Bias related to funding source in statin trials

One of the fundamental principles of pharmacology is that the effects of drugs (both efficacy and harms) are dose related. In a linked paper (doi:10.1136/bmj.g5741), Naci and colleagues use network meta-analysis as a technique to confirm that the effects of statins on serum levels of low density lipoprotein (LDL) cholesterol are dose related.1 They also conclude that industry sponsorship of statin trials is not associated with increased efficacy estimates and that there are no differences in risks of bias between industry and non-industry sponsored trials. They incorrectly argue that this finding differs from our previously published study examining industry bias in drug-drug comparisons of statins.2

Like Naci and colleagues, we also found no differences in the results of industry and non-industry sponsored studies comparing statins with statins or other drugs. We did not include placebo comparisons. In addition, we found no differences in the adequacy of concealment of allocation, blinding, and inclusion of all subjects in the analysis between industry and non-industry sponsored studies.

However, we went on to analyse a pre-specified subset of studies—all of which were industry sponsored—to determine whether the results favoured the drug of the company sponsoring the study.2 Naci …