Scribd
Upload
63 views

Confounding Variables: Ali Yassin and Bara'a Jardali Presented To Dr. Issam I. Shaarani

The document defines confounding variables as those that are associated with both the independent and dependent variables and can cause a false correlation. It provides examples of how smoking can act as a confounding variable in studies of lung cancer and alcohol consumption. Finally, it discusses methods like randomization, matching, stratification, and restriction that can be used to control for confounding variables in experimental studies.

Uploaded by

Ali Ghanem
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
63 views

Confounding Variables: Ali Yassin and Bara'a Jardali Presented To Dr. Issam I. Shaarani

The document defines confounding variables as those that are associated with both the independent and dependent variables and can cause a false correlation. It provides examples of how smoking can act as a confounding variable in studies of lung cancer and alcohol consumption. Finally, it discusses methods like randomization, matching, stratification, and restriction that can be used to control for confounding variables in experimental studies.

Uploaded by

Ali Ghanem
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 12

Confounding Variables

Ali Yassin and Bara’a Jardali


Presented to Dr. Issam I. Shaarani
Types of Variables
 Independent Variable:
 The condition that is intentionally changed by the investigator in an experiment
 There can be only one independent variable in an experiment
Example: Dyslipidemia

 Dependent Variable:
 The factors or conditions that will change as a result of the independent
variable.
 The factors that you measure or observe as data
 There can be one or more dependent variables in an experiment
Example: Coronary Heart Disease
i.e. Dyslipidemia is a risk factor affecting coronary heart disease.
Definition

 A confounding variable* is a variable that is associated with and


may influence both the dependent (outcome) and independent
(exposure) variable.

 It may cause the suggestion of a false correlation between the


dependant and independent variable. 

 Such a variable must usually be controlled for before looking at a


outcome–exposure relationship.

* Also known as the confounder, confounding factor, or lurking variable


Example
 Suppose we are interested in the relationship between lung-cancer incidence
and heavy drinking (≥2 drinks per day).
 We conduct a prospective study in which drinking status is determined at
baseline; The study population is divided into two groups:
1) Heavy drinkers
2) Non-drinkers
 The cohort is followed for 10 years

10 years
Heavy Drinking Lung Cancer
(Independent variable) (Dependent Variable)
Example (continued)
 This may give a false relation that heavy drinking is associated with Lung
cancer. However, this is a crude relationship that is presented without any
adjustment for possible confounding variables.
 One such confounding variable is smoking because smoking is related to both
drinking and lung-cancer incidence.
 One explanation of this crude relationship between lung-cancer incidence and
drinking may be that heavy drinkers are more likely than non-drinkers to be
smokers, and smokers are more likely to develop lung cancer than non-smokers.

10 years
Heavy Drinking Lung Cancer
(Independent variable) (Dependent variable)

Smoking
(Confounding variable)
Type of Confounders

 Positive Confounder:
 Positively related to both exposure and disease.
 Negatively related to both exposure and disease.

 Negative Confounder:
 Positively related to disease and negatively related to exposure.
 Negatively related to disease and positively related to exposure
Methods to Control Confounding

 In order to reduce confounding variables, make sure all the


confounding variables are identified in the study. Confounding is
prevented or minimized by several techniques of study designs
including:

1) Randomization
2) Matching
3) Stratification
4) Restrictions
Randomization
 Randomization means that
patients are randomly assigned to
different groups (i.e., to the
experimental and control groups).

 True randomization means that


the groups should be similar with
respect to race, gender, disease
severity, age, occupation, and
any other variable (including
variables that we do not even
recognize) that may affect the
response to the experimental
intervention.
Matching
 Randomization cannot guarantee that
the experimental and control groups
are similar in all important ways. An
alternative way of ensuring similarity
is by matching:

 each patient in the experimental


group is paired with a patient in the
control group who is a close match on
all relevant characteristics
 so if gender, race, age, and smoking
status were important factors
influencing the course of the disease
being studied, each experimental
patient would be matched with a
control patient of the same gender,
race, age, and smoking
Stratification
 This is a combination of
randomization and matching
techniques. The population under
study is first divided, or stratified,
into subgroups (“blocks”) that are
internally homogeneous with respect
to the important potential
confounding factors (e.g., gender,
race, age, disease severity).

 Equal numbers of patients from each


subgroup are then randomly
allocated to the experimental and
control groups. The two groups are
therefore similar, but their exact
membership is still a result of
randomization.
Restricition
 This involves restricting a study
to participants with certain
characteristics—if a study is
limited to people of just one
gender, race, age group,
geographical area, disease
severity, etc., then all these
factors are removed as potential
confounders.
Thank You

References:
 Fundamentals of Biostatistics 7th edition
 High-Yield Biostatistics, Epidemiology and Public Heath 4th Edition

You might also like