## Distinguishing Confounding from Bias **Key Point:** Confounding and bias are distinct phenomena in epidemiology, though both distort the true association between exposure and outcome. ### Confounding - A **real association** between a third variable (confounder) and both the exposure and outcome - The confounder is **not in the causal pathway** between exposure and outcome - Represents a **mixing of effects** - Example: Age confounding the association between smoking and cardiovascular disease (age is associated with both smoking prevalence and CVD risk) ### Bias - A **systematic error** in study design, conduct, or analysis - Leads to deviation of results from the true value - Not a real association, but a methodological flaw - Example: Recall bias in case-control studies where cases remember exposures differently than controls ### Key Distinction | Feature | Confounding | Bias | |---------|-------------|------| | Nature | Real association with extraneous variable | Systematic error in measurement/collection | | Mechanism | Mixing of effects | Methodological flaw | | Presence | Can exist in any study design | Can exist in any study design | | Control | Matching, stratification, multivariate analysis | Study design (blinding, standardization) | **High-Yield:** Both confounding and bias can occur in observational and experimental studies. The key is that confounding represents a **true but spurious association**, while bias is a **systematic deviation** from truth. **Clinical Pearl:** In clinical practice, recognizing confounders (like socioeconomic status affecting both exposure and outcome) is crucial for interpreting observational study results.
Sign up free to access AI-powered MCQ practice with detailed explanations and adaptive learning.