More on Atrazine

About a week ago, I posted on a New Yorker article about Atrazine - a widely used herbicide in agriculture.  After reading the article, I was a bit dismayed about the ability of science to resolve the controversy, writing:

More disconcerting still is what this says for the ability of science to resolve disputes about knowledge.  Whether atrazine causes abnormalities in a particular type of frog at a certain dosage should be one of those questions science can answer.  Either atrazine causes these effects or it doesn't.  This isn't one of the mysteries of the cosmos.  This isn't macroeconomics.  Yet, there are apparently findings on both sides of the issue (Note: I have't personally delved into the scientific literature on this particular matter in any detail). 

Well, I spent a bit more time digging into the issue, and the science on frog abnormalities seems much more settled than the New Yorker article let on.

It seems the defining study on the issue is the so-called Kloas study.  Here is what the New Yorker said about it:

The Kloas study was the most comprehensive of its kind: its researchers had been scrutinized by an outside auditor, and their raw data turned over to the E.P.A. But the scientists wrote that one set of studies on a single species was “not a sufficient edifice on which to build a regulary assessment.” Citing a paper by Hayes, who had done an analysis of sixteen atrazine studies, they wrote that “the single best predictor of whether or not the herbicide atrazine had a significant effect in a study was the funding source.”

While saying the Kloas study was the "most comprehensive", the article immediately  cites skepticism and immediately brings up the issue of industry funding. It seems to me an attempt to downplay the significance of this particular study and it's findings.

From what I can gather, the Kloas study (a long white paper about it is on the EPA web site) is by far the largest and most careful study on the issue (some additional background on it can be found in this article in the journal Environmental Science and Technology).  The study was mandated by the EPA because it felt that there were too many flaws in previous studies to make a definitive judgement.  Syngenta (the maker of atrazine) funded the study (because the EPA mandated they do so), but it was carried out by two independent laboratories, one in Maryland and one in Germany (under the supervision of the scientist Werner Kloas, which is where the study gets its name).  The study followed protocol set up by the EPA scientific advisory panel.  Data collection was audited and overseen by the EPA; the data was analyzed by a third party, which did not know the particular identify of each treatment, and the data itself is available to other researchers via the EPA.

The study employed over 1,000 frogs in each location.  There were five treatment groups, which varied the level of Atrazine from 0.01 to 100 parts per billion.  There was also a conventional control (with no Atrazine).  Interestingly, there was also a "positive" control - one group of frogs was given 17β-estradiol - a compound known to cause sexual abnormalities in frogs.  The idea with this "positive" control was to determine whether the conditions were such that abnormalities could develop (or were not being prevented via some other choice of the experiment environment).

Here are the results:  The "positive control" using the compound known to cause abnormalities, indeed caused a statistically significant rise in abnormalities.  However, there was no consistent statistically significant evidence that Atrazine caused abnormalities - either it's presence or in a dose-response fashion.  The EPA concluded (see page 115):

Because of the experimental design protocols and quality control of the [Kloas] studies, the data are sufficiently robust to outweigh previous efforts to study the potential effects of atrazine on amphibian gonadal development.

Based on the negative results of these studies, the Agency concludes that it is reasonable to reject the hypothesis formulated in the 2003 SAP that atrazine exposure can affect amphibian gonadal development. The Agency believes at this time, there is no compelling reason to pursue additional testing with regard to the potential effects of atrazine on amphibian gonadal development.

What about the previous studies showing effects?  Typically, such studies: 1) used a smaller sample size, 2) did not study whether there was a dose-response relationship, 3) did not utilize a proper control group (or "positive" control like the one described above), and/or 4) have not made their data available to other researchers.  In several places, the EPA has commented on these weaknesses, and on the inability to get specific data from researchers claiming ill effects.

One of the things that bothered me about the New Yorker article was the sense that science could not settle such an issue. I wrote:

But, I also fear this is part of an attempt to undermine the ability of scientific inquiry to settle empirical disputes. . . I hold out hope that science can, indeed, provide knowledge for those willing to follow the evidence where it leads.  Otherwise, every issue is simply a PR battle.

I have no idea whether Atrazine might cause other deleterious effects, and I cannot condone Sygenta's alleged actions against Hayes.  Nor can I condone his alleged actions against employees of Synenta. But, it seem to me the science has provided us a reasonably good answer to the question of whether Atrazine causes frog abnormalities at levels below 100ppb.  It is a shame that the New Yorker let telling a good story get in the way of this fundamental fact.