Criteria for authorship for statisticians in medical papers

We organize a statistician's potential scientific and intellectual contributions to a medical study into three types of activities relating to design, implementation and analysis. For each type, we describe high-level, mid-level and low-level contributions. Using this framework, we develop a point system to assess whether authorship is justified. Although we recommend discussion and resolution of authorship issues early in the course of any project, our system is especially useful when this has not been done.