Research misconduct is defined as behaviour by a researcher that falls short of accepted ethical and scientific standards – whether or not it was intentional. For example, the same data may be sent for publication to more than one medical journal, which might have the effect of their being counted twice in any META-ANALYSIS or systematic review, so biasing the final result; or the data may be ‘salami sliced’ to try to make the maximum number of publications, even though the data may overlap. Fraud in the context of research is defined as the generation of false data with the intent to deceive. It is much less frequent than carelessness, but its incidence is estimated at between 0.1 and 1 per cent. A figure of 1 per cent means that, in the United Kingdom at any one time, maybe 30 studies are being conducted, or their results published, which could contain false information. Examples include forged ethics-committee approval, patient signatures and diary cards; fabricated figures and results; invention of non-existent patient subjects; or sharing one electrocardiogram or blood sample amongst many subjects.

Research fraud should be first suspected by a clinical-trial monitor who recognises that data are not genuine, or by a quality-assurance auditor who cannot reconcile data in clinical-trial report forms with original patient records. Unfortunately, it often comes to light by chance. There may be suspicious similarities between data ostensibly coming from more than one source, or visits may have been recorded when it was known that the clinic was shut. Statistical analysis of a likely irregularity will frequently confirm such suspicion. The motivation for fraud is usually greed, but a desire to publish at all costs, to be the original author of a medical breakthrough, to bolster applications for research grants, or to strengthen a bid for more departmental resources are other recognised reasons for committing fraud.

Probably the most comprehensive regulatory body is the Office of Research Integrity in the US, although it can only sanction those whose misconduct involves federal funding. The four Nordic countries (Denmark, Finland, Norway and Sweden) have committees on research dishonesty that investigate all cases of suspected research misconduct. Editors of many of several thousand of the world's leading academic journals have united to form the Committee on Publication Ethics, based in the UK, which advises editors on proper practice and assists them in retracting or refusing to publish articles found or known to be false and encourages them in how to persuade universities and other funding institutions to investigate. (See ETHICS; ETHICS COMMITTEES.)