electric competition
pollution prevention

Initial Comments of Ashley C. Schannauer
EPA Ombudsman Hearing on the
WTI Hazardous Waste Incinerator


September 23, 2000
East Liverpool, Ohio

My name is Ashley Schannauer. My address is 600 Grant Street, Suite 2980, Pittsburgh, PA 15219. Iíd like to thank you for the opportunity to testify today.

I participated as a lawyer in the WTI litigation in the early 1990s. More recently, I wrote an article that analyzed the WTI risk assessment. The article was published in the Vermont Law Review last Fall. It highlights the extent to which policy decisions -- not science -- influenced the WTI risk assessment and recommends the need for greater public input to produce risk assessments that have any value. My comments today are drawn largely from that article.

There's a famous line about risk assessments Iím sure youíve heard:

Risk assessments are like spies; Torture them enough and they'll say anything. Well, the WTI risk assessment proves the truth of that saying. The EPA manipulated its standards for acceptable risk, and its risk assessment guidelines and produced a document that says the risks posed by the WTI incinerator are acceptable. The risk assessment reached those conclusions by excluding large numbers of risks and inventing new standards of how much risk is acceptable. For compounds that might not pass the conventional standards, the EPA devised new standards that they could pass.

Your June 16 questions to the EPA asked what steps were taken in the construction and testing of the WTI plant to protect the publicís health. EPAís answer, in its July 14 letter, stated that the permit is designed to protect human health and the environment and that testing, including the WTI risk assessment, was done to determine whether the permit and regulatory standards are sufficiently protective.

Permits are issued primarily on the basis of standards relating to combustion efficiency, not health risks. So, the only review of health risks was in the risk assessment.

My comments will show some of the ways in which the EPA tortured the WTI risk assessment. I address three major topics:

  1. I identify the risks that the EPA excluded from its risk calculations.
  2. I show how the exclusions violated EPA risk assessment guidelines and accepted risk assessment practice.
  3. I show how the EPA changed its standards of acceptable risk during the course of the risk assessment.

1. Risks not addressed in the WTI Risk Assessment

a. The EPA excluded risks from compounds lacking EPA-approved toxicity values (excluding the large majority of the 200 compounds initially selected as Substances of Potential Concern).

The EPA identified an initial list of more than 200 "Substances of Potential Concern" to include in the risk assessment. The list was based upon the EPAís risk assessment guidelines and the WTI trial burns. For a variety of reasons, however, the EPA actually analyzed only a small fraction of those compounds.

EPAís primary basis for excluding a compound was the lack of an EPA-approved toxicity value for the compound. The EPA only used EPA-approved toxicity values to assess the risks of compounds. If the EPA had not approved a toxicity value for a compound, the EPA excluded the compound from its risk calculations.

The lack of EPA-approved toxicity values, however, does not mean that the compound is not toxic nor that sufficient data doesnít exist to determine a value. It may mean instead, as in the case of dioxins, that the compound is extremely toxic and controversial. Plausible alternative toxicity values from other sources, including other agencies and the public, can be used. The federal Agency for Toxic Substances and Disease Registry (ATSDR), for example, has recommended a threshold dose (Reference Dose (RfD)) for dioxins, but the EPA does not use it.

The result of EPAís practice in the WTI risk assessment is that the risks of the large majority of the 213 compounds identified as substances of potential concern were not included in the EPAís risk estimates. Iíve attached a table to my testimony that totals the compounds that were assessed.

The table shows that of the 213 compounds initially selected for analysis --

  • Only 59 (or 27%) were assessed for cancer risks arising from inhalation.
  • 119 (or 56%) were assessed for non-cancer risks arising from inhalation.
  • 31 (or 15%) were assessed for cancer risks arising from food chain exposures.
  • 20 (or a full 9%) were assessed for non-cancer risks arising from food chain exposures.

Perhaps most significant, the non-cancer risks of dioxins and furans were totally excluded from the food chain calculations. This also meant that the EPA failed to assess the non-cancer dioxin risks to breast-feeding infants as was recommended in the EPAís December 1994 risk assessment guidance.

137 were omitted from the assessment of inhalation cancer risks, and 70 were omitted from the assessment of inhalation non-cancer risks. For food chain risks, the EPA says it omitted 31 compounds due to the lack of average emission rates. 77 more were omitted due to the lack of toxicity values for cancer and 72 were omitted from the non-cancer analysis due to the lack of non-cancer toxicity values. See Schannauer article at 66-67, 79-80.

b. The EPA conducted alternative qualitative assessments for dioxins, lead, other criteria pollutants and the risks of accidents.

For a few compounds lacking EPA-approved toxicity values, such as dioxins, lead, and other criteria pollutants, the EPA used alternative risk analyses separate from the agencyís numerical calculations of risk. The EPA also conducted a limited assessment of the risk of accidents, given the plantís location next door to the East End Elementary School.

This practice had two results. First, it gave the EPA the opportunity to devise alternative risk standards for compounds that might fail the usual standards. Second, the exclusion of the risks from the calculations compared to the usual standard gave the plant a greater chance to pass the usual standard.

Accordingly, the incremental cancer risk of 1 in 1 million calculated by the EPA excludes the risks of large numbers of compounds, as does EPAís claim that exposures are all well below threshold levels of toxicity for non-cancer health effects. If the excluded compounds were, in fact, included, the EPAís risk calculations would certainly be higher.

The most significant example of such an alternative test is the Margin of Exposure analysis that the EPA invented to assess the risks of non-cancer effects from dioxins -- instead of using the probabilistic and hazard indices that were used for other compounds. -- Margin of Exposure analysis is used for compounds which would cause an exceedance of acceptable risks. Instead of calculating the health risk of such compounds, the EPA merely compares the plantís emissions with the background levels of the compound in the ambient air. EPAís reasoning is that the plantís emissions create an acceptable risk if the plantís emissions are relatively small compared to amount of the compound already present in the ambient air.

Thatís nonsense, isnít it? The MOE analysis totally avoids the question of whether the emissions from the plant may cause health problems, whether the background levels are already creating health risks, and whether the resulting levels of the compounds in the ambient air will cause health problems. Instead, the MOE considers the risks acceptable, if the new emissions from the plant are merely small compared to the background.

The MOE also seems fairly illogical. It suggests that a dirty plant can be operated in an area whose air is already dirty. On the other hand, the same dirty plant cannot be operated in an area with clean air. The total exposures -- and the health risks -- may be far greater in the area with the dirty air; but the plant is allowed there. The MOE approach seems to be a clever device thatís used to assess the risks of the most toxic chemicals -- the ones that will otherwise cause the calculated risks to exceed the agencyís risk standards.

The MOE approach, therefore, was ideal, from WTIís point of view, to assess the non-cancer risks of dioxins. It was inappropriate if your purpose is to estimate the health risks of a compound.

In its 1994 Dioxin Reassessment, the EPA suggested that dioxins and furans were so toxic and their concentrations in the environment already so substantial that it was impracticable to develop a traditional noncancer reference dose. The EPA stated that the threshold dose would be 10-100 times less than the average national daily exposure. See Schannauer article at p. 80. Nevertheless, the federal Agency for Toxic Substances Disease Registry (ATSDR) had recommended a threshold dose for dioxins -- but the EPA had not approved it for use in risk assessments. Id.

Instead of using a toxicity value for the non-cancer analysis which would have produced a quantitative risk estimate (which may have produced an unacceptable risk estimate), the EPA developed a MOE analysis (which included a different standard of acceptable risk). Not surprisingly, the results showed acceptable risks.

The EPA also used a similar MOE approach for background exposures, discussed below. Cancer and non-cancer effects of lead were assessed using a "bio-kinetic uptake approach", and cancer and non-cancer effects of sulfur dioxide, nitrogen dioxide and particulate matter were assessed based upon comparisons to the NAAQS for such compounds. The EPA invented a totally new assessment approach for the risks of accidents. See Schannauer article at 79-80, 83-85.

c. The EPA reduced the number of compounds assessed in the indirect (food chain) analysis to "focus" on the greatest risks.

The EPA said it attempted to analyze the risks of all of the substances of potential concern for inhalation risks (except for compounds lacking EPA-approved toxicity values). For the indirect (food chain) analysis, however, the EPA said it wanted to narrow the initial list of 213 compounds to "focus" the risk assessment -- to assess only the compounds with the greatest emissions, potency and bioaccumulation potential.

Most of the chemicals excluded in this attempt to "focus" were, in fact, excluded because the EPA lacked approved toxicity values for the compounds. Nevertheless, the EPAís attempt to "focus" was contrary to the EPAís April 1994 risk assessment guidance. That guidance sought to maximize the number of compounds to be assessed in the food chain analysis. In particular, the guidance outlined a long tentative list of compounds to be assessed for food chain risks and did not provide for the exclusions attempted in the WTI risk assessment. The guidance stated that it was implementing two significant expansions in risk assessment policy established in the agencyís 1993 Draft Combustion Strategy -- (1) to increase the number of routes of exposure analyzed and (2) to increase "the number of compounds analyzed and used in the risk assessment . . . in order to identify as large a fraction of the emissions as is realistically possible." April 22, 1994 Risk Assessment Guidance. The Guidance did not suggest the "focusing" effort attempted in the WTI risk assessment. See Schannauer article at 66-68.

d. The EPA excluded risks other than normal emissions -- "upsets" and accidents.

Quantitative risk estimates were calculated only for normal emissions, not for upsets and accident risks. EPA did qualitative analyses of upsets and accident risks, which were not incorporated into the risk estimates compared to EPAís standards of acceptable risk. See Schannauer article at 63-66. The exclusion of upsets from quantitative analysis violates EPA risk assessment guidelines. (see section 2 below).

e. The EPA calculated risks based upon emissions levels less than WTIís permitted emission rates.

The WTI risk assessment did not evaluate risks based upon maximum permitted levels of emissions; it evaluated the risks based upon emissions levels "expected" by the EPA and WTI. One would expect it reasonable to base a risk assessment upon the emissions that a plant is legally allowed to emit. The EPA risk assessors considered that approach too conservative.

The omission is also troubling because two analyses conducted in the course of the risk assessment process which did address permitted levels of emissions concluded that the emissions would exceed the agencyís risk standards. The Phase I Screening Risk Assessment from 1992 showed that the permitted 4.7 tons per year of lead emissions would cause an exceedance of EPAís screening level standard for inhalation risk. More recently, the Screening Ecological Risk Assessment component of the 1997 risk assessment revealed excess risks with the use of permitted emissions levels. The EPA risk assessors, however, chose not to evaluate the risks of the plantís permitted levels of emissions in the human health risk assessment. See Schannauer article at 70-73.

Also, risk estimates were not calculated for "high end" exposures contrary to EPA Risk Assessment guidelines. (see section 2 below).

f. The EPA excluded the risks of background exposures and sensitivities.

The risk assessment calculated risks based upon the false assumption that the ambient air is clean and the population completely healthy. The risks calculated by EPA only represent WTIís incremental contribution to compounds in the ambient air. The actual risks faced by residents, however, include the total concentrations as supplemented by the emissions from WTI.

In addition, the risk assessment excludes the background of harmful compounds, such as metals, that are already present and being retained in residentsí bodies. It also excludes the analysis of the increased risks associated with the populationís existing health problems, which are likely to be aggravated by WTIís incremental emissions. See Schannauer article at 73-76.

Instead, the EPA also used a MOE approach to assess the risks of background exposures, which, of course, did not assess the risks of background exposures. The EPA merely compared the incremental emissions from WTI against a proxy measurement for background exposures, the areaís Toxic Release Inventory emissions. The EPA found, under this analysis, that WTIís incremental emissions were relatively small compared to the background and found the risks acceptable. See Schannauer article at page 75. The EPA didnít analyze the actual health risks of the background exposures or the combined level of exposures.

James Butler, one of the WTI peer reviewers recommended both that background levels in the ambient environment be calculated and incorporated into the risk estimates and that background levels already contained in peopleís bodies should be considered:

For some contaminants (e.g., mercury, lead) it is essential to factor in estimates of existing body burdens and intakes from other sources. The project plan needs to address the issue of background exposures in the population subgroups, especially for compounds that are retained in the body, have relatively low thresholds, or have other significant sources in the environment. 1993 Peer Review Report, App. 1 (Comments of James Butler).

g. The EPA excluded the risks of endocrine disruption.

Despite the EPAís acknowledgment of the potential risks of endocrine disruption, the WTI risk assessment completely omitted endocrine disruption as a risk to be assessed. See Schannauer article at 81-82.

h. The EPA excluded likely synergistic effects.

The WTI risk assessment acknowledged the likelihood of health risks that total more than the sum of the risks from multiple compounds, but excluded the potential for adverse synergistic effects. The EPA said the available data was insufficient to calculate risks. See Schannauer article at 90.

2. Violations of EPA Risk Assessment Guidelines.

The WTI risk assessment violated EPA risk assessment guidelines as follows:/p>

a. The evaluation of only a small fraction of compounds recommended in Risk Assessment guidelines for food chain analysis to "focus" the assessment of such risks violates the EPAís April 22, 1994 Risk Assessment Guidelines.

The risk assessment states that the EPA assessed all "substances of potential concern" for inhalation risks, but did not explain why it was important or justifiable to assess only a small fraction of those compounds for food chain risks. See Schannauer article at p. 66-69. This is contrary to the April 1994 risk assessment guidance which seeks to expand the number of compounds considered in risk assessments. See also section 1 c above.

b. Sole reliance on EPA-approved toxicity values violates 1990 Risk Assessment guidelines requiring independent analysis of hazards of compounds.

Traditional risk assessments involve four steps: hazard identification, dose-response assessment, exposure assessment and risk characterization. See Schannauer article at pp. 38-40. Instead of independently identifying hazards from compounds likely to be emitted and then determining the toxicity values of the compounds, the EPA used EPAís limited list of toxicity values to determine hazards. If compounds had no approved toxicity value, the compound was not considered a potential hazard.

The EPAís 1990 risk assessment guidance states that hazards are to be investigated in each risk assessment, suggesting that EPA-approved toxicity values are not expected to be the sole determinant of whether a compoundís risks are assessed or not. See Schannauer article at pp. 77-79.

c. The exclusion of non-cancer risks of dioxin exposures to breast feeding infants violates the EPAís December 1994 risk assessment guidelines.

The EPAís December 14, 1994 risk assessment guidelines established a process and risk management standard for the non-cancer effects of dioxins on breast-feeding infants. The WTI risk assessment failed to address these risks, without even mentioning the December 1994 guidelines. The guidelines state a procedure and a standard to evaluate these risks:

The dioxin exposure assessment document released by the Office of Research and Development in April 1994, presents procedures for calculating infant exposures to dioxins and other lipophilic compounds through ingestion of human breast milk. The procedures are based on the intake of the contaminant by the mother. The exposure to an infant from breast feeding can be presented as an average daily dose (ADD) or a lifetime average daily dose (LADD). The ADD to the infant over a one year averaging time is predicted to be much higher (e.g. 30 to 60 times higher) than the ADD for the mother. . . . Research is incomplete however in the area of calculating risk for infant exposures to dioxin-like compounds in breast milk. One method of risk characterization, and the method used in this document, is comparison of the ADD to the average adult background level for dioxin exposure. 0.5 pg/kg/day. EPA, Guidance for Performing Screening Level Risk Analyses at Combustion Facilities Burning Hazardous Wastes (December 14, 1994), at C-5-2; see also C-5-53 ("[T]his guidance suggests that the average daily dose for one year of breastmilk exposure be compared to the average adult background exposure level for dioxin of 0.5 pg/kg/day, as suggested in the Dioxin Exposure Document.").

The EPAís analysis refers to the EPAís April 22, 1994 guidelines. It states, in a single sentence, that non-cancer health effects from dioxins were not considered in the risk assessment due to the significant uncertainty in applying noncancer toxicity data to infants. Final Risk Assessment, Vol. V, at VIII-17 - VIII -18.

d. The exclusion of the risks of upsets violates EPA Risk Assessment guidelines. See Schannauer article, at pp. 63-66.

The 1993 Peer Review Group also recommended unsuccessfully that upsets be included in the quantitative risk estimates:

The group is very concerned that potential emissions from non-routine conditions be considered as a major part of the risk assessment and not an afterthought. These may be the most important exposures and they should be fully evaluated. Methodologies are available to provide estimates of risks for accidents and spills (failure analysis) and should be incorporated . . . For process upsets the model should include how "upset" the processes are and how this affects the different products emitted. Modeling of accidents is crucial. While accidents, especially severe accidents, may be unlikely events, the group is concerned that high exposures could occur away from the containment devices. 1993 Peer Review Report, at 3-32.

e. The exclusion of background exposures violates EPA Risk Assessment guidelines.

The EPAís 1990 risk assessment guidelines require that background levels be incorporated into the quantitative risk estimates. See Schannauer article at n. 224.

The 1993 Peer Review Group also recommended that background levels be incorporated into the quantitative risk estimates:

Exposure to emissions from the WTI plant occurs against a background of previous and ongoing exposures to a variety of other chemicals, including chemicals of concern for the WTI facility. The additional burden from the WTI plant may move total exposures to the steep part of the dose-response curve. Thus modeling just the exposure from the WTI plant would underestimate the impact of the plant. For those chemicals that are air pollutants, the total exposure in air should be modeled. For the other, existing body burden should be considered, especially for chemicals of concern that are retained within the body (lead, dioxin). 1993 Peer Review Report, at 3-28.

f. The failure to calculate risk estimates for "high end" of exposures violates EPA Risk Assessment guidelines. See Schannauer article at pp. 70-73.

3. Risks first observed after EPA Risk Assessment

Obviously, the WTI risk assessment also fails to address new information about health risks that has become known since the time of its release. This new information will be discussed by others. It includes new information about the health effects of dioxins, endocrine disruption, the number and nature of accidents, fires and explosions, permit violations, the cancer study performed by the Ohio Dept. of Health, and the notorious "purple plume".

4. The standards of acceptable risk used in the risk assessment and in the EPAís approval of final operating conditions after completion of the risk assessment were arbitrary.

One way to determine that the risks of a plant are acceptable is to torture or manipulate the assumptions used in the risk assessment to produce lower estimates of risk. That's the approach that most critics focus on. This was done with the WTI risk assessment.

A second way to determine that a plant's risks are acceptable is for the EPA to change the standards of acceptable risk as it does the risk assessment -- to effectively increase the size of the risk that's considered acceptable. That was also done with the WTI risk assessment -- as the EPA judged whether the calculated risks were acceptable.

a. The EPA commingled the distinct functions of risk management and risk assessment, destroying the credibility of results.

"Risk management" is the process of determining which risks to assess and the level of risks considered acceptable. Risk management, for example, determines the standards used to grant or deny a permit. "Risk assessment" is the process of calculating the chances that the risks selected in the Risk Management process will occur. This includes the processes of identifying hazards and emissions, determining toxicity values and calculating risk estimates. Risk assessments are supposed to answer the questions and satisfy the standards established in risk management standards.

These processes are distinct and should be performed by separate groups of people. The National Research Council (NRC) of the National Academy of Sciences stated in a 1994 book entitled "Science and Judgment in Risk Assessments" that the commingling of risk management and risk assessment functions destroys the credibility of a risk assessment and permitting decisions based upon the risk assessment results:

Risk management policy, by its very nature, must entail value judgments related to public perceptions of risk and to information on risks, benefits, and costs of control strategies for each substance considered for regulation. Such information varies from substance to substance, so the judgments made in risk management must be case-specific. If such case-specific considerations as a substanceís economic importance, which are appropriate to risk management, influence the judgments made in the risk assessment process, the integrity of the risk assessment process will be seriously undermined. Even the perception that risk management considerations are seriously influencing the conduct of risk assessment in an important way will cause the assessment and regulatory decisions based on them to lack credibility. National Research Council, Science and Judgment in Risk Assessment (1994) at 49 (Emphasis added).

With WTI, the same group of people performed both functions. These people adjusted -- or increased -- their standards of acceptable risk as they performed the risk assessment; and they made questionable and controversial assumptions in their calculations of estimated risks.

b. Ad hoc standards and manipulation.

The EPA was able to adjust its standards of acceptable risk in the course of the WTI risk assessment, because the standards are determined by informal guidance which is not generally binding on the agency and because the agencyís risk management standards leave important questions unanswered.

This ad hoc process gave the EPA the opportunity to define the scope of risks assessed and the permitting standards to conform with the expected risk assessment results. Conversely, the process also gave the EPA the opportunity to manipulate the results to satisfy the risk management standards.

Unlike formal regulations, guidance documents are not binding, so that the EPA can change the standards with each risk assessment and each permit decision at each plant. The EPA is also usually careful to call the guidance documents "drafts." The agencies are not definitively required to follow them. They can change them quickly and without the public comment period required for regulations. In this way, the EPA can change the scope of risks it evaluates, the methods of performing risk assessments and the standards of acceptable risk for each plant.

Further, the EPAís only formal risk management and permitting standard for hazardous waste incinerators is that the facility be "protective of human health and the environment." EPA fleshed this out somewhat more in risk assessment guidelines with a standard that appears definite -- 1 in 100,000 incremental risk of cancers and a 0.25 hazard index for non-cancer effects. On closer examination, however, the apparent definitiveness is illusory. The standard does not specify the compounds whose risks are evaluated against the standard. Which risks? Normal emissions, upsets and/or accidents? Which compounds? Obviously, risk estimates will be higher or lower (and will fail or pass the standard) depending upon the number and nature of the risks included. These vagaries give the EPA the flexibility to add or delete compounds in the quantitative estimates in each risk assessment.

Examples of this ad hoc approach are the use of the Margin of Exposure analysis for non-cancer effects of dioxins and the risks from background exposures discussed earlier. The EPA used the MOE approach after it excluded these risks from its calculated risk estimates. Other examples include the failure to assess the non-cancer dioxin risks for breast-feeding infants, and alternative analyses for lead, for other criteria pollutants and for accidents.

5. The risk management and risk assessment standards used in the WTI risk assessment were determined without public input in violation of EPA policy

The EPAís 1993 draft Combustion Strategy recognized the importance of public participation in the permitting process and recommended improved public participation opportunities and methods in the permitting process for hazardous waste incinerators, including risk assessments. The 1994 final Combustion Strategy was not as specific in these recommendations, but EPAís recent policy statements have encouraged greater public participation.

Public input, however, was not allowed in the WTI risk assessment process.

Public input should be allowed on the following if any further risk reviews are conducted:

  • To help develop the risk management standards that determine which risks will be assessed, the metrics in which the risks will be calculated and whether any calculated risks are acceptable.
  • To help develop the risk assessment standards which determine how to conduct a risk assessment.
  • To help develop the site-specific assumptions used in the risk assessment process.
  • To develop an alternative set of plausible assumptions to provide a reality check on the accuracy of the agencyís results.

6. Conclusion.

Finally, in answer to the question you posed to the EPA in June about what the EPA did to ensure that the plant is protective of human health and the environment, the answer is that the EPA did a flawed risk assessment and not much more. The EPA was, with one hand, manipulating its standards for acceptable risk, and, with the other hand, it was manipulating the risk assessment results -- all to produce a report that says that the risks of the incinerator are acceptable.

Iíll close with a quote from one of the EPA-designated Peer Reviewers of the WTI Risk Assessment, Halstead Harrison, Prof. of Atmospheric Sciences at the University of Washington. Halstead summarized his opinion of the risk assessment as follows:

This ex-post-facto Risk Assessment process has produced distracting verbiage that obscures a simple truth: it was wrong to permit the construction of a toxic-waste incinerator on a flood plain of a narrow river valley, adjacent to a school. May 12, 1997.