A database to record the impact of fraud and misconduct in studies included in systematic reviews

Tags: Oral
Urquhart B1, MacLehose H1, Foxlee R1
1Cochrane Editorial Unit

Background: Primary studies, including clinical trials, can be retracted or corrected due to the identification of fraud, misconduct or mistakes caused by honest error. It is therefore important to have a process for identifying these post-publication changes in the source literature and then taking appropriate action, so that the systematic review reflects the status of the current literature.

Objectives: To identify the most appropriate actions required when fraud or misconduct are identified in studies included in a Cochrane Review (CR).

Methods: We recently set up a database and have started to collect the details of cases of scientific fraud and misconduct that have been brought to the attention of the Cochrane Editorial Unit. We recorded how the fraud/misconduct was identified, which details were added to the Cochrane Register of Studies (CRS), and what action was taken to update the CR.

Results: Action in response to cases of scientific fraud/misconduct was generally taken after the publication of a retraction notice. It remains a challenge to identify post-publication changes, especially corrections and expressions of concern. The database of previous cases has been used to draft a Cochrane policy on fraud and misconduct, which will provide consistent processes for dealing with instances of fraud/misconduct in the future.

Conclusions: The establishment of a database of cases of scientific fraud/misconduct has been instrumental to the drafting of a Cochrane policy on dealing with scientific fraud and misconduct. The policy will standardise the decisions that need to be taken when fraud or misconduct is confirmed or suspected in studies included in Cochrane Reviews. Future work will include establishing a defined process for identifying and recording post-publication changes to studies, and will aim to establish the steps that can be taken to identify doubtful studies before their inclusion in a CR.