In oncology, hundreds of prognostic marker studies are published each year, yet few markers have been demonstrated to be clinically useful. Altman and Riley discuss the pitfalls associated with publication bias, inadequate reporting and retrospective studies, and advocate that an evidence-based approach should be adopted. The authors comment on the advantages of making individual patient data available and improving the reporting of the results of prognostic marker studies.
- Douglas G Altman
- Richard D Riley
