What is it? Why is it important?

Confounding variables are variables that can lead to the misinterpretation of study results. The reason is that these variables are related both to the outcome (dependent) and predictor (independent) variable(s),

In research, identifying confounding variables is essential in order to draw accurate and meaningful conclusions from the analysed data.

 

Example

A study investigates the relationship between coffee consumption (predictor variable) and the risk of heart disease (outcome variable = occurrence of heart disease).

  • Study results show a strong association between coffee consumption and heart disease.
  • However, researchers note that a substantial number of study participants who consume a lot of coffee are also regular smokers
  • Consequently, part of the observed effect may stem from tobacco consumption rather than coffee consumption

To draw accurate conclusion on the effect of coffee consumption, tobacco consumption (confounding variable) must be included in the analysis  

What do I need to do?

As a SP-INV, familiarise yourself with how to identify confounding variables in your study.

 

The impact of confounding variables on study results can be limited by:

  1. Adopting appropriate sampling/randomization strategies in Randomized Controlled Trials (RCT).
  2. Adjusting the statistical analysis: By including additional variables in the regression model to account for their potential influence on the relationship between the independent (predictor) and dependent variable (outcome(s))

 

When planning your study, in addition to the outcome and predictor variables, potential confounders are the third type of variables you should consider, and include in the evaluation of your study. 

More

Examples on how the impact of confounding variables on study results can be limited:

 

Example: Adopting appropriate sampling/randomization strategies

A Randomized Controlled Trials (RCT) assesses the effect of an anti-ageing cream (predictor) on the risk of skin cancer (outcome). Age was recognised as a potential confounding variable as age is related to both cancer risk and use of anti-ageing cream. Thus, randomization is stratified to ensure the age distribution of study participants in the intervention and control groups is similar (i.e. number of participants using and not using anti-ageing cream is similar between the two group)

 

Example: Adjusting the statistical analysis:  Adjusting the analysis for the effect of smoking allows you to derive the effect of coffee consumption on heart disease after controlling for the effect of smoking.

Where can I get help?

Your local CTU can support you with experienced staff regarding this topic

Abbreviations
  • CTU – Clinical Trials Unit
  • SP-INV – Sponsor Investigator
  • RTC – Randomised Controlled Trial
Basic ↦ Statistic Methodology ↦ Data Collection ↦ Confounding Variable
Study
Basic

Provides some background knowledge and basic definitions

Basic Monitoring
Basic Drug or Device
Concept

Starts with a study idea

Ends after having assessed and evaluated study feasibility

Concept Statistic Methodology
Concept Drug or Device
Development

Starts with confidence that the study is feasible

Ends after having received ethics and regulatory approval

Development Drug or Device
Set-Up

Starts with ethics and regulatory approval

Ends after successful study initiation

Set-Up Ethics and Laws
Set-Up Statistic Methodology
Set-Up Quality and Risk
Set-Up Drug or Device
Conduct

Starts with participant recruitment

Ends after the last participant has completed the last study visit

Conduct Statistic Methodology
Conduct Drug or Device
Completion

Starts with last study visit completed

Ends after study publication and archiving

Completion Statistic Methodology
Completion Drug or Device
Current Path (click to copy): Basic ↦ Statistic Methodology ↦ Data Collection ↦ Confounding Variable

Please note: the Easy-GCS tool is currently under construction.