Retracted Publications in Medical Imaging Literature: an Analysis Using the Retraction Watch Database

Abstract

It is currently unknown how many publications in the medical imaging literature are retracted and for which reasons. The purpose of this study was to perform an updated analysis on retracted medical imaging publications using the Retraction Watch Database. The Retraction Watch Database was searched for all retracted publications in the subject category "Radiology/Imaging" (no beginning date limit, search update until April 27, 2022). Reasons for retraction were extracted using standardized coding taxonomy. The number of citations per retracted publication was determined. Spearman's rho was used for statistical analysis. 192 retractions, originally published between 1984 and 2021, were included. Most retractions originated from China (31.3%), the United States (12.5%), Japan (7.3%), and South Korea (6.3%). The number of retractions increased over the years, especially since 2000 (Spearman's rho=0.764, p <0.001). Delay between original publication and retraction ranged from 0 days to 14 years and 3 months (median of 11 months). Most common reasons for retraction were duplication of article (7.1%), plagiarism of article (6.8%), concerns/issues about data (5.4%), investigation by company/institution (4.5%), and forged authorship (4.0%). Scientific misconduct was deemed present in 107 of 192 retracted articles (55.7%). Retracted articles (of which 138 were listed in Web of Science) received a median of 2 citations (range 0-148, IQR 5). The number of retracted medical imaging publications continues to increase over time, which could indicate that more compromised research has either been published or discovered. Scientific misconduct was the main cause for retraction.