Yes, conflicts of interest are a real cause for concern. [But] there’s not much evidence that industry-sponsored research is less valid or rigorous.
We need more research on that question. But yes, the evidence we have suggests that researchers with financial ties to manufacturers tend to interpret their data more favorably when they’re studying those companies’ products.
That doesn’t necessarily mean that the sponsored studies were biased. A drug company doesn’t fund a major clinical trial unless it strongly suspects the treatment is a good bet. If you were an industry exec and half your human trials failed, you’d be out of work in a hurry.
Besides biased interpretation, I think it’s the issue of selective reporting. A company will often publish its best trial results two or three times in different forms, but data that would provide a more complete picture of the risks and benefits don’t make it into the journals. We need public registries for trial data, and participation should be mandatory.
It’s not always clear what counts as a conflict of interest. Does a brief consultancy I had three years ago affect my judgment today on a different drug or trial? Does a speech I gave for Merck bear on any work I do in any of the fields that Merck is involved in? Those are often judgment calls.
There’s a difference. Journalists can avoid financial ties and still get their job done. Many of our best treatments for cancer, heart disease and diabetes are products of private companies. If scientists stopped working with drug companies, the whole enterprise would suffer.
I think the ones we developed at the American Society of Clinical Oncology provide a good model. Those rules say that in big clinical trials, the principal investigator and the executive committee supervising the trial should have no financial relationship with an interested party within the past two years. The key is to have untainted people running the most important trials–people who can make independent judgments and who are free to publish all of the data.
