Industry-sponsored studies are more likely to report favourable clinical outcomes than independently-funded studies. Dr. Joel Lexchin and colleagues provide explanations for the types of bias inherent to industry-funded research:
- Industry-sponsored trials often employ inappropriate choices of doses, dosing intervals, and comparators.
- Drug companies hide unfavourable trial data and publish, and re-publish, favourable clinical trial data. Overall, as a result of this practice, bias is introduced into the assessment of published literature on the effectiveness of drugs. The consequence of this bias in the literature is that any subsequent trials or studies on a drug, which is associated with only favourable literature, are based on incomplete data sets and cannot draw accurate conclusions. This selective publication of data influences not only physicians' prescribing choices, but also drug regulatory decisions regarding the safety of medications on the market.
- Upon submitting clinical trial data to regulatory agencies for approval, industry-sponsored studies often reinterpret either negative or questionable results to appear positive.
- Even when clinical trial results are reported accurately in publications, the conclusions drawn in industry-sponsored trials may misrepresent and distort the meaning and implications of the results to appear more favourable.
- Authors with financial conflicts of interest (COI) with industry are more likely to misrepresent data to appear favourable to the sponsoring company.
- Medical ghostwriters are often recruited to write manuscripts for publications that represent a drug favourably, when the data suggests otherwise. A well-known academic or physician is, then, recruited to sign the manuscript and publication as the author, without acknowledgement of the medical ghostwriter.
Dr. Joel Lexchin is an emergency physician at the University Health Network and professor of Health Policy & Equity in the School of Health Policy and Management, Faculty of Health at York University, Toronto.