Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
3 Some Approaches to Setting Cleanup Goals at Hazardous Waste Sites HALINA SZEJNWALD BROWN During the past decade the assessment and cleanup of haz- ardous waste sites has come to occupy a prominent position in the activities of federal, state, and local governments. Currently, EPA estimates over 23,000 potential sites nationwide and over 850 on the Superfund national priority list. In Massachusetts alone, there are about 400 confirmed hazardous waste sites, of which 21 are on the Superfund list. Cleanup of these sites raises a vexing question: How clean is clean enough? The question is neither new nor unique to haz- ardous waste sites. Yet compared to direct emissions of toxic materials into water or air, soil contamination presents a signifi- cantly more complex problem. As illustrated in Figure 3-1, human and nonhuman exposure to soil contaminants can occur through a variety of pathways. Also, because hazardous waste sites usually contain large numbers of toxic substances with a wide combined spectrum of adverse effects, cleanup standards must be sensitive to this multiple route/multiple agent exposure pattern. Finally, specific circumstances of human intake of the substances through multiple media are difficult to predict or measure. Determining the extent of cleanup of hazardous waste sites can be approached using one of two general methods: absolute or relative. The absolute approach is based on the assumption that we can define acceptable concentrations of hazardous materials in the environmental media from which no significant risk of adverse 34
APPROACHES TO SETTING CLEANUP GOALS iN WATER J WATER .',j \"". - ~ DRY -] SOIL ~ _ ~- FIGURE 3-1 (1980). 35 _ i~ ROOT CROP ~ ~ .,,.''' ~ J MOISTURE Jr: ) GROUND WATER +: WATER A A - PLANT CROP ~ Ale A,; _ FOLIAGE _ ~ I'm ANIMAL .; ~` I WATER ~14~- , I HYDRO SOIL ' :- FILTER/ BOTTOM FEEDER .,, ~ B -VAPOR) MOISTURE ATMOSPHERE (PARTICLES) J Pollutant pathways from soil to man. SOURCE: Dacre et al. effects to humans and the environment would be expected. For toxic effects to humans that have a threshold, this level would be somewhere below the expected threshold for the population at risk. For nonthreshold effects such as cancer, the definition of "clean" is often linked to some acceptable or, as some (Kasperson, 1983) argue, tolerable risk level. The common feature of absolute approaches is their search for universally acceptable numbers (i.e., standards, guidelines, and criteria). Once established, these numbers drive the cleanup process because they, in effect, define the term "clean." In contrast to the absolute, standard-based approach to man- aging environmental pollution, the relative approach defines "clean" for each particular situation. It may be driven by technol- ogy, costs, comparison with other current and historical hazards, or risk/benefit analysis, or it may be expressed as a percentage
36 HAZARDOUS WASTE SITE MANAGEMENT reduction of a hazardous material (for example, 99.99 percent). In essence, an acceptable level of contamination is defined as that as- sociated with the most acceptable option in a particular decision problem. Hence, the acceptable level is defined for each situa- tion through the risk management process rather than used as an absolute goal for hazard management. It has been argued (Fischhoff et al., 1981) that the absolute approach to acceptable risk (or, by analogy, to "How clean is clean?" ~ is simplistic and unworkable in most situations and that the issue should be viewed as a decision problem, unique for each specific situation. Despite the criticism, however, the standard- based approach has been consistently the favored one for risk managers. There are several reasons for this: . Once a standard is adopted, its application is simple and noncontroversial. It is easy to justify and defend in court. It provides a means of communication among all the techni- cal and nontechnical participants of the risk management process on both sides of the issue. It appears to be an objective process grounded in scientific analysis and free of value judgments. ~ It relieves policymakers from the cumbersome burden of dealing with uncertainty and from being charged with imposing their own values and beliefs on society. . It simplifies the problem by automatically determining the goals of risk management activities. It reflects a recurrent hope that we will find a scientific method for objectively resolving the problem of "How clean is clean?" The purpose of this paper is to review five currently used approaches to determining "How clean is clean?" at hazardous waste sites. The paper focuses on the general concepts that are used as well as on specific methods. The work of the follow- ing agencies is reviewed: U.S. Environmental Protection Agency, U.S. Army, California Department of Health Services, Washington State Department of Ecology, and the New Jersey Department of Environmental Protection.
APPROACHES TO SETTING CLEANUP GOALS TlIE EPA SUPERFUND PUBLIC EEAITlI EVALUATION MANUAL General Concepte 37 This document is a comprehensive manual for site assessment and the establishment of cleanup goals. Conceptually, the EPA methodology is similar to that of California in its view of the en- vironmental migration of chemicals, the role of chemical analysis and dispersion modeling in determining media-specific concentra- tions of chemicals, and the reliance on toxicity-based criteria to determine cleanup levels. There are, however, differences between the two methodologies. One notable difference is that California concerns itself with all chemicals found at a site, whereas the EPA manual recommends the use of indicator compounds, chosen on the basis of minimum effective dose (MED) for toxic effects, carcinogenic potency, environmental mobility, and persistence. The following terminology is used in the EPA manual. Critical Toxicity Value This is a property of toxic substances that reflects the quan- titative relationship between daily dose and magnitude of adverse effect of that substance. Three types of critical toxicity values are used: ~ Acceptable intake for subchronic exposure (AlS). The high- est human intake of a chemical, expressed as milligrams per kilo- gram (mg/kg) x day, that does not cause adverse ejects when exposure is short-term (but not acute). This AlS is usually based on subchronic animal studies. ~ Acceptable intakefor chronic exposure (AlC). The highest human intake of a chemical, expressed as mg/kg x day, that does not cause adverse effects when exposure is long term. The AIC is usually based on chronic animal studies. ~ Carcinogenic potency factor. A measure of carcinogenic potency of a chemical, derived from animal data. It corresponds to a lifetime cancer risk per unit dose (mg/kg x day)-t Estimated Daily Intake . This is a daily dose of a substance by a specified route of
38 HAZARDOUS WASTE SITE MANAGEMENT exposure under some particular exposure conditions related to the site. Two types of estimated daily intake values are used: . Subchronic daily intake (SDI). The projected human intake of a chemical averaged over a short period of time, expressed as mg/kg x day. The SD! is calculated by multiplying the peak short- term concentration (STC) in an exposure medium by the human intake factor for that medium and by the body weight factor. . Chronic daily intake (CDI). The projected human intake of a chemical averaged over 70 years, expressed as mg/kg x day. The CDI is calculated by multiplying the peak long-term concentration (LTC) in an exposure medium by the human intake factor for that medium and by the body weight factor. Critical toxicity values are derived from studies on animals or observations made in human epidemiologic studies. Each is spe- cific for the route of exposure specified in the experiment on which it is based. Thus, AlS (oral) is different from AlS (inhalation), and they cannot be used interchangeably. Acceptable intake val- ues and carcinogenic potency index are properties of a substance administered under specified conditions and are therefore appli- cable at any site for any exposure scenario. Estimated chronic and subchronic daily intakes (SD! and CDI) are calculated for a particular site and reflect conditions at that site as well as the estimated route, magnitude, and duration of human exposure. Derivation of Acceptable Intakes for Subchronic and Chronic Exposure A distinction is made between chemicals that produce carcino- genic effects and those that do not. Acceptable intake values are calculated only for compounds that do not exhibit carcinogenic properties. The evaluation manual is not specific on details of the deriva- tions of the AlSs and AICs beyond the fact that they are derived from no observed adverse effect levels (NOAELs) and that the protection of sensitive members of the population is considered. Based on that information, it is reasonable to assume that AlSs and ATCs are derived from quantitative toxicity data by applying uncertainty factors to experimentally derived NOAELs.
APPROACHES TO SETTING CLEANUP GOALS Estimation of Daily Stake 39 The methodology is based on the assumption that human ex- posure to toxic materials present at the site can originate from the following media: air, ground water, surface water, soil, and contaminated fish. Human intake of toxicants from these me- dia can occur through ingestion, inhalation, and skin absorption. Although soil as a medium and skin as a route of absorption are ac- knowledged, the methodology does not specify how human intake should be calculated for these. Instead, the manual recommends that the agency be contacted on a case-by-case basis when intake from soil and through skin (or both) is expected to be significant. Human intake is estimated separately for each indicator com- pound/route of exposure/duration of exposure/population ex- posed. Duration of exposure is divided into chronic and sub- chronic. Thus, for a particular population, SD! and CDI are estimated for each chemical X and route Y using the general formulas: and SDIX,y (mg/kg x day) = STCx,y x human intake factory CDIX,y (mg/kg x day) = LTCx,y x human intake factory, where SDIX y and CDIX y are subchronic and chronic daily intakes of chemical X by route Y; STCx y and LTCx y are short- and long-term concentrations of chemical X in a medium associated with route of exposure Y; and the human intake factor of the medium is associated with route of exposure Y. This is illustrated below for two routes and three media: SDIX,inha~ = STCX'air x human intake factorair, CDIX,inha~ = LTCX,air x human intake factorair, and SDIX,ora~ = STCX~wa~er x human intake factorwa~er + STCX,fi~h x human intake factoring. These examples show that for each route of exposure to a chemical, the total human daily intake is a sum of the daily in- takes from all media by the same route. The additivity applies
40 HAZARDOUS WASTE SITE MANAGEMENT only to the same population exposed at the same time and for approximately the same duration (chronic versus subchronic). For carcinogenic substances, CDI values are also used to cal- culate lifetime carcinogenic risk, according to the formula: Lifetime riskx,y = CDIX y x carcinogenic potency factory y. The value of lifetime risk is later used to determine cleanup levels for the site. Daily intake values for chronic and subchronic exposure, as well as carcinogenic risk, are calculated for specific exposure con- ditions and are therefore specific for each site. Exposure to Multiple Chemicals by Multiple Routes Noncarcinogenic Effects The methodology assumes that the effects of simultaneous exposure to several chemicab that cause the same type of toxicity are additive. Therefore, total daily intake of each chemical must be adjusted to meet the acceptable intake level. This is shown in the following: ~ CDI (route)) < 1 i 1 AIC (route))- SDI (route)) < 1 i 1 AIS (route))- ' where i is the substance number. Once again, the acceptable intakes for chronic and subchronic exposures are specific for the duration of exposure and the route of exposure (oral or inhalation). The methodology also assumes that the effects of exposure to a particular substance through several exposure routes are additive, as shown in the following: ~ CDI (subst)j I 1 AIC (subst) ~ SDI (subst) I 1 AIS (subs")` where j is a route number.
APPROACHES TO SETTING CLEANUP GOALS 41 The overall hazard index for multiple routes of exposure to multiple chemicals with similar toxic effects can be expressed as a sum of hazard indices for each route. Thus, for chronic exposure: m rat Hazard index = ~ ~ CDIij/AICij. i=1 j=1 No significant adverse effects would be expected in the population if the hazard index does not exceed 1 (hazard index < 1~. Carcinogenic Effects The assumption of additivity is also applied to compounds producing carcinogenic effects. For multiple carcinogenic com- pounds absorbed through a specific route, the total risk is: m Cancer risk for route Y = ~ CDIyi x carcinogenic potency factory). i=1 Likewise, the risks from multiple routes of exposure to substance X are additive: n Cancer risk for substance X = ~ CDIjX j=1 x carcinogenic potency factorjx. The total carcinogenic risk for multiple substances and multiple routes is: m n Cancer risk = ~ ~ CDIij x carcinogenic potency factorij. i=1 j=1 Only chronic, 70-year exposure duration conditions are used for calculating cancer risks. Cleanup Criteria Site Assessment Site assessment involves the following steps: Step 1. Selection of indicator compounds.
42 HAZARDOUS WASTE SITE MANAGEMENT Step 2. Estimation of concentrations of indicator compounds in environmental media at the points of maximum human expo- sure, both for short and long periods of time (STC and LTC). Step 3. Comparison of STCs and LTCs in specific media with environmental criteria such as drinking water standards and guidelines, ambient air standards, and water quality criteria. The assessment stops here if standards/guidelines are available for all indicator compounds. Otherwise, the process proceeds to Step 4. Step 4. This step involves the most comprehensive health assessment. Estimated human daily intakes (SDIs and CDTs) of indicator compounds are estimated for each substance/route of ex- posure/duration combination. Cancer risks associated with SDIs and CDIs are also calculated. Also in this step the hazard in- dex for multiple routes of exposure is calculated. Step 4 requires knowledge of critical toxicity values such as acceptable intake for subchronic exposure (AlS) and carcinogenic potency factors. Target Levels The goal of a cleanup is to meet target levels for indicator com- pounds. Target levels are defined differently for compounds with and without environmental standards. For a target concentration for compound with a standard, an acceptable target concentra- tion is one that does not exceed the specific standard for that medium (requirements). Target concentrations for compounds without standards are divided into two categories: potential car- cinogens and chemicals with noncarcinogenic toxic effects. For potential carcinogens, cleanup levels should maintain can- cer risk in the range from 10-4 to 10-7 for a lifetime exposure, with 10-6 as the desirable target risk level. This is a total risk for a particular population. The target concentration is that concen- tration that will produce chronic daily intake associated with this range of risks. If only one carcinogenic substance is present, the target concentration is calculated using the formula: target chronic daily intake Target concentration (medium) = . . intake factor (medium) Target concentration (medium) = acceptable cancer risk potency factor x intake factor (medium) -
APPROACHES TO SETTING CLEANUP GOALS 43 For multiple routes/multiple agents, the target chronic daily in- take (and therefore the target concentrations) can be apportioned between media and chemicals in any combination as Tong as the total cancer risk is within the 10-4 to 10-7 range. For chemicals with noncarcinogenic toxic effects, the target concentration is defined as that at which (1) chronic daily intake does not exceed the acceptable intake for chronic exposure for indi- vidual substances/routes; and/or (2) the hazard index for multiple routes/multiple substances exposures does not exceed unity; that IS .O, CDI (subst, route) < AIC (subst, route), Hazard Index < 1. As with carcinogenic substances, for multiple exposures the con- centrations of individual substances in specific media can be ap- portioned in any way as Tong as the two conditions are met. CAI:~?O11~IA SITE MITIGATION DECISION T1tEE General Concepts This document provides state decision makers with a stan- dardized approach to setting site-specific cleanup levels. It is based on the assumption that a toxicant deposited in the soil will be distributed among the environmental media in accordance with its chemical and physical properties as well as the properties of the media (air, soil, surface water, and ground water). It further assumes that the biologic receptors (humans and terrestrial and aquatic biota) will be exposed through contact with one or more of these media. The system relies on environmental monitoring and predictive formulas and models to estimate the actual concentra- tions of toxic agents in each medium. The emphasis is on defining acceptable concentrations of toxic materials in environmental me- dia at points of contact with the biologic receptors. Three terms are essential to understanding the system: . The maximum exposure [eve! (MEL) is a daily dose (mg/ day) of a substance that is not expected to produce adverse health effects in a 7~kg adult chronic exposure.
44 HAZARDOUS WASTE SITE MANAGEMENT . The applied action level (AAL)is a concentration of a substance in a particular medium that, when exceeded, presents a significant risk of adverse impact to a biologic receptor. AALs drive the cleanup process for a site. . The cleanup level is a site-specific criterion that a remedial action would have to satisfy in order to keep exposure at the biologic receptor level at or below the AAL. The maximum exposure level provides the toxicologic basis for the derivation of AALs and is substance specific. AALs are derived from the MEL and calculated for each medium (water, air, soil) using the average daily human exposure level to that medium as their basis. Like MELs, AALs are substance and species specific. Thus, for a particular agent, human AAL(soil~is different from human AAL (air) or AAL (water). Likewise, human AAL (water) is most likely different from aquatic AAL (water). In essence, AALs define "How clean is clean?" Derivation of MEIs for Humans For the purpose of developing MEI.s and AALs, toxic sub- stances are divided into two groups: (1) threshold agents, which produce effects for which there is a threshold; and (2) nonthreshold agents, which produce effects for which no threshold level can be assumed, such as cancer, mutations, and genotoxic or teratogenic effects. Threshold Substances The following sources of quantitative and/or qualitative data on the toxic properties of substances are recommended, in a de- scending order of preference: human or animal toxicity data, drinking water standards and guidelines, and occupational ex- posure limits, which are used by the American Conference of Governmental Industrial Hygienists to determine threshold limit values (TEVs). These undergo internal review by professional staff before being used as the basis for MEL derivation. From human or animal toxicologic dose-response curves. The derivation of MELs from toxicologic dose-response curves follows a classic method of acceptable daily intake (ADI) derivation, which is illustrated in the following formula:
APPROACHES TO SETTING CLEANUP GOALS TABLE 3-1 Uncertainty Factors Used for the Derivation of Maximum Exposure Limits (MELs) Uncertainty Factor Basis for MELs 10 10 or 100 100 1,000 100,000 Large controlled epidemiological studies Occupational standarde--this range of uncertainty factors accommodates the background behind the various occupational standards NOAELs derived from chronic animal studies NOAELe extrapolated from subacute animal studies NOAELe extrapolated from acute animal studies NOAEL = no observed adverse effect level. MEL ( /d ) NOAEL (mg/kg x day) x adult body weight (kg) uncertainty factor 45 where NOAEL is a no observed adverse eject level and body weight is 70 kg for an adult. The NOAEL can be derived either from human epidemiologic data, which are preferable but rarely available, or from animal laboratory data. As shown in Table 3-1, different uncertainty factors are assigned, according to the source of the data. From occupational TIV8. MELs are derived from occupational TEVs according to the formula: MEL (mg/day) = TLV (mg/m3) x 20 m3/day x 8 hr x 5 days x 47 years uncertainty factor x 24 hr x 7 days x 72 years . As shown here the 8 hours per day/5 days per week occupational limit for 47 years of exposure is extrapolated to a 24 hours per day/7 days per week for 70 years environmental exposure. The
46 HAZARDOUS WASTE SITE MANAGEMENT uncertainty factor is 10 or 100, depending on the uncertainty associated with a particular occupational limit. Nonthreshold Substances For nonthreshold agents, the MEL is defined as the level of exposure that ensures an incremental maximum excess risk (above background risk) of affecting one individual in a million, on a lifetime exposure. Thus, for these agents the acceptable level is derived from an estimated quantitative risk and is equated with an individual lifetime excess risk of one in a million (10-6~. The quantitative risk assessment is performed in-house using a multistage linearized model for Tow-dose extrapolation and a 95 percent upper bound of dose-response data. The methodology relies on the system developed by the International Agency for Research on Cancer (lARC) to classify carcinogenic properties of substances. In the California system, all substances classified by lARC as "probable" or "possible" human carcinogens are treated as nonthreshold agents (carcinogens). Derivation of AAls Acceptable action levels for each medium are derived from MELs using the following formula: AAL (medium) = . x PF, intake factor where the intake factor is the average daily intake of the medium. The pharmocokinetic factor (PF) is an adjustment factor to ac- count for differences in absorption, distribution, and elimination between the different routes of exposure. For air and water, AALs are calculated as follows: AAL (water) = MEI' (mg/3ay) x PF AAL (alI) = 20 (3/g/ Y) X PF ., . Cleanup Level Determination Determination of the cleanup level consists of comparing the
APPROACHES TO SETTING CLEANUP GOALS 47 predicted concentration (C) of toxic material at the biologic recep- tors with those considered toxicologically safe (AAL). The method considers exposures to individual agents in a single medium, indi- vidual agents in multiple media, and multiple agents in multiple media. The following criteria must be met by a cleanup action. Single Agent/Single Medium Single Agent/Multiple Media If a substance is present in more than one medium, the com- bined dose to the biologic receptor is assumed to be additive. Thus, the sum of the ratios of C/AAL in each medium cannot exceed one if the MEL is not to be exceeded. Thus, n ~ AAt < 1. i=1 Multiple Agents with the Same Toxic Action/Multiple Media In this scenario, both the total dose from each medium (a sum of media-specific doses) and the combined toxic effect on the biologic receptor are assumed to be additive. A cleanup action must proceed until CL < 1. ~ ~ AA1 < 1, where i is a medium number and j is a substance number. U.S. GRIMY APPllOACH General Concepts This methodology, which has been used for a number of years by the Army's technical personnel even though it is not officially endorsed by the Army, has been used to assess numerous sites
48 HAZARDOUS WASTE SITE MANAGEMENT Kwa \ '~7-,.£- \ KSP_ /_ / "a FIGURE 3-2 Pollutant pathway from soil to man through water, plant, and animal compartments. (Small, 1984~. Its primary emphasis is on the environmental fate of chern~cals. The methodology views the environment as a set of compartments and a substance as being in equilibrium between these compartments but not between the final compartment and human receptor. This is illustrated in Figure 3-2. The Army approach uses the following terminology: . The acceptable daily dose (DT) (mg/kg x day), is a dose of toxic substance, per kilogram of body weight, that is not expected to produce significant adverse health effects in a population upon chronic exposure. The preliminary pollutant limit value (PPI,V) is a concen- tration of a chemical in soil that will not produce adverse health effects on chronic exposure either directly to the soil or to one or more secondary environmental compartments, assuming equal partitioning of a chemical among all the environmental compart- ments, including soil. When the chemical is partitioned only be- tween soil and one other compartment, this soil concentration is referred to as a single-pathway preliminary pollutant limit value (SPPPLV). Derivation of the Acceptable Daily Dose For the purpose of DT derivation, substances are divided into threshold and nonthreshold agents. The nonthreshold agents are carcinogens. Although not explicitly stated in the document, the threshold substances can be assumed to include all those that produce toxic effects other than cancer.
APPROACHES TO SETTING CLEANUP GOALS TABLE 3-2 Information Sources from Which to Derive Values of Acceptable Daily Doses (DT) of Toxic Pollutants for Human Beings (in order of priority) Input Information Calculation Required Existing Standards Acceptable daily intake (ADI) None Maximum contaminant level (MCL) in drinking water Threshold limit value (TLV) for occupational exposures FDA guidelines for concentrations in foods Experimental Results in Laboratory Animal Studies Lifetime no-effect level (MEL) 90-Day no-effect level (MELgo) Acute toxicity (LD50) Adjust for water consumption factor Use factors for breathing rate, exposure time, safety factor of lO 2 Use factors for consumption of particular foods Use safety factor of 10 2 Use safety factor of 10 3 Use safety factor of 1.155 x 10 5 Threshold Agents 49 The acceptable daily dose applies to chronic toxicity in hu- mans. It is derived either from toxicologic dose-response curves by applying a safety factor to no-effect levels (NEIJs) or from existing standards or guidelines. The NEL used in this system is concep- tually analogous to the better known NOAEL. The term "safety factor" is equivalent to the "uncertainty factor" used in other systems. Table 3-2 lists seven sources of data for DT derivation and the corresponding safety factors. As shown in Table 3-2, the conversion of standards (threshold limit values, maximum concen- tration levels, U.S. Food and Drug Administration [FDA] guide- lines) to DTS requires application of daily intake factors (transfer factors) appropriate to the specific route of exposure. The conversion factor of 1.155 x 10-5 from animal acute tox- icity data lethal dose in 50 percent of animals (LD50) is based
50 HAZARDOUS WASTE SITE MANAGEMENT on the assumption that a safe limit for the maximum body con- centration of a toxic substance is 5 x 10-4 x LD50 and on the assumption that the disappearance rate of a toxic ant from a body is 2.31 percent per day. Thus, DT = 2.31 X 10-2 X 5 X 10-4 X LD50 = 1.155 X LD50 X 10-5 Carcinogenic Substances . For carcinogenic substances, the acceptable daily dose is that corresponding to an excess lifetime risk of 1 in 100,000 (10-51. The quantitative data on carcinogenic properties is derived chiefly from EPA water quality criteria documents. Derivation of Single-Pathway Prellm;nary Pollutant Limit Values SPPPLVs are calculated from DTS using the following formula: body weight SPPPLV (medium) = DT X - transfer factor x K K is the partition coefficient or the product of intermedia partition coefficients between the medium from which an agent originated and that through which the actual human exposure occurs (for example, when a substance is deposited in the soil but human exposure occurs through ground water or through fish from contaminated surface water). Derivation of Preliminary Pollutant [iInit Values In most cases, a toxicant deposited in soil is sufficiently mobile in the environment that the actual human exposure occurs through several media (soil, water, the food chain). In order not to exceed the allowable daily dose through all three pathways, the permissi- ble concentration of the solvent in the original medium, soil, must be adjusted downward from that allowed by any one route. The resulting PPEV is calculated using the following formula: PPLV= 11/(SPPPLV), + 1/(SPPPLV)2 + 1/(SPPPLV)31-1
APPROACHES TO SETTING CLEANUP GOALS 51 If a chemical is distributed in only one medium, the formula is reduced to: PPLV = 1/(SPPPLV)~1 = SPPPLV; that is, the preliminary pollutant limit value equals the single-pathway pre- liminary pollutant limit value. This formula does not apply to situations in which several independent sources of a particular pollutant exist. Cleanup Level Although not explicitly stated in the document, the implied goal of any cleanup is not to exceed the PPEV value. It is also recommended that, for multiple sources of a particular pollutant, the cleanup level must meet the DT X body weight value. The document does not specify whether the cumulative DT X body weight value is calculated by addition or multiplication. Because the equilibrium state cannot be assumed for a chemical that par- titions itself between soil or water and ambient air, the PPLV calculation excludes air as a route of exposure. Finally, although the system does not address simultaneous exposure to multiple toxicants, it is assumed that similar toxic ejects are additive (Small, 1984; Rosenblatt et al., 1982~. NEW JERSEY CLEANUP IEVEIS 1?011 CONTAMINATED SOIIS General Concepts This methodology is designed to identify a range of allowable concentrations of organic compounds in soil. For inorganic com- pounds, acceptable soil concentrations are multiples of background concentrations in New Jersey or U.S. soils (personal communica- tion from R. Dime, New Jersey Departrrlent of Environmental Protection, 1986~. The methodology is similar to the U.S. Army methodology in that the authors view the environment as a set of compartments and the chemical being in equilibrium between them. Exposure through ambient air is not addressed. Like the EPA manual, the New Jersey methodology focuses on indicator compounds, selected for their toxicity, mobility, and persistence, rather than all chemicals identified at a site. Acceptable soil contaminant level (ASCL) is the key term used. It is a concentration of a chemical in soil that meets one or more of the following conditions:
52 HAZARDOUS WASTE SITE MANAGEMENT . does not present a significant risk to health under average conditions of chronic human exposure to soil; . is protective of aquatic life in surface water impacted by migration of a chemical from soil; and . does not present a significant risk to health under average conditions of chronic human exposure to ground water impacted by migration of a chemical from soil. Derivation of ASCIs to Protect Human Health Mom Contaminants in Ground Water ASCEs to protect human health from the effects of drinking the ground water impacted by the leaching of a chemical into an aquifer are derived from either (1) EPA ambient water qual- ity criteria (WQC) for humans or (2) drinking water guidelines, according to the following formula: ASCL = KD(standard) (depth factor) (mobility factor), where ED is the soil/water partition coefficient; depth and mobility factors are soil parameters; and the standard is the WQC for humans or the EPA drinking water guidelines, whichever is lower. Derivation of ASCIs to Protect Human Health Tom Cont~rninants in Soil ASCEs for direct human contact are based on the assumption that contaminants enter the human body through the ingestion of contaminated soil. For the purpose of ASCL derivation, substances are classified into one of two groups; carcinogens and noncarcino- gens. The ASCL for each group is derived from a health-based acceptable daily intake of a substance by applying an average daily soil intake factor. For carcinogenic substances the health- based acceptable daily intake is that which corresponds to an excess lifetime risk of 10-6. For noncarcinogens, it is equivalent to ADIs published in EPA water quality criteria documents. The following formulas are used:
APPROACHES TO SETTING CLEANUP GOALS Carcinogens ASCL = (acceptable cancer risk) (carcinogenic potency, kg x day/mg) (1, 000 g/kg) (lifetime avg. daily soil intake, g/kg x day) ~ 53 where the acceptable cancer risk is 10-6; carcinogenic potency is a slope of dose-response curves in animal bioassay, as calculated by the EPA Carcinogen Assessment Group; and lifetime daily soil intake is 0.0028 g/kg x day. Noncarcinogens ASCL = ADI (mgiday) x 1'000 g!kg x 10 kg daily sol1 Intake by a child x 70 kg where ADI is acceptable daily intake; 10/70 is a child/adult body weight conversion factor; and soil intake is for a log child with pica. Determination of Cleanup Levels According to the methodology, site assessment is conducted in two steps. In Step 1, indicator compounds are selected on the basis of the total score, using the following formula: Score = relative amount score + toxicity score + volatilization score + leachability score + persistence score + bioaccumulation score + aquatic toxicity score In Step 2, ASCEs for indicator compounds are derived for each environmental pathway (soil, ground water, and surface water). The selection of a cleanup level starts with a listing of ASCEs associated with human exposure through three media and ASCEs associated with aquatic life (two values). The ASCL associated with the most sensitive pathway is selected. (The document does not define "most sensitive pathway." ~ No consideration is given to multiple route/multiple chemical exposures.
54 HAZARD O US WASTE SITE MANA CEMENT WASHINGTON STATE I~^ CIEA~ POlICY This is a short in-house manual for the assessment and cleanup of hazardous waste sites. The methodology is based on the as- sumption that contaminants may migrate from the point of origin to other environmental media although no guidance is given on methods for determining the levels of toxicants in environmental media. The cleanup levels for each medium are derived by one of three methods: 1. specified multiples of existing standards namely, drink- ing water standards, ambient air quality standards, occupational standards, and, for chronic air exposure, dangerous waste limit values (not defined in the document); 2. specified multiples (including one) of background levels of the toxicant in the same medium; or 3. biologic tests for water quality (not defined). These methods are illustrated below. For soil, the method uses 10 times the appropriate drink- ing water or water quality standard. If no standard exists, then 10 times water quality background is used. If the water quality background is not detectable, then soil background is used. For ground water and surface water, the appropriate drinking water or ambient water quality standard is used; if no standard exists, then background is employed. For air, the method uses U.S. Occupational Safety and Health Administration/Washington Industrial Safety and Health Admin- istration (OSHA/WISHA) limits for air quality over the site prior to backfilling or ambient air quality standards at the site bound- aries prior to backfilling. If no standards exist, then background levels are used. COMPARISON OF TlIE METHODS A review of the five methods (EPA, U.S. Army, California, Washington State, and New Jersey) for defining levels of cleanup at hazardous waste sites reveals that their key goal is the pro- tection of public health. Implicitly or explicitly, all assume that chemicals deposited in the primary medium, soil, will migrate into secondary environmental media according to their properties and those of the media. The concentrations of chemicals can be
APPROACHES TO SETTING CLEANUP GOALS 55 determined by direct sampling and analysis or by predictive meth- ods. All five approaches recognize that human exposure can occur through more than one medium. Once a site investigation indi- cates that human exposure to toxic materials present at the site is likely, the goal of each cleanup action is to prevent significant ad- verse health ejects in the exposed population. In each of the five methods the goal of the cleanup is defined through a set of media- specific numerical permissible concentrations of toxic substances at the points of human exposure to them. Thus, the methodologies described here are consistent with the general preference for an absolute standard-based rather than relative approach to defining environmental cleanup levels for chemicals. In addition to conceptual similarities among the five methocI- ologies, there are also some profound differences among them. These differences are grounded in different applications of the gen- eral concepts and include such variables as choice of simplifying assumptions, degree of reliance on the principles of toxicology, sources and interpretation of toxicity data, level of detail, termi- nology, definitions, acceptability of carcinogenic risks, and others. A number of these variables are discussed in the sections that follow. Terminology It is immediately apparent that each approach uses a unique set of terms and acronyms that are incomprehensible to all but those who are very familiar with the documents. Table 3-3 provides some clarification of terminology. Environmental Media Addressed As shown in Table 3-4 the five methods differ in this area. All methods consider drinking water, but air, soil, and foodstuffs are not universally included by the five methods. Environmental Partitioning The common assumption implicit in the five methodologies is that the chemicals deposited in the primary medium, soil, will migrate into secondary media according to their properties and
56 So 4) TIC v o L. 4} ._ :^ I In so o v al CO CO m :^ DO o be . In ~ Cal U] . U) Cal ._ 4,0- ._ v o ._ 0c a ~ ~ O _ ~ ._ ~ ~ ~ ~ V O ~rats ~ ~ 3 3 ~ ~ ~3 ~ ~V V ~o ~z ¢ ~ ~4) a' al A co on oo . - . - . - . - . - o o o o o - uto zc _ ~ ~ ~0 ~ _ sot o _ ~o o ~. - o E-~ ~ ~3 ~ - ~o ~- o ~.~ Z~ o C~ ~ ~ ~ _ · ~c x~ E° ~ ° ~ ~ ~E V `'s V .~ ,V~ ~ 9 ~ ~ ~ ~ _, ~e ~ >. ~V ~
APPROACHES TO SETTING CLEANUP GOALS 57 those of the media. The methodologies differ, however, in their ap- proaches to estimating the media-specific concentrations of chem- icals. The Washington State methodology does not address the topic in any detail. In the EPA and California approaches, media- speciDc concentrations of chemicals in secondary media are deter- mined by direct sampling and by environmental modeling. Thus, the knowledge of current and future concentrations of chemicals in the primary and secondary media is as close to the reality as analysis and modeling permit. The goal of site cleanup under these approaches is to ensure that these concentrations do not exceed previously established chemical-/media-specific numerical criteria. The U.S. Army and New Jersey methods take a different tack. First, both view the environment as a set of compartments in equi- librium with each other so that the concentrations of chemicals in secondary media can be calculated from soil concentrations by using a set of equilibrium constants. Of course, because in reality equilibrium conditions occur only at compartmental boundaries at best, the calculated concentrations of chemicals are often signif- icantly overestimated. Further, the equilibrium assumption does not apply to assessing the ambient air concentrations of contam- inants. Second, by centering around the question "what cleanup level is necessary in the primary medium such that the predicted concentrations in the secondary media do not exceed the health- based acceptable levels?" the two methods attempt to use math- ematical formulas that link the last point in the environmental pathway of a chemical to the first one. The EPA and California methods do not do that. Instead, they rely only on comparing concentrations of chemicals in individual media at the points of human exposure with the acceptable health-based levels in these media, with the implicit understanding that cleanup of the pri- mary medium should somehow lead to acceptable levels in the secondary media. So, whereas the U.S. Army and New Jersey methods may be overly simplistic and stringent, the EPA and California approaches are narrower in scope. Derivation of Media-Specific Numerical Criteria As stated earlier, in each of the five methodologies reviewed here, media-specific numerical criteria play an essential function in defining cleanup levels at hazardous waste sites. In short, these
58 ID c o ~0 ._ ~ 4, In og c~ 4, - v v o o 4, ~c o L. 4, ._ ~c o c o ._ c~ o v 03 m c o c p" C . _ L~ ao 4, .o ~} C ~C 4, v E ' 0 ' E ~_ ~ 2 a i' ~0 E c ~ ~4) S ~ ~ ~ ,a, v O E n E ° 3 ~0 2 ~° :t ~v <: 0 e E, V :r: ~V 4, ~E ,. ~ ~:: S ~ n ~ E ~4, "Q ~ O ~. 05 x O ._ ~4, ~ ~c ~O O ~c o ~,= ~ O D m ~cn
59 o oil ~·o c' ~ ~Go ._ ~. On ~Z Z ~5 ~5 IS ° al ~V ~ O ~ ~L. L. ~id O ~ be ~5 ~5 ~ a: ~ s ~ ~ be._ ~o O O O ~ ' ~As He ~v v s P~ .> ,,o ~E ~ ~ ~=. S ~V 4, ~5 .~ 47 a. ~ ~ o a. 4} - ~E l E ~do, ~ == Van ~ ~ ~ ~ ~S -.~: E ~ ' ~ ~o V , V o L o ·O o ,, ~ ~V ° ~ ~V · - ·- ~ ~ o o ~° c E C ,., o ~t ~o ·,, = ~os , ~V ~z
60 HAZARDOUS WASTE SITE MANAGEMENT numbers determine "How clean is clean?" Therefore, the method of derivation of these numbers is a cornerstone of each method- ology. There are three main conceptual approaches to this task: (1) use media-specific background levels of chemicals or their mul- tiples; (2) use chemical-specific existing standards for air, soil, and water; and (3) develop chemical-/media-specific criteria from toxicity data. The first approach is simple, but in practice it may be un- achievable. Only one of the five methodologies, that of Washington State, uses it. The second approach is also simple and does not require a knowledge of toxicology, but it suffers from three major limi- tations. First, environmental standards and guidelines, derived under different laws and based on different sets of requirements and assumptions, are a mixed bag of numbers that are not nec- essariTy protective of the public health of a diverse population. Second, because these numbers are meaningful only when applied to a particular substance in a particular medium, they can not be used to address the multiple media/multiple chemical exposure scenarios that are prevalent at many hazardous waste sites. Third, the number of chemicals for which air and water standards, guide- lines, or criteria have been developed is small. Perhaps for these reasons, the use of ambient standards is limited. Only the Wash- ington State methodology makes extensive use of them to define cleanup levels. The EPA approach uses environmental standards to a limited extent; namely, when all indicator compounds in all media have them, a very rare event. The third approach to deriving numerical criteria from tox- icity data is the most popular (used by EPA, the U.S. Army, California, and New Jersey) and the most difficult. In essence, it consists of the derivation of a chemical-specific AD! (or its conceptual analog), followed by its modification by media-specific intake factors, according to the following formula: criterion (chem, medium) = ADI (chem)/intake factor (medium). Because AD! is a chemical-specific toxicity parameter, it can be modified accord- ing to particular exposure conditions, which is the advantage of this approach. Hence, multiple chemical/multiple media exposure conditions can be considered. There are two ways by which the acceptable daily intake is calculated: 1. Exclusive reliance on toxicity data. Here, the acceptable
APPROACHES TO SETTING CLEANUP GOALS 61 daily intake is calculated by applying an uncertainty factor to a threshold daily dose. In the California and EPA methods the NOAEL (no observed adverse effect level) serves as a threshold dose. In the U.S. Army method the NEL (no-effect level) is used. Only the EPA methodology relies exclusively on this approach. California, the U.S. Army, and New Jersey use it in conjunction with another approach, which is described in the next paragraph. 2. Conversion of existing guidelines, standards, and criteria into acceptable daily intakes, according to the formula: Acceptable daily intake (chem) = criterion (medium, chem) x intake factor (medium). For instance, in the California and U.S. Army methods, occupational exposure limits are converted into the MEL and DT, respectively. Likewise, drinking water standards and food residue limits are converted into DTS by the U.S. Army. The conver- sion methods vary. In the California method, the conversion into maximum exposure levels {MELs. expressed in m~/daY) is cer ~1 · ~. . 1 ~ ~1- ~ · ~· formed oy in-house experts turougn tne appllcarlon OI UnCerFaIn£Y factors, pharmacokinetic factors, intake factors, and professional judgment. The U.S. Army method relies on uncertainty factors and intake factors. The New Jersey method relies only on intake factors to convert numerical criteria into allowable daily doses. (See Tables 3-1 and 3-2 for a list of the uncertainty factors used in the California and U.S. Army approaches.) Clearly, there are differences among the methodologies. The advantage of approach 1 is its firm reliance on toxicity data and principles of toxicology. Its disadvantage is that it re- quires extensive data and sophisticated scientific expertise and is · ~ resource Intensive. The advantage of approach 2 is its efficiency. Its main disad- vantage is that, as stated before, standards and guidelines, derived under different laws and based on different sets of requirements and assumptions, are a mixed bag of numbers that are not necessarily related to toxicity data for a particular chemical. Furthermore, by converting these numbers into acceptable daily doses (MELs, DTS), this approach erroneously implies that these are toxicity- based numbers. Despite these clear limitations, approach 2 is used extensively by California, New Jersey, and the Army. It is apparent that there are significant differences among the four methodologies (excluding Washington State, which uses a to- tally different approach) in the derivation and use of chemical-/
62 HAZARDOUS WASTE SITE MANAGEMENT media-specific numerical criteria. They diner in both their toxi- cologic data bases and methods of conversion. It is thus unlikely that criteria developed by different methodologies for the same medium/chemical should be the same or even comparable to each other. It is also evident that it is inappropriate to use numbers originating from more than one source to solve a particular prob lem. Estimation of Carcinogenic Risks In all four approaches the lifetime excess cancer risk is a prod- uct of carcinogenic potency factor and dose. Where the approaches differ, however, is in the interpretation of carcinogenic potency and the data base used. The New Jersey and EPA methods use the EPA Carcinogen Assessment Group's slope factors (expressed as kg x day/mg). These are 95 percent statistical upper bounds risk estimates that are derived mostly from animal experiments and are not converted to human unit risk values. California relies on its own in-house quantitative risk assessment. The potency fac- tor is based on animal or human data and reflects a 95 percent statistical upper bound of raw data, extrapolated to humans and extrapolated to low doses using the multistage model. The U.S. Army approach uses the unit risk values from EPA water quality criteria documents. These are 95 percent statistical upper bounds estimates, extrapolated to humans and extrapolated to Tow doses using the one-hit model. Given the above differences one may expect that carcinogenic risks calculated by each method for the same substance/exposure conditions may differ by one or more orders of magnitude. Acceptability of Carcinogenic Risks In the four methodologies reviewed here that use chemical-/ media-specific criteria to define "How clean is clean?" separate treatment is given to substances with and without carcinogenic properties. For substances with carcinogenic properties the cri- teria are based on some cancer risk level set as a goal. The three methodologies that address cancer risks for multiple sub- stances/multiple media exposure conditions (California, EPA, and the U.S. Army) assume additivity of cancer risks. The methods
APPROACHES TO SETTING CLEANUP GOALS 63 vary in what they consider a goal risk level. New Jersey and Cali- fornia use a total risk of 10-6, the U.S. Army uses 10-5, and EPA uses a range of from 10-7 to 10-4 with 10-6 being a preferred goal. Multiple Chem~cal/Multiple Route Exposures The Washington State methodology, which relies mainly on existing media-specific standards, does not address this issue. Nei- ther does the New Jersey approach. Both California and EPA consider cancer risks from multiple routes and/or multiple chemi- cals to be additive. Also, the adverse effects of multiple chemicals with similar types of toxic response are additive. Finally, the total dose from multiple routes of exposure to a substance is ad- ditive. The U.S. Army approach also assumes that multiroute doses of a substance are somehow cumulative but does not specify their exact mathematical relationship (additive, multiplicative, or other). Multiple chemical and multiple carcinogenic risks are not addressed. SUMMARY AND CONCLUSIONS Hazard management at waste sites is more complex than at other locations because it involves multiple pathways of exposure. All of the methods reviewed in this paper focus on the protection of public health from the adverse effects of exposure to single tox- icants as well as their mixtures, through single or multiple routes of exposure. The most favored approach to defining Chow clean is clean?" for hazardous waste sites is that based on chemical-/ media-specific numerical ambient acceptable concentrations for specific toxic materials. These criteria are derived separately for substances with and without carcinogenic properties, a practice consistent with many past experiences in regulating air and water contaminants. The rationale used by each method to derive these health-based numbers, however, is unique to each method; thus the results are not comparable. The similarities and differences among the five approaches were summarized in Table 3-4, which shows that, despite the sim- ilarities in defining cleanup levels for hazardous waste sites, the differences in applying the general concepts are vast. The con- fusion in terminology, although frustrating, is the least of the problem. The most serious differences stem from variations in
64 HAZARDOUS WASTESITEMANAGEMENT the basic assumptions about the environmental fate of chemicals, stringency of application of principles of toxicology, data base, use of existing standards/guidelines, use of safety factors, interconver- sion among routes of human exposure, acceptability of cancer risk, and extent of reliance on expert judgment. Because of this diver- sity, acceptable ambient concentrations derived by one method are not comparable with those from another. Furthermore, the adop- tion of numbers derived through one method for use by another is inappropriate. Finally, it is instructive to Took at the results of this analysis in the context of the current emphasis on the separation of risk assessment from risk management. The application of numerical criteria to the "How clean is clean?" question, all related to toxicologic properties of compounds, would imply that this is a risk assessment issue. An examination of the basis of these criteria and the methods for their derivation shows, however, that none of the five methodologies succeeds in the task of separating risk assessment from management. In general, the practice of converting the existing "numbers" into chemical-/media-specific criteria, the need to simplify the complex scenarios, and the need to fill the lack of data with assumptions make it clear that the separation, however desirable, cannot be maintained. REFE1lENCES Department of Health Services, Toxic Substances Control Division, Alterna- tive Technology and Policy Development Section. 1985. The California Site Mitigation Decision Tree. Draft working document. Dacre, J. C., D. H. Rosenblatt, and D. R. Cogley. 1980. Preliminary pollutant limit values for human health effects. Environmental Science and Technology 14:778-783. Dime, R., and W. Greim. 1986. Calculation of Cleanup Levels for Con- taminated Soils. New Jersey Department of Environmental Protection, Hazardous Sites Mitigation Administration. Fischhoff, B., S. Lichtenstein, P. Slavic, S. Derby, and R. Kenney. 1981. Aceeptablc Risk. Cambridge: Cambridge University Press. Kasperson, R. E. 1983. Acceptability of human risk. Environmental Health Perspectives 52:1 5-20. Rosenblatt, D. H., J. C. Dacre, and D. R. Cogley. 1982. An Environmental Fate Model Leading to Preliminary Pollutant Limit Values for Human Health Effects. Pp. 474-505 in Environmental Risk Analysis for Chemicals, ed. Richard Conway. New York: Van Nostrand Reinhold.
APPROACHES TO SETTING CLEANUP GOALS 65 Small, M. 1984. The Preliminary Pollutant Limit Volume Approach: Pro- cedures and Data Base. U.S. Army Medical Bioengineering Research and Development Laboratory, Ft. Detrick, MD 21701. Technical Report 8210. U.S. EPA, Office of Emergency and Remedial Response. 1985. Superfund Public Health Evaluation Manual. Washington, D.C. Washington Department of Ecology. _ Guidelines. July. 1984. Final Cleanup Policy Technical PROVOCATEUR'S COMMENTS David Miller Because the preceding paper is an excellent survey of state approaches to cleanup goals, ~ would like to spend my time as a provocateur discussing the basic concept of using numerical criteria or setting standards for determining "How clean is clean?" at hazardous waste sites. The thought ~ would like to get across is that numerical criteria or standards, or whatever you want to call them, are diversions. They are an impediment that removes science from the process of developing rational solutions to soil and ground water contamination problems. As one who has been involved from the start in negotiations on "How clean is clean?" ~ have watched the numbers and the criteria become more and more stringent. It is not worth arguing over the numbers because almost none of them is achievable. The natural characteristics of the soil and ground water system at each particular site determine the effectiveness of pumping and treating, capping, or flushing the soil. Aquifers do not give up contaminants either uniformly or completely. Yet most sites can be managed to minimize health and environmental impacts without spending tens of millions of dollars to clean them up to background levels. Contaminated portions of aquifers will never be developed by the waterworks industry as potable water supplies anyway, and further contamination of ground water and surface water sources can be prevented. Our real challenge is not how to set the standard but how to educate the legislator and the public to the reality of the cleanup process. The money and effort presently being expended to accom- modate impossible cleanups should be spent on determining and implementing the best way to protect the rest of the resource.
66 HAZARDOUS WASTE SITE MANAGEMENT For example, ground water pumping operations should be located downgradient and not within the boundaries of waste sites where treatment costs are highest and the time required to achieve cleanup standards is greatest. Otherwise, the legacy of the Su- perfund effort will be the endless operation and maintenance of remedial action systems that originally were justified on the basis of artificial criteria and unscientific risk assessments. Finally, let me relate some statistics that perhaps can be used later. The average proposed cleanup cost for key Superfund sites has risen from $5 million to about $20 million. This rapid escalation in cost over the past few years is principally driven by a preoccupation with achieving numerical cleanup standards. The potential number of such sites ranks in the thousands. Investigating Superfund sites has become a million-dolIar pro- cess, with a million more going into litigation. These expenditures have created a giant data base describing the extent of the problem but very rarely shed much light on the technical and economic fea- sibility of remedial alternatives. Endless negotiations over "How clean is clean?" have delayed the initiation of remedial actions for more than 3 years at some of the better-known Superfund sites. During these delays, plumes of contamination increase in size as does, proportionately, the ultimate cost of the cleanup. In conclusion, ~ am not advocating no action, but ~ am propos- ing source control and the treatment of contaminants with the principal objective of protecting what is left and reaching achiev- able cleanup goals over a reasonable length of time.
