invalidates the analysis and thus the claims in the paper's conclusions.
Washington, DC (PRWEB) April 24, 2007
Experts who have reviewed a paper claiming a link between phthalates exposure and obesity say that the study should be taken with a large grain of salt.
"Based on our review, conducted by member company toxicologists and statisticians, this study is hard to swallow - pun intended," said Marian Stanley, manager of the Phthalate Esters Panel, a unit of the American Chemistry Council. "The authors themselves advise that there are 'several important limitations' to their findings - seven, to be exact. Perhaps the most important flaw is the selective way they presented their results. They reported results on their studies of males only. The results of their study of females are not shown because, they concede, the 'dataset gave dissimilar results'!"
The paper, published in Environmental Health Perspectives, claimed that phthalate exposure correlated with "abdominal obesity and insulin resistance in adult U.S. males." Its principal author is Richard Stahlhut of The University of Rochester School of Medicine. The paper does not claim to show that phthalates actually contributed to obesity, and in fact states that additional study would be need to establish such a link. But the paper was publicized by the university, and gained much media attention.
The data was drawn from single examinations of thousands of men, women and children in a government health and nutrition survey. The authors concede that obesity is a syndrome gathered over time, and a single-point snapshot cannot reliably predict the impact of anything. They also acknowledge that obesity has many causes, and trying to single out one cause is difficult. The Panel scientists also criticize the study for bending or breaking some of the basic rules of statistical analysis. For example, they grouped phthalate exposure data, because no association with any single phthalate could be found. Grouping data is done when the amount of data is limited, but in this case, they were drawing on a mountain of data from the government study. Grouping it has the effect of reducing the level of information, not improving it. The authors also applied linear regression analysis to non-linear data, which exaggerates the differences in the data, and in the judgment of the Panel's experts, "invalidates the analysis and thus the claims in the paper's conclusions."
"The Phthalate Esters Panel takes all science seriously, and we had this paper reviewed carefully by experts in the field," said Ms. Stanley. "It is our reviewers' conclusion that the authors performed their statistical analysis poorly and then over-interpreted it, and excluded data that disagreed with their conclusions. It is fair to ask whether this paper even deserved to be published."
Phthalate Esters Panel
The Phthalate Esters Panel (the Panel) of the American Chemistry Council is composed of all major manufacturers and some users of the primary phthalate esters in commerce in the United States. The Panel is committed to continued research and testing of phthalate esters and will continue to work closely with government agencies so that these materials can continue to be used safely in a wide array of products consumers find valuable in everyday life. Web site: http://www.phthalates.org