Why do you think you should be the author on this manuscript? Analysis of open-ended responses of authors in a general medical journal

BACKGROUND To assess how authors would describe their contribution to the submitted manuscript without reference to or requirement to satisfy authorship criteria of the International Committee of Medical Journal Editors (ICMJE), we analyzed responses of authors to an open-ended question "Why do you think you should be the author on this manuscript?". METHODS Responses of authors (n=1425) who submitted their manuscripts (n=345) to the Croatian Medical Journal, an international general medical journal, from March 2009 until July 2010 were transcribed and matched to ICMJE criteria. Statements that could not be matched were separately categorized. Responses according to the number of authors or their byline position on the manuscript were analyzed using Mann-Whitney U test and Moses test of extreme reactions. RESULTS The number of authors per manuscript ranged from 1 to 26 (median=4, IQR=3-6), with the median of 2 contributions per author (IQR=2-3). Authors' responses could be matched to the ICMJE criteria in 1116 (87.0%) cases. Among these, only 15.6% clearly declared contributions from all 3 ICMJE criteria; however, if signing of the authorship form was taken as the fulfillment of the third ICMJE criterion, overall fraction of deserving authorship was 54.2%. Non-ICMJE contributions were declared by 98 (7.6%) authors whose other contributions could be matched to ICMJE criteria, and by 116 (13.0%) authors whose contributions could not be matched to ICMJE criteria. The most frequently reported non-ICMJE contribution was literature review. Authors on manuscripts with more than 8 authors declared more contributions than those on manuscript with 8 or fewer authors: median 2, IQR 1-4, vs. median 2, IQR 1-3, respectively (Mann Whitney U test, p=0.001; Moses Test of Extreme Reactions, p<0.001). Almost a third of single authors (n=9; 31.0%) reported contributions that could not be matched to any ICMJE criterion. CONCLUSIONS In cases of multi-author collaborative efforts but not in manuscripts with fewer authors open-ended authorship declaration without instructions on ICMJE criteria elicited responses from authors that were similar to responses when ICMJE criteria were explicitly required. Current authorship criteria and the practice of contribution declaration should be revised in order to capture deserving authorship in biomedical research.