Foundations of Relative Risk

Understanding Relative Risk What is Relative Risk? Relative risk (RR) is a fundamental measure in epidemiology that compares the likelihood of an outcome occurring in two different groups: those who have been exposed to something and those who have not. It answers the question: "How much more (or less) likely is the outcome in the exposed group compared to the unexposed group?" More formally, relative risk is the ratio of the probability (or incidence) of an outcome in the exposed group to the probability of the same outcome in the unexposed group. The Mathematical Formula Relative risk is calculated as: $$RR = \frac{I{\text{exposed}}}{I{\text{unexposed}}}$$ where $I$ represents the incidence rate—the number of new cases of the outcome divided by the total number of individuals at risk in each group. For example, if smoking exposure leads to lung cancer in 10% of smokers but only 1% of non-smokers, the relative risk would be: $$RR = \frac{0.10}{0.01} = 10$$ This tells us that smokers are 10 times as likely to develop lung cancer as non-smokers. Interpreting Relative Risk Values The value of the relative risk tells you the direction and magnitude of the association between exposure and outcome. When RR > 1: The exposure increases the risk of the outcome—this is called a risk factor. An RR of 2.5 means the exposed group is 2.5 times as likely to experience the outcome. When RR = 1: There is no association between exposure and outcome. The probability is identical in both groups. When RR < 1: The exposure reduces the risk of the outcome—this is called a protective factor. An RR of 0.5 means the exposed group has half the risk of those unexposed. This might occur, for example, when studying the effect of vaccination (the exposure) on disease (the outcome). Relative Risk Reduction When an exposure is protective, we often quantify the benefit using relative risk reduction (RRR): $$\text{RRR} = 1 - RR$$ If a treatment reduces the relative risk of a bad outcome from 100% to 50% (RR = 0.5), then the relative risk reduction is $1 - 0.5 = 0.5$, or 50%. This means the exposure reduced the risk by half relative to the baseline. Absolute Risk: The Important Context A concept that's frequently confused with relative risk is absolute risk—the actual probability of the outcome occurring in a group, stated without comparison to another group. This distinction is critical because it reveals a common pitfall in interpreting statistics. Imagine two scenarios: Scenario A: A rare disease affects 1 in 10,000 unexposed people and 2 in 10,000 exposed people (RR = 2) Scenario B: A common outcome affects 400 in 10,000 unexposed people and 800 in 10,000 exposed people (RR = 2) Both have the same relative risk of 2, yet the actual increase in risk is very different. In Scenario A, your absolute risk only increased from 0.01% to 0.02%. In Scenario B, it increased from 4% to 8%. Understanding both the absolute and relative measures is essential for meaningful interpretation. The Base Rate Fallacy The base rate fallacy occurs when people interpret a relative risk without considering the underlying prevalence (how common the outcome already is). A relative risk of 3 sounds alarming, but if the outcome is extremely rare to begin with, the absolute increase in risk may be negligible. Always ask: "What was the baseline probability, and how much did it actually change?" Relative Risk vs. Odds Ratio Another important comparison measure is the odds ratio (OR). While both RR and OR quantify associations between exposure and outcomes, they are fundamentally different. The Key Difference Relative risk compares probabilities (the likelihood of outcomes) between exposed and unexposed groups. Odds ratio compares odds (the ratio of the probability an event occurs to the probability it doesn't) between groups. For example: If 20% of smokers develop a disease, the probability (risk) is 0.20, and the odds are $\frac{0.20}{0.80} = 0.25$ If 10% of non-smokers develop the disease, the probability (risk) is 0.10, and the odds are $\frac{0.10}{0.90} = 0.11$ The relative risk is $\frac{0.20}{0.10} = 2.0$ The odds ratio is $\frac{0.25}{0.11} = 2.27$ Notice they're close, but not identical. When the Outcome is Rare: The Rare Disease Approximation When an outcome is rare (typically when incidence is less than 5-10%), the odds and probability become nearly equivalent. Consequently, odds ratio approximates relative risk. This is why in studies of rare diseases (like a specific cancer), researchers often report odds ratios as if they were relative risks—the difference is negligible. Case-Control Studies and Why We Use Odds Ratio In case-control studies, researchers start with people who already have the outcome (cases) and those who don't (controls), then look backward to see who was exposed. Because the outcome is already determined by the study design, you cannot calculate a true incidence rate or relative risk. Instead, you can only estimate odds ratio, which is always calculable from case-control data. This is why case-control studies inherently produce odds ratios rather than relative risks. When Outcomes Are Common: A Critical Pitfall For common outcomes, odds ratio can dramatically exaggerate the apparent effect size compared to relative risk. Consider an outcome that occurs in 99% of the unexposed group and 99.9% of the exposed group. The relative risk is only: $$RR = \frac{0.999}{0.990} = 1.009$$ barely above 1. However, the odds ratio is: $$OR = \frac{0.999/0.001}{0.990/0.010} = \frac{999}{99} = 10.09$$ This is more than 10! The odds ratio makes a trivial increase in risk appear massive. This is why odds ratio should not be interpreted as relative risk for common outcomes—you must convert OR back to relative risk, or use RR from the start if possible. <extrainfo> Logistic Regression and Odds Ratios When analyzing data with logistic regression, the model produces coefficients that represent odds ratios, not relative risks. This is because logistic regression models are linear in the log odds, not in the probability scale. If your research question requires relative risk estimates, you must either use different statistical methods (like log-binomial regression) or convert odds ratios to relative risks—particularly important if outcomes are common in your study. </extrainfo> Summary Relative risk is a direct, intuitive measure of how exposure changes the probability of an outcome. It's calculated as the ratio of incidence rates and interpreted by comparing it to 1. When comparing this measure to odds ratio, remember that while they approximate each other for rare outcomes, odds ratio can dramatically overstate effects for common outcomes. Always consider both the relative and absolute changes in risk when interpreting results.