Maisonneuve H. The management of errors and scientific fraud by biomedical journals: they cannot replace institutions. La Presse Medicale 2012;41(9):853-860
Journals do not have the aim to assess research integrity: that's the institutions' roles. Journals discover research misconduct when articles are reviewed, or after the article is published. The peer review system is criticised, including the anonymous peer review: it has never been proved that quality of anonymous reading was better than quality of open reading. When errors and fraud are identified, journals can publish 3 statements: erratum for errors, expression of concern for errors or fraud when evidence is not established, and retraction when evidence is obvious.