Conflicts of interest 


Going for gold
May 17th 2001
From The Economist print edition (May 19th 2001)

Scientists should disclose their financial interests, and let the
world decide what to think about them

“SCIENTISTS at the McDonald’s Centre for Obesity Research suggest
that eating a hamburger a day actually reduces cholesterol levels.”
Well, we made that up. However, this kind of company-backed research
is becoming more common, and is undermining the fragile trust in
science held by a public that has been fed a few too many whoppers. 

There is good reason for concern. Encouraging private money for
science not only gives the boffins more resources; it also creates a
useful incentive for them to press on in their quest for truth in
ways that will yield measurable benefits to humanity. Unfortunately,
money can also create an enticement to stretch the truth somewhat.
Several studies have shown that researchers with a financial interest
in the outcome of their work tend to publish conclusions that are,
oddly enough, positively aligned with their affiliations. That may
not surprise cynics—or indeed economists. But it would be a sad thing
for science if its sponsors were able to dictate the outcome of

The editor of the British Medical Journal recently protested strongly
against the donation of money by a tobacco firm to a British
university (see article). Yet last year, he defended the right of the
journal to publish research sponsored by tobacco money. This creates
a puzzle. If tobacco-sponsored research is acceptable, as long as it
passes the bar of peer review, surely the sponsorship of research
centres by tobacco companies should be acceptable, too?

Some publications, such as the American Thoracic Society’s journal
and the British Journal of Cancer, refuse outright to publish
tobacco-sponsored research. This is a consistent position, but a
dangerous one. Scientific journals are not meant to serve as arbiters
of the morality of particular industries. They are meant to publish
good research. But as companies increasingly pay researchers’ grants
and expenses, how can journals sift out the unbiased from the biased?

The answer is, they cannot. All they can do is to report whatever
conflicts a scientist discloses, and take his word that the list is
complete. This is not as naive as it sounds. Science has always
relied on trust. When researchers submit a paper to a journal, they
must vouch that they performed the research they say they did, that
they followed experimental protocols correctly, and that they did not
tamper with the data or the analysis. It ought not to strain the
limits of credibility to ask them to own up to their sources of
income too. Science has always had to cope with conflicts of
interest. The most awkward one a researcher can face is his own
interest in the correctness of his hypotheses. 

Donald Kennedy, the editor of Science, points out that scientists
often come to believe so strongly in the validity of their theories
that they cease to examine them objectively. But this loss of
objectivity is overcome by the efforts of other researchers, who are
properly sceptical in their outlook. Eventually, faulty ideas are
discarded and sound ones prevail.

Achieving truthful science depends not on the eradication of bias but
on its gradual correction. Printing disclosures of interests,
publishing only research that passes rigorous peer review, and then
letting the readers decide for themselves whether the research is
worthy should be enough. The vigilance of rival scientists, which is
up to catching even the tiniest flaws in experimental design, will
surely detect any omissions of interest that occur.