Incidence of Data Duplications in a Randomly Selected Pool of Life Science Publications

摘要

Since the solution to many public health problems depends on research, it is critical for the progress and well-being for the patients that we can trust the scientific literature. Misconduct and poor laboratory practice in science threatens the scientific progress, leads to loss of productivity and increased healthcare costs, and endangers lives of patients. Data duplication may represent one of challenges related to these problems. In order to estimate the frequency of data duplication in life science literature, a systematic screen through 120 original scientific articles published in three different cancer related journals [journal impact factor (IF) < 5, 5-10 and > 20] was completed. The study revealed a surprisingly high proportion of articles containing data duplication. For the IF < 5 and IF > 20 journals, 25 % of the articles were found to contain data duplications. The IF 5-10 journal showed a comparable proportion (22.5 %). The proportion of articles containing duplicated data was comparable between the three journals and no significant correlation to journal IF was found. The editorial offices representing the journals included in this study and the individual authors of the detected articles were contacted to clarify the individual cases. The editorial offices did not reply and only 1 out of 29 cases were apparently clarified by the authors, although no supporting data was supplied. This study questions the reliability of life science literature, it illustrates that data duplications are widespread and independent of journal impact factor and call for a reform of the current peer review and retraction process of scientific publishing.