Frequency, risk and impact measures in epidemiology

The results of epidemiological studies should be expressed in terms of measures of health or disease. This article reviews the key frequency, risk and impact measures, which can be estimated using proportions, ratios or rates, depending on the specific context. It discusses which measures are appropriate for a study based on its design. Cross-sectional studies serve to estimate the prevalence; cohort studies allow calculation of the incidence, relative risk and attributable fractions; case-control studies use the odds ratio and clinical trials determine the relative risk, absolute and relative risk reductions and number needed to treat (NNT). The article also outlines criteria for the interpretation of these measures, supported by specific examples.