Document NEQpXzd46mVJ9Ge74aJYBokBb

TESTIMONY OF Dr. Rudolph H. Stehl Senior Analytical Specialist The Dow Ch-^mical Co. U.S.A. Midland, Michigan 48640 PREPARED FOR PRESENTATION OF DEPARTMENT OF LABOR HEARINGS ON VINYL CHLORIDE OCCUPATIONAL EXPOSURE STANDARD June 25, 1974 i 'TioW 365 This testimony is presented as a.result of d. .cassions among various members of the Society of Plastics Industry # Task Force on Monitoring and Jnalytical Methods for the determination of vinyl chloride in the work environment. It does, we feel, reflect the current "state-of-the-art" wiih respect to analytical procedures, techniques, and instrumenta tion for the accurate measurement of vinyl chloride in a variety of work environments. Procedures for the determination of vinyl chloride can generally be divided into two rather different groups: 1. Procedures applicable to the examination of plant or production atmospheres and reporting rhe results of that analysis im;.. diately. Devices which collect a sample from a large area for a short time and report the results of that analysis immediately are termed area monitors. 2. Procedures which are applicable to the examination of the atmosphere directly surround'ng work personnel. Devices 2- - which collect samples from this small area (usually in the worker's breathing zone) over a long period of time (e.g. 8 hours) and retain any collected material for subsequent analysis are termed personnel monitors. In general, area monitoring provides information on what the concentration of suspected compounds is "right now" while personnel monitors provide information on what concentrations workers'have encountered in the past. Obviously, the time requirements for these two types of analyses are very different. The first, area monitoring, requires procedures capable of detecting vinyl chloride accurately, quickly, and reliably, since the area monitor generally is used to sample many locations (sequentially) and must operate with very little attention or maintenance. The second, personnel monitoring, requires that the method be able to accurately measure vinyl chloride in the presence of the many other gases and vapors with which the worker may come in contact 367 -3over the course of a day. Since the results are always ex post facto, speed of analysis is less important, but the accuracy and quantitative requirements are no less demanding than for area monitoring. Ajain, since examination of personnel monitors from many people may be required, and the analysis may be repeated at periodic intervals (annually, quarterly, etc.), the analytical procedure foil, the examination of personnel monitor samples must also not require high maintenance and must be very reprocible, since the most recent results are to be compared to previously analyzed samples (last month, last quarter, last year, etc.). Thus, the demands for accuracy, . V selectivity (the measurement of vinyl chloride with little interference from other compounds), sensitivity (to be able to detect concentrations as low as 1 part of vinyl chloride in a million parts of air), and reliability (to be able to make the measurement again and again and again without fail and without requiring extensive servicing or repair) are prime -4requirements for both area monitor and personnel monitor procedures. Additionally, area monitors have the demands to be able to do the analysis as quickly as possible so that any hazardous situations can be recognized and the situation corrected. The methods and procedures which I will describe are procedures which have been used, either in the laboratory or in production environments or both for the determination of vinyl chloride in "air" (i.e. work environment). I will describe each procedure in some detail with particular attention to the differences between them as regards the points I mentioned earlier: selectivity, sensitivity, speed, and reliability, I will discuss them more or less in order of their applicability, although some particular situations will recommend one procedure over the others. In order to be consistent with the goal of accurately 369 -5reporting concentrations of vinyl chloride throughout thte work environment, each of the techniques discussed* as "area monitoring" methods has the capability of analyzing "air" samples from a number of different locations sequentially and automatically. 370 -6Analytical Methods 1. Gas-liquid chromatography (GLC) This is a procedure which physically separates and quantitatively detects most individual components in mixtures. The principle upon which this method is based is that different compounds are absorbed into an organic silicone liquid to different degrees. If the silicone is deposited as a thin film on,the surface of granular particles in a long tube and a carrier gas is forced to flow through the tube, when a sample mixture containing ,rinyl chloride and other compounds is introduced into one end, each compound, in the course of its travels through the packed column, will be absorbed by the stationary liquid and desorbed by the moving carrier gas in a unique manner. The result is that each compound will emerge from the tube at a unique time. If the temperature of the tube and the carrier gas flow rate in the tube are constant, each compound (including vinyl chloride) will emerge at its unique time reproducibly. Thus, when the emerging gas forms a part of -7the fuel source of a small flame, the individual compounds can be detected and quantitatively measured by their."effect" on the flame. If the main fue?. source is hydrogen gas, organic compounds will be detected by virtue of formation of carbon ions in the flame and are observed electrically. These types of detectors are termed Flame Ionization Detectors (FID) and are widely used for the detection of organic compounds. The resulting instrument, a gas chromatograph with flame ionization detector, is an extremely useful device for detecting and measuring vinyl chloride in work environments. There are many different models and styles available from many different manufacturers. They have been in use in laboratories and plants for many years and have a good record of reliability. They do require pure hydrogen gas and an uncontaminated source of air (as well as electricity) for their operation. The "open" flame used as the detector does not allow them to be used in areas where explosion hazards exist. Generally a separate room is advisable when they are used in production plants. Csov 372 -8There are several outstanding advantages of gas chroma tographic methods. Primary among them is the ability of this technique to selectively determine vinyl chloride in the presence of many other kinds of compounds. The ulitmate sensitivity of this method is also very good: about 1 ppm and perhaps as low as 1/10 ppm. This method is also applicable to both the analysis of "air" samples as an area monitor as well as being useful for the examination of personnel monitors. The major disadvantage of this procedure is that somewhat longer times are required for the analysis; usually 5 to 15 minutes per analysis. This means, for area monitoring with 10 different areas or locations being sampled and the sample analyzed by one GLC-ITD, each location is only monitored each 150 minutes. There is good evidence to suggest that the time required for the GLC-PID analysis could be reduced to about 5 minutes by means of additional valves but the long-term operation of these valves at these fast cycle times is not known and some maintenance problems can certainly be expected. 373 -92. Infra-red Spectroscopy (IR): Infra-red spectroscopy is another technique which has considerable applicability in the determination of vinyl chloride in "air" (i.e. work environments). This procedure employs a device called an Infra-red spectrometer to irradiate the gaseous sample (contained in a sample cell) with radiation in the infra-red portion of the energy spectrum. Different molecules, by virtue of the chemical bonds comprising the molecule, will absorb energy at specific different wavelengths. For example, vinyl chloride shows rather strong absorption of energy at wavelengths of 6.15p, 9.8y, 10.9y, and 13.9y. Since other common organic atmospheric contaminants have different arrangements of atoms, there is relatively little interference from most other compounds when the energy absorption is monitored at one of these wavelengths. A major advantage associated with the use of IR as an -10- area monitor derives from the fact that a relatively short time is required to make the energy absorption measurement. In general, only about two minutes is required for the measurement, although some additional time may be required to remove the previous uample and introduce the next one. With long pathlength cells, sensitivities of 2-5 ppm can be expected and lower concentrations are obtainable with some additional modification of the sampling procedure. A major limitation of IR tor area monitoring is associated with the fact that the air sample can vary considerably with respect to content of other components, primarily water vapor. Water vapor does show some.absorption at 6.15y and, although this could be compensated, results in a decrease in sensitivity. There are some procedures which have been used to reduce the amount of water vapor in the sample stream but they are still being tested in several laboratories and their long-term reliability is yet unproven. 375 -11A secondary effect of water vapor as well as some other compounds is to slowly form a deposit or "mist" on the transparent end-windows of the sample cell. This too reduces the sensitivity and causes the instrument to be unstable. Again, the demonstration of applicability of several of the water vapor "traps" would" reduce the magnitude of the'problems. At the present time, there are several manufacturers of IR monitoring equipment. 3. Combustion Conductivity The technique called combustion conductivity has also been employed as an area monitor for vinyl chloride. This method is based on the fact that when vinyl chloride is heated in air at 1000C (1800F), almost all of the vinyl chloride is converted to carbon dioxide and hydrochloric acid (HC1). If the combustion products are bubbled into a "pure" water stream, the HC1 may be detected by virtue of 3'?3 -12tbe increase in electrical conductivity of the water (due to the hydrogen ions and chloride ions). This method is very rapid, only about 30 seconds being required to measure the conductivity. It does not require a large amount of maintenance and is a relatively simple device to both install and maintain. The major disadvantage of this method is that there are many other compounds which would be detected as well as vinyl chloride. Particularly severe interference would be expected from such compounds as the various FREON ^ gases, common refrigerant gases. However, this device, wnen used to monitor plant atmospheres where interferences are known to be low, can provide a rapid indication of the concentration of vinyl chloride. The simplicity of the data collection lends this device to easy computerization to provide supervisory personnel with calculated time-weighted average exposures for different types of jobs within the area being monitored. 0,0V 377 m -13- * 4. Total Flame Ionization Detectors (FID) This method is similar to the previous combustion conductivity method in that the total sample is subjected to burning and the combustion products monitored (as in Method 1, GLC-FID). However, the sample is not physically separated into constituent compounds. This procedure, like number 3, is relatively fast; about 2 minutes per sample. It too has a relatively simple data presentation and is easily amenable to a variety oi computer generated report formats (calculated TWA, etc.). The major disadvantage, like number 3 above, is that - it too is a non-specific analyzer. In general, any organic compound (i.e. containing carbon) will be detected and generally with a sensitivity equal to that j * i .; > i l f i \5iW 378 -14for vinyl chloride. Thus, any contamination of the area being monitored- from automobile exhaust, hydrocarbon vapors, etc., will be detected as vinyl chloride. Some correction of' the signal from these interferences is possible but generally at the expense of simplicity and reliability. 5. Gas Chromatography - Mass Spectrometry (GC-MS) This analytical method is not generally applicable to the problem of area monitoring of vinyl chloride. However, it is probably the most specific method for the determination of vinyl chloride. It also amenable to a wide variety of types of samples and is capable of sensitivities well below 1 ppm. It is discussed briefly here since it appears to be useful as a REFERENCE METHOD. As such, it can be used to determine the concentration of vinyl chloride in a variety of different situations and the results compared to one or more of the above area monitoring methods. t |- * f 7 [ i [ 1 r j csW 379 -15- The principle of this method is the same as number 1, GC-FID; physical separation of sample components in a packed tube. However, in this procedure the compounds emerging from the tube are detected with a device called a mass spectrometer. This instrument subjects the sample gas in a rather high vacuum to bombardment by a stream of electrons. These electrons cause the formation cf ions of the sample gas. These ior.s are then separated according to their mass by a magnetic field. Since each compound produces ions which are peculiar to its own atomic and molecular structure, the resulting ions are extremely characteristic of each different compound. When this characteristicion- spectrum is combined with the specificity obtained from the gas chromatographic column, the result is that there are almost NO interferences in the determination of vinyl chloride. A major limitation to the use of GOMS for area monitoring is that it is a fairly complex instrument which requires a 380 -16trained analyst for operation. Also, the maintenance is greater than any of the previously mentioned methods. Finally, the instrument is' expensive to buy-and maintain ($50,000 minimum, $5-10,000/yr. maintenance). 6. Personnel Monitoring Techniques and Procedures As mentioned earlier, personnel monitoring is extremely useful for the examination of the immediate environment around workers. The normal procedure consists of collecting a sample of "air" from the worker's breathing zone continuously as the employee goes about his normal tasks. If the collected sample is then analyzed, the result yields the total amount of vinyl chloride to which the worker was exposed. At this time, there are no continuously-indicating dev ces which can be worn to provide an immediate indication of vinyl chloride concentration. Since this sort of monitoring serves the very useful purpose of permitting an actual "encounter" record, the analysis must be as fast and accurate as possible. tov 3 -17Since samples are collected over a long period of time (up .to 8 hours), the collection device must retain essentially all vinyl chloride-until.the time of analysis and then permit essentially all the collected vinyl chloride to be removed for analysis. Additionally, the collection device must be sufficiently comfortable to wear so as to not interfere with the normal safe performance of each employee's job. % Components necessary for accurate personnel monitoring include: a) sample collection system, comprised of an "air" pump, sample tube, and adsorbent to hold the collected vapors.` b) sample elution and analysis system consisting of equipment for removing vinyl.chloride from the adsorbent and determining the amount. The system components for collection should be components of well-tested reliability and performance. For example, silica 382 -18gel adsorbent must be used with caution, since high relative humidity.will cause, incomplete sorption of vinyl chloride, especially if high sampling flow rates are used. Activated carbon has shown extreme variability in its ability to adsorb vinyl chloride. Two types of carbon which appear to be successful are Pittsburgh PCB, 12/30 mesh and Westvaco WV-H, 8/30 mesh. Elution of the vinyl chloride from the activated carbon for analysis can be achieved either by solvents or by heating. If solvents are used, carbon disulfide is the preferred eluting solvent. It, however, is extremely flammable and precautions must be taken to avoid exposure of fumes to open flames, sparks, and heat. Alternately, any vinyl chloride collected can be desorbed by heating the adsorbent tube to 430C and flushing with 12 liters of dry nitrogen gas, collecting the gas in an appropriate bag. Both elution procedures appear to produce accurate, reproducible results. If the sample collection tubes 383 -19are not analyzed immediately/ care must be taken to avoid additional contamination or loss of vinyl chloride. It is preferable to analyze the samples within several days of their collection. Obviously, all sampling and analytical instruments should be calibrated and tested before use and the initial results compared to duplicate samples taken for analysis by the reference method. Sampling Frequency: The number and frequency of samples required to insure a safe and healthful work environment requires judgement. Continuous area monitoring can be used as a processing aid to allow for early detection of leaks and losses so that equipment can be maintained, it can notify the employee of a contaminated area so that proper protective equipment can be worn to prevent exposure. With adequate continuous monitoring of the work area, less frequent personnel monitoring is needed to document employee exposure. If the continuous monitors \'^qV 384 -20indicate that the work environment is in compliance with established standards, an annual personnel monitoring program should suffice.' If continuous area monitors demonstrate that potentially hazardous areas exist, a quarterly personnel monitoring program by job classification should be done. Summary and Recommendations 1. Reference analytical methods should be established in the permanent OSHA vinyl chloride standard. 2. Continuous area monitoring should be installed in those areas with potential employee exposure. Several different instruments and methods are acceptable, depending on each particular location and manufacturing situation. Whichever instrument or method is used, it should be checked no less than quarterly against the Reference Method using air samples as similar as practicable to those extant in the work environment. 3. Annual Personnel Monitoring should be done in those plants with continuous area monitors. Personnel monitoring 385 -21should cover each job classification, with potential exposure under typical operating conditions. Quarterly personnel monitoring should be done, by job classificaitons and under typical operating conditions, in those plants without continuous area monitors. '4. Gas chromatography, or where it is available, gas chromato graphy-mass spectrometry should be established as the instrument in the Reference Method. Specific analytical procedures which have been successful are given in the attached documents, particularly Dow Method ML-AM 74-12. 5. All results should be recorded as the time-weighted average (TWA) for the period of actual sampling time. 3S6 TABLE OF CONTENTS Foreword.................................................................................................... xi Introduction................................................................................................ 1 I. Basic Policy Problems in Toxic Substances Control: De ciding Under Uncertainty and Balancing Incommensurable Interests................................................................................................. 9 ! A. The Limits of Cancer Risk Assessment................................ 9 B. The Limits of Economic and Technological Assessment... 15 C. The Balancing Problem: Dilemmas of "Socially Accept able Risk"................................................................................. 19 D. Summing Up the Decision Problem...................................... 21 II. Vinyl Chloride Case Study................................................................ A. An Introduction to Vinyl Chloride......................................... 1. Vinyl Chloride's Uses and the Associated Industries 2. Health Risks and Sources of Exposure to Vinyl Chloride............................................................................. B. Vinyl Chloride in the Workplace--The Occupational Safe ty and Health Act...................................................................... 1. The Key Issues for OSHA............................................. 2. The Statute: Substantive Criteria for Standard Setting............................................................................. 3. OSHA's Early Regulation of VC: The National Consensus Standard....................................................... 4. Regulation of VC as a Carcinogen: The Emergency Temporary Standard..................................................... 5. The Proposal for the Permanent Standard................... 6. Comments on the Proposal............................................ 7. The Permanent Standard.............................................. 8. Judicial Review of the Permanent Standard.............. 23 23 23 28 36 36 37 45 45 49 52 55 58 4iO vii Contents 9. Industry Compliance with the Permanent Standard... 62 10. Evaluating OSHA's Actions......................................... 64 Vinyl Chloride Emissions from Factories--The Clean Air Act.............................................................................................. 66 1. The Key Issues for EPA................................................... 66 2. The HazardousPollutant Section and its Early Use.... 70 3. EPA's Search for Alternatives to the Use of Section 112..................................................................................... 4. EPA's Reinterpretation of the Hazardous Pollutant Section.............................................................................. 74 76 5. The Designation of VC as a Hazardous Pollutant and the Proposed Standard...................................................... 77 6. Promulgation of the Standard........................................ 80 7. EDF Settlement and the Proposed Revision of the Standard............................................................................. 82 8. Evaluating EPA's Action................................................ 86 Vinyl Chloride in Food--The Food, Drug, and Cosmetic Act............................................................................................. 89 1. The Key Issues for FDA................................................. 89 2. The Statute: Classification of VC as a Food Addi tive..................................................................................... 91 3. FDA's Early Regulation of PVC Food-Contact Mate rials .................................................................................... 95 4. FDA's Aborted Regulation of PVC in 1973 ................ 5. FDA's 1975 Proposal for PVC....................................... 6. Subsequent History of the Proposal............................... 95 96 99 7. Correcting the Approach to Migration: The Regula tion of Acrylonitrile......................................................... joi The Aerosol Bans--The Food, Drug, and Cosmetic Act, the Federal Environmental Pesticide Control Act, and The Federal Hazardous Substances Act........................................ 102 1. The Key Issues in the Regulation of Aerosols.............. 102 2. The Statutes Governing Aerosol Products.................... 104 3. Agency Actions Against VC as an Aerosol Propel lant ...................................................................................... 109 4. Evaluating the Aerosol Regulations............................... H2 Emissions from Transportation of VC--Four Transporta tion Statutes and the Occupational Safety and Health Act .. 113 1. The Key Issues Regarding VC Transportation............ 113 2. The Safety of the General Public Near Transportation Facilities............................................................................. 115 3. The Safety of Transportation Workers ........................... 119 4. Evaluating the Control of VC Transportation.............. 124 Contents G. Miscellaneous Sources of Vinyl Chloride Exposure and the Statutes for Their Control................................................. 1. VC Entering Air and Drinking Water From Factory Effluent, Water Transport, Sludge Waste, and PVC Pipe.................................................................................... 2. Drug and Cosmetic Packaging, Medical Devices, and Consumer Products.......................................................... 3. The Application of the Toxic Substances Control Act..................................................................................... III. General Observations and Future Directions................................. A. Mitigating the Problem of Jurisdictional Fragmentation.... 1. Overview of the Fragmentation Problem...................... 2. Intra-agency Coordination--The Example of EPA.... 3. Interagency Agreements and Memoranda of Under standing ........................................................................... 4. Cooperative Regulation of Particular Substances....... 5. Broader Forms of Cooperation....................................... 6. The Possibility of Executive Reorganization............... B. Legal Tools for Controlling Toxic Substances Control De cisions........................................................................................ 1. Balancing Rules and Health-Only Rules....................... 2. Precautionary Regulations.............................................. 3. Allocating the Burden of Persuasion............................. 4. Judicial Control Over Agency Decision Making......... Conclusion.................................................................................................... 125 126 138 141 143 144 144 147 148 149 151 154 155 156 160 165 168 177 Appendix: Ecology Law Quarterly Contents...................................... 179 FOREWORD This book on the law and public policy of toxic substances control is the second product of a modest grant made in 1976 by RFF to the Ecology Law Quarterly, Boalt Hall School of Law of the University of California at Berkeley. Our grant supported the extra expenses involved in preparation and publication of a special issue of the Quarterly on "Hazardous Substances in the Environment." This ambitious issue was published in the summer of 1978, just as the continuing public debate on this most important topic was heating up. In reviewing the special issue, several of us at RFF concluded that the paper by David Doniger constituted such an excellent introduction to this evolving field that its wider circulation could be very useful. Ac cordingly, we have arranged to bring out this study alone, with the hope that it will be used in legal, economic, and other courses on environ mental problems. The publication may also serve as an announcement of RFFs interest in this field for, while several staff members have writ ten individually on toxic substances, a formal program on toxic sub stances problems is now developing here under the direction of Allen V. Kneese. Doniger is currently an attorney with the Natural Resources De fense Council and was formerly with the Environmental Law Institute, both in Washington, D.C. He holds a J.D. and a master's degree in city planning from the University of California at Berkeley. While at Berke ley he was special issue editor of the Quarterly. In organizing the project, he had the support and cooperation of Richard H, Cowart (editor-inchief 1976/77), Karl E. Geier (managing editor 1977/78), and mem bers of the board of editors. For readers whose interest goes beyond this paper, we have reprinted the complete table of contents from the journal issue on page 179. Copies of the entire issue are available for $10.00 from the Ecology Law Quarterly, School of Law (Boalt Hall), Univer sity of California, Berkeley, California 94720. The case study of vinyl chloride regulation, the observations about the federal agencies' coordination efforts, and the analysis of the devel- pi,;. jfiWv ?' 1 Wflf "Vs! wy&ijt* ),<* i j&j?/' IS*'! Wi'l 11 INTRODUCTION jp4!$$Oi January 22, 1974, the B.F. Goodrich Company revealed that three i at its plant in Louisville, Kentucky, recently had died of angiosark of;the liver, an extremely rare and incurable cancer, and that a fourth isbfithe tame illness five years before.1 The plant converts vinyl `petrochemical gas, into polyvinyl chloride, the second most f<tped plastic in the United States.1 The grouping of these rare cancers plant immediately raised the suspicion that vinyl chloride was the UThis suspicion was soon confirmed by reports from other companies i,of their workers exposed to the chemical had developed the same r, and by disclosures that since 1970 vinyl chloride had induced a wide of cancers in experimental animals.*3 I * *. jV St# Within a week of the Goodrich disclosure, the Occupational Safety and IfWiith Administration and the National Institute for Occupational Safety |!a<ri| Health began preparing a workplace standard for vinyl chloride.4 *T&his the beginning of the still incomplete regulation of this pervasive iMtmical, an endeavor that has involved five major federal agencies operat13 separate health and environmental statutes. F't'i vTbfii event* ere summarized in Occupation*! Safety end Health Administration, d fyr Exposure to Vinyl Chloride, 39 Fed. Reg, 35,890, 35.89041 (1974) [hereinafter I OSHA Permanent Standard for VC]. See also Hearings on Dangers of Vinyl Chloride tih*S*bc0mm.onEnvimnmentoftheSennteComm.onCommercf,9'}dCong.>`ld$**i. SfelJWti (1974) (testimony of Dr. Marcus Key, Director, National Institute for Occupational S'vSAty and Health) [hereinafter cited u VCHtaringt], '-p4' See teat accompanying notes 81-82 infra. J, See teat accompanying notes 124-129, 139-141 infra. t -* 0*P'of Labor, Possible Hazards of Vinyl Chloride Manufacture and Uset Request $ t fotUformation and Notice ofFact-Finding Hearing, 39 Fed. Reg. 3874 (1974) [hereinafter cited m^8NA Hearing Notice]. pf, & I)< 1 !fwit?|-' s 413 . Toxic Subetetncee Comm/i . J\ >. i'-A This Article examines Ac federal regulation of vinyl chloride. and;)(<^ through that experience, thecomplex law and policy of toxic, substances m control in the United Statet^Attempts to control toxic substances were begun seriously only in thei:,1970s. The law and policy in this area are in rapid growth and transition and are still deeply disorganized. The vinyl chloride problem is one of several chemical crises that has strongly influ enced the growth of the field.5 Vinyl chloride gas (VC) and polyvinyl Chloride plastic (PVC) permeate modem American living. PVC has hundreds of widely different uses, some very important and others completely frivolous. PVC is a major construction material, used in products such as water pipe, floor tile, and exterior siding. > It is a major food packaging material. PVC is used to make consumer ,, products ranging from household furniture to auto interiors,, from credit Cards to baby pants. Ironically, it provides a valuable coating for pollution control equipment, due to its resistance to corrosion. VC gas itself has been used as an aerosol propellant and a refrigerant. It was once even tested for use as a general anaesthetic,* , Hundreds of thousands of people are exposed to VC at their places of work. Millions are exposed to VC from living and working near the factories where it is made and processed, and near the routes over which it is,, transported between factories.' VC leaches from many of the PVC products with which the consumer is in contact daily--from packaging into food, from pipe into drinking water; from latex paint into indoor air, and from many other sources.57 86 The primary threat from VC is to the workers. To date there have been at least 68 known cases of liver angiosarcoma among the roughly 30,000 workers most heavily exposed to the chemical in the three decades prior to 1974. The illness is occurring in these workers at a rate as much as 3,000 times higher than in the general population. VC is also suspected of causing an equivalent number of more common cancers in these workers. Moreover, many other workers are expected to develop liver angiosarcoma and other cancers as a result of their exposure in the 1950s, 1960s, and early 1970s.* 5. There have been several other studies of the regulation of VC: Krause, Environmen ts Carcinogenesis: Regulation on the Pronelers ofScience, 7 Envt'l L. S3 (1976); N. Ashford, E. Zolt, D. Hattis, & i. Kau, The Impact of Governmental Restrictions on the Production and Use of Chemicals: Draft Final Report (appendices concerning regulation of VC In the workplace and as a food additive) (Dec. 1976) (report prepared (or the Council on Environmen tal Quality): G. Adams, Toxic Substance Control: Vinyl Chloride (unpublished mailer's thesis, Washington University, St. Louis. Mo., Dec. 1976). 6. For a more thorough description of the uses of VC and PVC, see text accompanying notes 84-87 and Figure I infra. 7. For a more complete description of the sources of exposure, see text accompanying notes 148-167 infra. 8. See text accompanying notes 98. 175-135 Infra. vR;- r?;:v $ MbdVictims known to date have been exposed to VC at levels ' Ada those experienced by most workers, consumers, and others. jedkldM of a carcinogen are a matter of serious concern, ^hf.not possible to identify safe levels of exposure to carcinol.aogoe scientists believe that as little as a single molecule of such a it interacting with the appropriate portion of the genetic material of cell, can cause a fatal cancer many years later.* i effects of VC are not limited to cancer. At relatively high lllaulet a variety of degenerative symptoms. At low doses it may bJnhdefects and mutations.10 ^Unfortunately, VC is not an oddity. It is representative of thousands of tare capable of causing cancer, other long-term illnesses, and |fojtf'*Ajtle Adverse environmental effects. Like VC, many of these chemrWMia once thought to be safe, are important components of significant ; and are manufactured, transported, consumed, and discarded in Nearly everyone is exposed to them in complex, possibly i combinations; nearly everyone is at some degree of risk." rThaae hazardous substances are only a fraction of the estimated 65,000 pin bommeree and the more than four million chemicals that are s.%Bubin absolute terms toxic substances are large in number, and 'Brefesponaible for a substantial portion of the illnesses, deaths, and I insults experienced today.13 The vast number of chemicals to Jested; evaluated, and regulated makes toxic substances control in many [reaprrtt the major challenge of the health and environmental movement. Onp focu* of this Article is the staggering complexity and fragmenta- of,mu federal programs to control these toxic substances. About 20 environmental statutes empower five major agencies to instances. (See Table 1) The patchwork of statutes grew re mostly over the last two decades, as Congress perceived |dditioaad, relatively narrowly defined needs for controls. Together the laws virtually all the avenues through which people can be exposed to !<dangaoujtj chemicals, and most of the avenues through which such subtaaces eph harm natural systems.14 , .... -1 . < A Swim1 aeoempauyingnotes 22-80, 148-167 infra. i -- "-- Sm**** tawrenyins now 138 infra. K;j-Wg'M-ljl. Oarhe WW Number of hazardous substances needing regulatory attention, re* Sltiin t '' 3 A ignOer, CSegotUadon of Chemicals Under the Toxic Substances Control Act, 7 Ecolooy _ L> Q-159<I97S); hat,A Generic View of Toxic Chemicals and Similar Risks, 7 Ecology L. Q. | (1978), As an example of the numbers problem, see National Institute poa Occutation- ; al garerv A Hialth, Suspected Carcinogens (2d ed 1976), which identities t coo-2,000 -OtWufcals that animal experiments have implicated as possible carcinogens, i 12. Maueh, Chemicals: How Many Are There? 199 Science 162 (1978). :'.i ^Ij. Between 60 and 90 percent of human cancers are estimated to be caused, wholly or in > pert, by chemicals In food, water, workplaces, cigarette tmoke, and the general environment. ,* Council ON Environmental Quality. Sixth Annual Retort 32-33 (1975). See also text Mempanyiag note 33 Infra. 14. Two possible exceptions are the components of cosmetics and of substRnces harmful fc i t'l QT ^ r- TABLEU . > Federal AwtfSUTY Over Vinyl Chloride Aoency Statute ''"''f'"'"'* Year'" Uses or SoURCEi^*1' v;^ Enacted Exposure Covered ! Occupational Safety and Health Administration Occupational Safety and Health Act 1970 Exposure in factories and in transportation (workers only), Environmental Protection Agency Clean Air Act Federal Environmental Pesticide Control Act Federal Water Pollution ` Control Act Safe Drinking Water Act < 1970 1972 1972 1974 Resource Conservation and 1976 Recovery Act f', Toxic Substances Control Act 1976 -l , tj' Factory emissions Insecticide aerosols VC discharges to water ' VC in drinking water, PVC wa ter pipe PVC sludge wastes Possible to use in lieu of mul tiple separate actions .upder other laws :v, 6$, ,,*'" Food and Drug Administration Food, Drug, and Cosmetic Act Food Additives Amendment New Drug Amendments Medical Device Amehdments 1938 1958 1962 1976 Cosmetic aerosols, PVC cosmetic packaging PVC food packaging, PVC water pipe Drug aerosols, PVC drug packaging ' PVC medical devices Consumer Product Safety Commission Federal Hazardous Substances Act IV*1'' 1966 Household aerosols, house hold plastics and paints' Department of Transportation Hazardous Material* Transportation Act Federal Rail Safety Act 1973 1970 Ports and Waterways Safety Act 1972 Dangerous Cargo Act 1940 Rail and truck tank vehicles Rail tank cars and roadbed Barges and tank vessels S the coverage of these statutes, taken together, is nearlyitf* highly fragmented. As noted above, to address all of the kfAMntinan exposure to VC would require action by all five agencies funder 15 of the statutes. Few chemicals are as widely used as VC, J abdiaiercfore lew will fall under so many authorities. Nonetheless, there are i, Sfalready .mote than 20 chemicals undergoing regulation by two or more ; ff agencies." This is likely to be the rule rather than the exception, at least for -vs! chemicals currently well established in commerce.16 A serious problem is that in many cases the boundaries between . agencies' jurisdictions are not clear. Sometimes authorities overlap, empowering two agencies to regulate a given source of exposure.17 In some ^ sreM where two agencies have abuning jurisdiction--i.e., where together their Statutes cover a type of hazard without overlap--the dividing line is .uncertain, and it is uncertain into which bailiwick a particular source falls.1* ^Jurtadktiooal complexity and confusion often discourage agencies from ^'Stepping forward to deal with a problem; each waits for another to act. Even when jurisdictional issues are resolved, the fragmented system M discourages comprehensive assessment and balancing of all of a substance's |^f|friakAand benefits. No agency has the responsibility to consider the net social gaiit'Or loss from different levels of control. Such a holistic consideration :migl)f yield a different result than the sum of partial analyses. Moreover, in 'jfcifttbyC. some risks have been seriously, understated and tome 4 conttoi Costs seriously exaggerated because of the fragmentation of assess- iX* t` nmi of both hazards and economic benefits.19 t . The final drawback to the current balkanized system is the duplication of decision making. Many separate, essentially identical proceedings must 4 be held. The duplication wastes the resources of all concerned--govern^ifrmenij' regulated industries, and health and environmental groups. Some " .fl)industries may prefer the fragmentation because it slows the speed with 7 which the government places controls on the profitable use and sale of mW: JSL to ItW WWral environment which are component! of food and drugs. Sit generally Page A Blackburn. Behind tht Looking Glass Administrative. Legislative, and Private Approaches to Catenate Safety Substantiation. 24 U.C.L.A. L. Rev. 795 (1977); EPA, Fully Halogenaltd OHrofluoroaikanes. Proposed Rule, 42 Fed. Reg. 24.544. 24,545 (1977). -..yM, ..See Interagency Regulatory Liason Group, Joint Regulatory Developments, March 1, l9l ragvfnMd In I BNA Chew. Reg Rep.--Cub*. Rep. 1916-21 (1978) [hereinafter cited at ' TM* of Subtunces of Common Concern], The list includes VC. For some of theie i> tueh u VC, some regulations are already in effect and more are under con* r problem may be avoided for new chemicals, which under the Toxic Sustances. I Act (TSCA) must be tested by manufacturer* and evaluated by the Environmental Agency prior to their entry into commerce. TSCA ft 4, 15 U.S.C. ft 2063 (Supp. V See text accompanying noiev 672-674, 753-771. 793-795 Infra. f.-. See (ext accompanying notes 557-561, 608 infra. See text accompanying notes 347-348, 470-472, 003-804 infra. Toxic Snbitancts Control M hazardous substances. But in many cases uncertainty over the ultimate scope of regulation probably outweighs 'the industries' advantages in delay. In response to problems encountered in the control of VC and several other substances through this fragmented system, the agencies recently havebegun efforts to coordinate theirmany programs and to increase the consist tency of their actions. The promise, and pitfalls of these efforts are surveyed at the end of the Article, on the basis of the lessons of the VC experience.30 , tUm The second focus of thit Artkle ls how agencies cope with problems of> uncertainty and competing^ dissimilar interests that are inherent in >toxiie substances control decisions. Aa is explored in Part I, the regulation of any? of these substances involves- ootnplex decision making under uncertainty, and controversial value judgments concerning the weighing of health and? environmental values against economic interests. The VC case study per-i mits one to see how the many statutes and agencies approach these prob lems, in a relatively constant scientific, technological, and economic' context. v| m "Si Part II, the case study, makes up the major portion of this Article. It begins with a survey of the uses of VC and PVC, the technology and economics of associated industries, and what is known of VC's toxicity. This is the factual background for all the regulatory proceedings; additional< data peculiar to individual areas is given in those discussions. The next three sections in Part II analyze in depth the actions taken to; dale respecting three of the major sources of VC exposure. The first section considers the development of a standard for workplace exposure by the > Occupational Safety and Health Administration (OSHA). The second ad dresses standard setting for emissions from factories to the surrounding air, by the Environmental Protection Agency (EPA). The third considers the proposal by the Food and Drug Administration (FDA) to regulate the use of PVC food packaging. Each of these sections illustrates central legal and policy problems in toxic substances control and a range of responses by agencies and interested parties. Considered together, the OSHA and EPA actions also illustrate the underweighting of risks that can result from jurisdictional fragmentation. The extremes of jurisdictional complexity are illustrated in the next two sections, on VC-propelled aerosol products and the transportation of VC between factories. Authority over aerosols is divided among three agen cies: FDA, EPA, and the Consumer Product Safety Commission (CPSC). Authority over hazardous material transportation is divided among OSHA and three agencies within the Department of Transportation (DOT). Because very small economic interests were at stake over aerosols, regulation pro 0 ceeded relatively smoothly. However, the aerosol episode reveals the poten 1 tial for serious jurisdictional conflict in those future cases when more money20 20. See text accompanying notes 836-868 infm. O ' > ' :-iy-. (. when decisions are more difficult, and when the results for a > might depend on which statute and agency it falls under. In area the economic interests are greater and the balancing fit is more difficult. The jurisdictional lines also are unclear, Storically none of the agencies concerned has regulated the tjof carcinogens and other substances with long-term or subtle . is reluctant to assert its authority vigorously and is content to Jto step forward. , ,, jn of the case study briefly .surveys theat|tWiti<3 for ^(emaining sources of VC exposure. Principally, these are i connected with water, consumer products, packaging other pthenfnjfrml- sod medical devices. The agencies involved are EPA, FDA, No controls of these exposure sources have yet moved beyond ' stage. The control of these sources leads to several more of jurisdictional overlap, of underweighting of risks, and of ; statutory responsibilities. };finel section also considers how the Toxic Substances Control Administered by EPA, relates to the control of VC and to future which authority over a dangerous chemical is seriously frag` J ;,The Act was passed only in 1976, after some standards for VC i been set under prior laws or had been proposed, and therefore it ipf ^felnal use to control this chemical. But EPA might use the Act in the [^ftyture^reduce the number of separate, fragmented actions that are needed dnular substances. J^tt TO of this Article has two purposes. First, in light of the VC ode and similar experiences, it surveys past and ongoing efforts to deal pHith jurisdictional fragmentation and to coordinate federal action and policy 1'iMt.loiric substances control. It discusses the agencies' internal coordination pr#ffoiflh5 interagency agreements, cooperative regulation of particular sub- . ----and two new interagency groups created to foster broader forms of iation. Serious coordination efforts are only just beginning. . $|tond, Part III of the Article confronts the capacity of certain charac" of the legal framework for toxic substances regulation to deal with Ojber central policy problems noted above: decision making under sty and balancing of dissimilar, competing interests. The statutory lions, which are analyzed individually in detail in the appropriate of the case study, are considered here in general terms. Rom the case study it becomes clear that in view of the problems of ^factual uncertainty and conflicting values, each agency faces a wide range of ''x*ti0f!pi choices in setting exposure limits and other standards, but no one oice can be said to be objectively "correct." Each agency must make dgments about what to assume when the true facts are uncertain and about ,11. IJ U.S.C. li 2601-2629 (West Supp, 1978). TodcSutmmtomCtmtt 'fyjjif which interests to favor among competing ones. The policies or norms that. ,m drive these judgments ate difficult to state clearly and difficult to apply precisely to the circumstances at hand. *1 The statutes that delegate responsibility to the agencies for making these judgments provide certain limitations, along with some measure of guidance for decision making. The statutory formulations differ in many particulars, but certain themes and basic alternatives emerge deafly.4 * ' -^ i' First, virtually all the statutes ate precautionary; l.e., they direct the ,, agencies to act on the basis of uncertain, suggestive indications that a substance is dangerous. Because many substances can be shown definitively to be harmful only after serious harms have already occurred, these statutes reject the view that control measures must await proof of actual, past harm. On the balancing problem there is substantial division among dm statutes. Most of the statutes require agencies to weigh health and environ- i r:; mental concerns against economic considerations, but several statutes pro- hibit agencies from considering anything but health factors. Depending upon which approach is used, the results of regulation will differ dramat ically. Because of the severe results of the health-only rule, die prohibition is acknowledged more often in the breach than in the observance. ' The toxic substances control statutes also differ in the allocation of the 'burden of persuasion. Mpd place the burden of showing that a substance , 'should be regulated on me government. Several important statutes', how ever, place the burden of showing that a substance should be allowed in commerce on the proponent of use. The difference in the burden of persua sion can affect fine judgments at the margin, such as where an agency is deciding between alternative levels of exposure. Finally, the statutes call for searching judicial review of agency deci sions, although ultimately they leave to the agencies a large measure of discretion. Although the statutory formulations regarding standards of judi cial review vary, the courts are developing consistent principles for scrutinizing decisions. Through these devices the statutes give some guidance and offer some control, but they leave the agencies a wide range of legitimate choices. This Article contends that these legal tools are not instruments of finely-tuned > control, and that the agencies and advocates waste considerable energy in unproductive attempts to draw from the statutory differences fine (Untinctions in the agencies* obligations. Ultimately, regulations emerge from the interplay of available facts, the advocacy of interested parties and the predispositions of the agencies. The dynamics of this process are best seen in operation, as in the VC case study. Before analyzing the actions of the agencies in the regulation of VC, it is necessary to examine the central policy problems of uncertainty and balancing that the agencies have had to 1 . face. These problems are considered in the next section. 9 PROBLEMS IN TOXIC SUBSTANCES CONTROL! DBCIDINO UNDER ^Uncertainty and balancing incommensurable interests ordinal problems are endemic to any scheme for regulating i that cause cancer or other long-term, serious health or environeffects of relatively low probability. First, all decisions must be pibadhtaider substantial uncertainty about the medical and ecological risks, difficulties, and economic costs associated with different egrtea of exposure. Second, all decisions involve trade-offs among groups at that are not readily comparable. These two problem* form p t^boundaries of analysis" that prevent regulatory agencies from Attract, objective, and noncontroversial decisions.11 This section i the problems of uncertainty and balancing generally, as a preface to slfcpiocxtion of agency response to the VC hazard, and to the discussion Nhe legal framework within which the agencies make decisions. '^Cancer is the primary adverse effect of VC, and the disease is the p||chemcal hazard most on the public mind. For these reasons, this section ~ explores the nature of scientific uncertainty in the management of hazardous ^substances through the example of cancer risk assessment. This section also ` ~ * ^discusses die technological and economic uncertainties of estimating the rand cost of toxic substances control. Finally, the section examines f flf determining what risk-benefit trade-offs ore acceptable to* i and to the society as a whole. Only with an understanding of the boundaries of our scientific, technological, economic, and ethical knowl edges in ft possible to evaluate fairly the analytical efforts and normative JNhojbai Of the agencies that have regulated VC. . A. The Limits of Cancer Risk Assessment ni-'Cancer is a group of illnesses characterized by the unrestrained multiplication of cells that somehow have lost an essential self-regulatory raechanism.a The uncontrolled growth of these cells eventually threatens the life i.of the host organism. Presently, cancer is the second leading cause of death J in the United States.14 One American in four is expected to contract some if .type of cancer,15 and one American in five is expected to die of it.16 .. 32. The quoted phrase is taken from the title of a study of water project planning and, I'' tnofstpasttculariy, from an article on similar issues in cost-benefit analysis in that context. Set I A Peiveson. Benefits and Costs, Winners and Lasers in Boundaries op Analysis: XINQUUtY INTO TOE Tocks Island Dam ContiioverSy 123.144 (H. Feiveson, F. Slnden, 4R. iobw ads. 1976). ,, See Council on Environmental Quality, Sixth Annual Repost 13 (1973) (ch. I, Wf'CdAttnotens In the Environment) [hereinafter cited as CEQ Sixth Annual Report]. For ^'expttnatiorts of the essentials of cancer directed to the lay reader, see M. SHtMKIN, Science ' ANtrCANCER 1-6. 43-34, 87-98 (1973) [hereinafter cited as Science and Cancer]: Cairns, The Cancer Problem, Scientific American, Nov. 1973, at 64 [hereinafter cited as The Cancer - "AoWrm). ; ' 24. CEQ Sixth Annual Report, supra note 23, at 9. table 2. Id. at 12. 26. The Cancer Problem, supra note 23, at 66. 418 Toxic Substances Control Most forms of cancer are difficult or Impossible to cure; lets than one- htlfof all cancer patients survive longer than five years from the discovery of their illness.77 The elusiveness of cures largely is due to the fact that cancer's basic biological mecjuuvsm&at the cellular level are not well 'understood.1* ;, > --U' The causes of cancer are, however, somewhat better understood than the cures. Studies of cancer incidence in particular groups have shown strong statistical connections, between exposure to certain chemical sub stances and particular cancers. The connection between tobacco smoke and lung cancer is the most widely known.19 Markedly elevated cancer rates are also found among certain occupational groups in the United States and in other highly industrialized countries.30 Cancer rates are elevated where air and drinking water are contaminated with industrial organic chemicals.31 In general, cancer rates are higher than average in American urbanized areas.31 From comparisons of different rates of different cancers throughout the world, the World Health Organization and other prominent institutions and individual experts have concluded that 60 to 90 percent of all human cancers are caused by exposure to chemical substances (and, to a lesser extent,. radiation) present in our air* workplaces, food, water, and die rest of our environment.33 The causal relationships underlying the statistical connections observed in humans have been confirmed for many substances by controlled experi Ht ments on animals. With one possible exception, all substances related to cancer in humans have been shown to cause cancer in animals.34 In addition, animal experiments have implicated 1,500-2,000 other chemical substances; 27. Cancer Patient Survival, Ref. No. 5, U.S. Dep't Heaith, Educ., A Wei.f. Pu. No. (N1H) 77-992, at 3 (1976). 2R. The Cancer Problem, supra note 23, at 72. 29. See Hammond, Tobacco in Persons AT Hioh Risk of Cancer: An AfmoaCH TO Cancer Etiology and Control 131 (J. Fraumeni ed. 1975) [hereinafter cited ai Persons at Hioh Risk op Cancer]. 30. CEQ Sixth Annual Retort, supra note 23, at 23-26: Cole A Goldman, Occupation in Persons at Hioh Risk of Cancer, supni note 29, at 167, 31. EPA, Interim Drinking Water Regulation,i: Control of Organic Chemical Contami nants In Drinking Water, 43 Fed. Reg. 5756, 5758 (1978), citing National Academy of Sciences, Drinking Water and Health (June 1977); Pike, Air Pollution in Persons at Hioh Risk of Cancer, supra note 29, at 215. 32. - CEQ Sixth Annual Retort, supra note 23, at 19: Hoover, Mason, McKay, A Fiaumeni, Geographic Patterns af Cancer Mortality In the United States in Persons at Hum Risk OF Cancer, supra note 29, at 343-44 A table I, at 345. 33. CEQ Sixth Annual Refort, supra note 23, at 17; Higginton. Importance of Bnvbonmeutal Factors In Cancer in Environmental Pollution and Carcinogenic Risks 15.17 (C. Roacnfeld ft W. Davis, eds. 1975) (hereinafter cited as Environmental Pollution); Boyland. The Correlation of Experimental Carcinogenesis and Cancer In Man, 11 Progress M Experimental Tumor Research 222, 223 (1969); Epstein, Environmental Determinants af Unman Cancer, 34 Cancer Research 2425 (1974). See also R. Doll, Prevention of Cancer; Pointers from Epidemiology (1967) [hereinafter cited as Prevention of Cancer]. 34. CEQ Sixth Annual Retort, supra note 23, at 30-32. The apparent exception Is arsenic. pmrUn.Dtmltar -j* po6*atul human carcinogens.33 Many of these substances are synthetic chemicals that have been in commercial use only since the 1930s. cancer is a latent disease that typically manifests itself only 15 to 40 exposure begins, it is too eariy to know the effects of chemicals `hhawbeun in widespread use for only this short period. dvidence suggests that cancer rates could be cut significantly by exposure to the disease's chemical causes. Even though it -ttiylM^uite expensive, preventing human exposure to carcinogens often is arnoreeffective and economically efficient method of reducing cancer rates ton attempts to cure patients who already have the disease.36 In order to make to best use of the resources available to prevent cancer, precise data which substances are carcinogenic and on how dangerous they are at sariout levels of exposure would be helpful. Unfortunately, the causal relationship between a chemical and cancer is often difficult to establish. 'Bwn where a qualitative relationship is visible, precise quantitative esti- risks to humans cannot be made reliably particularly for low risks j&inth^nfder of one case in 10,000 or more subjects.37 I#tbe first place, not enough is known about how chemical carcinogens `.'especially at the cellular level. There is general agreement that the cause Changes in the genetic material of an individual cell or in through which the genetic material controls a cell's behav- thduclng It to multiply wildly.3* There is uncertainty and disagreement >*totor;Only one such "hit" need occur or whether a certain sequence of Jepnadant hits by the same or different substances is needed.39 Further |toOrtafaity ttems from the complexity of cellular metabolism--the system1 Irif'ChWBical and physical processes that occur within a living organism. ffipinfapS differ on whether there are chemical reactions that detoxify certain .^IjAiountt of a carcinogen by converting it into a harmless substance, or that ^repair genetic changes after they have occurred.40 The metabolic "path way" of a substance from its point of entry (e.g., lungs, skin, or digestive i fyttem) to its point of damage is also often uncertain.41 35. Occupational Safely and Health Administration, Identification, Classification and gulMhm af Toxic Substances Posing a Potential Occupational Carcinogenic Risk, 42 Fed. 54,141 (1977), . 36 Schaddermn. Sources. Resources, and Tsouris in Persons at High Risk of CIR, sstpra note 29, it 451, 452, 459. generally Schneiderman, Mantel, A Brown, From Mouse to Man--Or How to Get I the Laboratory to Fork Avenue and S9th Street, 246 Annals N.Y. Acad. Set. 237. 243 3) Qiiraianfler dud ns From Mouse to Man], 3a. J, SCIRMCE AND Cancer, supra note 23. at 45-54, 87-92. I.Jfe1 IVt'ijSPF generally Mantel A Schneiderman, Estimating "Safe" Levels, a Hazardous Underrating. 35 Cancer Research 1379 (1975) [hereinafter cited as Estimating "Soft" UueD], l 40* . Sm Cornfield, Carcinogenic Risk Assessment, 198 Science 693 (1977) [hereinafter id as Carcinogenic Risk Assessment]., 41. Set, t.g.y Watmnabe A Gehrfng, Dose-Dependent Fate of Vinyl Chloride and fts ; Fosslhle Relationship to Oftc0f*/t/city In Rats, 17 Envt'l Health Perspectives 145 (1976). Toxic Subttancaa Control 1 Whether or not there ape,tleftiiise or repair mechanisms has profound implications for strategies |w eaqeaf; pneyentioii. If the "onpMt'*jnodeMs accurate, and if there are jMrffetoxificatton, repair, or other defense meehih -# X- nisms, then as little as one moleculeof a carcinogenic substance, interacting / with the appropriate portion of, the susceptible cell, can cause an irreversible cancer. If multiple hits by different substances are needed, any one sub stance alone may not be,,carcinogenic (or may be only weakly so) .but , ,V together these substances may be potent causes of the disease. If detoxifies-. Y, V tion or repair mechanisms or other defenses exist, there may be safe dosesrr "thresholds"--below which no cancers will be caused. More important * than the question of whether a threshold exists is the question of what risks to expect from a range of doses. Different propositions about cancer causa tion lead to different conclusions about the rate of cancer to expect from each dose.41 The second major source of uncertainty is a result of the limitation* of available research techniques. Current methods for investigating the car cinogenicity of substances do not permit the verification or disproof of alternative theories of cancer causation. The methods are themselves also the subject of great controversy. Observation from direct human experiences is of limited utility. Purposeful experimentation on humans is ethically unacceptable, since the results often would be fatal. Human evidence of carcinogenesis usually comes from observation of occupational groups ex posed, often unwittingly, to chemicals in the industrial economy.41 Some connections can be drawn in the general population, but for the most part humans are exposed to tqo many different substances at unknown doses for unknown periods to permit statistically reliable conclusions to be draw*.4^ .j>< Moreover, there are synergistic and antagonistic interactions between chant- > teals that drastically complicate drawing conclusions about the effects of `: each chemical. Finally, because latency periods run 15 to 40 years or 42. Some assumptions lead to the conclusion that threshold doses exist and that the risk ai doses approaching the threshold declines to zero. See Kotin, Dose-Response Relationship and Threshold Concepts, 271 Annals N.Y. Acad. Set. 22,25-27 [1976). See alio Cardssogenic Risk Aitfiinnt, iupra note 40; DNA Repair: New Clues to Carcinogenesis, 200 SctSNCS 518 (1978). Other models decline to decide whether there are thresholds. One major model has dose and response in a logarithmic relationship, with risk declining more rapidly than dote at tow doses. This model yields higher risks for given doses than those which posit thresholds. Sen Estimating "Soft" Levels, safest note 29. Other researchers argue that the dose-response relationship at low dose* is likely to be linear, i.e., that decreases in risk we probably proportional to reductions of dose; This approach yields a risk for a liven dose higher than the risks estimated by the other two approaches at least at low doses. See Crump, Hoel, Langley, A Peto, Fundamental Carcino genic Processes and Their Implications /or Low Dose Risk Assessment, 36 Cancer RmtaCH 2973 (1976) [hereinafter died as Fundamental Carcinogenic Processes], The issues are summarized briefly in National Academy op Sciences, Principles roe Evaluating Chemicals in the Environment 86-88 (1975) [hereinafter cited as Principle* roe Evaluating Chemicals). 43. CEQ Sixth Annual Report, supra note 23. Rt 23-26. 44. Id. at 26-28. < .Desttgar g definitive studies of effects on humans are impracticable.41 ^ rodents are the major source of data on, the carcinogenicity ,Their response characteristics are considered essentially yio those pf humans, so that a substance carcinogenic to one is likely `-*v':nogemc to the other.47 But although the qualitative inferences are 1,.there are limitations on the ability of the animal tests to indicate i magnitude of human risks. It is difficult both to detect small risks in test Valid to translate risks for animals into risks for humans. difficulty in detecting small risks is statistical in nature. For fri/>rjpracj$cal and financial reasons, nearly all experiments on animals involve small.numbers of subjects, usually no more than a few hundred.41 In so * Wia^a group, a chemical must cause an effect at a relatively high rate for i i: the relationship to be confidently distinguished from random occurrences of jjWjjppjine event.49 The dose of a substance that induces cancer at rates ^f.^|jflet^pble in such tests is often far higher than most people experience.10 ^^l^icri.ues**on *s Aether lwer doses cause cancer, and at what rates. ' " '*' Hj(VIfJsSect occurring at a very low rate stands a good chance of not being . in so small a test group, so that the failure to observe an effect in test is not a reliable indication of the substance's safety for a larger 51 Thus, no test has confirmed the existence of any threshold or B r. J, detoxification mechanism or resolved any other basic aspect of the theoreti- If '' } Wl controversies discussed above.51 tSiA . ': ----------------- --------------------------------------------------------------------------------------------- ---- -----------------4). Id. If the effeci is munition raiher than cancer, the effect will not manifest itself for OWf.V RKM't gCIWTRlkHIS. For I genenil description of (he nature of epidemiological evidence and of the utet and of such research, see Prevention op Cancer, supra note 33, at 15-29. CEO Sixth Annual Report, supra note 23, at 28-32; In re Shell Chem. Co., 6 ERC w. ffiPA. FIFRA Docket. 1977); Principles for Evaluating Chemicals, supra note (135-39; T. Loomis. Essentials of Toxicology 206 (2d ed. 1974) [hereinafter cited a* iBNttALS of Toxicology). r-T, % CEQ Sixth Annual Report, supra note 23, Rt 30-32; Estimating Safe Levels, supra `;`f `' ,x'df4. See From House to Man, supra note 37, On the trade-off between the sensitivity and Mf tests, see Bates. Laboratory Approaches to the Identification ofCarcinogens, 271 Ann. N-YJ-AcaD, Sci. 29, 30-32 (1976). For an explanation for laymen of the statistical issues, see W. Lowrance, Of 'Able Rout 60-64 (1976) [hereinafter cited as Of Acceptable Risk], pi, This is not always the case, however. Some carcinogens are potent enough to cause CWNpt fat experimental animals when administered at the dose levels to which people have 64eft exposed. See text accompanying notes 133-146 Infra. , Ji. Three examples illustrate the point. In a test involving 100 animals each in an axperimwaal end e control group, if no tumors are detected in either group, there is a 1.096 chMlqa that the teal rate of cancer is as high as 4.596. With 1,000 animals in each group and no luarorl. theta remains a 1.0% chance that the real rate is as high as 0.46%, or 4.6 animals Out of each 1.000. If only ten animals are used in each group and the results show no turnon, the potential error increases drastically: there is a 1.0% chance that the real rate is as high as 37.096. See Or Acceptable Risk, supra note 49, at 62. 52. For a clear discussion of ihese points. sec Saffiotti, Comments on the Scientific Basis for the "Delaney Clause," 2 Prfventivf Medicine 125 (1973) [hereinafter cited as Comments on the Scientific Basis for the "Delaney Clause'']. See also World Health Organization, Assessment of the Carcinogenicity and Mutagenicity of Chemicals. Technical Report Series, No. 546, at 9-11 (1974): Estimating "Safe" Levels, supra note 42. at 1382-83. Toxtc Substances Cotetrel il To investigate the effecH of low doses directly would requite experts nrenfi'lhvolvlng enormoblitibers of animals. To demonifWiA'?5 perriml'confident* that dose of Just one subsinie*li#ftiW dt^'ore cancer In a nuUldlrtil&jeiH* would require a test involving at least i!ri* million animals. Sudl^frtejs-mousd" experiments generally are con sidered impracticably expemVe and vulnerable to laboratory Crrors that caa destroy the statistical reliability of the results.53 ir f Limited to observation* at unrealistically high doses in Unrealistically low numbers, the scientist's recourse is to use mathematical models of doseresponse relationships to extrapolate from experimental results downward to the effects of low doses. However, like the theories on which they are based, the models yield Widely divergent estimates of the risk associated with each low dose. The extent of the differences is astounding. For /example, the major models differ by a factor of 100,000 on the size of the dose that creates a risk of one canCer in a million subjects.5543The models dd provide credible outer limits for the risk associated with each dose,55 and Vsi they do permit the ranking of carcinogens in rough order of their potency, But they cannot provide the regulator with precise estimates of the risks of low doses. ia,, y,L , ' 'si More uncertainty is added to risk estimates by our ignorance of how to translate dose-response data across species lines. There simply is not enough ' known to determine if humans are more or less sensitive to a given dose of a carcinogen than the test animals.56 Several new techniques for assessing carcinogenicity are developing, but these do not yet hold out the promise of yielding quantitative risk estimates or of answering the basic questions about how cancer is caused. There are "quick" tests--such as the Ames test--of chemicals' abilities to mutate bacteria or other single-celled organisms. There is a high correlation between the ability to cause such mutations and carcinogenicity.57 *Current ly, however, the value of the "quick" tests is primarily qualitative; they * the agency does not know how great a difference in risk is ^ changes in dose. If there is a large.difference,then email r;||t 'allowable exposure will have a'sigmficant impaction knd must be considered carefully. If the difference is small, are not so important, and extensive efforts to obtain /with a small change in a standard might not be worthwhile. problems of risk assessment are aggravated to some fegreffor regulatory agencies by their incomplete access to information. Moct.taticological research is carried out or sponsored by the industries that qikaihe.rerket the substances being evaluated; industrial researchers have I'^ICeafhrea to withhold negative information or to perform poorly designed pKl encuted experiments incapable of revealing negative information.56 To aoma'extent, this behavior can be controlled by the use of standard test ,,1-,'iprotoodU and other means.59 The problem of unequal access to data and of vilneentlVes to misinform or misrepresent is more serious with regard to the ; ^wtrtmumof costs.60 faftum; because the nature of chemical carcinogenesis is unknown, and available research techniques are limited in their ability to predict ilHskS from exposure to carcinogens, the only conclusion that may be complete certainty is that no level of exposure to a chemical that in animals is sure to be safe. Neither experimental nor gj^Jbeore^ical analysis can give the agencies precise estimates of the risks ,y.j'ttoaociaied-with low doses of substances that are known to cause cancer in |;lmmuu*Cjr animals at higher doses. Nor can regulators be sure how sensitive /^ffska are to changes in dose. At present, the best available techniques yprodtiea only broad estimates of the outer limits of risk. The Limits of Economic and Technological Assessment The costs of controlling exposure to a toxic substance are shrouded in may be able to distinguish strong from weak carcinogens, but cannot give more precise risk estimates. One important consequence of the uncertainty about the size of small risks is that a regulator agency does not know the marginal risk at each U g if* 53. Estimating "Soft" LevsIs, supra note 42, at 1383. See also From Mouse to Msm, supra note 37, at 241. 54. Carcinogenic Risk Assessment, supra note 42, at 694. 55. For example, Schnetderraan and his colleague! estimated on the basis ot the animal experiments on VC completed by May IS74 "that a dose as low as 1 ppm [part per million] is almost certain to have a risk of leas than I in 10,000" for animals. From Mouse to Man, supra note 37, at 241-42. 36. See Rail, Problems of Low Doses of Carcinogens, 64 J. Wash. Acad. Set. 63 (1974). 57. McCann A Ames, A Simple Method for Detecting Environmental Carcinogens as Mutagens. 271 Annals N.Y. Acad, Set. 5 (1976). See also Note. From Microbes to Men: The New Toxic Substances Control Act and Bacterial MutagenlcUy/Carclnogenidty Tests, * Envt'l L. Re?, 10,243 (1976). ,,-( 'iijl SKlSee National Research Council, Decision Making in the Environmental Pro- 34-57 (,977) [hereinafter cited as Decision Marino in EPA). - W- As an example of standard protocols for cancer testing, see Sonlag, Page, & Saffiotti, ChndeBaas for Carcinogen Bioassay in Small Rodents. National Cancer Institute, Car.cbiosMiaais Technical Report Series No. I (1976), summarized In Shubik A Clayson, Applica tion ofme Results ofCarcinogen Bioassays to Man in Environmental Pollution, supra note 343- See also Principles for Evaluating Chemicals, supra note 42, at 134-55. On* other M for controlling the quality of data is to certify laboratories that meet mdards for their performance. See Decision Making in EPA, supra note 58, at 54- addition, there are penalties in certain of the regulatory statutes for a company's Bfea," ,ii|hitepniaentiai or withholding toxicological data Currently, the Food and Drug Administra- llW ; tWvoloptai such certification procedures as part of "Good Laboratory Practice" regula- ' - rions, god the auency has referred at least one case involving misrepresentation of data to the - Justice Department with a tecontmendation for criminal prosecution. See Creative Penmanship Pn,mPl* FDA Controls. 198 Science 1227 (1977). In addition, Velsicol Qumicil Corporation has been indicted for withholding data from EPA regarding animal cancer tent on heptechlor/chlordane. See Indictment Charges Velsicol, Six Persons, Withheld (Mordant, HtptacMordane Data, I BNA Chem. Reg. Rep.--Curr. Rep. 1413-14 (1977). 60. See text accompanying notes 67-71 infra. O If