

One-group Pre-test/Post-test :

Advantage Disadvantage
Perfect design Unethical if exposure
is harmful
Unbiased if blinded Unethical if sure that treatment works
Can study all aspects of the relation risk- Expensive
exposure
Shows causality Difficult if outcome is rare