STATS ARTICLES 2007

2009 | 2008 | 2007 | 2006 | 2005 | 2004 | 2003


Washington Post Skews Story on Chemical Obesity Risk
Trevor Butterworth, March 12, 2007
Limitations in scientist’s research not addressed, contrasting risk evaluations by Harvard, others ignored.

After the Economist reported that scientists had, at the annual meeting of the American Association for the Advancement of Science last month, discussed the possibility that exposure to certain chemicals might be a factor in obesity, it was only a matter of time before other news organizations decided to scare their readers.

While the Economist reported that the theory was “controversial,” it misrepresented some of the scientific research, as we noted last week. Now, the Washington Post has entered the fray with Chemicals May Play Role in Rise in Obesity, which claims that the evidence is “preliminary,” and cited other scientists concern that the risk was “plausible and possible.”

The problem with the Post’s piece, however, is its reliance on one scientist to attest to the risks that bipshenol-A presents:

"Exposure to bisphenol A is continuous," said Frederick vom Saal, professor of biological sciences at the University of Missouri at Columbia. Bisphenol A is an ingredient in polycarbonate plastics used in many products, including refillable water containers and baby bottles, and in epoxy resins that line the inside of food cans and are used as dental sealants. In 2003, U.S. industry consumed about 2 billion pounds of bisphenol A.

Researchers have studied bisphenol A's effects on estrogen function for more than a decade. Vom Saal's research indicates that developmental exposure to low doses of bisphenol A activates genetic mechanisms that promote fat-cell activity. "These in-utero effects are lifetime effects, and they occur at phenomenally small levels" of exposure, vom Saal said.”

The Post “balanced” vom Saal’s points with a comment from an industry spokesman – a bizarre decision, not only because industry sources do not have the rhetorical equivalence as “independent” scientists, but because the paper could have gone to the scientists who participated in two peer-reviewed risk assessments of bisphenol-A, one by the Harvard Center for Risk Analysis, and the other by the Gradient Corporation – or, for that matter, to other independent scientists.

In any news report, a scientists tagged as independent will always trump an industry source in reader credibility. Vom Saal criticizes the industry spokesman’s claims to no ill-effects  as “a blatant lie” – even though the two risk reviews did not find any risk to human health from low dose exposure to bisphenol-A. The Post says many scientists agree with vom Saal, but fails to name any.

The Post also failed to either understand or note is that vom Saal’s position is controversial within toxicology. Though he may well be a visionary for waging a campaign to warn the public about the effects of low-dose exposures to chemicals like bisphenol-A, this is not yet a consensus position, and there is much to disagree on. But perhaps the most significant problem with vom Saal is that the evidence he has marshaled is limited for the following reasons.

In his last literature review vom Saal appears to accept any study that found any result pointing to a risk from bisphenol A, no matter what the dosage or endpoint, and ignore any study that didn’t find a negative outcome.

There appears to be no attempt to analyze the negative studies for accuracy or applicability of method or results, or to see if the results of these studies cohere with each other in a way that points to some underlying causal mechanism, or to see if the results can be replicated. Vom Saal also doesn’t appear to address, at a methodological or causal level, why there are disagreements between studies on different endpoints. Nor does he discriminate between studies that, for methodological reasons, have little or no relevance to human health.

This is in marked contrast to the methods used by the Harvard Center for Risk Analysis and the Gradient Corporation, neither of which found evidence for bisphenol-A being linked to obesity.

There is also the problem that some of research vom Saal has engaged in on bisphenol-A could not be replicated by other scientists. As John Gierthy notes in Testing for Endocrine Disruption: How Much is Enough? “there is considerable controversy over the existence and/or relevancy of these low dose estrogen effects.” And as STATS noted, vom Saal has clearly overplayed the evidence of one Japanese study that apparently found a link between elevated body mass index and bisphenol-A.

Vom Saal has charged that the studies claiming no effect were industry-funded, and that the studies he has referenced which found problems are all independent, thus suggesting, at least to the ordinary reader, an industry conspiracy. This is possibly why vom Saal has found much favor among environmental activists campaigning for tougher chemical regulation. But in light of the limitations we have noted about vom Saal’s claims about bisphenol-A, one must equally ask whether he has come to believe in his thesis to the point where he is unwilling to look at data that undermines it.

It's also not fair to do no more than count up studies that appear to have found negative results and then claim that the mere fact of their independence gives them more credibility than industry-funded reviews that were undertaken by such prestigious institutions as Harvard using careful methods of analysis. The protocols for such reviews do not give industry a say in how the research is carried out.

This is why we need the media to break the mold on the way it reports chemical risks, by taking the underlying toxicological problems seriously. Determining the existence and effect of low-dose exposure to chemicals presents a huge scientific challenge, and if confirmed, the consequences are potentially enormous. When it comes to reporting the work of vom Saal, we need the press to do better than turn to an industry spokesman for the only source of balance. We need scientists who are less invested in a pre-determined outcome to analyze the data.