Epidemiology bias · BEGINNER LESSON
Confounding
Confounding is a type of bias in an epidemiologic study where a third variable distorts the observed association between an exposure and an outcome. It happens when that third variable is linked to both the exposure and the outcome, but is not part of the causal pathway between them.
What it is
Confounding is an error source in epidemiology, the study of patterns and causes of disease in populations. It is measured by how much the observed exposure-outcome association is distorted by a confounder, a third variable associated with both exposure and outcome. This distortion can make the association look too strong, too weak, or even create a false one. [0,1]
[1] Reference Guide on Epidemiology - Reference Manual on ... [2] Confundidores y modificadores de efecto - PMC - NIH
How it happens
A confounder influences the outcome and is also related to the exposure, but it is not on the causal pathway. For example, smoking can confound a study of alcohol use and emphysema if smoking is related to both drinking and emphysema. In observational studies, exposure is not assigned by chance, so these third variables may be unevenly distributed across groups. [0,1]
[1] Reference Guide on Epidemiology - Reference Manual on ... [2] Confundidores y modificadores de efecto - PMC - NIH
How researchers try to reduce it
Researchers can limit confounding in the study design or in analysis. Randomization aims to balance third variables by chance, matching can make groups similar on a chosen variable such as age, and stratified analysis or multivariable regression can adjust for measured confounders. Causal diagrams called directed acyclic graphs, or DAGs, can help identify likely confounders. [0,1]
[1] Reference Guide on Epidemiology - Reference Manual on ... [2] Confundidores y modificadores de efecto - PMC - NIH
Limitations
Confounding control works only for factors that are recognized and measured well. If the confounder is measured poorly, residual confounding can remain after adjustment. Matching may also reduce how representative the control group is, which can affect how generalizable the results are to the wider population. Confounding is different from effect modification, where a variable changes the size of the association rather than distorting it. [0,1,2]
[1] Reference Guide on Epidemiology - Reference Manual on ... [2] Confundidores y modificadores de efecto - PMC - NIH [3] Consequential (and inconsequential) environmental epidemiology
SEE THE IDEA
How confounding distorts an observed association
The factor being studied, such as alcohol use.
A third variable linked to both exposure and outcome, such as smoking.
The disease or result being studied, such as emphysema.
The measured link that may be exaggerated or diluted by the confounder.
The confounder connects to both exposure and outcome, so the measured association may not reflect the true exposure-outcome relationship.
Confounding is a third-variable problem: the study sees an exposure-outcome link that is partly shaped by another causal factor.
TRY IT YOURSELF
Test your understanding
Choose an answer, then check the explanation.
PUT IT TOGETHER
Match each concept to its meaning
Drag a label onto its description, or select a label and then select a description. This also works with a keyboard.
4 pairs to match
Sources and further reading
This general lesson was generated from public source material. It explains a concept; it does not interpret a private paper or provide medical advice.