A Study on studies.

[PopeTom]

Dr. David Ludwig has lead a study that seems to show a bias in beverage studies that seems to favor the funder of the study itself.

Does milk lower blood pressure? Does juice prevent heart disease? Beverage studies were four to eight times more likely to reach sweet conclusions about health effects when industry was footing the bill, a new report contends.

"We found evidence that's strongly suggestive of bias," said Dr. David Ludwig, an obesity specialist at Children's Hospital Boston who led the work, which was published Monday in the online science journal PLoS Medicine. The consumer advocacy group Center for Science in the Public Interest also participated.

A spokes person for the beverage industry sees it another way.

"This is yet another attack on industry by activists who demonstrate their own biases in their review by looking only at the funding source and not judging the research on its merits," says a statement by Susan Neely, president of the American Beverage Association. "The science is what matters — nothing else."

Is this a sad example of how the scientific process can be abused by someone who has a stake in what is being studied, or are the findings not really a big surprise and one should assume that a study on an industry funded by the industry being studied is probably not as good as it could be (if good at all)?

Yahoo News

 

[El Greco]

There have been other reviews too that have showed such biases. Indirect funding is even more treacherous. What can we say, this can't be stopped since people will always love money. Peer reviewers unfortunately don't always have the ability to distinguish carefully manipulated results. One can only look for more studies in order to draw safer conclusions.

 

[Kaylee]

Privately funded research has its problems due to conflict of interest.

Every few years I try to research the latest in the hearing aid market -- not easy. Most of the research in that area is privately funded and it's difficult to get the complete results (e.g., How did the competitors products do? And exactly which of the competitors products were being compared?). And I've no doubts that the file drawer effect exists. I've never seen a negative study released by a hearing aid manufacturer.

But expecting complete results to be released from privately funded test to the public isn't realistic. The best that can be done, I suppose, is that privately funded research should be clearly labeled as such and all the fund sources revealed. I personally regard the value of privately funded research as somewhere between advertising and publicly (or independently) funded research results. I'm just not sure where on the continuum it falls.

ETA: El Greco, by indirect funding do you mean private funding that is not revealed as such?

 

[casebro]

I believe the same holds true for statin drug studies .

And saturated fat studies too.

 

[El Greco]

ETA: El Greco, by indirect funding do you mean private funding that is not revealed as such?

Yes. I happen to have personal experience with a couple of such cases.

 

[Modified]

Overall, studies funded entirely by industry were four to eight times more likely to be favorable to their sponsors.

They may be largely measuring industry bias in the selection of studies to fund. I'm sure that in general, they only fund studies for which there is some initial indication that results will be favorable. From the link you gave, it seems there is no accounting for this effect.

 

[Art Vandelay]

This was given as a hypothetical as a paradox. If studies showing an effect that the funder was looking for are invalid because of the possibility of bias, then a study showing such (which, presumably, would be funded by someone looking for evidence of bias) would be biased in favor of finding such, so it would be invalid, so there wouldn't be evidence of it, so we shouldn't reject the study, so there is bias, so the study is biased, so it is invalid, so there isn't evidence...

 

[Slimething]

I'm sure that in general, they only fund studies for which there is some initial indication that results will be favorable. From the link you gave, it seems there is no accounting for this effect.

Right on. The study on studies is itself biased. Science is expensive, especially when a human or animal effect is being measured. The scientists needed to administer and track these effects are very highly trained and good talent costs money. Companies don't spend that kind of money on research unless they are fairly well already convinced that measuring a chosen characteristic would be beneficial to their product line.

On the other hand, there is no law that states that unflattering data must be published. Only in the case of finding an adverse effect does a company have a legal obligation to disclose the information and, even then, only to the appropriate government regulators.

 

[athon]

While not commenting directly on the study itself, I do agree that the research is what is important, rather than making an overarching statement about all studies that have been sponsored by a group.

For example, perhaps the reason why it's four times more likely to have positive results for a sponsored study is that the specific questions asked by such a company are more open to positive outcomes. The study itself isn't biased, even if the limitations of the study are to begin with.

For instance, a non-sponsored study on a brand of hot-dogs might be more inclined to ask how nutritious they are, while the sponsored study might ask how many vitamins they have. The latter study might not be biased, if you actually look at the study itself...even though its direction might be misleading.

In summary - argue the study itself.

Athon

 

[SezMe]

News bulletin: The sky is blue.

A recent egregious example: studies on smoking sponsored by the tobacco industry. The most obvious current example: studies on global warming sponsored by the energy industry.