Epidemiologist David Michaels describes the problem with industry-funded research in today’s Washington Post. His point is one that needs emphasis — it’s not that companies interfere directly with the research they fund, it’s that they ensure the questions the research answers are biased in their favor:
At first, it was widely assumed that the misleading results in manufacturer-sponsored studies of the efficacy and safety of pharmaceutical products came from shoddy studies done by researchers who manipulated methods and data. Such scientific malpractice does happen, but close examination of the manufacturers’ studies showed that their quality was usually at least as good as, and often better than, studies that were not funded by drug companies.
This discovery puzzled the editors of the medical journals, who generally have strong scientific backgrounds.
Richard Smith, the recently retired editor of BMJ (formerly the British Medical Journal), has written that he required “almost a quarter of a century editing . . . to wake up to what was happening.” Noting that it would be far too crude, and possibly detectable, for companies to fiddle directly with results, he suggested that it was far more important to ask the “right” question.
What Smith and other researchers, such as Lisa Bero of the University of California at San Francisco, have found is that industry researchers design studies in ways that make the products of their sponsor appear to be superior to those of their competitors.
This bias effect even has a name — the “funding effect.” And now it’s becoming more well-known and public knowledge as researchers are being questioned about the industry funding of their past research. Hiding behind a university research policy isn’t going to be sufficient to answer questions related to this effect. And why researchers, fully aware of this effect, continue to do research funded in this manner.
Indeed, one of the first things I look at when evaluating research is what specific questions or hypotheses are the researchers testing. If the questions look skewed or framed to detect statistical significance (but ignores clinical significance or patient report), then I already know the study may be suspect. Furthermore, I look at the measures used — are they only clinician-based or esoteric assessments, or are they a broad mix of such measures along with patient measures and even third-party measures (such as family members’ report).
There is an answer:
It has become clear to medical editors that the problem is in the funding itself. As long as sponsors of a study have a stake in the conclusions, these conclusions are inevitably suspect, no matter how distinguished the scientist.
The answer is de-linking sponsorship and research. One model is the Health Effects Institute, a research group set up by the Environmental Protection Agency and manufacturers. HEI has an independent governing structure; its first director was Archibald Cox, who famously refused to participate in President Richard Nixon’s “Saturday Night Massacre” meant to help cover up the Watergate scandal. HEI conducts studies paid for by corporations, but its researchers are sufficiently insulated from the sponsors that their results are credible.
Sounds like a model that the entire pharmaceutical industry should pursue. Before the decision is made for them.
Read the full article: It’s Not the Answers That Are Biased, It’s the Questions