, , , , , , , , , , , , , , , , , , , , ,

There is still time to comment by Thursday night, 1159 US Eastern Time (NY-DC time zone). It becomes public record https://www.regulations.gov/docket?D=EPA-HQ-OA-2018-0259 Docket ID: EPA-HQ-OA-2018-0259

Excerpt from “Comments of the International Society for Environmental Epidemiology on EPA’s proposed rule on Strengthening Transparency in Regulatory Science (EPA-HQ-OA-2018-0259-0001)

4. The EPA’s proposal in regards to dose-response analysis is inappropriate and not scientifically based EPA’s proposal details very specific criteria for judging what types of analyses provide valuable information about what concentration or dose-response curves look like, instead of relying on the judgment of the scientific community and its own science advisory committees. The proposal states “When available, EPA shall give explicit consideration to high quality studies that explore: a broad class of parametric dose-response or concentration-response models; a robust set of potential confounding variables; nonparametric models that incorporate fewer assumptions; various threshold models across the dose or exposure range; and models that investigate factors that might account for spatial heterogeneity.” But this ignores studies that sequentially exclude observations with exposures (doses) above a certain level, although they can unambiguously demonstrate that an association exists below that level, and subject to power limitations, that it does not exist below that level. It also puts an inappropriate emphasis on studies looking at broad classes of parametric forms for dose response, which is an approach little used in environmental epidemiology, because those parametric forms by their nature can force dose-response curves into shapes not indicated by the data. Moreover, giving higher priority to a study because it tried a large number of mostly inappropriate parametric forms compared to another study that considered a smaller number of well thought out parametric forms creates major multiple comparison issues and stands science on its head. Multiple bad analyses do not make a study more reliable. Instead, environmental epidemiology prefers penalized splines and similar non-parametric approaches. We know of no epidemiology society that recommends EPA’s approach.

The emphasis on threshold models is also questionable. A threshold model is a form of a spline, a model where for example, the slope of a linear association is allowed to change at a certain point. In a threshold model, one considers a piecewise linear dose-response, where the slope is assumed to be zero below some dose T, and some positive number above T. But that is a very restrictive assumption. Spline models are quite common in environmental epidemiology, but rather than, in the example above, forcing the slope below T to be zero, they allow the slope to change at T, but allow the slope below T to take on any value the data dictates. EPA provides no reason to justify the approach they propose, or for constraining the slope to be zero when the data might differ.

Moreover, when EPA states that there is growing evidence for nonlinear dose-response relationships they imply that science is finding more evidence of thresholds. In fact, in human studies the opposite is true. For lead, for example, the nonlinearity is a flattening out of the dose-response relationship at the high dose end, not at the low dose end. This is true for many other toxins as well, including PM2.5.

And there is a clear biological reason— most of these toxins act on pathways in the human body that produce the diseases of aging. These pathways have already had their coping abilities overwhelmed, and hence we age.

Therefore, any incremental stimulus along these pathways would be expected to produce incremental damage to the body. It is for this reason that a National Academy of Sciences committee on risk assessment stated “The committee recommends that cancer and non-cancer responses be assumed to be linear as a default60.”

The methods for assessing the shape of a concentration-response relationship is a scientific question, and better left to scientists and EPA’s science advisor, without stating preferences with no scientific justification given, and no apparent input from either the scientific community or EPA’s own scientific advisors“. Emphasis our own. https://www.regulations.gov/contentStreamer?documentId=EPA-HQ-OA-2018-0259-1973&attachmentNumber=1&contentType=pdf

Many radioactive materials poison as both heavy metals and radiologically. There is even radioactive lead!

Entire comment here:

Yellow highlight added. Original here: https://www.regulations.gov/contentStreamer?documentId=EPA-HQ-OA-2018-0259-1973&attachmentNumber=1&contentType=pdf