Objective: To determine the contributions of each author to multiauthored biomedical research papers.
Design: Mailed, self-administered survey.
Participants: A total of 184 first authors from a consecutive sample of 200 papers with four or more authors published in 10 leading biomedical journals.
Main outcome measures: First authors' ratings of which authors had made substantial contributions to the following: initial conception of the study, design of the study, provision of needed resources, collection of data, analysis and interpretation of data, and writing the first draft of the paper or revising drafts for important intellectual content.
Results: The contributions of nonfirst authors varied greatly within and among papers. Even second and last authors--though they generally contributed more than other nonfirst authors--were markedly inconsistent in the extent and pattern of their contributions. Time spent on the research differed among authors by orders of magnitude. An appreciable number of authors made few or no substantial contributions to the research.
Conclusions: The nature and extent of contributions of nonfirst authors to biomedical research reported in multiauthored papers cannot reliably be discerned (or discounted) by authorship or order of authors. The two core purposes of scientific authorship--to confer credit and denote responsibility for research--are not adequately being met by these authorship practices.