Beyond Democratization Of Risk Assessment:
An Alternative To Risk Assessment
Mary H. O'Brien
On December 9, 2001, a small piece on the precautionary principle appeared in the New York Times Magazine. In the first paragraph, author Michael Pollan (2001) wrote,
For the last several decades, American society has been guided by the ''risk
analysis'' model, which assesses new technologies by trying to calculate the
mathematical likelihood that they will harm the public. There are other
ways, however, to think about this problem. Indeed, a rival idea from
Europe, the ''precautionary principle,'' has just begun making inroads in
Pollan is here noting a contrast between risk analysis and the precautionary principle. On its face, however, the phrase "risk analysis" or "risk assessment" would seem to imply precaution. That is, isn't it cautious to consider various human health risks before issuing a permit to emit toxics into the air, or to consider the risks to a fishery before establishing a particular harvest level? But instead, risk assessment is cited in opposition to precaution. Why?
Let me cite another declaration of opposition of risk analysis to the precautionary principle. Last month, on November 16, 2001 Secretary of Agriculture Ann Veneman spoke to the National Association of Farm Broadcasters, two days after returning from the Qatar trade meetings. In her speech, she points out all the successes the U.S. team had at those meetings. One was to shut the precautionary principle out.
Veneman (2001) said:
We also were successful in precluding the Europeans from including what's called the precautionary principle, so to speak. Now that principle they wanted in there to be able to use, we believe, it could have undermined the whole sanitary and phytosanitary agreement, allowing countries to say, well, in the interest of precaution we're going to take this action, undermining the sound science that's contained in the sanitary and phytosanitary agreement, and that we could not live with, and that was not allowed to go into the text.
In Veneman's talk, the phrase "sound science" is a code word for "risk analysis," because, as you know, the free trade sanitary and phytosanitary agreements require that a country wishing to bar imports of a hazardous product, must first show by risk assessment that a certain level of harm will occur and cannot be mitigated. For instance, the U.S. took the European Union to the World Trade Organization court, and won a ruling that the European Union had not proved that harm would occur from importing and eating artificial hormone-treated beef from the U.S. (Wallach and Sforza 1999).
The three WTO trade lawyers who debate in secret for their rulings are the judge of when sufficient risk of harm has been demonstrated. For instance, Australia wanted to bar import of raw salmon from Canada because some 20 bacteria not present in Australia are present in Canadian and U.S. salmon (Wallach and Sforza 1999) They demonstrated that Canada had not developed a treatment to eliminate the bacteria. They also demonstrated that food prepared for human consumption has already in a number of cases ended up in the animal food supply and in this manner the Canadian salmon bacteria could infect Australian salmon.
But the WTO court ruled that Australia could not say "no" to imports of raw salmon because they had not demonstrated the precise likelihood of salmon disease entry and transmission. They wrote, "It is not sufficient that a risk assessment conclude there is a possibility of entry, establishment or spread. A proper risk assessment must evaluate the likelihood, i.e., the probability of entry, establishment and spread" (WTO 1998). It thus ruled against Australia. Meanwhile, Canada does not have to do anything to ensure that its salmon exports are free of bacteria that are known to afflict North American salmon.
In other words, under the WTO, precaution is opposed and trumped by risk analysis. It is clearly important to examine why precaution and risk analysis are viewed in opposition to each other. To many of us who have worked for the past two decades in grassroots advocacy for environmental protection and environmental health, the reason why is clear: The vast majority of risk assessments that we encounter, whether in a local school district, or at the federal level, or in international trade, are not prepared in order to consider risks; they are prepared to estimate the level of a hazardous substance or activity that is safe, or of insignificant harm, or at least of "acceptable" risk. In other words, they are prepared in such a way as to obviate the need for precaution whenever information is less definitive than a dead body count.
Twenty-one years ago, I believed that the preparation of risk assessments was a good idea, "sound science," as it were. I believed that all that was necessary to correct their frequent downplaying of risk was insertion of more accurate numbers. But I soon became disavowed of this belief.
For instance, I became involved in a suit against the federal Animal Plant Health Inspection Service for their nationwide environmental impact statement (EIS) for aerial insecticide spraying for gypsy moth. The EIS included risk assessments for four insecticides, and for a deposition, I analyzed the risk calculation for the carbamate insecticide, carbaryl. There were about ten elements to the risk formula, and I found that each one was an estimate, based on some kind of guess or literature. But as I looked at the scientific literature, I found reason for more conservative numbers for each of the ten elements. For instance, the estimate had been made that five percent of carbaryl would be absorbed through exposed human skin. I found a study using human volunteers (and I do not approve of such studies) by Howard Maibach of UCLA in which he had documented over 70% absorption of carbaryl through human skin. So I re-analyzed the entire risk assessment for carbaryl, showing that with more valid numbers in a number of the model's elements, the risk of harm could be orders of magnitude higher. The judge dismissed this claim of our case, saying I was "nitpicking." (We did win the case on other grounds, however.)
Around this same time, in the mid-1980s, the herbicide dacthal, was found to be contaminating many groundwater wells in an eastern Oregon, onion-growing area where the water table is 9-11 feet below the sandy soil surface. Dacthal is always contaminated with 2,3,7,8-TCDD. Dacthal was found in these wells in amounts up to 431 ug/L. The only existing "standard" for dacthal was a draft EPA 1982 "health effects guidance" of 500 ug/L based on a 1963, wholly inadequate, two-year rat study (USEPA 1982). In this experiment, a lung infection had been present in control and exposed rats, and so the rats had been dosed with antibiotics throughout the experiment. Both experimental and control rats had died at a high rate during the experiment, and some rats had been sacrificed midway through the study. The numbers of animals from which the experimenters could draw conclusions was inadequate. However, the study had found effects at all doses and a "variety of tumors" in the exposed rats.
The standard of 500 ug/L, which used a safety factor of 1,000, had been calculated as if the lowest dose had not caused any adverse effects (which was not true), the tumors had not been observed (which was not true), and a 20-kilogram child were drinking the water. And even so, the drinking water in the Ontario area of eastern Oregon was approaching, at 430 ug/L, this 500 ug/L standard.
A year after dacthal had been found at 431 ug/L in the eastern Oregon wells, the EPA issued a new "health advisory" for dacthal in drinking water that set a standard of 3500 ug/L (USEPA 1987).
Why did dacthal suddenly look seven times safer? Had a new study emerged? No, the EPA was still using the same 1963 rat study, but now the formula assumed that only 70-kg adults, not children, would drink the dacthal-contaminated water, and it applied a 100-fold safety factor rather than the 1000-fold safety factor. The 1987 dacthal standard still treated the risk of cancer as zero, even though tumors had been found in the treated animals. Neither standard mentioned that dacthal is contaminated with the carcinogen, hexachlorobenzene (Wapensky 1969, Cabral et al. 1977). Additionally, the 1987 health advisory did not mention that dacthal had been found in 1985 to be contaminated with 0.16 ppb 2,3,7,8-TCDD (USEPA 1985a).
At this point, I realized that life was too short to spend it trying to assist with making better risk assessments when clearly, the goal of most risk assessments, was not to think about harm that could occur, but to instead establish some standard for how much of the hazardous substance or activity would be considered "safe" or "acceptable."
At a similar point in time, I was working as staff scientist for Northwest Coalition for Alternatives to Pesticides (NCAP), whose coalition members had five times obtained court rulings against forest herbicide EISs of the Pacific Northwest Region of the US Forest Service and Oregon Bureau of Land Management (O'Brien 1990). These risk assessments were claiming that aerial spraying the phenoxyacetic herbicides, 2,4,5-T and 2,4-D, over national forests and streams, and in the vicinity of rural residents, would be safe. We had won the cases on the basis of a National Environmental Policy Act (NEPA) regulation that has since been gutted, as a result of our court cases. The regulation was Sec. 1502.22 of the NEPA regulations, and it required preparation of a worst-case analysis. A worst case analysis clearly requires considering harm that could happen. In their risk assessments, however, the Forest Service and BLM had repeatedly denied that cancer could be caused by exposure to 2,4,5-T or 2,4-D, and the courts had repeatedly said, there is evidence for both of these herbicides that, in a worst case analysis, they could cause cancer, and so the EIS is illegal. (It is a telling comment on the goal of risk analysis that the federal government subsequently gutted the worst case analysis provision of NEPA; in other words, the one provision that didn't allow for best-case analysis calculations of safety.)
By the time of their fifth decision, the courts had lost patience with the Forest Service, and shut down all of their herbicide spraying, for even experimental purposes, throughout Washington and Oregon (NCAP v. Lyng 1987). So the Forest Service had to develop a new EIS if they were going to regain use of any herbicides. They came to our organization, Northwest Coalition for Alternatives to Pesticides, and asked, "What do we have to do to not see you in court again?"
We said they needed to do two things. First, stop using Oregon State University pesticide advocates as their risk analysts; and second, develop a credible, least-herbicide use alternative to be considered alongside other alternatives in the EIS. As you are probably aware, section 1502.13 of NEPA requires that in an EIS for any federal action that may cause significant social or environmental harm, "all reasonable alternatives" must be considered and analyzed for their positive and negative impacts.
The regulation writes that this consideration of alternatives is "the heart of the environmental impact statement....It should present the environmental impacts of the proposal and the alternatives in comparative form, thus sharply defining the issues and providing a clear basis for choice among options by the decision maker and the public."
At any rate, the Forest Service agreed to consider our input as they developed a least-herbicide alternative. For our input throughout the next two years, we relied heavily on the experiences of those forest managers in the Northwest who had been managing both national and private forests well without herbicides. Ultimately, the Forest Service selected an alternative that was very much like the least herbicide-use alternative, and I realized this: The assessment of the benefits and drawbacks of a full range of alternatives, not assessment of the acceptable level of a hazardous activity, is not only the heart of an environmental impact statement, it is the heart of wise decisionmaking in a democracy.
In the subsequent 15 years, I have never been disavowed of this belief. Two days ago, on December 10, 2001, I was once again sitting in NCAP's office, with the newly-appointed team leader for developing the Pacific Northwest Forest Service Region's upcoming EIS on invasive species management. It is a different experience this time: The main issue we discussed is how the Forest Service can best develop, in conjunction with the public, an alternative based on landscape-scale goals, least pesticide use, and prevention of invasive species as well as control or treatment of invasive species. Prevention necessarily means reining in livestock grazing, off-road vehicles, roads, and forest thinning, which play major roles in introducing, establishing, spreading, and reintroducing invasive species. Once again, this alternative will be only one among a number that will be considered in the EIS, but the hardest part of all is to get an agency, company, community, nation, or international body to seriously consider reasonable alternatives to hazardous activities Once reasonable alternatives are seriously considered, health and environmental protection is much more possible..
Does risk of harm need to be analyzed when assessing alternatives? Of course, but so-called "safe" amounts of a given hazardous activity don't have to be ginned up. The various potential harms of each alternative are considered in a context of comparative harms (and benefits). If the process is broadly participatory, as it should be, given that potential harms tend to be broadly distributed, a diversity of perspectives will provide for a range of types of harms to be considered. Relevant harms, of course, are not only the biological harms generally considered by risk assessments, but also economic, and social harms posed by the various alternatives. Equally as relevant are biological, economic, and social benefits posed by each of the alternatives. Within alternatives assessment, the meaning of risk is thus broadened, at the same time that its monolithic power is reduced. And finally, in the context of alternatives assessment, the consideration of risks, of potential harm, becomes compatible with the precautionary principle.
Secretary of Agriculture Ann Veneman in her November talk to the farm broadcasters, presents risk analysis (what she labels "sound science") as opposed to the precautionary principle, because she wishes to forbid the European Union from making decisions based on alternatives assessment, for instance, considering whether hormone-laced beef is necessary or desirable. She wants to limit risk analysis to ginning up no-proven-excessive-death-or-extinction levels. She wants to exclude consideration of the potential social death of a nation's local agriculture and self-sufficiency, or consideration of the potential extinction of species through collateral spread of genetically modified organisms. Only by pretending hazardous products are safe, can Veneman and the WTO justify forcing countries to import hazardous products. And only by risk analyses can Veneman and the WTO pretend they are safe.
As I noted in the stories above of dacthal, carbaryl, 2,4-D and 2,4,5-T, this use of risk analysis to claim safety is not new to environmental and health activists.. This has been the attraction of risk analysis from the beginning for the William Ruckelshauses, Ann Venemans, John Grahams, Grace Corporations and Monsantos of the world. It is why so many citizens oppose risk analysis. They do not oppose consideration of risks, they oppose risk analysis. They do not oppose sound science, they oppose risk analysis. They want us all to take more care for the world and each other.
Biologists speak of insects or fish or mammals developing a "search image" for nectar sources or prey. With a search image, pollinating insects will be able to hone in on their most productive nectar species in a meadow full of many species; a lynx will see its favored prey, the snowshoe rabbit, among heavy brush. I have a friend who has a search image for arrowheads and other small hand-worked artifacts; he sees them even when they are three-quarters buried in sand in the Southwest, after I have passed over them all, unseeing Over the past twenty years, I have developed a "search image" for processes that are unnecessarily risk-dominated.
This has made me uncomfortable at times even with formulations of the precautionary principle. And I finally realized, a few months ago, that many formulations of the precautionary principle have incorporated obeisance to risk-based burdens of proof. Let me show you a scheme that would seem to diagram implementation of the precautionary principle as many advocates articulate it (Figure 1).
In the usual harm-driven scheme, implementation of the precautionary principle begins, as does risk assessment, with a hazardous activity or substance in the driver's seat. Then, the first question to be asked is, "Is this activity or substance harmful enough to trigger implementation of the precautionary principle for possible restrictions, banning, or alterations?" Now this is a different question than is asked by nearly all risk assessments, which is, "How much of this activity or substance is safe enough to be permitted, used, exported, etc.?"
But this first step in this harm-driven scheme, making a decision whether to implement the precautionary principle, is essentially a risk assessment threshold, as it requires coming up with a sufficiency of suggestive evidence that points toward sufficiently significant harm to warrant precautionary action. It is here that the John Grahams of the world try to wrench the precautionary principle back into risk assessment, and reduce the principle to a mere nod toward being more conservative in risk assessments which the John Grahams of the world have already claimed are extremely conservative and health-protective. (As you know, John Graham is the former Director of the Harvard Center for Risk Analysis and current Director of the federal Office of Information and Regulatory Affairs in the Office of Management and Budget). Critics of the precautionary principle can point out that this first step is infinitely vague. As Tom Webster of Boston University says (and Webster is NOT one to justify unjustifiable harm), "If the precautionary principle were the guiding principle, we would have said no to agriculture." In other words, almost every activity or substance results in some kind of harm. So how are we not to be paralyzed?
Activists rightfully respond that implementation of the precautionary principle does not necessarily lead to "No," but can lead to any number of outcomes, including, if there are no feasible alternatives, an unqualified "Yes."
But sometimes the harm-driven scheme for implementing the precautionary principle will, like risk analysis, focus primarily on the harmful activity or substance, as when the precautionary principle is invoked for phasing out a substance or activity, such as clearcutting. The consideration of alternatives will then be relegated to the sidelines, as in what to do after the phaseout.
The harm-driven process for implementing the precautionary principle is generally a struggle between defense of particular commercial or government activities (such as barging salmon around Snake River dams) on the one hand, and, on the other hand, public pressure to avoid damage (for instance, extinction of the remaining Columbia River salmon runs). The harm-driven process thus tends to ignore the question of whether the activity is even necessary to accomplish any positive community or social goal.
However, as Michael Pollan points out in the third paragraph of his New York Times Magazine piece, the term "precautionary principle" is only one translation of the German word, Vorsorgeprinzip, which had began to be implemented in the 1970s in Germany in response to their forests dying. The root word, Vorsorge, has the connotation of caring into the future, or "fore-caring." Thus, the precautionary principle could just as accurately have been translated as the forecaring principle.
Accordingly, a second concept of implementing the precautionary principle perhaps further lessens bondage to risk assessment thinking (Figure 2). A goal-driven process, a forecaring process, places some public health and environmental goals, rather than harmful activities in the driver's seat, and alternatives for reaching the goal are developed and considered. A goal might be the virtual elimination of childhood asthma, or no toxics in breast milk, or, as Sweden is aiming for, zero annual highway mortality.
In the goal-driven process, there is no all-or-nothing threshold decision to invoke or not invoke the precautionary principle. Instead, the precautionary principle is a screen, as it were, through which all decisions pass: Deciding what goals to adopt, which alternatives to consider for reaching the goal or goals, which alternatives are preferable, when to correct course after a given alternative has been implemented and there are unintended harms. There is no way to avoid all harm at all points; the question is always one of considering alternatives.
Given that the alternatives are being proposed for achieving some positive goal, then questions will be more likely raised about the "necessary-ness" of activities that retard accomplishment of the goal, or that are likely to cause harm. And, since most social goals cannot be achieved without changes in habits of multiple sectors of society, the goal-driven alternatives are more likely to be examined for their comparative "fairness" in burdens for change.
The science involved in analysis of potential harms and benefits of alternatives within the goal-driven process is more likely to include social sciences in addition to biological sciences, for instance in considering how consumers will accept a particular change, or how the work force will be affected under two different alternatives. The science is necessarily of a more complex type, as well, because a combination of activities are generally involved in accomplishing positive goals such as near-elimination of childhood asthma.
As I was traveling here from Oregon yesterday, I had to shut down my laptop as we were descending into Boston. So I picked up a document I had been meaning to read. It was an October 3, 2001 memorandum to the Dale Bosworth, Chief of the U.S. Forest Service, from Jim Furnish (2001), as he was departing from his position as Deputy Chief of the National Forest System. He had become Deputy Chief during the last administration and does not feel that he can be of help to the current administration's intentions for the Forest Service. In this memorandum, he argued, in effect, for a goals-driven process for the Forest Service:
I have been deeply troubled that most agency officials, in my experience, are averse to principled, in-depth, and progressive discussions about values — not ours, but those of the public we serve. Research points to a few simple and understandable values that are broadly supported today for management of public lands — these are naturalness, clean water, abundant wildlife, personal renewal (re-creation), and maintaining choices for future generations. [Note: Naturalness, clean water, abundant wildlife, personal renewal, and maintaining choices for future generations could equally be called goals.] These are not the values that provided the framework for Forest Service programs in the past. Nor have they been unequivocally embraced so as to define our agenda for the future. I believe that values are a fundamental foundation and premise for management. We could benefit greatly from an explicit articulation of the values that will drive our policies, as well as a commitment to the principle that we will pursue actions that will achieve outcomes consistent with those values." [Emphases in original.]
I was struck with the conjunction of writing on my laptop about a goals-driven process and then picking up a memorandum I had not earlier read, that urges a goals-driven process. It is extremely important to note that Furnish was urging the Forest Service not to aim for their own values or goals as a bureaucracy - but to aim for the values and goals that are those of the public they serve. A goals-driven process for implementing the precautionary principle depends on broad public participation from the beginning, including setting the goals.
To bring this all back to risk analysis, it is interesting to note where risk analysis fits within this goal-driven process of Figure 2. It is in the third step, after selecting a positive goal, after laying reasonable alternatives on the table. The third step involves analysis of the alternatives. Analysis of potential harms of each of the alternatives (i.e., "risk analysis"), is just one among other analyses, including analyses of opportunities offered by each of the alternatives; barriers to implementation of any of the alternatives; resources that could be mobilized for implementation of the alternatives; and what we don't know or aren't sure of regarding the alternatives. All of these analyses should employ sound science.
So, this is risk analysis dethroned. This is risk analysis playing one essential role among many others to protect and restore environmental and social health. This is risk analysis in the service of good.
Presented to the Society for Risk Analysis, New England Chapter. Boston, MA, December 12, 2001.
Figure 1. The Precautionary Principle as a Harm-Driven Process