Australasian Science: Australia's authority on science since 1938

Trust Me, I Have a White Hat

By Tim Olds

Can you trust obesity research funded by the interests of Big Food?

In August, an article in the New York Times attacked one of the world’s leading health experts, Dr Steven Blair, for his involvement in the Global Energy Balance Network, an organisation committed to fighting obesity and “healthier living through the science of energy balance”. This sounds innocuous enough, but the GEBN is heavily funded by Coca-Cola, and the article argued that Coke was trying to shift attention away from the role of food, and particularly sugar-sweetened drinks, in obesity. Marion Nestle, professor of nutrition at New York University, reportedly said that “the Global Energy Balance Network is nothing but a front group for Coca-Cola”.

Public health advocates have long claimed that corporate interests – particularly Big Food, Big Tobacco and Big Pharma – have been recruiting scientists as hired guns and “merchants of doubt”. A number of reviews have shown that studies funded by the food industry tend to produce results that downplay the connection between food and obesity. One study, for example, found that studies funded by soft drink companies are five times more likely to find no association between soft drink intake and obesity than studies without such funding. Another found that soft drink-funded studies found associations about five times smaller than studies with other funding sources.

The influence of Big Food may not be overt. It is unlikely companies would deliberately falsify data or veto the publication of unflattering results, but there are many ways to finesse results. You wouldn’t have to be a superstar researcher to know that if you publish something contrary to the sponsor’s interests, you may not be their first choice in their next funding round.

But let’s look at this a bit more closely. Mark Cope and David Allison have introduced the concept of “white hat bias” based on Western movies: the good cowboys wear the white hats. White hat bias refers to selectively manipulating, suppressing or sponsoring scientific studies for ideological, non-commercial purposes. White hat bias may be “fuelled by feelings of righteous zeal, indignation toward certain aspects of industry” or a leftist anti-capitalist ideology.

How would white hat bias work? And how would we know if it was there?

White hat bias may occur when researchers refuse to submit for publication results that go against their ideological beliefs, or when zealous journal editors or reviewers refuse to accept them for publication despite their solid scientific methodology.

Cope and Allison proved that there must have been many small studies that found no association between soft drinks and obesity and were never published. This conclusion is possible because of a statistical gem known as a “funnel plot”, which plots the strength of a finding against the size of the sample. We know that when samples are small we are more likely to get wild underestimates or overestimates of the strength of an association, while larger samples tend to gravitate towards the true association. With a sample of children in the thousands we should find a correlation between soft drink intake and fatness that is close to the true value (in the case of children it’s just 0.03). But with a small sample size I’m likely to get much larger correlations (say 0.2 or 0.3) or much smaller correlations, including some negative correlations (that soft drink intake is associated with less fatness).

When we look at published studies of soft drink intake and fatness, one whole side of the funnel plot — where the small studies showing negative associations would be found – is entirely missing. These studies have never been published because of white hat researchers, editors or reviewers. When we allow for the fact that there are unpublished studies out there showing low or negative associations, the apparent bias in funded studies is much reduced – but not eliminated entirely.

Beyond the soft drink example lie two bigger issues. First, not all scientific bias arises from venal commercial motives. It can also arise from political correctness, zealotry, ideology and commitment to a cause – even a good cause. Second, we should not rely on criticisms of the researcher or the funder rather than the science. If critics feel a study is biased because of the funder, they should be given open access to the original data (which is becoming a requirement of journals) and be obliged to show how the data have been manipulated. If the researchers or funders refuse access we can draw out own conclusions. If studies are rejected out of hand without consideration of their scientific worth, we are losing a substantial corpus of work of scientific value.

Ironically, the New York Times article itself betrays traces of white hat bias. It notes that soft drink companies may be becoming increasingly proactive in sponsoring merchants of doubt because their sales have fallen by 25% in US in the past two decades, with similar, though less dramatic, falls in Australia. But adult obesity has been increasing in both Australia and the US over this time, which would lessen our confidence that soft drink intake plays an important role in obesity. The New York Times also evinces as their “obesity expert” a University of Ottawa academic with a very modest publication record. Dr Yoni Freedhoff is a prolific video blogger whose recent book The Diet Fix is not mentioned by the article as a potential source of conflict of interest.

It turns out the guys with the white hats have guns, too, and they’re more and more willing to use them.

Professor Tim Olds leads the Health and Use of Time Group at the Sansom Institute for Health Research, University of South Australia. He has received modest funding from Coca-Cola to present at a conference in Dubai, and Coke once sent him a nice basket of fruit in appreciation of a presentation he gave to executives. He is part of a multinational study funded by Coca-Cola, but neither he nor his institution is directly remunerated by Coca-Cola or any other food company.