Objective: It has been proposed that the scientific literature purges itself of articles known to be fraudulent. To test this, an investigation was carried out of post-retraction citations over a 19-year period in the Breuning case.
Methods: On 10 March 2008 a cited reference search was conducted (all languages, all document types) using the name 'Breuning SE*'. The time limit was 1989-2007 with an option to exclude self-citations. The search included the ISI Web of Science Database including the Science Citation Index Expanded, the Social Sciences Citations Index and the Arts & Humanities Citation Index. To ascertain the citation context, citations of Breuning were classified by two raters as affirmative, negative or neutral.
Findings: For the period 1989-2000 both negative and affirmative citations were found. For the period 2001-2006 only affirmative citations (even to retracted articles) were found, some in journals with higher impact factors than those citing the case as fraudulent. In spite of the small number of citations of Breuning's articles, it is alarming that the affirmative citing of fraudulent research has not completely ceased but continues 24 years post-retraction (retracted 1982, cited 2006). While the limitations of a single case study are conceded, the results challenge the belief of scientific literature purging itself of fraudulent material.
Conclusions: Retraction databases and widespread availability of computer software to check lists of references free of charge in any database or the internet are called for. Moreover, if a paper is never formally retracted, software for searching author names in the internet for fully investigated and proven scientific misconduct might be developed. The ethical guidelines on duplicate publication for purposes of disseminating the information as widely as possible should be reviewed.