Panel Discussion

JOEL S. HIRSCHHORN, Ph.D., Seniro Associate, Office of Technology Assessment, Washington, D.C.

ELLEN K. SILBERGELD, PhD., Chief Toxics Scientist, Environmental Defense Fund, Washington, D.C.

DAN BEARDSLEY, Ph.D., Staff Director, Interagency Risk Management Council, Washington, D.C.

PETER B. HUTT, Esq., Moderator, Covington & Burling, Washington, D.C.

DR. JOEL HIRSCHHORN: First, I would like to support the view that there is, in fact, a lack of public trust in the Environmental Protection Agency (EPA). This lack of trust extends to both risk assessment and risk management.

Second, when it comes to chemical pollution in the environment, I think that the risks are underestimated. Typically, the data we are shown do not represent a board range of health effects. And we hardly ever see any data on strictly environmental effects and damage. There are many significant health effects of chemical pollution other than cancer and death, and people certainly have a right to be concerned about strictly environmental effects as well.

Let me also say that the scientific community has hardly begun to study synergism and interaction. When it comes to a finding that one gets cancer from a food or some other source, I do not know how one can conclude that this result has not been affected by an exposure to chemical pollution in the environment.

[16 ELR 10205]

I believe that risk stemming from chemical pollution will increase over time. This will happen for two principal reasons: exposures will increase, and we will see more and more people, who are controlling their risks voluntarily — I am thinking of the individual who is not overweight, who does not eat meat, who does not smoke or use caffeine, and so forth — being exposed involuntarily to environmental risks. I suspect that over time we will see better management of risk, and we should be more concerned at this stage in a precautionary sense with involuntary risks. The number of people involved is enormous.

EPA has demonstrated an extremely narrow view of risks and risk management. Risks are routinely shifted over time and space. We have abundant evidence in the Superfund program, for example, that cleanups have consisted of shifting waste from one community to another. In my view, this does not constitute a valid form of risk management.

Let me give you another example from the Superfund. Wastes are sometimes left in the ground and called "contained." Walls are built around the waste, and they are capped. I will say unequivocally that this technology will fail. Everyone who knows anything about technology knows that this technology cannot be effective over the lifetime of the toxic wastes. All we do is shift the risks to future generations. Eventually, the technology will fail, and the toxic chemicals will spread into the environment.

EPA has been attempting to minimize short-term costs. I find it fascinating that EPA has systematically ignored more effective technologies that are available to deal with these kinds of environmental risks.

There are three main areas of activity in the management of toxic wastes: source reduction (the reduction of generation of hazardous wastes); the regulation of newly generated wastes under the Resource Conservation and Recovery Act (RCRA) regulatory program; and the cleanup of toxic waste sites under the Superfund program. Let me assure you that our work, as well as the work of every other analytical group that has looked into these areas, has reached one conclusion: technologies available to mitigate risks are absolutely, unequivocally underutilized. They are underutilized solely because they increase the costs of those who must bear the costs. Our entire view of risk management in the area of chemical pollution attempts to minimize the immediate cost but does not deal with long-term cost.

I must agree wholeheartedly with Professor McGarity about the influence of the Office of Management and Budget (OMB). The discounting of future risk is a widespread problem. It is a very poor economic analysis that does not take into account that wise cost-effective spending on higher-cost technologies in the present can deal permanently with these risks.

I also wholeheartedly with Professor McGarity's comment about the need for public involvement. We have suggested in our recently released Superfund Report that the local community should be involved in decisions on toxic waste cleanup, because reuse, restoration, and rehabilitation of a site should be viewed as a land use decision.

DR. ELLEN SILBERGELD: As we have heard from Professor McGarity, the actual picture of how risk assessment and risk management are conducted is not particularly encouraging. I would like to use my allotted time to speak about how they ought to be conducted.

The answer to whether agencies should use risk assessment has to be, "yes." There are policies why risk assessment must be undertaken. As Professor McGarity mentioned, societal forces compel collective management of risk, because of the externalities that risks impose beyond those who create them. For a long time, we did not question the collective management of risk in areas such as the innoculation of the populace to prevent epidemics, the removal of counterfeit medicines, and, more recently, the control of salmonella-infected milk.

Agencies must require that more scientific data be put into the risk management process. The minimal requirements for data vary tremendously from agency to agency, from law to law, and even within a single agency. Of all regulatory statutes, the Food, Drug and Cosmetics Act requires the maximum data; the Toxic Substances Control Act, as administered by EPA, requires the least.

In our drive to achieve coordination and consistency in risk assessment, we should be guided by scientific principles. As practiced in the twentieth century, science has been a mixture of a dialectic and a consensus-building approach. Too great an emphasis on one approach or the other would be destructive to the acquisition of scientific knowledge. Overzealous pursuit of consensus for political or other reasons would overlook the fact that science is not always predictable by consensus. Indeed, an individual may discover a scientific truth, even though he or she is outside the mainstream of scientific thought on any particular issue.¹

In terms of flexibility, I think there is a major conflict between science and policy. It is in the nature of science to protect its flexibility zealously in order to increase knowledge and achieve change. Yet, there is an overwhelming impetus for risk assessment to develop numbers and to build defenses around those numbers in order to withstand legal and other challenges.

The recent proposal to develop uniform inference guidelines sends a shudder up the scientist's back. It sounds too much like an attempt to establish science by fiat.

We have already heard about some of the difficulties inherent in improving the risk assessment process. As Joel Hirschhorn mentioned, EPA is developing guidelines for looking at other end points. The guidelines, however, are still preliminary and do not have the methodological scientific basis that we have seen with respect to cancer studies. Of course, it is the large base of existing molecular biological dogma, as well as manipulative mathematical techniques that drive much regulation towards management of carcinogens rather than other substances or health effects.

Dr. Nelson touched upon the problems involved in coordinating epidemiology and toxicology to deal with risk assessment. I think it is pretty clear that regulatory agencies do not make much use of epidemiology, and those that do, use it poorly.

There are, of course, great differences between epidemiology and toxicology, and these are not always recognized or acknowledged. There are, for instance, major problems in using epidemiology to determine individual versus group risks. Judge Jenkins addresses this point in his important decision, Allen v. U.S.²

There are uses of epidemiology that are totally neglected in the regulatory process. For instance, epidemiology could be used to test whether regulatory theory or practice has indeed worked. To date, there has been only one example of this use: the study of the effects of lowered levels of lead in gasoline on the American public. That study provided an extraordinary [16 ELR 10206] vindication of the decision to regulate leaded gasoline. This use of epidemiology can validate risk assessment and allow for appropriate use of cost-benefit analysis and cost-effectiveness criteria in dealing with risk assessment and risk management issues. Epidemiology can also be used to identify those issues that are important to managing risks.

In conclusion, I believe that there are ways to improve risk assessment methods that have not yet been attempted. To use the current flaws as a reason for rejecting the entire process is no longer a valid option. We do not have that luxury, because risk assessment is fundamentally a method of expressing disparate data in consistent, objective, and uniform terms that various parties can then use to understand how different decisions are reached.

DR. DAN BEARDSLEY: As the EPA representative here, I am put on the defensive, and I am now going to change that. Professor McGarity's paper, particularly the first twenty pages, constitutes a splendid discussion of what occurred at EPA and why. My only quarrel is with the answers he proposes.

The notion of rebuilding trust has three main aspects. One is the problem of who is responsible for science policy. As Professor McGarity mentioned, the legislature ultimately has to make that policy choice. But, as the Professor also stated, the legislature is not a good risk manager. I see no reason why Congress should do any better at deciding science policy issues than it does at deciding risk management issues.

The second aspect of rebuilding trust is that of inference guidelines. Inference guidelines are connected with the idea of an organic statute, which has been discussed at EPA ever since Earth Day. Anyone who is familiar with the history of EPA may know that William Ruckelshaus planned a third phase of reorganization that would have been more functional and that, presumably, would have built a better base for an organic statute across the government. That plan, however, is moot, and I do not know anyone who seriously believes that it could ever be implemented.

Inference guidelines do seem important. One must, however, uncouple the idea of inference guidelines from that of the uniform statute. Inference guidelines are attractive for a reason that Professor McGarity did not mention: they would help to cure the inconsistency with which different federal regulatory agencies make decisions about the same substance. As long as the Occupational Safety and Health Administration, the Consumer Product Safety Commission, the Department of Housing and Urban Development, and EPA come to different judgments about the risks associated with formaldehyde, for example, the public receives a picture of federal regulatory agencies as wildly inconsistent.

The Interagency Regulatory Council would seem a reasonable body to draft the inference guidelines. With a sufficient period of time set aside for public comment and review, as suggested by Professor McGarity, the agencies are the safest bet to manage risk. While the Office of Science and Technology Policy (OSTP) has done good work on cancer policy, the OSTP operates out of the White House and must be careful about policies. Regulatory agencies do not need to worry about politics quite as much.

The third aspect of rebuilding trust involves the use of risk assessment. I wonder whether Professor McGarity has perhaps misconstrued the real political policy question in Washington. There is a difference between a lack of trust in agencies as risk managers and a lack of trust in risk assessment and science as decisionmaking tools.

Thus far, participants in this conference have assumed that the use of risk assessment as a basis for decisionmaking enjoys a consensus in Washington. For the past year, I have been involved in a six-month study of air toxins, and let me assure you that no such consensus exists.

PROFESSOR THOMAS McGARITY: In theory I believe that an organic statute is a a good idea; in practice, I do not believe that there is any chance of getting one enacted.

I did not mean to avoid the issue of whether risk assessment should be used in decisionmaking. I have some qualms about that question and about economic analysis, as well. I have spent the last two years working on a project for the Administrative Conference on Regulatory Analysis. I began that project very skeptical, knowing how easy it is to manipulate risk assessment, and expecting economic analysis to be no different. I emerged from that project more optimistic, but I still had qualms about the importance that the public attaches to the risk assessment number.

PETER B. HUTT: Is there any realistic alternative, either now or in the near future, to quantifying the carcinogenic risk through the use of mathematical models?

BEARDSLEY: Along with Dr. Anderson, I am one of the chief advocates of quantitative risk assessment at EPA. As a matter of public policy, the issue of public trust is not caught up in the subtleties of quantitative risk assessment. The question is whether the public and the Congress have any faith in the use of risk assessment itself.

SILBERGELD: I endorse risk assessment, primarily because I see it as a Trojan Horse that brings within it the awareness that there is an alternative to the Delaney approach — an approach that has not proven useful in environmental regulation. Through risk assessment, scientists are able to persuade lawyers, legislators, and administrators that they must pay attention to toxicological data and act in the absence of human data. Without that discipline that enables one to convert the results of toxicological studies into a number that makes some sense of the likely risk to humans, one is thrown back into the hodge-podge of epidemiology. There is then a risk that one may fail to act at all because of a lack of data.

That is my fundamental reason for endorsing risk assessment. I think that if we do not endorse it, it will persist in a subterranean way, as it does at the Food and Drug Administration (FDA), despite the Delaney Amendment.

HIRSCHHORN: The Office of Technology Assessment notes the need to limit the use of risk assessment. Risk assessment is so time-consuming and expensive to do properly, that we must delineate situations where it will not be necessary. In the Superfund program, for example, there are many instances where one may conclude that risk assessment is unnecessary, because cost-benefit analysis and consideration of alternatives to cleanup have already led to a decision not to do a cleanup.

HUTT: I would like Dr. Silbergeld to respond to Dr. Beardsley's defense of inference guidelines on the ground of consistency. Dr. Silbergeld suggested that the guidlines might have a stultifying effect on science. Over the years, I have encouraged FDA to adopt inference guidelines for precisely the reasons suggested by Dr. Beardsley.

SILBERGELD: Emerson's observation that consistency is the hobgoblin of little minds still holds true, and nowhere more so than in science. As it stands now, risk assessment numbers take on a force of their own that are difficult to contradict. This would even be more true of inference guidelines.

I believe that the existence of certain methods for estimating [16 ELR 10207] cancer risk has not had a beneficial effect upon the molecular biology of chemical carcinogenesis. To extend that approach into other areas of risk assessment — by which I mean other kinds of health effects — would be quite dangerous. I think EPA, in its proposals for risk assessment methodologies for other end-points, such as developmental toxicology and mutagenesis, has preserved a more wary attitude about the specific stipulation of methods. To enlarge this approach to the level of uniform inference guidelines enacted by Congress would do further damage to the kind of scientific flexibility that we need.

HUTT: Are you more concerned about the use of inference guidelines in risk assessment or in risk management?

SILBERGELD: I am concerned about both areas, because I share Professor McGarity's difficulty in understanding where one begins and the other ends.

HUTT: Would you be opposed to a system that adopted one mathematical model and set of inferences that must be used by everybody but that also allowed an alternative series of assessments using other models? This system would permit comparison from substance to substance, because there would always be one single methodology used, along with others.

SILBERGELD: It would be preferable to recommend use of a range of a models and to make it quite explicit where the models came from, as well as the results of each model.

HUTT: That recommendation raises the possibility that one would never be able to perform comparisons, because if one were to compare ten different substances, one might be relying on ten different models.

SILBERGELD: That is unlikely to happen. Usually, people select an overlapping group of three or four models, as in the area of carcinogenicity.

PARTICIPANT: I would like to suggest another use for risk assessment. Perhaps we could perform an assessment of what will happen if one does nothing at all, as opposed to what will happen if one takes a particular action. This process is somewhat analogous to what physicians tell their patients when prescribing medicine. The doctor will usually tell the patient what progress the disease will take without the medicine, as well as what benefits and side effect to expect from the medicine itself.

McGARITY: I think that is a useful suggestion. It certainly facilitates comparison and does not require the valuation of lives and other activities that generate bad feeling among the public.

SILBERGELD: That raises the issue of the value of the thing one is looking at to begin with. It is the problem embodied in the FDA's "generally recognized as safe" list and EPA's Industrial Biotest Laboratories list of shame, both of which include compounds that have their own infrastructure, economic and otherwise, built up around them. There are also new compounds seeking to enter the marketplace. How does one deal with the different structures associated with these compounds?

HUTT: It does not make any sense to use the same kind of risk analysis to discuss industrial wastes and food products. FDA knows that the biggest risks come from natural foods, such as mustard and peanuts, which contain natural toxins and carcinogenic substances. To perform a risk-benefit analysis on peanuts containing aflatoxin has never seemed a particularly useful activity.

Before banning sassafras tea because its principal flavoring constituent is a powerful carcinogen, I performed my own personal version of a benefit-risk analysis. I considered that, because I never drank sassafras tea, it could not be all that important. I then heard from 20,000 people who drank nothing but sassafras tea. To them, it was extremely important. They asked the FDA to label the tea with a warning similar to that provided on saccharin-containing products. Remember that FDA had planned to ban saccharin, but because of public outcry, Congress intervened and ordered FDA to label it instead, leaving it up to the consumer to decide whether to use the products.

One cannot, of course, use the labeling approach to deal with industrial wastes.

SILBERGELD: Frankly, I believe that labeling is of limited value, and not just because of studies that dispute the efficacy with which any particular message gets through to the public, or because of what happens to the warning after various competing interests have fought over its wording. In terms of their mildness, cigarette warnings are a good case in point. I would also like to state that we do not use the labeling approach when it comes to counterfeit drugs and unsafe food. We did not, for example, allow the Jewel Tea Company to continue selling milk labeled, "Warning: this product may contain salmonella."

In the pesticides program, it has long been EPA's practice to allow pesticides to remain on the market long after they have been found to have teratogenic properties. Instead of removing or changing the registration of the product, a label is added. The effect of that label is quite clear. In the agricultural work force, women of childbearing age have been excluded from jobs that involve handling those products. When we consider labeling as a solution in the area of environmental products, there is thus a broader equity issue that must be confronted.

HIRSCHHORN: At one time, we did suggest that consumer products be labeled to indicate the level of toxic waste created by production of the product. Although the idea was technically feasible, no one took us seriously. Currently, there is a debate in Congress over a waste end tax that would specifically place an economic burden on some manufacturers and, therefore, on some products. A counter-position urges that we spread the burden of the Superfund so we do not differentiate among companies and their products.

McGARITY: Over the past four or five years, I have worked with various environmental groups to advocate statutory changes to the Federal Insecticide, Fungicide, and Rodenticide Act. At the very first meeting, one person suggested that we could solve the entire problem by requiring grocery stores to indicate above various kinds of produce whether the food was contaminated by a pesticide, whether the safety studies on the pesticide were fraudulent, and whether the pesticide had been determined to be carcinogenic. Even among a fairly radical group, that suggestion was considered unrealistic.

BEARDSLEY: I agree that labeling is not terribly practical. I also do not think that it is equitable. Some people are concerned enough to read labels; others are not.

HUTT: What do we do about mustard, for example? It is now clear that the principal constituent of mustard is a carcinogen. Do we just withhold this information from the public? Do we label the product, or do we ban it?

McGARITY: That is definitely an issue in the realm of individual risk management. Individuals could use some help from professionals with respect to assessment, for instance, from an article in Consumer Reports and then make an [16 ELR 10208] informed choice about whether to eat mustard. Even assuming the risk was extremely high, I am not sure that I would ban the product.

HUTT: The reason FDA did not ban mustard or peanuts is because of the damage to the agency's credibility that would result from such an action.

SILBERGELD: There is a difference between natural substances that we have always had with us and those that we have introduced relatively recently. We are probably right to apply a double standard. In the past, FDA has banned natural components of food. The only action that is more controversial than regulating environmental chemicals — that is, synthetic chemicals — is trying to advise the American public on diet.

As Norton Nelson reminded us, we exist against a background of hazards, many of which we did not create. It certainly behooves us to act with a special vigilance with respect to those background risks that we can expect realistically to control and also with respect to those risks that we can somehow modulate at the point of industrial conception.

PARTICIPANT: We can all agree that involuntary risk is immensely large and difficult to manage. Perhaps we should stop worrying about voluntary risk and concentrate solely on involuntary risk. I also think that the agencies should focus first on those risks that pose the greatest threat to the individual.

McGARITY: It is when we come to voluntary risk that we most need the professional risk assessor to do whatever technical work needs to be done, and also, more importantly, to educate us about the risks so that we can decide how to manage them. I also disagree with the suggestion that voluntary risks are not a major source of the risks we face. Driving an automobile is still a fairly risky business.

1. Ruckelshaus, Risk, Science and Democracy, ISSUES IN SCIENCE AND TECHNOLOGY, Spring, 1985.

2. Supra, p. 1.