Much heat continues to be generated by the topic of scientific fraud. In the United States, the Office for Research Integrity identifies papers in the biomedical sciences declared to be fraudulent after an official inquiry ( and publishes their bibliographic data.

In addition, papers are often retracted from journals after publication for other reasons. But are these data being generally disseminated?

There are some grounds for pessimism. Kochan and Budd1, for example, showed that retracted papers by John Darsee continued to be positively cited even though a considerable amount of time had passed since retraction, and even though the case generated much publicity. Pfeifer and Snodgrass2 recorded citations to 82 completely retracted articles and found that, although retraction reduces subsequent citation compared with a control group, retracted papers were often cited to support claims. Finally, Budd et al.3, using Medline to identify articles retracted between 1966 and 1997, found that many retracted articles were still being cited as valid.

It seems, therefore, that in at least some cases, authors are not aware of retractions.

A systematic screening method is required to prevent the citation of fraudulent or retracted papers. This could be done for some disciplines via databases such as Medline (which can be searched for retracted publications), but would not cover all fields of research.

Another approach would be for an organization, for example a scientific publishers' group, to run a web-based database of retracted and/or fraudulent papers covering all fields of research. Authors could then search this database before submitting their papers for publication. This search could be a requirement for submission of the paper to a journal.

Such a database could also help people to avoid doing new research based on useless claims in the literature.

How such a venture would be funded, and how it would work in practice, are the next questions to address.