Environmental Protection Agency

The Environmental Protection Agency (EPA) was created by executive order by President Nixon in 1970. The EPA was set up as an independent Federal agency to be the administrative home for a number Federal environmental programs that had previously been scattered over the Executive Branch. The programs out of which the EPA was cobbled had their own legislative authorities and histories. Because the consolidation was by executive order (and not through a new environmental act specifying a melding and recasting of these programs), the various components of the new EPA retained their different legislative mandates, regulatory powers, and scopes. Many of the laws were amended during the early years of the EPA, tailoring their treatment of issues of particular concern. In addition, new laws were added to bring additional environmental problems into the ambit of the Federal environmental effort.

The result is that, even twenty-five years later, the EPA represents a collection of environmental programs that has only partly been consolidated and centralized. Risk analysis is used in support of regulation and rulemaking under a half-dozen major environmental laws and a number of minor ones. Although the role of risk assessment, particularly quantitative risk assessment, has grown largely since EPA's founding, the separation of regulatory programs has had an effect on risk assessment practices in various parts of the agency. The history of risk assessment at EPA has been marked by ongoing issues of consistency versus case-specificity of risk assessment methods and analyses, and consolidation versus dispersion of the conduct of risk assessment.

The dispersion of risk assessment activity over parts of the EPA makes the issue of coordination and maintenance of consistency particularly important to this agency. There are several means in place toward this end. They include the publication of a series of risk assessment guidelines, development of methodology documents, the chartering of several cross-agency groups to coordinate and harmonize practices and to resolve methodologic and policy questions that may arise, the reliance for advice and scientific guidance on external experts through the EPA Science Advisory Board, and the maintenance of a computerized, publicly available data base of agency-wide consensus on risk assessments.

The risk assessment methods employed by the Environmental Protection Agency have much in common with those used elsewhere, reflecting the general practices, standards, and precepts of the field. Risk assessment is a practical field, and the principles that have evolved reflect the concerns and ends of practitioners, including regulatory agencies and public health institutions, both national and international. The EPA has been an influential player in this development because of its major role in environmental regulation, the growing role of risk assessment in that regulation, and because the agency has made special efforts to define and develop the underpinnings of its methods through the promulgation of risk assessment guidelines and promotion of scientific discussions about risk assessment methods.

EPA's risk management practices are guided primarily by President Clinton's executive orders (Executive Order 12866 on Regulatory Planning and Review and Executive Order 12875 on Enhancing Intergovernmental Partnership). These revoke and replace executive orders from President Reagan, but include many provisions on similar matters. The EO 12866 directs agencies to conduct cost-benefit analysis for all "significant regulatory actions" and to promulgate regulations only when necessary due to "compelling public need." Regulatory approaches should be chosen to maximize net benefits, minimize the overall regulatory burden on society, and to be the most cost-effective means of achieving the desired end.

• Office of Pesticide Programs

The regulation of pesticides is carried out by EPA's Office of Pesticide Programs (OPP), which is a part of the Office of Prevention, Pesticides, and Toxics (OPPT). Pesticides are different than other potentially toxic compounds in that they are intended to be poisonous, at least to the pests they are designed to control, and they are intentionally introduced into the environment for that purpose. This situation naturally calls for the consideration of both costs and benefits, and the statutes under which pesticides are regulated provide for such analysis.

Pesticide regulation falls into two parts, and each part is accomplished under a different statute. The registration of pesticides (i.e., licensing for sale and use in agriculture or extermination) is carried out under the Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA). No chemical may be sold in the U.S. as a pesticide without such registration, which establishes the conditions of legal use. The question of tolerances for pesticide residues on foods as encountered by the consumer is regulated under the FFDCA.

FIFRA (7 U.S.C.A. §§136 to 136y) provides for the regulation of sale and distribution of pesticides, where pesticide is defined as "any substance or mixture...intended for preventing, destroying, repelling, or mitigating any pest, [or]...intended for use as a plant regulator, defoliant, or desiccant" [FIFRA §2(u)]. No pesticide may be introduced into commerce without obtaining a registration from the EPA. Registration is obtained through petition to the agency, with the petitioner providing information on the intended use, data on efficacy of the pesticide and its toxicological properties. The agency is empowered to ask for the provision of additional data, including the requirement for more toxicological testing, if the information is deemed necessary for the registration decision.

The Administrator may approve the petition if the pesticide "will perform its intended function" [§3(c)(5)(C)], and "when used in accordance with widespread and commonly recognized practice it will not generally cause unreasonable adverse effects on the environment" [§3(c)(5)(D)], which are defined as "any unreasonable risk to man or the environment, taking into account the economic, social, and environmental costs and benefits of ...use" [§2(bb)]. Pesticides are registered either for general use or for "restricted use" [§3(d)], with the latter category specifying conditions of use such as application methods, amounts used, target pests, geographic restrictions and so on.

Once granted, registrations expire after 5 years, at which time the petitioner can apply for renewal of registration [§6(a)]. There are provisions for EPA to cancel a registration early [§6(b)] if the Administrator finds adverse effects could indeed be caused, but a decision to cancel must take into account "the impact...on the agricultural economy." Much of the modern registration framework was introduced into FIFRA by 1972 amendments (the Federal Environmental Pesticide Control Act, 7 U.S.C.A. §136), and a large number of previously registered pesticides had been "grandfathered in" under the lax pre-1972 procedures. Further amendments in 1988 required re-registration (or cancellation) of these within 9 years, a large burden on the agency's risk assessment apparatus.

In sum, the registration process under FIFRA amounts to the granting of a license for sale and distribution of a potentially dangerous chemical. The license is not unlimited; it specifies the conditions of use that are permitted, potentially including restrictions on the target pests, the amounts of pesticide used, the application method, frequency, and timing of use, training of applicators, the time that must elapse after application before workers can reenter a treated field, and the time that must elapse after application before the crop can be harvested. Importantly, the registration also includes restrictions on which specific crops may be treated. Once registration is granted, however, all uses that fall within the specified restrictions become legal and permissible. That is, the regulatory power of registration is over permissible uses, not over actual practice within the permissible range.

To be granted a registration, the petitioner must demonstrate that the pesticide, when used on the proposed crops at the proposed levels, is effective at controlling pests and that, when used according to the restrictions, it will not cause unreasonable risk to humans or the environment. The definition of such adverse effects in FIFRA is very vague, but in practice it includes risk to the applicators and farmworkers, ecological risks, risks to homeowners from extermination procedures, and (through interaction with the tolerance setting process of the FFDCA, as discussed below) risks to consumers of treated foodstuffs. The mandate in FIFRA for balancing costs and benefits is similarly vague, comprising only the statement that "economic, social, and environmental costs and benefits" are part of the definition of what adverse effects are to be deemed "unreasonable." (The FFDCA is at least somewhat more specific on matters of both costs and benefits in regard to tolerances for residues on food.)

The FFDCA (21 U.S.C.A. §§321 to 394) provides for regulation of permissible contents of toxic substances in or on food, and pesticides are explicitly considered in its provisions. While primarily an FDA statute, the parts of the FFDCA applying to pesticides are administered by the EPA. The FFDCA is discussed in the section on FDA, but some key provisions are briefly reiterated here.

Tolerances are the concentrations (on a per weight basis) permitted to remain in or on food as it is available to the consumer. The process of setting tolerances is also by petition, with the petitioner submitting proposed tolerance levels along with toxicological information to demonstrate that such tolerances will be sufficiently protective. Tolerances of pesticides on raw, unprocessed agricultural commodities are regulated under FFDCA §408, which mandates that tolerances should be set "for pesticide chemicals which are not generally recognized, among experts qualified by scientific training and experience..., as safe for use, to the extent necessary to protect the public health." However, "appropriate consideration" must be given "to the necessity for the production of an adequate, wholesome, and economical food supply."

The processes of petitioning for registration and petitioning for tolerances are interconnected, and in practice they often occur concurrently. Although regulated under separate laws and following different procedures, the two processes have a practical linkage in that the conditions and limitations for use of the pesticide established during registration must clearly lead to residues experienced by the consumer that will be below tolerances that can be approved on health grounds. The approval of tolerances is based on exposure from the total diet, so each new approved use of a pesticide in the registration process leads to potential residues that "use up" part of the total allowable intake. Because each use of a pesticide must employ enough of the agent to be effective against pests, a registrant must carefully choose the particular crop and use restrictions for which registration is being sought to ensure that the sum of resulting residues will be below the level for which a tolerance can be approved on consumer health grounds.

Because registration is regulation of a prospective activity, much of the analysis of exposures, use levels, benefits, and costs must be based on professional judgment. In many cases, the rigorous analysis of costs and benefits, and the economic and agricultural effects of using various alternative pesticides and pest control practices, arises when a registration renewal is in question or when a cancellation of registration is being considered.

OPP considers three categories of exposure: to consumers, to those occupationally exposed (which in practice focuses on applicators, but also includes farmworkers generally), and the general public exposed via non-dietary means (i.e., through environmental contamination). As with most regulatory programs, there is no written rule or policy regarding the level of risk that must be deemed acceptable, but (also as with most agencies) there is understood unwritten practice that is revealed in the examination of regulatory decisions taken by the agency.

OPP generally tries to ensure that individual risks in all three categories do not exceed 10^-6 for lifetime exposure. Until recently, the goal for occupational exposures was somewhat higher, closer to 10^-4, but this was lowered to match the other categories during the tenure of Assistant Administrator Linda Fisher, and has remained so since. In the case of consumers, the 10^-6 risk applies to cumulative exposure to the pesticide from all dietary sources, with these estimates usually being based on conservative residue estimates but population average rates of consumption of food types. As noted earlier, it is difficult to determine when this combination is conservative, especially vis-à-vis the high end of levels of consumption of particular foods. For pesticide applicators, the exposure assumptions are not particularly conservative in terms of exposure per treatment, but there may be assumptions about maximum allowable use of the agent that are not met in reality.

These risk criteria are nominally for individual risk levels. However, the fact that consumer risks are calculated based on consumption levels averaged over the entire population makes these risk calculations apply to the whole population (at least on average, and bearing in mind the conservative residue assumptions). Thus, the criterion really hinges on a kind of population risk measure. High individual cancer risks that result because of high consumption of the affected food products is not captured because of the nature of the exposure analysis.

For non-cancer risks, many of the same considerations apply; high end individual exposures are not captured by the exposure assessment. However, differences in average exposure in each of 22 demographic subgroups are considered.

The consideration of costs and benefits is vaguely specified in the pesticides statutes, but registrations and tolerances are set bearing in mind the balancing of the risks engendered with the costs to agriculture and food prices. As registrants tailor their petitions for which crop treatments are to be approved, limitations on uses, and tolerances, they consider the economic and agricultural benefits to be gained by different combinations of uses that might be approvable. Those specific uses that are most efficacious and economically favorable to agriculture are more likely to be proposed by the registrant because they will lead to a better market for the pesticide once registered.

• Office of Pollution Prevention and Toxics

The EPA's Office of Pollution Prevention and Toxics (OPPT) is a relative newcomer among EPA regulatory programs, having been founded (under the original name of the Office of Toxic Substances) to implement the 1976 Toxic Substances Control Act (TSCA). In addition to its original role as implementer of TSCA, OPPT has been given responsibility for pollution prevention programs, regulation of certain abatement programs (such as that for asbestos), and the administration of the Toxics Release Inventory, mandated under amendments to the Superfund law. The focus of risk assessment in OPPT, however, is under TSCA.

TSCA (15 U.S.C.A. §§2601 to 2692) was conceived of as a "gap-filling" statute; Congress recognized that the existing array of environmental legislation covered risk posed by chemicals only under those particular exposure conditions each program was mandated to regulate. Moreover, this regulation was often in reaction to existing pollution, and its efficacy was hampered by lack of information on the chemicals in question. TSCA was passed in 1976 as an attempt to take a comprehensive approach to regulation of toxic substances, stressing properties of the chemical rather than of particular exposures to the chemical, and encouraging the development of information regarding toxic properties and exposures. The aim was to prevent risks from toxic substances that might "fall through the cracks" between other environmental statutes. This cross-cutting role has meant that throughout its history, there have been ongoing questions about TSCA's overlap with other environmental statutes.

The provisions of TSCA implement a set of policy statements set out at the beginning of the act [TSCA §2(b)]. First, "adequate data should be developed with respect to the effect of chemical substances and mixtures on health and the environment." Moreover "the development of such data should be the responsibility of those who manufacture and...process such chemical[s]." Second, the government should have adequate authority "to regulate chemical substances...which present an unreasonable risk of injury to health or the environment," including imminent hazards. Finally, exercise of this authority should "not...impede unduly or create unnecessary economic barriers to technological innovation while fulfilling the primary purpose...to assure that...such chemical substances...do not present an unreasonable risk." Section 2(c) goes on to require that "the Administrator [of EPA] shall consider the environmental, economic, and social impact of any action" taken under the act.

Section 4 of TSCA relates to testing and gathering of information on chemicals. It authorizes rulemaking requiring manufacturers to conduct toxicological testing for "carcinogenesis, mutagenesis, teratogenesis, behavioral disorders, cumulative or synergistic effects, and any other effects with may present an unreasonable risk of injury to health or the environment" [§4(b)(2)(A)]. The burden is on EPA to show that such testing is necessary, however. (This is unlike testing mandates under FIFRA or FFDCA, in which the agency can without rulemaking call for all information needed to grant or deny petitions.) The substance must present possibilities of unreasonable risk, "enter the environment in substantial quantities," or be likely to have "substantial human exposure" [§4(a)], all criteria that require the agency to do some preliminary risk assessment. An Interagency Testing Committee is established to set testing priorities. (Through this means, §4 is a vehicle for various Federal regulatory groups to obtain testing mandates, as long as their interests parallel those of EPA.) In practice, testing is done through enforceable negotiated consent agreements ever since a lawsuit challenged the earlier practice of negotiated voluntary testing [NRDC v. EPA, 595 F.Supp. 1255 S.D.N.Y.1984)].

TSCA makes a distinction between new and existing chemicals. The latter are those on a "list of each chemical substance which is manufactured or processed in the United States", which EPA is required to compile and maintain. Anyone proposing a new chemical (i.e., one not yet on the list), or to undertake a "significant new use" of an existing chemical, must give notice to EPA, along with test data and information bearing on its potential risk. EPA reviews the submission and permits the chemical's manufacture, suspends its manufacture or distribution, restricts its use pending the provision of further data, or initiates rulemaking to regulate its manufacture or distribution. Once a chemical enters commerce, it becomes an "existing" chemical.

In essence, the Toxic Substances Control Act aims at establishing a system of both public and private vigilance against health and environmental risks from chemicals in commerce that might not be noted or covered by other regulatory authorities. The mandate is to avoid "unreasonable risk of injury to health or the environment," while balancing the benefits of any controls against "unnecessary economic barriers" [§2(b)]. The onus is on EPA to show that unreasonable risk exists, but if it does so, controls are to "protect adequately" against the risk [§6(a)]. In promulgating any such rule, the Administrator must "consider and publish a statement with respect to...the effects...on health and the magnitude of the exposure of human beings,...the effect on the environment,...the benefits of such substance...for various uses and the availability of substitutes..., and...the reasonably ascertainable economic consequences of the rule, after consideration of the effects on the national economy, small business, technological innovation, the environment, and public health" [§6(c)(1)].

In other words, EPA is given rather general authority to seek out and regulate any "unreasonable risk" wherever it may be found, but what might otherwise be sweeping authority is reigned in by the requirement to consider economic and social impact. The act also offers a myriad small checks on this authority in addition to one major one--"If...a risk of injury to health or the environment could be eliminated or reduced to a sufficient extent by actions taken under another Federal law" [§6(c); §9(a)(1)], that other law must be deferred to unless it can be shown to be in the public interest to regulate under TSCA. In practice, this "hand-off" to another regulatory authority almost always happens, and most assessments of risk due to major "existing" chemicals (as opposed to "new" chemicals, as discussed above) are referred to the CPSC, OSHA, or another part of EPA.

In the analysis of new chemicals, OPPT generally seeks margins of exposure relative to NOAELs of 100. Cancer risks are generally ruled acceptable if they fall below 10^-4 lifetime individual risks for occupational settings and below 10^-5 for general population exposures. It should be borne in mind that these are rough criteria, given the screening nature of new chemical assessments.

TSCA is a cost-benefit balancing statute, but a rigorous analysis of costs and benefits is usually only possible for actions contemplated under §6. The much more frequent new chemical analyses and development of risk justifications for test rules employ a more qualitative consideration of costs and benefits.

• Office of Air and Radiation

Until about the 1950's, air pollution regulation was framed in terms of control of public nuisances; local and state laws aimed to control particular emissions sources that created visible and direct public annoyance. Growing awareness of the chronic health effects of air pollution, and a growing concept of unsullied air as a public resource held in common and in need of public protection, led to various control measures, including the passage of the Clean Air Act (42 U.S.C.A. §§7401to7671q) in 1963. Initially, the Federal role was largely limited to research, with primary responsibility for control left to the states. It became evident, however, that state control alone was insufficient to deal with cross-boundary movement of polluted air. Moreover, states varied widely in the vigor of their enforcement, prompting fears that states would vie to attract industry by providing lax regulatory environments. The inherent conflict is that the sources of air pollution are local, and hence properly in the realm of state and local regulatory control, but the effects are on the common resource, so that irresponsibility of the few despoils the air for all--a classic "commons" problem.

This initial, desultory phase of air pollution control ended in 1970 with the passage of amendments to the Clean Air Act that for the first time created a strong Federal role. Implementation of pollution control plans, issuance of emissions permits, and enforcement were still the province of the states (as they continue to be today), but these state activities had to accomplish the meeting of Federally mandated and uniform standards for air quality, with provisions to ensure that the states would rigorously enforce the standards.

The Federal standards are of two basic kinds: standards for air quality and standards for the performance of pollutant sources in terms of allowable emissions. Standards for air quality specify uniform national definitions of what constitutes acceptably clean air, and regulatory programs (much of which occur at the state level with EPA oversight) covering the spectrum of sources of the pollutant by a variety of means are then aimed at achieving air quality at least up to those standards. Performance standards for sources are aimed at establishing uniform national limits on the emissions from particular kinds of sources, including motor vehicles (mobile sources) and stationary sources. (For some purposes, the CAA distinguishes among "major" and "minor" sources based on amounts of emissions, and on "point" and "area" sources based on whether the emissions come from a specific, identifiable facility or from more general human activity not easily localized to a few geographic coordinates.)

Sections 108 and 109 of the CAA call for the development of air quality criteria for the widespread "criteria pollutants." Criteria pollutants include sulfur dioxide, particulates, ozone, nitrogen dioxide, carbon monoxide, and lead. (The criteria pollutants are not named in the statute, but are those with "emissions which...may reasonably be anticipated to endanger the public health or welfare...[and] result...from numerous or diverse mobile or stationary sources" [§108(a)(1)]. Over time, lead has been added to the list and hydrocarbons dropped.) The criteria "shall accurately reflect the latest scientific knowledge useful in indicating the kind and extent of all identifiable effects on public health and welfare" [§108(2)]. So-called "primary" ambient air quality standards are to be standards which, "allowing for an adequate margin of safety, are requisite to protect the public health" [§109(b)(1)]. (There are also "secondary" standards that consider non-health effects.) Legislative history has led the "ample margin of safety" mandate to be interpreted as requiring protection of most of the population, including sensitive population groups (e.g., asthmatics, the elderly) but not the most exposed individual or the most sensitive member of a sensitive group. These are to be purely health-based criteria, and are not dependent on costs or technical feasibility.

It is up to the states to provide plans for controlling pollution so as to attain these National Ambient Air Quality Standards (or NAAQSs); section 110 calls on each state to submit to the EPA for approval "a plan which provides for implementation, maintenance, and enforcement of such primary standard in each air quality control region (or portion thereof) within such State" [§110(a)(1)]. Such State Implementation Plans (SIPs) are to include "enforceable emission limitations and other control measures...(including economic incentives such as fees, marketable permits, and auctions of emissions rights)...as may be necessary" [§110(a)(2)(A)] and must provide for monitoring and enforcement. Section 111 provides for Federal standards of performance for new sources of criteria pollutants "which may reasonably be anticipated to endanger the public health or welfare." Sections 160-169B provide for the prevention of significant deterioration of air quality in regions that are already in attainment of the NAAQSs.

Mobile source emissions are addressed in §202; emissions standards for new motor vehicles may be set for "any air pollutant...which may reasonably be anticipated to endanger public health or welfare" [§202(a)(1)]. Although the main concern has been motor vehicles as a source of criteria pollutants, mobile source toxics are also addressed in §202(l), which calls for study of "emissions that pose the greatest risk to human health or about which significant uncertainties remain" and calls for standards for these, including explicit requirements for regulation of benzene and formaldehyde. Fuel formulation may be regulated under §211, and manufacturers of additives may be required to conduct "tests to determine potential public health effects...including...carcinogenic, teratogenic, and mutagenic effects." Such regulations must consider technical and economic feasibility.

Air toxics are regulated under CAA §112. The amendments of 1990 added a list of 189 compounds designated as hazardous air pollutants [§112(b)]. Chemicals may be added to this list by rule if found to "present...a threat of adverse human health effects." Compounds may be deleted from the list by petition if "adequate data" determine that "emissions, ambient concentrations, bioaccumulation or deposition of the substance may not reasonably be anticipated to cause any adverse effects to human health or adverse environmental effects" [§112(b)(3)(C)]. The EPA must build and maintain a list of the principal areas sources and of "major sources" of these pollutants (i.e., those emitting more than 10 tons/year of any one listed chemical or 25 tons/year of any combination). §112 mandates that emissions of compounds on its specified list be controlled to the extent feasible on technical and economic grounds, regardless of the risk they may pose (excepting the de minimis delisting). Section 112(f) calls for the examination of risks that may remain after such technical controls are in effect; EPA must develop methodology to estimate such "residual risk" and recommend legislation to address any such risk that may be found. If Congress does not act on this recommendation, the EPA must promulgate emissions standards "with an ample margin of safety to protect the public health." That is, if residual risks exist after Maximum Available Control Technology (MACT) standards are in effect, there is a fallback to the pre-1990 basis for air toxics regulation. In particular, the promulgation of such standards is triggered if MACT controls "do not reduce lifetime excess cancer risks to the individual most exposed to emissions from a source...to less than one in one million" [§112(f)(2)(A)]. The standards adopted need not protect this maximally exposed individual (MEI) to the 10^-6 level--the criterion is instead the "acceptable risk" policy developed after the vinyl chloride decision [54 FR 38044]--but the existence of a 10^-6 risk triggers the consideration of residual risk regulation.

Thus, despite the fact that the 1990 amendment of §112 was designed to reduce the role of risk assessment in air toxics regulation (and the consequent questions and delays as uncertainties in those assessments are debated), there are several places where risk assessment is called for in evaluating the technology-based controls. These include (1) the listing and delisting of hazardous air pollutants, which depends on whether a chemical may "present...a threat of adverse human health effects;" (2) the de minimis delisting of source categories, which requires less than a 10^-6 risk to the MEI; (3) the triggering of post-MACT standards to address residual risk, which also requires less than a 10^-6 risk to the MEI; and (4) the offset trading of one pollutant for another based on whether the increased emission is "more hazardous." Of these, the third (the residual risk determinations) is the one based primarily on EPA initiative, but it is one that will require extensive analysis, because each source of each hazardous air pollutant should in principle be evaluated at a level of detail such that the individual near each source at highest risk can be characterized.

For criteria pollutants, standards are set by using a complex characterization of the distribution of exposure levels in the population that would be expected under a specified air quality criterion. When combined with the exposure-response relationships, this gives a projection of the number of health effects incidents to be expected in the exposed population. Both the exposure and dose-response components are estimated based on extensive data; they require little extrapolation and few default assumptions, and the estimates of health impact are thus characterized as unbiased estimates without added conservatism. Point estimates rather than "upper bounds" are used. Ranges of risk are estimated corresponding to the experience of sensitive groups.

The risk mandate for protection of public health with an adequate margin of safety is accomplished by setting air quality criteria such that most of the population is protected, including sensitive sub-groups and highly exposed individuals, but not necessarily the most sensitive or most exposed person. There is no fixed level of acceptable risk, which depends on the nature of the health effect in question, the size of the group potentially affected, and the degree of uncertainty about effects and exposure. These decisions are prohibited from considering costs and feasibility.

Although the effects in question are non-cancer health effects, they are generally held not to display a practical threshold exposure for effects. The methods recognize that even quite protective standards do not banish the possibility of some few people being affected. In this way, the situation is similar to that of carcinogens, where "safety" cannot be absolute, and so a reasonable degree of protection must be defined. For criteria pollutants, the risk characterization focuses on population risk, that is, on the health impact on the population as a whole, recognizing that that impact is most likely to appear among the most sensitive and most exposed. There is no real individual risk criterion.

In the analysis leading up to the development of a proposed ambient air quality standard, an analysis may be done of the effects that would be expected if the whole population were exposed to air just at the limit of the standard. Although this is not the primary decision criterion, such an analysis provides an idea of potential impact if all the air were indeed as polluted as is being allowed. This situation is unlikely to occur in practice in a compliant area, since the air quality criteria represent the allowable maximum in what is always in reality a variable level of air quality. (It is interesting to compare the minor role this analysis plays for criteria pollutants to the major role that a similar analysis plays in the regulation of pesticide residues, as discussed in the section on the pesticides office. In that case, the regulatory decision is made on an analysis presuming that all foods contain their maximally allowed residues, even though a distribution with mostly lesser values is likely to be true. The chief difference, of course, is that pesticide residues are more readily manipulated up to their allowable level than is ambient air quality.)

In the case of air toxics, the application of analysis as now being formulated to regulatory decisions is still in the future, and so it is difficult to characterize with confidence. The presumption is that for most sources of most hazardous air pollutants, the maximally available control technology will be sufficient and further regulation not needed. Actual regulations of residual risk, where necessary, will be made under the criteria prevailing before the 1990 amendments, that is, the criteria mandated by the D.C. District Court's 1987 "vinyl chloride decision" [NRDC v. EPA, 824 F 2d 1146]. These criteria have an individual risk component, that an individual exposed to the maximum fenceline concentration for 70 years should not have a risk exceeding 10^-4. They also have a population risk component, that as few people as possible should have a risk greater than 10^-6. The 10^-4 level is the policy definition of "safe," fulfilling the mandate for a regulation that "protects the public health." It is intended that this level of safety be guaranteed even to someone who chooses to fulfill the fenceline exposure scenario, whether or not someone actually does so. The aim to protect as many people as possible from the 10^-6 risk level is interpreted as the provision of an "ample margin of safety" as provided for in the CAA. In the case of non-cancer effects, it is presumed that exposures below the reference concentration (RfC) fulfill both the mandate for safety and for an ample margin of safety. Given the amount and site-specific detail of exposure analysis required to trigger post-MACT regulation, it is likely that the exposure assessments for such regulations will be much less conservative and "worst-case" than may have been the case prior to 1990. Although the regulatory criteria are nominally the same, the risk outcome and the stringency of regulation may end up being somewhat different.

• Office of Water

Regulation of water pollutants is carried out by EPA's Office of Water (OW). The Office of Water administers two major statutes, the Federal Water Pollution Control Act (better known as the Clean Water Act or CWA 33 U.S.C.A. §§1251 to 1387) and Title XIV of the Public Health Service Act (better known as the Safe Drinking Water Act, or SDWA 42 U.S.C.A. §§300f to 300j-26). The Clean Water Act has as its goal to maintain and improve the cleanliness and biological integrity of the nation's waters, including lakes, rivers, and navigable waters. The aim is to make these waters "fishable and swimable." In many ways, the nature of the pollution problem and the nature of the statutory approach parallel that of the Clean Air Act, discussed in an earlier section; the nation's waters constitute a broadly distributed common resource the quality of which is impinged upon by the activities of many local sources of contamination. Each source of effluent is not solely responsible for the resulting water quality, but the collective burden of discharges may result in unacceptable deterioration of the resource as a whole. The regulatory approach is the promulgation of nationwide uniform criteria defining the degree of water quality that is compatible with intended uses and states of different water bodies. (The criteria are health-based, but they are not rules, and are themselves unenforceable.) These water quality criteria are coupled with enforceable technology-based standards for allowable discharges from point sources, which (also like the Clean Air Act) are implemented through permitting regulations by the states. It is the responsibility each state to conduct regulation of discharges such that the applicable water quality criteria are met for the state's waters.

The Clean Water Act opens with a "Congressional declaration of goals and policy" [CWA§101] that sets ambitious goals for the nation, declaring "it is the national goal that the discharge of pollutants into navigable waters be eliminated by 1985" and that "the discharge of toxic pollutants in toxic amounts be prohibited." The history of amendment of the CWA has been in part the history of rescheduling and delaying the milestones and timelines for achievement of the mandated complete solution to the nation's water pollution problems, as issues of feasibility and practical impediments are encountered. Nonetheless, the act has provisions for citizen lawsuits that has led to the agenda of water regulation being driven largely by court orders and consent agreements.

The CWA distinguishes "conventional" pollutants from "toxic" pollutants. The former are largely those associated with discharge of sewage and nutrients, such as fecal coliform bacteria, suspended solids, and sources of biological oxygen demand. In some ways, they are analogous to the criteria air pollutants, the inevitable, widespread products of human activity that are dangerous by virtue of their overproduction if uncontrolled. The present report will concentrate on the "toxic" water pollutants, analogous to the air toxics, that are treated and analyzed as exposures to toxic chemicals.

As enacted in 1972, the CWA required implementation of standards for toxic pollutants providing an "ample margin of safety;" that is, feasibility considerations were not allowed. For reasons similar to the difficulties seen in regulating air toxics under a similar standard, the CWA was amended in 1977 to include a named list of chemicals [§307(a)(1)] to be regulated within three year with regulation to be based on "best available technology" (abbreviated BAT, a feasible technology approach similar to the 1990 revision of the Clean Air Act). A residual risk-like provision permits the Administrator to set a more stringent "ample margin of safety" standard if necessary [§307(a)(4)].

Section 304 of the CWA calls on EPA to establish "criteria for water quality accurately reflecting the latest scientific knowledge...on the kind and extent of all identifiable effects on health and welfare," including ecological effects. That is, the criteria are to be entirely health- and effect-based. For carcinogens, no level can be named that fulfills the designation of "safe," so the criteria are presented as water concentrations that would be expected to lead to lifetime cancer risk levels of 10^-5, 10^-6, and 10^-7 when consumed at the standard rate for a lifetime. For non-carcinogens, water quality criteria are developed that will not violate the RfD. (Cancer risks and RfDs are calculated by the standard methods.) These calculations are based on individual risk, but the criteria are to apply nationwide, so it is presumed that any criterion will apply to a significant number of people. Actual exposures for many people will of course be less, but exposures will be higher for a significant number, both because of the midrange nature of the consumption assumptions and because much surface water in the country is not in compliance with the water quality criteria which (despite the policy statements set out at the beginning of the CWA) remain goals to be striven for in many cases.

The second major statute administered by the water office is the Safe Drinking Water Act, which regulates the contamination of drinking water provided by public water systems. The act took its current form after 1986 amendments that followed a report from the Office of Technology Assessment documenting widespread serious incidents of contaminated drinking water (Findley and Farber, 1992). As with the Clean Water Act, there are a number of statutory timelines for promulgation of regulations that set a very ambitious schedule, one that has been difficult to meet in practice. Regulation is based on the permissible levels of contamination of finished water, that is, as it appears to consumers at the end of the tap. These standards, called national primary drinking water regulations, are promulgated by EPA [§1412(b)(3)] and enforced by the states, which can opt to set more stringent standards [§1413]. The standards apply to all public water supplies serving at least 25 people. Section 1412(b)(3)(A) calls on the EPA Administrator to "promulgate national primary drinking water regulations for each contaminant...which...may have any adverse effect on health of persons and which is known or anticipated to occur in public water systems." The standards are set on a health basis alone, but the requirement is to come as close to meeting them as is technologically feasible. Primary enforcement authority is with the states, which can opt for more stringent standards.

A standard specifies two levels of contamination of drinking water by the compound in question: a "maximum contaminant level goal" is set "at a level at which no known or anticipated adverse effects on the health of persons occur and which allows an adequate margin of safety." For each standard with such a goal there is also specified "a maximum contaminant level which is as close to the maximum contaminant level goal as is feasible" [§1412(b)(4)], where "feasible" means "feasible with the use of the best technology, treatment techniques and other means which...are available (taking cost into consideration)." (In practice, achievability is judged by affordability of control technology to larger public water suppliers; smaller suppliers may have economic difficulty complying. If contaminant levels cannot be measured, a standard can specify a treatment technique to be used.)

In other words, maximum contaminant level goals (known as MCLGs) are to be set solely on health grounds to protect with an adequate margin of safety. Maximum contaminant levels (known as MCLs) are levels that are practically achievable. It is the technically feasible MCL, and not the health-protective MCLG, that is the enforceable standard. The level set for the MCL depends on available technology, and the appropriate level can change with technological advance. Section 1412(b)(9) provides for periodic revision of MCLs to address this.

The main reason for the MCLG/MCL distinction is that carcinogens, being presumed to be without a threshold, have no safe level. (Clearly, it is also possible that an agent with a threshold has that threshold level lower than is technically achievable.) That is, the common problem faced under all statutes requiring "safety" (especially with an "adequate margin") when dealing with non-threshold toxicants is addressed under the SDWA by controlling contamination to as low a level as technically and reasonably possible without particular regard for how much risk is estimated to remain. This is similar to the "carcinogen policy" at OSHA as it existed before the Supreme Court benzene decision and practice at the EPA Office of Air and Radiation before the vinyl chloride decision, both of which policies were overturned by those decisions, as discussed in the sections on those groups. The chief difference is that the SDWA explicitly decouples the risk and the feasibility issues.

It is important to remember that MCLs are set on a technical feasibility criterion, with the feasibility issue being affordability of controls by public water providers. In some cases, other regulatory programs (notably Superfund and Solid Waste) use the water office's MCLs as though they were health-based criteria, for example as standards to be attained for cleanup of or release into water. The entirely reasonable rationale is that requiring concentrations to be lower than allowable in tap water seems to be unwarranted, but the inappropriate implication is sometimes made that attainment of the MCL is a standard of health protection.

• Office of Solid Waste

The regulation of hazardous solid waste is the responsibility of EPA's Office of Solid Waste (OSW). The office implements the 1976 Resource Conservation and Recovery Act (RCRA 42 U.S.C.A. §§6901 to 6992k), which amended the Solid Waste Disposal Act. The purpose of the act is to develop mechanisms for ensuring stewardship over hazardous compounds from their generation to their proper disposal. The act's provisions set up an extensive set of requirements for reporting and record keeping in addition to standards for generators and transporters as well as treatment and disposal practices. That is, the aim is to ensure that hazardous wastes are kept track of--and that ownership and responsibility for those wastes are not lost or obscured--during storage, transportation, and disposal. The provisions can be seen as a means to avoid the processes leading to dangerous hazardous waste sites, especially those at which responsibilities for the wastes are no longer assignable.

RCRA declares it to be "the national policy of the United States that, wherever feasible, the generation of hazardous waste is to be reduced or eliminated as expeditiously as possible. Waste that is nevertheless generated should be treated, stored, or disposed of so as to minimize the present and future threat to human health and the environment" [RCRA§1003(b)]. Hazardous waste is defined as solid waste that may "cause or significantly contribute to an increase in mortality or an increase in serious irreversible, or incapacitating reversible, illness" or otherwise present a potential hazard to human health or the environment [§1004(5)].

Section 1003 requires EPA to identify hazardous wastes and to list those wastes that should be subject to RCRA's provisions. (There is a provision for delisting a waste as well [§3001(f)].) Listing is to take into account "toxicity, persistence, and degradability in nature, potential for accumulation in tissue" as well as factors such as corrosiveness and flammability. EPA is empowered to issue standards "as may be required to protect human health and the environment" in three broad areas: generation, transport, and disposal. Generation is covered by §3002, requiring standards for record-keeping, handling, labeling, and use of appropriate containers. Section 3002 sets up a manifest system to ensure that the waste is kept track of and responsibility for it assigned, from its generation to eventual disposal, even if this involves transactions and transfers of ownership of the waste. Transport standards are mandated in §3003, which also incorporates the manifest system, as does §3004, which governs storage and disposal. Disposal standards are largely framed in terms of technology that must be used. Land disposal is prohibited unless "to a reasonable degree of certainty,...there will be no migration of hazardous consituents from the disposal unit...for as long as the wastes remain hazardous." RCRA also provides for EPA regulation of cleanup of currently active industrial sites that hold RCRA permits and requires permits for waste incineration and other disposal methods in addition to land storage.

Given the largely technical and procedural nature of its provisions, RCRA has relatively little to say about risks and risk assessment. It simply calls for EPA to act to ensure that hazardous waste management practices "are conducted in a manner which protects human health and the environment" [§1003(a)(4)]. Section 3019(b) states that when, in the Administrator's judgment, "a landfill or a surface impoundment poses a substantial potential risk to human health, due to the existence of releases of hazardous constituents, the magnitude of contamination,...or the magnitude of the population exposed", a request may be made for the Agency for Toxic Substances and Disease Registry (ATSDR) to conduct a health assessment of the site. Such a health assessment is not a risk assessment per se, but it contains many of the elements of one, including characterization of the exposures and potential exposures around the site, identification of potential exposure pathways, review of the known health effects of the hazardous constituents present, surveys of health complaints in the population in the vicinity of the site, and the review of applicable health-based exposure standards that may exist.

In practice, the evaluation of toxicity information and the potencies of substances is largely drawn from other EPA sources outside of OSW, including information on the IRIS database, reports produced by the EPA Office of Research and Development, maximum contaminant levels taken from the EPA Office of Water, and methods borrowed from the Office of Emergency and Remedial Response (Superfund). In fact, the analysis of hazards posed by inadequate waste disposal sites has much in common with the analysis conducted by Superfund for abandoned sites. OSW combines this information with its own exposure analyses and conducts risk characterization appropriate to its uses of risk analysis.

Risk calculations represent individual risks under exposures that are calculated with conservatism tempered where possible by the use of distributional and Monte Carlo analysis. Individual lifetime cancer risk levels of 10^-5 or so from unregulated disposal trigger listing of a waste as a hazardous substance and hence subject to RCRA controls on handling and disposal. Newer methods are adopting a range of 10^-4 to 10^-6 as a range in which this decision can be made. Delisting a substance as a hazardous waste requires a risk estimate less than 10^-6 for unregulated disposal. Incinerator permits have usually been granted if risks are below 10^-5. Remediation of active waste sites depends on many non-risk technical and other factors, but a post-remediation risk level of 10^-4 to 10^-6 is aimed at.

RCRA also has little to say about costs, neither requiring nor prohibiting their consideration (Schierow, 1994).

• Office of Emergency and Remedial Response

The Superfund program was created by the Comprehensive Environmental Response, Compensation, and Liability Act (CERCLA, 42 U.S.C.A. §§9601 to 9675) of 1980 and its subsequent amendments, the 1986 Superfund Amendments and Reauthorization Act (SARA), to address the need for cleanup of the nation's hazardous waste sites. The program is administered by the EPA Office of Emergency and Remedial Response (OERR). With no state unaffected by past hazardous waste disposal practices, the Superfund program has perhaps done more than other programs to make the use of risk assessment a local issue. At the same time, it has become a lightning rod for criticism of the U.S. EPA's use of risk assessment for regulatory decision-making in general.

Neither CERCLA nor SARA specifically mention risk assessment, when it is to be used, what procedures to follow, or what levels of risk warrant remedial action or (in the case of specific action) define what actions are to be deemed "protective." The statutes provide a broad mandate to pursue action on contaminated sites that "may present an imminent and substantial danger to the public health or welfare" [§102]. Risk assessment is used under Superfund to define hazardous substances and the amounts of release that must be reported to EPA ("reportable quantities"), rank the risks posed by hazardous waste sites and identify the action priorities among them, including the addition of sites to a National Priorities List (NPL) of high-priority sites, and evaluating the effectiveness of options for remediation (which are chosen on various non-risk grounds in addition to considering risk reduction effectiveness).

Specific policies on risk assessment have been laid out in the National Contingency Plan (NCP, the body of regulations implementing CERCLA and its amendments) and in numerous guidance and policy directives issues pursuant to the NCP. The NCP, like the statutes themselves, does not specifically define the use and form that risk assessment takes in the Superfund site assessment and remedy selection process. However, especially in the area of remedy selection, the NCP interpretation of SARA sets the criteria which must be met and balanced in remedy selection and can profoundly affect the role that risk assessment plays in cleanup of hazardous waste sites. It is important to recognize that although regulatory policy has given risk assessment a role in the evaluation and remediation of hazardous waste sites, it is one of many considerations in the selection of a final remedial alternatives. The NCP establishes nine criteria by which remedial alternatives must be evaluated:

The first two criteria are considered threshold criteria that must be met before a remedy can be evaluated fully by the other criteria. The "overall protection" includes consideration of risks that may be generated as a result of the remedial action (e.g., risks to remediation workers or to the public surrounding a site). However, the strong preference for permanent remedies voiced in SARA and codified in the NCP creates a more technology-based approach to remedy selection, which critics argue can override the implications of a risk assessment.

Nominal decisions about cleanup are influenced (to the degree they are based on risk at all) on individual risk levels. These risks are based on standard scenarios of exposure depending on the anticipated future land use, and on estimates (often upper end estimates) of the concentration of contaminants currently at the site. Exposures are often figured as RMEs, or reasonable maximum exposures. RMEs correspond to exposure scenarios in which some contributing variables are set at conservative, upper-bound values, but most are set at population average values.

Policies regarding the level of risk that constitutes a hazard have evolved in the Superfund Program. At the outset of the program, a 10^-6 lifetime cancer risk was frequently the benchmark against which estimated risks for a site were judged. Under the current NCP and subsequent policy directives, estimated risks at a site are evaluated against a risk range of 10^-4 to 10^-6. The NCP states: "For known or suspected carcinogens, acceptable exposure levels are generally concentration levels that represent an excess upper bound lifetime cancer risk to an individual of between 10^-4 to 10^-6 using information on the relationship between dose and response." The NCP does not address the definition of "protective" in the context of exposure to non-carcinogens. In practice, however, exposures to contaminants resulting in hazard quotients or hazard indices exceeding 1 are considered to carry an increased potential for adverse noncancer health impacts.

An important and unique feature of Superfund risk assessments is the consideration of exposure to many chemicals simultaneously. This practice is attributable to the need of risk assessment to evaluate waste sites as health hazards, and not particular chemicals. Superfund does not consider the possible exposure of some people to multiple hazardous waste sites, however.