Guidelines for publishing papers on cancer clinical trials: responsibilities of editors and authors.

Guidelines are proposed for the reporting of clinical trials. These are meant to be used for both authors and journal editors. Specific proposals are made for: reporting the population under study, actual therapy used, describing the study design, accounting of patients, follow-up statistics, and describing the quality control of the data. Nonrandomized studies require special discussion. Such studies potentially have substantial biases which can lead to incorrect conclusions. Six areas of biases arising in nonrandomized studies are pointed out which should require a discussion in an acceptable published paper. The presentation of statistical results and the reporting of early studies present unusually difficult problems. Guidelines are presented for statistical analyses which, although simple to implement in practice, are not generally used. A proposal is made for the publication of early results of a study which require that the authors agree to follow and periodically report on their trial to the journal editor. If the early conclusions change with additional follow-up, the editor should notify its readers. The adoption of these guidelines can be an important force in raising the scientific level of the reporting of clinical trials.