The US Environmental Protection Agency (EPA) released unrealistic, scientifically flawed, and arguably illegal drinking water health advisories for perfluorooctanoic acid (PFOA) and perfluorooctane sulfonic acid (PFOS) on June 15, 2022. The health advisory levels are so low that drinking water suppliers cannot even test for these levels of PFOS and PFOA as they are hundreds of thousands of times lower than can be detected using EPA’s own test methods (see below).
So how did we get here?
The EPA based its health advisories for PFOA and PFOS on a 2018 study by Budtz-Jorgensen and Grandjean that looked at the association between PFOS concentrations in blood and diphtheria antibodies in children. This study calculated a PFOS concentration that the authors state is not likely to have an adverse effect on diphtheria antibody production. And that’s true — it appears that Budtz-Jorgensen and Grandjean did not see diphtheria antibody levels that were in the adverse health range for any subjects (read a different way: none of the children in the study were experiencing diphtheria antibody levels that were too low to yield basic immunity — more on that later).
The problem is: EPA failed to realize that Budtz-Jorgensen and Grandjean violated EPA guidelines and best practices for analyzing the data they had. As a result, when EPA adopted the study findings at face value, EPA violated its own guidelines and best practices.
Below I’ll dive into what actually happened and where the problems are.
EPA’s Critical Study Did Not Observe an Adverse Effect
The Safe Drinking Water Act (SDWA), as amended by the SDWA Amendments of 1996, requires that the EPA must use adverse effects for setting policies like drinking water health advisories and safe drinking water regulations. That means that a scientific study that EPA is using as the basis of an health advisory or regulation must demonstrate an adverse effect.
The problem is that the study by Budtz-Jorgensen and Grandjean did not show an adverse effect.
There’s actually an international consensus on what an adverse effect is when considering diphtheria antibodies. The World Health Organization, US Centers for Disease Control and Prevention, and US Food and Drug Administration all agree that the basic immunity threshold for diphtheria antibodies is 0.01 IU/mL. However, Budtz-Jorgensen and Grandjean are not looking for decreases in diphtheria antibodies that are below this level as they claim “this limit is somewhat arbitrary…”
When we look at Budtz-Jorgensen and Grandjean’s analysis (Figure 1 and reproduced below) we see that their model only goes down to a diphtheria antibody concentration of 0.444 IU/mL — that’s 44x larger than the basic immunity threshold of 0.01 IU/mL. So clearly their model is not capable of establishing the range where toxicity occurs, because that toxic level is not even covered by their model.
In other words: Budtz-Jorgensen and Grandjean’s analysis does not demonstrate an adverse effect.
So, when EPA wrote up the interim Health Advisories for PFOA and PFOS they did not base them on actual adverse effects. They based them on completely safe values. But recall that the SDWA requires Health Advisories to be based on adverse outcomes.
Bottom Line: EPA’s Interim Health Advisories for PFOA and PFOS are a clear violation of the Safe Drinking Water Act.
EPA Used Correlation, Not Causation, As Basis of Health Advisories
The Budtz-Jorgensen and Grandjean’s study is associative or correlative, meaning it cannot state PFOA or PFOS caused any effects. Rather, they are looking at PFOS and PFOA levels in people, and they are looking at diphtheria antibody levels, and calculating a correlation. Because they are only looking at correlations, the authors do not know what is actually causing the alleged non-clinically relevant decreases in diphthetia antibodies in children. These decreases could be due to some other factor that they are not measuring, perhaps an epigenetic issue, or exposure to a different chemical.
I won’t get into all of the requirements for a proper causal analysis, but among other things, you need to satisfy the “but-for” criterion– “but for PFOS/PFOA, the diphtheria antibody levels would be normal.” This study cannot address that. There are ways of structuring human studies to get at that but-for criterion. However, this study is not that.
Add on the fact that Budtz-Jorgensen and Grandjean’s study does not actually identify an adverse outcome and you have:
Bottom Line: EPA has no causal evidence that PFOA and PFOS cause an adverse outcome.
EPA Violated Its Own Guidance
Violating its own guidance seems to be a recurring theme lately at the EPA. For instance, the 9th Circuit Court of Appeals caught EPA violating their Cancer Guidelines.
In this case, the EPA is violating its own Benchmark Dose Technical Guidance that specifies how EPA and its contractors analyze dose or concentration-response data. Budtz-Jorgensen and Grandjean’s analysis violates EPA’s guidance, and uses parameters that the EPA states are not best practices.
But what EPA states in their Draft Approaches to the Derivation of an MCLG for PFOS in Drinking Water echoes what Budtz-Jorgensen and Grandjean give as a justification for their improper use of a 5% BMR:
EPA misapplied the BMD guidance by confusing the data they actually have in Budtz-Jorgensen and Grandjean, which is continuous, for quantal data, thereby misapplying the quantal guidance to the continuous data.
Or more precisely, Budtz-Jorgensen and Grandjean are misapplying The Benchmark Dose Technical Guidance, and EPA is going along for the ride, but chooses to justify the 5% benchmark response (BMR) in a slightly different way.
What EPA and Budtz-Jorgensen and Grandjean did was to read the wrong part of the Benchmark Dose Technical Guidance; specifically they both are referencing the (dichotomous) data section “…most reproductive and developmental studies with nested study designs easily support a BMR of 5%. Similarly, a BMR of 1% has typically been used for quantal human data from epidemiology studies.” Budtz-Jorgensen and Grandjean reasoned that a 5% BMR was more appropriate because it was the “lower BMR”, which defies logic as a 1% BMR is also mentioned; however, more importantly EPA improperly concluded that a BMR of 5% is appropriate “[g]iven the range of health outcomes includes fatality and the effect on children…”
However, Budtz-Jorgensen and Grandjean data are continuous. That means EPA must apply the continuous data guidance, not the quantal data guidance. The continuous data guidance begins in the very next section and paragraph of the Benchmark Dose Technical Guidance. Therefore, as the data in Budtz-Jorgensen and Grandjean is continuous data, and not quantal data, EPA has misapplied the BMD Guidance, and has used an inappropriate BMR.
These points taken together demonstrate that EPA/OW has grossly misapplied the BMD process and has made a significant error.
What EPA should have done is clearly stated in the Benchmark Dose Technical Guidance because they have continuous data and a consensus basic immunity threshold for diphtheria antibodies (page 22) :
This means that EPA, and not Budtz-Jorgensen and Grandjean, must use the consensus, clinically validated antibody threshold, established by the CDC, FDA, and the WHO in their benchmark dose analysis. That level is 0.01 IU/mL.
Bottom Line: EPA should have used the basic immunity threshold of 0.01 IU/mL as the BMR. Therefore, EPA erred when it stated that a 5% BMR is reasonable and appropriate.
EPA Bases Health Advisories on Draft Assessments Labeled, “DO NOT CITE, QUOTE, OR DISTRIBUTE”
I get it, I used to work at EPA, and sometimes people make mistakes. However, it is rather peculiar that the Agency would draft Interim Health Advisories, communicate them widely, and base them on documents labeled “DO NOT CITE, QUOTE, OR DISTRIBUTE.” The EPA did just those things — they cited these draft assessments in the interim Health Advisories, they quoted them, and then they distributed them.
Not a big deal, but rather strange.
Here are the Draft Assessments:
The Health Advisories Are Simply Impractical
The Health Advisory levels for PFOA and PFOS are 130,000-250,000x lower than the limit of detection of EPA’s own test methods. This is simply impractical. There is no way for water system operators to achieve the levels in the Health Advisories.
The Safe Drinking Water Act Health Advisories for PFOA and PFOS are significantly flawed. I am still not sure why the EPA decided to just use the Budtz-Jorgensen and Grandjean benchmark analysis and take it at face value. That was not something I ever recall doing when I was at the Agency. It’s not only sloppy, but it is a huge risk.
I think the act that puzzles me most is that EPA would so willingly misapply its own guidance, and it would just take at face value the determination of a couple of academics when it comes to something as important a the Health Advisories for PFOA and PFOS.
The impracticality of the Health Advisory levels is also simply astonishing. And given that EPA messed up their Health Advisories, they will have an opportunity to do better.
Here are the bottom-line conclusions:
- EPA failed to realize that Budtz-Jorgensen and Grandjean violated EPA guidelines and best practices for analyzing the data they had.
- EPA’s Interim Health Advisories for PFOA and PFOS are a clear violation of the Safe Drinking Water Act.
- EPA has no causal evidence that PFOA and PFOS cause an adverse outcome.
- EPA should have used the basic immunity threshold of 0.01 IU/mL as the BMR. Therefore, EPA erred when it stated that a 5% BMR is reasonable and appropriate.
EPA needs to rectify this, and generate scientifically defensible SDWA Health Advisories for PFOA and PFOS. Taxpayers deserve better.