Retracted Publications Within Radiology Journals.

OBJECTIVE: The purpose of this study was to characterize trends related to retracted publications within radiology journals. MATERIALS AND METHODS: PubMed was queried to identify all articles with the publication type "retracted publication" or "notification of retraction." Articles published within radiology journals were identified using Journal Citation Reports' journal categories. Available versions of original articles and publication notices were accessed from journal websites. Citations to retracted publications were identified using Web of Science. Overall trends were assessed. RESULTS: Forty-eight retracted original research articles were identified within radiology journals since 1983, which included 1.1% of all PubMed "retracted publication" entries. Distinct PubMed entries were available for the retracted publication and retraction notification in 39 of 48 articles. The original PDF was available for 37 articles, although the articles were not watermarked as retracted in 23 cases. In six cases with a watermarked PDF, further searches identified nonwatermarked versions. Original HTML versions were available for 13 articles but 11 were not watermarked. The mean (± SD) delay between publication and retraction was 2.7 ± 2.8 years (range, 0-16 years). The mean number of citations to retracted articles was 10.9 ± 17.1 (range, 0-94 citations). Reasons for retraction included problematic or incorrect methods or results (although it typically was unclear whether these represented honest errors or misconduct) in 33.3% of cases, complete or partial duplicate publication in 33.3% of cases, plagiarism in 14.6% of cases, a permission issue in 8.3% of cases, the publisher's error in 6.3% of cases, and no identified reason in 6.3% of cases. One or no retractions occurred annually from 1986 to 2001, although two or more retractions occurred annually in nine of the 12 years from 2002 through 2013. CONCLUSION: Retraction represents an uncommon, yet potentially increasing, issue within radiology journals that publishers have inconsistently and insufficiently addressed. Greater awareness and training in proper biomedical research conduct, as well as establishment and enforcement of standardized publishers' policies, are warranted.