Confounding bias
Welcome! In this demonstration, we will explain the concept of confounding bias. Back to Home
In an imaginary cohort study, we are evaluating whether consumption of red wine (exposure) is associated with fewer risks of Covid-19 (outcome). This exposure-outcome association is confounded by economic status (confounder), which is associated with both the exposure and the outcome. In this example, we assume all variables are binary, and people with higher economic status are more likely to drink red wine and also have better access to healthcare, which may reduce their risk of Covid-19. In the section below, you can enter the risk (in percentage) of Covid in people who do and do not drink red wine according to their economic status, as well as the proportion of people with high economic status within each exposure group. A chart will show how the confounding bias affects the observed association between red wine consumption and Covid-19 risk.
| Parameter | Value |
|---|---|
| Risk of Covid-19 | |
| Red wine, high economic status (%) | |
| Red wine, low economic status (%) | |
| No red wine, high economic status (%) | |
| No red wine, low economic status (%) | |
| Proportion of people with high economic status | |
| Among Red wine (%) | |
| Among No red wine (%) | |
Scenario
Based on the data, the risk difference (RD) and risk ratio (RR) of Covid-19 comparing red wine drinkers to non-drinkers are:
- Among people with high economic status: RD=%; RR=
- Among people with low economic status: RD=%; RR=
- Overall (crude): RD=%; RR=
Assuming an overall prevalence of drinking red wine of %, the prevalences of drinking red wine among people with high and low economic status are % and %, respectively, resulting in a PR measuring the confounder-exposure association of .