Abstract
CONTEXT AND OBJECTIVE: Increased frequency of retractions has recently been observed, and retractions are important events that deserve scientific investigation. This study aimed to characterize cases of retraction within general and internal medicine in a high-profile database, with interest in the country of origin of the article and the impact factor (IF) of the journal in which the retraction was made. DESIGN AND SETTING: This study consisted of reviewing retraction notes in the Thomson-Reuters Web of Knowledge (WoK) indexing database, within general and internal medicine. METHODS: The retractions were classified as plagiarism/duplication, error, fraud and authorship problems and then aggregated into two categories: "plagiarism/duplication" and "others." The countries of origin of the articles were dichotomized according to the median of the indicator "citations per paper" (CPP), and the IF was dichotomized according to its median within general and internal medicine, also obtained from the WoK database. These variables were analyzed using contingency tables according to CPP (high versus low), IF (high versus low) and period (1992-2002 versus 2003-2014). The relative risk (RR) and 95% confidence interval (CI) were estimated for plagiarism/duplication. RESULTS: A total of 86 retraction notes were identified, and retraction reasons were found for 80 of them. The probability that plagiarism/duplication was the reason for retraction was more than three times higher for the low CPP group (RR: 3.4; 95% CI: [1.9-6.2]), and similar results were seen for the IF analysis. CONCLUSION: The study identified greater incidence of plagiarism/duplication among retractions from countries with lower scientific impact.