When to use the odds ratio or the relative risk?

The relative risk (RR) and the odds ratio (OR) are the two most widely used measures of association in epidemiology. The direct computation of relative risks is feasible if meaningful prevalences or incidences are available. Cross-sectional data may serve to calculate relative risks from prevalences. Cohort study designs allow for the direct calculation of relative risks from incidences. The situation is more complicated for casecontrol studies. If meaningful prevalences or incidences are not available, the OR provides a valid effect measure: It describes the ratio of disease odds given exposure status, or alternatively the ratio of exposure odds given the disease status. Computationally, both approaches lead to the same result. The OR for a given exposure is routinely obtained within logistic models while controlling for confounders. The availability of this approach in standard statistical software largely explains the popularity of this measure. However, it does not have as intuitive an interpretation as the relative risk. This is where problems start: OR’s are often interpreted as if they were equivalent to relative risks while ignoring their meaning as a ratio of odds. It is for instance common to describe an OR of “2” in terms of a “twofold risk” of developing a disease given exposure. This inaccuracy entails potentially serious problems because the OR always overestimates the RR. This can easily be deduced from the mathematical formulas as depicted in Table 1 because of the way the denominators differ.