Interpreting measures of treatment effect.

A confusing number of measures are used to describe the effect sizes from clinical trials or systematic reviews. Absolute measures (absolute risk reduction, number needed to treat) and relative measures (relative risk reduction, relative risk or odds ratio) may give not just different numerical answers but convey different messages. This paper describes the role and meaning of the different measures, advises on their interpretation, and highlights the importance of taking into account the initial risk when assessing effect sizes from published studies.