Background: Bias in industry-sponsored trials is common and the interpretation of the results can be particularly distorted in favour of the sponsor's product. We investigated sponsors' involvement in the conduct and reporting of industry-sponsored trials.
Methods: We included all industry-sponsored trials published in The Lancet in 2008 and 2009 and corresponding trial protocols provided by The Lancet. For each protocol and publication, we extracted information on trial conduct and reporting.
Results: We identified 169 publications of randomised trials and included 69 (41%) that were industry-sponsored, and 12 (7%) industry-funded but seemingly independently conducted as a subsample. Entry of data into the study database was done independently by academic authors without the involvement of the sponsor or a contract research organisation in one of the 69 trials. Two trials had independent data analysis and one independent reporting of results. In 11 of the trials, there was a discrepancy between the information in the protocols and papers concerning who analysed the data. In four of the 12 seemingly independent trials, the protocol described sponsors' involvement in writing the report while the published paper explicitly stated that the sponsor was not involved.
Conclusions: The sponsors are usually involved in the analysis and reporting of results in industry-sponsored trials, but their exact role is not always clear from the published papers. Journals should require more transparent reporting of the sponsors' role in crucial elements such as data processing, statistical analysis and writing of the manuscript and should consider requiring access to trial protocols, independent data analysis and submission of the raw data.