Content
Summary
The odds ratio is a quantitative metric used to estimate the strength of association between an exposure and an outcome, particularly in case-control studies. It is distinct from correlation, which describes how variables change together without implying causation. Understanding the difference between association and causation is critical to avoid common misinterpretations in biostatistics.
Intuition
Think of correlation as observing that two things happen together, like a longer wait for surgery and higher perforation rates in appendicitis. The odds ratio goes a step further by quantifying how strongly an exposure (like a specific treatment) is linked to an outcome. However, just because two variables move together or are associated via an odds ratio does not mean one causes the other. For instance, while low BMI is associated with lower cardiac arrest mortality, it does not mean low BMI prevents cardiac arrest; other factors may be at play.
Derivation
- Identify the study design, typically case-control, where the odds ratio is most applicable.
- Observe the relationship between exposure and outcome variables.
- Calculate the odds of the outcome in the exposed group versus the unexposed group.
- Recognize that this metric estimates association strength, not causation.
- Distinguish this association from simple correlation, which only indicates direction of change.
Worked example
A study finds that patients with lower BMI have lower cardiac arrest mortality. Does this mean low BMI prevents cardiac arrest?
- Identify the association: Lower BMI is associated with lower mortality.
- Check for causation: The text states this association does not imply low BMI prevents cardiac arrest entirely.
- Consider alternative explanations: The second parameter may not be the cause of the first.
- Conclusion: The relationship is an association, not necessarily a causal prevention mechanism.
Answer No, the association does not prove causation; low BMI may not prevent cardiac arrest.
Common mistakes
- Interpreting correlation as causation.
- Assuming that because two variables change together, one causes the other.
- Believing that an association between low BMI and lower mortality means low BMI prevents cardiac arrest.