The perils of state air toxics programs Second of a three-part series Edward J. Calabrese Elaina M . Kenyon Universiry of Massachusens Amherst, MA OIW3 Over the past decade there has been a major effort at the state level to develop air toxics programs. These programs are essentially designed to address health problems resulting from exposure to contaminants other than the seven that EPA regulates with ambient air quality standards (I). The argument for additional regulatory action to control toxic air pollutants has been based most notably on a higher incidence of lung cancer in the urban environment and on ambient air monitoring Studies. The latter studies, especially of urban
to reduce the uncertain area. (See the October 1
W13-936x189/G923-1323$01.50/0
0
1989 American Chemical Society
areas, show many contaminants, including carcinogens such as benzene, vinyl chloride, and chloroform. The claim that air pollution causes the higher incidence of lung cancer in uhan settings is highly controversial, given the multifactorial nature of cancer. The variation in factors such as diet, stress, smoking patterns, and indoor air pollutant levels between urban and IUAareas makes consensus on the role of the “urban factor” in lung cancer difficult. Biostatisticaland epidemiologicala p proaches of different investigators reflectthe lack of consensus on this point. Karch and S c h n e i d e m (2) have estimated that at least 1196 and possibly as much as 21% of lung cancer may be attributable to air pollution after c o w Enuiron. Scl.Techno1.. Vol. 23, No. 11, I989 1923
tion for smoking and occupational effects. Doll and Pet0 (3) estimate that the contribution of all forms of pollution (air, water, and food) to cancer incidence is 2% with a range of 1-5%. Doll and Pet0 also indicate that the contribution of air pollution to the observed incidence of lung cancer after correction for smoking is minimal and that it does not account significantly for differences in observed lung cancer rates between rural and urban areas (3). Despite the dispute over the causes of the urban factor, there is little argument that urban air contains numerous pollutants, many of which are mutagens and carcinogens in animal systems. What is controversial is whether the levels observed pose significant health risks to the general population. In the absence of an unequivocal answer, many states sought guidance from EPA to develop new ambient standards. The response from EPA during the Reagan years was to provide data on exposure levels based on specific studies and to encourage technical information ”fer. In addition to concern at the state level over the impact of air pollution on public health, the states were also re quested by EPA to develop their own air toxics programs (4). This process has led to a bewildering array of a p proacbes for deriving ambient air levels (AALS), which has evolved into decentralized risk assessment for air toxics. This paper will examine this deceotralked process and will emphasize its public health Implications.
I
Federal regulation The first federal air pollution legislation in the United States was adopted in 1955. This legislation granted the federal government the authority to conduct research, training, and technical assistance programs. Further legislation in the frvm of the Clean Air Act of 1963, which was subsequently amended several times, was ultimately passed. The concept of national ambient air quality stan(NAAQS) did not come into being until the clean air amendments of 1970 were passed; these amendments required EPA to develop and promulgate NAAQS for pollutants for which health criteria had been issued. This requirement represented a significant change from previous (1%7) amendments to the Clean Air Act, which had provided that the srates develop their own air quality standards (1). The process of NAAQS development begins with the identificationof air pollutants that pose the most serious threat to human health based on known health effects and extent of use, occurrence, and release. This is followed by an exhaustive review of the literature on the 1324
Environ. Sci. Technol., Voi. 23,No. 11, 1989
National Ambient Air Quality Standards Sulfur dioxide Carbon monoxide Nitrogen dioxide Hydrocarbons Oxidants (e.g., ozone) Particulate matter (PM,d
-
to public health from air toxics is embodied in a four-part plan. This includes: maintaining a regulatory posture for contaminants of national concern; assisting states in dealing with the control of local air toxics; improving the information base on multimedia contaminants from multiple sources; and developing an improved program for emergency preparedness and response, including the enhancementof information systems and training of
ciatilon of ~ocalAir Pollution Control Officials. The clearinghouse collects, classifies, and disseminates air toxics federal, state, local, and’other agencies
.
V
LA
substances regarding health effects on humans and on other biota as well as analytical techniques and environmental fate and distribution. Before a final criteria document is published, the information collected and the conclusions drawn are subject to extensive internal and external review. Because NAAQS are standards that are legally enforceable (as opposed to guidelines, which are not), considerations such as economic and technical feasibility also apply. partially because the review process is extensive, the development of NAAQS has proceeded very slowly and at present there are only six substances for which NAAQS have been developed. These are referred to as criteria pollutants and are listed in the box (lj. While efforts to develop NAAQS are lagging at the federal level, many states have been faced with mounting public and legislative pressure to address perceived problems with noncriteria air pollutants. As a result, and with the encouragement of EPA, most states have developed or are in the process of d e veloping air toxics programs. The EPA strategy for reducing risks
State and local regulation The scope of and approach to air toxics regulation varies widely among agencies with respect to the pollutants regulated, whether emissions guidelines, ambient air guidelines, or both are developed, and whether emissions standards or guidelines will be applied to existing or modified industrial sources emitting a given pollutant. In addition, requirements for the application of best available control technology or reasonably available control technology vary considerably in terms of the types of sources covered (new, existing, or modified), what controls are required, and when they must be applied (5,6). The state and local entities that have developed AAL guidelines have used three general approaches. In the case of substances classified as carcinogens, some state and local agencies have performed quantitative risk assessment, usually using the linearized multistage model applied to animal bioassay data and setting the AAL at a preselected de minimis risk level (usually one in a million or 1W). More often, however, agencies have used preexisting risk assessments performed by the Carcinogen Assessment Group of EPA (5,6). A second approach used mainly for noncarcinogens, but in some cases a p plied to carcinogens, has been to select a no-observed-adverse-effect level (NOAEL)-or lowest-observed-ad-
verse-effect level (LOAEL) if a NOAEL is unavailable from a suitable animal or human study-and divide it by a series of multiplicative uncertainty factors. The uncertainty factors are intended to account for animal-to-human extrapolation (interspecies variation), sensitivity of high-risk individuals (interindividual variation), use of a LOAEL rather than a NOAEL, and extrapolation from less-than4nmic exposure to chronic exposure. This is the general approach used by EPA and other federal agencies for the develop ment of health-based guidelines for noncarcinogenic substances in a variety of media such as food, water, and air (1).
The third and most common approach involves the application of uncertainty factors to occupational exposure levels (OELs). The OELs typically used are: recommended exposure limits (RELs), which are developed by the National Institute of Occupational Safety and Health; permissible exposure l i t s (PELs), which are developed by the Occupational Safety and Health Administration; and threshold limit values (TLVs), which are developed by the American Conference of Government Industrial Hygienists. Unlike RELs, PELs are standards rather than guidelines and as such must be based on economic and technical feasibility in addition to health effects. Although both RELs and TLVs are primarily health-based, they are also based on the analytical methods and practical detection limits for the particular substance in question. Because OELs are designed to protect workers during a typical 40-h work-week, when they are modified to protect the general population they are often divided by a factor of 4.2 (168 h/ 40 h) to adjust for continuous exposure (7). Additional uncertainty factors, usually in the range of 10 to 100, are incorporated to account for interindividual variation and relative severity of effect-for example, carcinogens versus noncarcinogens (8).
Use of OELS to derive AALS Because TLVs are the most commonly used occupational exposure level in AAL derivation, the discussion here will focus primarily on TLVs, although many of the points made apply to other OELs. There are several types of TLVs: TLV ceiling limits,TLV short term exposure lits, and TLV timeweighted averages (TLV-TWAs). The latter are the most commonly used by states as a starting point in AAL derivation. The TLV-TWA is defined as “the
time-weighted average concentration for a normal 8-hour workday and a 40hour workweek, to which nearly all workers may be repeatedly exposed day after day, without adverse effect.” With respect to the appropriateness of using TLVs to develop AALs, the American Conference of Government Industrial Hygienists has stated the following: “These limits are intended for use in the practice of industrial hygiene as guidelines or recommendations in the control of potential health hazards and for no other use, e.g., in the evaluation or control of community air pollution nuisances, in estimating the toxic potential of continuous uninterrupted exposures . . .” (9).The use of TLVs as a starting point in AAL derivation is clearly inconsistent with their intended use. Other problems with the use of TLVs for AAL derivation are summarized in the following box. Ideally, complete risk assessments should be performed on a chemical-bychemical basis to develop AALs, but this is not an option for many state and local agencies faced with resource l i i tations and a pressing need for air toxics programs. In spite of the problems associated with the use of OELs as a starting point in AAL derivation, it remains an attrac-
-
tive option for many states for several reasons. OELs and their documentation constitute the largest data base available on toxic substances in air that has been peer-reviewed by highly qualified professionals from relevant disciplines. Thus, they are an extremely valuable resource for agencies with limited staff and funding. To ignore the wealth of data on health risks to occupationally exposed workers that has been collected, analyzed, and critiqued by committees of nationally and internationally recognized experts would be a waste of limited organizational resources. In addition, the many OELs are a convenient starting point for agencies needing to regulate a large number of chemicals in a short period of time. We recommend that OELs not be used as a basis for AAL derivation when the critical toxic effect is carcinogenicity. The reason for the latter distinction is the presumed nonthreshold dose-response relationship for carcinogenic effects compared to the generally accepted threshold dose-response relationship for most other toxic endpoints. However, if examination of the background documentation reveals that the numerical basis for the OEL is sufficiently documented health effects, and
Disadvantages in the use of occupational exposure levels (OELs) for ambient air level (AAL) derivation’
Typically state and iocai agencies assume that threshold limit values (TLVs) are the equivalent of human no-observed-adverse-effect levels (NOAELs), when in fact TLVs may be based on human or animal experimental data, industrial experience, or chemical analogy based on similar structure. The degree to which the type and quality of data are factored into the limit is inherentlyvariable because the final TLV is based on consensus judgment by the Chemical Agents TLV committee of the American Conference of Governmental Hygienists. TLVs are intended to prevent or minimize a given effectin generally healthy workers between the ages of 18 and 65. However, the health effectof major concern in workers may not be the same as that for the general population, which contains high-risk subpopulations (e.g., the very old or young or those with preexisting disease states, particularly respiratory disease). TLVs are set assuming a “zero-exposure” recovery period, Le., a time period during which there is no exposure. Correction factors (e.g., 4.2) may appropriately correct for this in the case of cumulative effects, but they may be overly conservative for noncumulative threshold activity agents such as primary irritants. In addition, recovery times allowed by different OELs vary (e.g., TLVs allow 16 h between workdays and 64 h on weekends: the National Institute of Occupational Safety and Health recommends 14 h between workdays and 68 h on weekends). Use of TLVs or other OELs cannot account for factors such as environmental fate of the compound and multiple sources of exposure. Use of OELs does not give the state or local agency t h e flexibility to change AALs to reflect new data until t h e TLV is changed, and substances for which OELs have not been set cannot be dealt with in a system that is based on occupational limits Source: References 7 and 10.
Environ. ~~i.Technoi., voi. 23,No. 11. i9m 15%
the endpoint on which it is based is appropriate for the general population, then the OEL may be a reasonable starting point for AAL derivation. The operative question to ask is: Do available experimental data indicate that the OEL is a reasonable surrogate human NOAEL or LOAEL (8)? Apprnaches to AAL derivation Benzene, which is considered a known human carcinogen by both the International Agency for Research on Cancer (IARC) (Group 1) and EPA (Class A) weight-of-evidence classifications, is used in Table 1 to illustrate the the large AAL variation between agencies. The large disparity in derived values is characteristic of AALs for carcinogens and reflects the use of a risk-based approach, which tends to result in more conservative AALs, rather than an OEL-based or uncertainty factor approach, which tends to result in less conservative AALs. Further examples of the range of AALs suggested for known and probable human carcinogens (as classified by IARC) are presented in Table 2. Table 3 illustrates that AALs for noncarcinogens also vary widely, though not as much as those for carcinogens. In addition to the obvious potential for differential protection of public health that these differing guidelines provide, other potential problems also exist. Examination of Table 1 reveals that two neighboring states, Massachusetts and CoMectiCut, have an approximately 5Gfold difference in their benzene guidelines after standardization to a 24-h averaging time. Because these are ambient air guidelines rather than emission standards, a source in one state could emit levels of a contaminant that are acceptable there but that are unacceptable in a neighboring state. Widely differing guidelines among states might also pose problems for industrial entities that have facilities in a number of states. Table 1 also illustrates that there are large differences among averaging times for different guidelines. If concentrations are the same, the shorter the averaging time the more stringent the standard. The guiding principle in selecting an appropriate averaging time is that it should correspond to the expected duration of exposure required for the manifestation of the endpoint of concern. Thus, longer averaging times should be used when chronic low-level exposure is of greatest concern and shorter averaging times when shortterm effects are most relevant. However, AAL development efforts do not exist in isolation from other facets of air toxics programs, and the necessity of tying AALs to practical considerations
t air levels (AALs) for benzene by state or local agency
,.sa
~(and.ldked
nowy
Connecticut Massachusans
nnm 8h 24 h
150 1.2
Bask
M L
53.4 2
Michigan
Annual
0.14
4
North Carollna
Annual
0.12
1
Nevada
8h Annual Annual
New York
Phliadelphia Virginia
24 h
714 1M)
72 300
254
1718 1237 300
RELa/2W
Quantitative risk asSeSsment Quantitative risk assessment Quantitative risk assessment TLVS/42 TLV1300 TLV/42U TLVllW
.To lepltlmirecomparlwn of AALs, they w m standardized to a unifwm averaging time of 24 using the method described in R e f h c e 11. 'Recommended ex 08um limit. 5hreshoid limit vai?m. rcs Refereme-
of air monitoring at the state and local levels sometimes requires the use of shorter averaging times (8). Although it might be argued that decentraliiation of regulation, as has happened for noncriteria air pollutants, encourages greater creativity and acceptance of responsibility at the state level, it is apparent from a comparison of approaches to air toxics regulation at the state level that the disadvantages far outweigh the advantages. The current situation in air toxics regulation often results in state and local agencies diverting their limited fiscal resources and expertise toward air pollutant regulation development effortsthat are being duplicated in orher states. This is not only wasteful, but it encourages the development of an unresponsive bureaucracy. Variation in AALs among states offers the potential for differential protection of public health. Variation in requirements for control technology places an unfair burden on industries, particularly those with facilities in more than one state. In fact, one of the compellig arguments for the adoption of national occupational health standards at the time the Occupational Safety and Health Administration (OSHA) Act was passed by Congress was the extreme variation between states with respeet to industrial exposure standards (1). For example, the 6rst attempt at banning an industrial carcinogen in the United States occurred in 1961, when the state of Pennsylvania banned the use of p-napthylamine. However, not all states followed the example of Pennsylvania, thereby permitting exposure of workers to this carcinogen in some states but not in others. Such a piecemeal approach to regulating the use of potentially harmful substances in the workplace led to competition among states to attract industry and to the differential protection of workers' health. This approach was ul-
1326 Environ. Sei. Technol., Vol. 23, No. 11, 1989
timately a critical factor in the development of the subsequent fedcral OSHA Act in 1970. It is ironic that EPA has recently encouraged a process that has led to different state-specific AALs when, in 1970, this same decentralized approach was viewed as a major national public health inequity that Congress tried to overcome. At present, the diversity of state and local approaches to AAL derivation and noncriteria pollutant regulation exists because the federal agencies have not taken a more active role in formulating a comprehensive national air Loxics policy that would promote more uniformity among states. For the immediate future,EPA will probably continue to assist the states in guideline development by providing technical information through mechanisms such as the National Air Toxics Information Clearinghouse. Although facilitation of information exchange is an important role for federal agencies in their relationships with the states, it is clear from the problems discussed previously that a more concerted approach to guideline development and air toxics regulation in general is needed. In the absence of leadership at the federal level, the most viable alternative is regional cooperation among the states. Such regional ccoperation together with input from industry would promote the development of more uniformity among state air toxics guidelies and allow more efficient use of l i i t e d resources. One of the critical questions that emerges from the recognition that regulatory decentralization has led at times to large differences in AALs among states, is, How does it affect the regulated community? In theory it appears that such regulatory disparity could be very burdensome to industries with facilitieslocated in states with widely differing AALs. In practice, the AALs have generally not been used in the
same manner as standards for criteria pollutants; in the screening and permitting process, the AALs have not been employed as strict ambient exposure standards.
In general, the states have used the
AALs as guides, and if the limit for a pollutant is exceeded then the industry and regulatory agency tend to develop a mutually acceptable plan to reduce exposure. One difficultyis that the AALs are often derived by a unit or agency other than that which handles the permitting process (e.g., the Department of Public Health versus the Department of Environmental ProtectiodManagement). The permitting group is usually not firmly bound by the AALs; the AAL is just one of a number of factors to be consided in the permit process, which essentially is a risk management process (8). Another aspect of the air toxics issue is the implementation of SARA Title 313, which quires reporting of routine emissions (in pounds per year) to communities. What do these numbers mean in t e m of human health? Risk communication is likely to be a major challenge for the industries and states, and it is likely that major differences of
opinion will emerge over the public health implications of these emissions. How this may then affect the derivation of AALs and the permining process remains to be seen. In summary, the EPA air toxics strategy has led to the development of a highly decentralized approach for the regulation of air toxics at the state level. This in turn has led to the derivation of highly variable acceptable-exposure guidelines for mutagens, carcinogens, teratogens, and systemic toxicans. Such interstate variability in AALs for toxic substances may lead to differential protection of the public health from air toxics, confuse the public about air pollution and health concerns, and undercut the credibility of public health and environmental regulatory agencies. It is interesting to note that while EPA enmuraged the development of such divergence in air toxics regulatory approaches and implementation at the state level, the Food and Drug Administration funded a National Academy of Sciences (NAS) assessment of the federal process of risk assessment. The goal was to determine if greater consistency could be achieved across agen-
cies, thus avoiding controversial decisions-especially those concerning the regulation of chronic health hazards. This effort resulted in the publication of the highly influential work, Risk Assessment in the Federal Govemmeni: Managing the Process (11). Thus, while the NAS report addressed the lack of agreement in assessing risk at the federal level and recommended ways to minimize it, EPA was encouraging just the opposite with respect to air toxics regulation at the state level. For example, the NAS committee strongly recommended “that uniform inference guidelines be developed for the use of federal regulatory agencies in the risk assessment process” (12). Although there can be compelling reasons for different emission regulations in different regions and states, EPA should strongly encourage the development of consistent risk assessment methodologies that assist the risk manager in the final decision-mdcing process.
References E. 1. Merhodologic Approaches to Deriving Environmntol and Occupntioml Health SrMdardr: Wiley: New York, 1978.
(1) Calabrese,
BLE 3
‘The orlglnal AALS _,” Information pmvlded in hlgheror lowar orlplnal
Envlron. Scl.Technol., Vol. 23, No. 11,1989 1827
Practical Statistics for the
Physical Sciences / A
Chemical S t ”
Software for
Personal Computers
Unique-A user’s guide to
available software! -/
/’
(2) Karch. N. 1.: Schneiderman. M. A. “Explaining the Urban Factor in Lung Cancer Mortality“: report to the Natural ReSOUIECE Defense Council prepared by Clement Associates, Washington. DC. 1981. (3) . . Doll. R.: Peto. R. J. Norl. Cancer Inst. 19RI.66. 1191-1308. (4) 77w EPA Air 7hxicxics Stratep (reprint): U.S. Environmental Protection Agency. National Air Air Toxies Information Clearinghouse. U.S. Government Printing Office: Washington. DC. 1985. (5) “NATICH Data Bare Report on State. Local and EPA Air Toxicr Activities”: U.S. Environmental Protection Agency. U.S. Government Printing Office: Washington. DC. 1988: EPA-45015-88-007. (6) “Rationale for Air Toxicr Control in Seven State and Local Agencies“: U S Environmcntal Protection Agency. U.S. Government Printing Office: Washington. DC. 1988: EPA-45015-86-005. (7) Rowan. C. A,: Connolly. W. M.; Brown; H. S.J. Environ. Sci. Heolrh 1984, 819. 618-48.
(8) Calabrese. E. 1.: Kenyon. E. M. Air Toxi a and Risk Arrerrment: Lewis Publishers: Chelsea. MI. in press. (9) “Threshold Limit Values and Biological Exposure Indices for 1988-89”: ACGIH: Cincinnati. OH. 1988. (IO) Chemical Manufacturers Association. Chrmirals in the Communiry: Methods to Evoluatr Airborne Chcmicol Lew1.v:
CMA: Washington. DC. 1988. (11) National Research Council. Risk Assessment in the Federal Government: M a n n ~ inn the Proc.css: National Academy
Press: Washington. DC. 1983. I
tatistical procedures can make an important COntlibUtlOn to evaluating scientific research. Now you can get a m e r understanding of how to use and apply the% prwedures to obtain the maximum amount of information from your data Wth this dynamic new “how-to” book you’ll learn the concepts underMng the use of statistics through understandable. easy-to-followsteps. You’ll cwer the basic pnnciples and assumptions required for using statistical anaips. Then you’ll leam how t o select and apply statistical techniques-and to properly interpret your results. A COmpWkW% refeMlE.thls b k iMlUdes wrkdd-outexamples illustrating each pmce dure. plus. commonty used formulas have been printed on the inSide cwers and a separate COlOred section Of statistical tables has been included for easy reference. h&aI Statrsba for the P h p i z l %em offen a useful and pracbcal approach to working with and understanding s t a t i m 1 p r m dum. NO prior lolowledge Of statistia is necessary. and only a basic knowledge of mathemabcs and algebra is required. Larry 1. Havllcek, UniversiQ of Kansas Ronald D. Crain. MicroAge Computer store ACS ROfeslonal Reference B w k 512 pages !I9881 ClothWuna ISBN08412 14530 L8810573 L S 8 Canada $59 95 Export $71.95 Order from: ~mrhnchrmiai hieq DlStnbution Wfce Oept 99 1 I55 Sufeenth St. N W Washinqton. K 20336 w uu mu FREE
S
800-227-5558
and use your credit cardl
1328 Environ. ScLTechnol.. VOl. 23,NO. 11. 1%
r tne firs t m+ce aoe IC compare qxq anc eas ’y a1 ava aD e software
P
dea tng w i n cremical str.ctJr6
Focs
N n tne Dasc deta Is ana hmcrions of over
(12) Bierma. T. 7h Use of Occupaiionol Erporure Stmdurdr to Estimate Safe Amhie m C o n c m ~ i i o n for s Air Pollurontr: A Compororiw Srudy: Illinois Environmental Protection Agency: Springfield. IL, 1982.
70
ofmare packages Fnd O L wnat ~ s ava abk Nna HMIpacages are best ces gnw IO ftl 0.1 news ana rexILrces Learn a m t turn t
wmmrrlrl xulu hardwn r q ” rcd
t
basrtuMlonroluchIomnnQ”
t
n 11s nmery user’s quae. cwer programs hat nave been groupm accommg to hnctwn: tnme arawnq. qrapha termina emJa ion. struR.re management. moetular model. 19. s w i a app ital ons. and more Exp‘ore tne vano.8 packaages available mth ‘nemca Srrdcr~rebfnvarefor P e r ” bmpJren and mane informed tnoices when e m n q software laniel E. Meyer. E d m AdVancedRwKh ’ffnnolcqes. lor Vendy A. Wan. Edrtw, IC1 FlwrtnawC& lichard A. Lave. M a t e Edhor. Ameftan KemicalSooety iCS Professional Reference B m k 07 pages 11988)Clothbound SBN 0-8412.1538-3 LC 88.8153 ~
~~~
~
~
~
~
~
S 8 Canaaa $49.95 Export $59.95 07 paqes (1988, Paperbound SEN 08412-1539-1 LC88.8153 .S 8 Canaaa $39 95 Export 547.95 lrder from: Ampriaan Chrmial Wiy DiRnbution me.Deot 07 1155 Sufeenth SI.. N W Washington. K 2W36 rWWFREE
Edward J . Calabrese is professor of roxicoloRy ar rhe Uiiiversiy of Massachuserrs School of Piihlic Healrh. Amhersr. His research and puhlicarions in,nl,,e animal exrrapolarion and rhe causes of human &ferenrial susceprihilirv to roxic suhsrances. He has been a conrulranr to the pvernmenr and to rheprivare sector
Elaina M. Kenyan is a research associate in roxicologv ar rhe Universiry of Marsochuserrs. Division of Public Healrh, Ahersr. She received her E. S. in resource development from rhe Universify of Rhode Island. an M.S. in epidemiology from Texas A&M Universiry, and is presenrly a Ph.D. candidare at the University of Massachusetrs. She workedfor several years at the €PA Marine Warer Qualify hborarory in Narraganserr, RI. Her research interests and publications concern animal exrrapolarion and risk assessment.
