Last week, the 9th Circuit Court of Appeals remanded EPA’s approval of two nanosilver pesticides for use in textiles. The case, NRDC v. EPA, is a fascinating application of the issue of “how safe is safe” and, in particular, how much conservatism must be applied to risk estimates when there is significant uncertainty in the analysis.
EPA sets the acceptable exposure to pesticides under FIFRA by determining the risk and then addressing uncertainties. In this case, EPA had to account for three different sources of uncertainty: interspecies uncertainty, since the underlying data was based on studies in animals other than humans; intraspecies uncertainty associated with differing sensitivities among different people; and “database” uncertainty relating to the minimal available data on nanosilver toxicity. EPA then applied a factor of 10 to each uncertainty.
In other words, EPA determined what level of exposure would pose a risk. It then applied each of these uncertainty factors, resulting in a number 1,000 times higher than the calculated risk level. EPA’s rule of decision was that it would conclude that the risk was of concern as long as it was less than or equal to this “margin of exposure” of 1,000.
Unfortunately, EPA’s own analysis of the aggregate risk to toddlers from use of nanosilver as a surface coating on textiles resulted in a margin of exposure of 1,000. Not 999. Not 1001. Since EPA’s rule was that there is a risk requiring mitigation if the MOE is less than or equal to 1,000, the Court concluded that EPA’s approval was arbitrary and capricious. EPA simply ignored its own decision rule.
EPA had two arguments in support of its approval. First, it asserted that the MOE of 1,000 resulted from rounding, and that if it did not round, the MOE would be 1,006. The Court gave EPA a short lesson in significant digits and concluded that EPA was in fact required to round, leaving the MOE at 1,000.
EPA’s second argument was more important, and it is the basis for the title of this post, which will strike a chord with anyone who has ever been in the unenviable position of challenging an EPA risk analysis. According to the Court, EPA argued that:
Its MOE calculations are based on very conservative assumptions so an actual MOE in the neighborhood of 1,000 does not mean that consumers are at risk.
Indeed, as EPA noted, its uncertainty factor of 1,000 means that toddlers would only be exposed to 1/1,000 of the amount of nanosilver shown to produce toxic effects. The Court’s response was succinct:
Although EPA’s point is well taken as a practical matter, it is irrelevant as a legal matter. This is EPA’s rule, not ours.
In other words, if EPA chooses a foolishly overconservative approach to risk assessment, it cannot then ignore the results when it realizes that it has been too conservative. Hoist on its own petard.
Anyone want to predict whether EPA will in fact go back and change its rule of decision?
From my experience in this discipline, I believe the recurring problem illustrated by this case would be mitigated by a requirement for the risk assessments to also include “most likely” endpoints in addition to whatever “worse than worst case” endpoints are produced by the currently required overly conservative assumptions. Consistently better discussion of uncertainties (and their implications, and how to resolve them) wouldn’t hurt either.