Objective: The purpose of this study is to evaluate the rate of duplicate publication in radiology journals. The secondary objective is to evaluate the sensitivity of iThenticate.
Materials and methods: From January 1993 to December 2013, Déjà Vu (a database of highly similar citations) and PubMed were used to search for similar citations in 53 radiology journals. Citations were screened independently by two reviewers and verified by a third using predefined criteria to determine true cases of duplicate publication. The overall rate of duplicate publication was calculated; analysis of rate by journal, impact factor, and publication year was performed. The sensitivity of iThenticate was evaluated by analyzing all identified duplicate publications.
Results: From 128,818 citations in the included journals, 1786 (Déjà Vu) and 104 (PubMed) were flagged as potential duplicates. Of these, 248 (226 from Déjà Vu and 22 from PubMed) were classified as true duplicate publications after application of our criteria. The overall rate was 1.92/1000 citations; it varied widely across journals from zero to over 10/1000 citations, showed no correlation with impact factor (R(2) = 0.06; p = 0.093), and no change over time (R(2) = 0.28; p = 0.515). iThenticate flagged 153 of 248 (61.9%) duplicates as "possible duplicates" (defined as overall percentage match > 30%) and identified the corresponding duplicate citation pair in 140 of 248 (56.7%) cases; in 98 of these, the duplicate citation pair was the highest percentage similarity match.
Conclusion: Duplicate publications in radiology journals are uncommon. The rate varies widely between journals, but was not associated with journal impact factor and did not change over time. iThenticate shows promise for identification of duplicate publications; however, refinements may be necessary to maximize its effectiveness.
Keywords: duplicate publication; ethics; informatics.