Meta-Analyses and Conflict of Interest

Abstract

The reliability of reports of studies funded by the pharmaceutical industry has been seriously questioned. Researchers with financial conflicts of interest are more likely to publish articles (original investigations, editorials, systematic and non-systematic reviews, meta-analyses) that support the products of the companies with which the researchers have financial ties. Simple disclosure of financial conflicts of interest is not regarded as sufficient for original studies funded by pharmaceutical companies, and strategies for minimizing biases have been suggested, such as ensuring that at least one author who is not employed by a commercial firm has full access to all of the data and the use of an independent biostatistician. Surprisingly, however, little has been proposed to minimize bias in other types of papers, and particularly meta-analyses. When these latter papers have been supported by the industry, bymeans of funding or authors’ ties, they have been found to reach conclusions that favoured sponsors’ interests more than independent meta-analyses. Furthermore, Roseman et al. found that only 7% of meta-analyses concerned with pharmacological trials reported the funding sources of the studies that were selected for analysis, despite the fact that about two-thirds of all trials are industry funded. This is interesting given the fact that meta-analysts are expected to assess the risk of bias related to various aspects of the trials they select for analysis and to exclude unreliable investigations. The reader trusts that this check has indeed taken place, but it is difficult to believe that it has if conflict of interest in the trials was not even reported in the paper. If we accept the principle that studies to be included in a metaanalysis should be screened for potential conflict of interest, simple disclosure of funding of those who perform the analysis, as suggested, appears to be rather inadequate and should be supported by other strategies, as described below. The first step, however, is to understand how conflict of interest may affect the technical decisions underlying meta-analytic procedures. There is a wide spectrum of potential financial interests in a meta-analysis and these interests are not necessarily disclosed, despite journal policies. They range from a lack of any conflict to a meta-analysis commissioned by a pharmaceutical industry or private firm with the presence of ties in all authors of the analysis. A meta-analysis by Baldwin et al. on drug treatment of generalized anxiety disorder that recently appeared in the British Medical Journal may illustrate the degree of conflict of interest that may ensue. A pharmaceutical firm conceptualized and designed the study, and commissioned two of the authors, who work in a private medical communications company, to conduct the systematic review and meta-analysis, and prepare the manuscript. Two other authors, who were university employees, performed a critical review of the results and assisted in the development of the manuscript. The study was allegedly independent, yet all authors had financial ties to the funding firm and other pharmaceutical companies that manufactured drugs that were included and discussed in the meta-analysis. The expert who was commissioned to write the accompanying editorial also had financial ties with a number of the pharmaceutical companies that manufactured drugs that were included and discussed in the EDITORIAL CNS Drugs 2012; 26 (2): 93-96 1172-7047/12/0002-0093/$49.95/0