Use of Comparative Risk Methods in
Regulatory and Common Law

Michael S. Baram*

I. Introduction ........................................ 1

II. Problems in Risk Decisionmaking ................... 2

III. Proposals to Use Comparative Risk .................. 5

IV. Comparative Risk for Setting Regulatory Priorities... 8

V. Setting Risk Limits in Regulatory Actions ............ 13
A . Background .................................... 13
B. Agency Efforts .................................. 15
C. Comparative Risk Concepts in Use .............. 20
D. Judicial Review ...... : .......................... 23
E. Problem s to Solve .............................. 33

VI. Evidentiary Use in Common Law Litigation .......... 37
A. Background .................................... 37
B. Expert Testimony ............................... 38
C. Rules of Evidence and Judicial Concerns ........ 45

VII. Conclusions ......................................... 49

I. INTRODUCTION

Several persistent problems afflict risk decisionmaking. In the
regulatory context, agencies confront the problems of how to pri-
oritize risks for best use of their limited resources and how to de-
termine "how safe is safe enough," or a risk limit, when action is
to be taken on a particular risk.

In the trial courts hearing toxic tort actions, the jury must often
determine whether an activity is "unreasonably dangerous" or a
product is "defective" because of its risk attributes.

To resolve these problems, many have proposed the use of risk
comparisons. Now that we can quantify risks, why not compare
them when regulatory priorities and risk limit issues are raised in
the agencies and courts, in order to facilitate decisionmaking?

* Adjunct Professor and Director of Law and Technology Center, Boston University

Law School; partner, Bracken and Baram, Boston, MA. The author wishes to acknowledge

support for this article by the National Science Foundation and Decision Research, Inc.

(grant number SES-8796182).



2 COLUMBIA JOURNAL OF ENVIRONMENTAL LAw [Vol. 13:1

This Article discusses these proposals, reviews cases in which
risk comparisons have been offered or relied on to deal with the
problems and evaluates this experience from a legal and policy
analysis perspective. Several issues are identified which obstruct
wider acceptance of risk comparisons, and recommendations are
presented for resolving these issues.

II. PROBLEMS IN RISK DECISIONMAKING

Decisions for controlling technological risk are made in two
legal domains: regulatory law and common law.

In regulatory law, decisions are made to prevent future risks
and reduce existing risks in accordance with statutory authoriza-
tions. Thus, legislatures select which categories of risk merit reg-
ulatory action and enact statutes to authorize agency regulatory
efforts. Agencies, in turn, select specific risks within the author-
ized category for regulatory action and then set standards or
make licensing decisions to control these risks consistent with the
statutory mandate. Courts then review those agency decisions
which are appealed (the judicial review function) and either affirm
or invalidate the agency decisions.'

In the common law, decisions are made to abate existing risks
and compensate for harms. State and federal courts hear com-
mon law actions brought by persons who claim personal injury or
property damage and seek compensation, injunctive relief and
other remedies as a result of the risky activities or products of
another. These actions are premised on tort theories of negli-
gence, strict liability and products liability, and on various con-
tract and breach of warranty theories. 2

Several persistent problems plague risk decisionmaking in both
domains. In regulatory law, a critical issue is how to determine
which risks deserve priority for action and the allocation of lim-
ited regulatory resources. This issue may arise at several levels:
(1) when the legislature is faced with several categories of risk as

1. See generally J. FIKSEL, M. BARAM, L. Cox & J.R. MIYARES, PRINCIPLES FOR USE OF DE
MINIMIS CONCEPTS IN RISK REGULATION 18-32 (1984) (report to the National Science
Foundation describing the role of statutory language in the regu.latory process) [hereinaf-
ter FIKSEL & BARAM] (copy available by writing Joseph Fiksel at Teknowledge, 1850 Em-
barcadero Road, P.O. Box 10119, Palo Alto, Cal. 94303). See Public Citizen v. Young, 831
F.2d 1108 (D.C. Cir. 1987), infra note 43, for a particularly relevant example.

2. See generally G. NOTHSTEIN, Toxic TORTS: LITIGATION OF HAZARDOUS SUBSTANCE

CASES §§ I I & 12 (1984) (discussing theories of liability and remedies).



Comparative Risk

candidates for statutory enactment of a regulatory program and
must choose which categories merit legislative action and fund-
ing; (2) after the legislature has chosen a particular risk category
for enactment, when it seeks to provide language in the statute to
direct the agency on how to choose which of the numerous risks
within the selected category deserve priority of agency attention
and resource commitment; and subsequently, (3) when the
agency seeks to use its limited resources in the most effective
manner and thereby seeks to prioritize the many specific risks
which fall within its statutory mandate, in order to choose which
risks it will act on. Since legislative directions to the agency are
often incomplete or ambiguous as to prioritization, the agency
usually has considerable discretion in structuring its prioritization
of risks. 3

A second critical issue in regulatory law is how to determine the
"risk limit," which is the maximum extent to which a selected risk
should be reduced. Most statutes provide agencies with direc-
tions as to the process and criteria to be used in regulating risks,
but few provide complete guidance on how stringently to control
the risks. As a result, in most cases the "risk limit" issue is left to
agency judgment and judicial review, and the outcomes embod-
ied in standards and licensing decisions are variable in their strin-
gency. Thus, an agency may set different risk limits for each of
several toxic chemicals under the same statutory mandate; and
the same toxic chemical, subject to regulation by several agencies
(e.g., Environmental Protection Agency, Occupational Safety and
Health Administration, Consumer Product Safety Commission)
acting independently under different statutory mandates, may be
controlled to differing risk limits and lead to differential protec-
tion of exposed persons (e.g., community residents, workers, and
consumers). 4 This lack of uniformity in setting risk limits creates
an aura of arbitrariness about agency decisionmaking and proba-
bly promotes controversy and appeals for judicial review. 5

3. For discussions of these and related matters, see generally Comparative Risk Assessment:

Hearings Before the Subcomm. on Science, Research and Technology of the House Comm. on Science and

Technology, 96th Cong., 2d Sess. (1980). See infra notes 25-34 and accompanying text.

4. Hattis, Goble & Ashford, Airborne Lead: A Clearcut Case of Differential Protection, 24 ENVI-
RONMENT 14 (Jan/Feb. 1982) (discussing the disparity of health protection afforded by

EPA and OSHA regulation of airborne lead).

5. See FIKSEL & BARAM, supra note I, at 10 1-33 for general discussion as well as detailed

analyses of agency decisionmaking on risk limits under six statutes. See generally W.D.

ROWE, AN ANATOMY OF RISK 3 (1977), for a discussion of"three major questions:" (1) what

1987]



4 COLUMBIA JOURNAL OF ENVIRONMENTAL LAw [Vol. 13:1

Finally, in the common law, there is a counterpart "risk limit"
issue: how to determine if an activity has been unreasonably risky,
or if a product is "defective." 6 In other words, what level of fore-
seeable risk must be shown to prove defendant's negligent or
other tortious behavior or breach of product warranty? In each
state common law system, these are generally viewed as factual
issues for jury determination, or for the presiding judge in non-
jury cases. But theories of liability, precedents and rules of evi-
dence which govern the risk testimony of expert witnesses differ
in each state. Further, jury and judicial attitudes and understand-
ing of risk matters are variable. As a result, court decisions reflect
a diversity of views about risk limits and, in the view of many risk
experts and other observers, these decisions also reflect irrational
jury expectations that technological activities and products be vir-
tually risk-free. 7

These problems of risk decisionmaking can therefore be char-
acterized as follows:

(1) prioritization of risks for regulatory action, a function of
legislatures and agencies;

(2) setting risk reduction limits, a function of legislatures,
agencies and courts.

Various methods have been proposed and used to address
these problems, chiefly the "rational" methods of engineering ec-
onomics such as cost-benefit and risk-benefit analysis. Proponents
have included Presidents Carter and Reagan and their Offices of
Management and Budget, various agencies (e.g., Nuclear Regula-
tory Commission), industry and economists generally.8 But use
of these techniques requires the arbitrary monetization and dis-
counting of unquantifiable health and environmental considera-
tions and other methodological problems, yields a range of

is risk, (2) how may different types of risks be compared for making risk decisions and
(3) how can acceptable levels of risk for specific activities be determined.

6. See generally W. KEETON, D. DOBBS, R. KEETON & D. OWEN, PROSSER AND KEETON ON
THE LAW OF TORTS §§ 78, 99 (5th ed. 1984) (abnormally dangerous things and activities,

dangerously defective or unsafe products) [hereinafter PROSSER AND KEETON ON TORTS];
G. NOTHSTEIN, supra note 2, § 1 I (theories of liability).

7. Determinants ofjury behavior is an extensively studied subject. See generally Research

on Juries, II THE JUSTICE SYSTEM JOURNAL 1 (1986) and Hinchcliff, Portrait of a Juror: A
Selected Bibliography, 69 MARQ. L. REV. 495 (1985/1986).

8. See, e.g., Baram, Cost-Benefit Analysis: An Inadequate Basis for Health, Safety and Environmen-
tal Regulatory Decisionmaking, 8 ECOLOGY L.Q. 473, 479-80 (1980). See also OFFICE OF TECH-
NOLOGY ASSESSMENT, U.S. CONGRESS, TECHNOLOGIES FOR ASSESSING CANCER RISKS

SUMMARY at 19-20 (1981) (noting support for increased use of cost-benefit analysis).



Comparative Risk

"justifiable" results without specifying any precise ordering of
risks or precise limit for any risk, and conflicts with various statu-
tory mandates and public values. 9 As a result, these methods
have been widely rejected by the public and in several instances
by the courts.' 0

Others have looked beyond these "rational" or quantitative
methods to more qualitative concepts which may appeal to com-
mon sense, be more amenable to public understanding and prove
more consistent with the statutory mandates for regulation: the
concepts of de minimis risk and comparative risk. Both concepts
are now being considered, and in some instances are being used,
in regulatory and common law proceedings. Since uses of the de
minimis risk concept have recently been evaluated elsewhere," l

this paper focuses on the comparative risk concept and its poten-
tial for solving the problems of prioritizing risks and setting risk
limits.

III. PROPOSALS TO USE COMPARATIVE RISK

As its name implies, the comparative risk concept provides that
the importance of a particular risk can be better understood and
communicated if it is compared to other risks which are already
known or familiar.

Over the last decade, academics and risk analysts have put forth
the concept of comparing risks, along with tabulations of numer-
ous risks to facilitate the comparisons, for several purposes: to
advocate certain technologies over others, 12 to illuminate the ar-
bitrariness of government efforts at risk reduction,' 3 to improve
public perception and understanding of risk,14 to foster public ac-

9. See, e.g., Baram, supra note 8, at 526-27.
10. Id. See also American Textile Mfrs. Assoc. v. Donovan, 452 U.S. 490, 506-22 (1980)

(rejecting argument that section 6(b)(5) of the Occupational Health and Safety Act re-
quires OSHA to perform cost-benefit analysis).

11. See FIKSEL & BARAM, supra note 1. See also Mumpower, An Analysis of the De Minimis
Strategy for Risk Management, 6 RiSK ANALYSIS 437 (1986).

12. See E. CROUCH & R. WILSON, RISK/BENEFIT ANALYSIS 103-63 (1982) (nine examples
of government or private studies using risk analysis to compare, among other things, the
efficacy of various auto safety devices, the risks and benefits of coronary artery surgery
versus more conservative treatment and the risk of conventional copper mining versus
mining assisted by nuclear explosions); Inhaber, Risk with Energy from Conventional and Non-

conventional Sources, 203 SCIENCE 718 (1979) (comparing risks to human health from five
conventional and six nonconventional energy-producing methods).

13. See Kletz, What Risks Should We Run?, 74 NEW SCIENTIST 320 (1977).
14. See Slovic, Informing and Educating the Public and Risk, 6 RisK ANALYSIS 403 (1986).

1987]



6 COLUMBIA JOURNAL OF ENVIRONMENTAL LAW [Vol. 13:1

ceptance of "expert" approaches to risk,1 5 to develop a profes-
sional consensus as to a systematic approach for agency risk
analysis 16 and to stimulate public concern and self-help
measures. 17

To support their arguments, these proponents have compared
diverse risks expressed in various formats, including: volun-
tary/involuntary risk of death per person per year, accident fre-
quency rates, estimated loss of life expectancy, increased chance
of death in any year, public perceptions, acute oral toxicities of
various substances and various consequence scales. The tabulated
risks cover a broad spectrum from smoking to parachuting to coal
mining to death from being struck by meteorites. The risk esti-
mates are usually provided without reference to their sources or
discussion of the variable quality of the studies and diversity of
assumptions which produced the numbers.

In recent years, the concept has been put to use in several ad-
versarial contexts (e.g., in the agencies and courts), either as part
of the advocacy strategy of the proponents of a risky technology
seeking less stringent regulatory decisions, or as support offered
by regulatory officials to defend an intended decision from critics.
For example, in agency standard-setting proceedings on particu-
lar risks, industry advocates have put forth risk comparisons to
argue that the levels of regulation for these risks should not be
more stringent than the levels tolerated or accepted for other
risks. Risk analysts and other experts are frequently hired to tes-
tify on risk comparisons in these proceedings.

Inevitably, these advocacy uses have generated criticisms about
the objectivity and validity of the comparisons offered, the qualifi-
cations of the experts and the relevance of the concept. Some
critics range more broadly to deal with larger issues of profes-
sional responsibility, the elitism and disdain for public concerns
inherent in risk analysis and its advocacy uses, and the need to
reinforce informed public consent and social choice in a techno-
cratic society. Thus, Otway damns the whole enterprise:

15. See Keeney & von Winterfeldt, Improving Risk Communication, 6 RISK ANALYSIS 417

(1986).
16. Lawless, Jones &Jones, Methodsfor Comparing the Risks of Technologies, in RISK EVALUA-

TION AND MANAGEMENT 157 (V. Covello, J. Menkes &J. Mumpower eds. 1986).
17. For example, recent EPA efforts to publicize radon hazards. See Radon: Threat Is

Real, but Scientists Argue Over Its Severity, N.Y. Times, Sept. 2, 1986, at CI, col. 1.



Comparative Risk

"Acceptable risk" was a numbers game in which we tried to de-
fine quantitative criteria by which the social acceptability of
risks and, implicitly, of technologies could be judged. These
studies were inspired by the fact that, contrary to the expecta-
tions of technical experts, the public were not convinced by
probabilistic risk calculations that their concerns about new
technologies were groundless. The methods used ranged from
primitive comparisons of estimated risks with dissimilar risks
accepted in everyday life, to those involving sophisticated
mathematics, eccentric philosophies, or both. Public fears were
often ridiculed as being "irrational" ....

Later research showed there are other, objective characteris-
tics of risk besides death and injury which matter to people
(such as voluntariness, control, delay, catastrophic potential),
and that it is perfectly normal to care about them. The view of
decision-making implicit in acceptable risk studies could be
called technocratic, elitist or maybe just "perfect-world" analy-
sis, but it did nothing to further democratic process because
the judgment of acceptability was seen as a matter for risk ex-
perts-that we could tell people what was best for them.18

Others have sought to redeem comparative risk by evaluating
its potential for improving risk analysis, communication with the

public and social choice. For example, Slovic argues that:

Doing an adequate job of communicating means finding com-
prehensible ways of presenting complex technical material that
is clouded by uncertainty, and is inherently difficult to
understand .... 19

The right of citizens, patients, and workers to be informed
about the hazards to which they are exposed from their daily
activities, their medical treatments, and their jobs, provides the
motivation behind much of the efforts to communicate infor-
mation about risks .... [T]here is a need for a deeper under-
standing of the concept of consent as well as for a theory of
informed consent that sets out criteria for evaluating the ade-
quacy of information presentations .... 20

... Perhaps we can learn, by studying people's understand-
ing of commonly used measures, such as distance, time, and
speed, whether and how their understanding of quantitative
risk can be improved. 2 1

18. H. Otway, Experts, Risk Communication and Democracy 3 (Nov. 9-12, 1986) (key-
note address, annual meeting of the Society for Risk Analysis, Boston, Mass.) (copy avail-
able at Columbia Journal of Environmental Law).

19. Slovic, supra note 14, at 403.
20. Id. at 411.
21. Id. at 412.

19871



8 COLUMBIA JOURNAL OF ENVIRONMENTAL LAW [Vol. 13:1

Given these problems and opportunities, the comparative risk
concept is now evaluated to determine if it can be responsibly
used to solve the persistent problems of risk decisionmaking-
prioritization of risks for regulatory action and the setting of risk
limits in agency and judicial proceedings.

IV. COMPARATIVE RISK FOR SETTING REGULATORY PRIORITIES

A regulatory agency must comply with its statutory mandate.
The numerous health, safety and environmental statutes contain
diverse provisions and directions but commonly delegate four es-
sential functions to the agencies for implementation:

(1) to identify and select hazards for regulatory management;
(2) to make the necessary findings of fact about these hazards
in order to support regulatory actions (e.g., using risk analysis
and other means);
(3) to evaluate the available regulatory options; and
(4) to choose and define the regulatory option and the risk
limit to be achieved.

The comparative risk concept is of obvious relevance to the se-
lection of hazards for regulation, the first function, and the set-
ting of the risk limit, the fourth function (to be discussed in a
subsequent section). In selecting hazards for regulation, compar-
ative risk can be used as a method for screening a multitude of
hazards that have been identified, and thereby enable an agency
to prioritize these hazards according to their respective risk at-
tributes, and select some for regulatory action on the basis of
their risk significance.

At the outset, it should be recognized that such prioritization
on the basis of risk attributes will be a rough estimation, since the
agency has not done risk analysis and has a very incomplete data
base at this early point in the regulatory process.

Another condition that should be noted at the outset is that use
of a comparative risk approach for screening and prioritization
may involve matching the hazards and their risk attributes against
other known or familiar risks, but will more likely involve a varia-
tion on this theme-the comparing of each identified hazard and
its risk attributes without reference to any tabulation of any other
risks. Thus, the comparative risk approach for prioritization is
transformed into a rank ordering of identified hazards and risk
attributes for agency use in choosing its regulatory priorities.



Comparative Risk

Certain statutes restrict this first regulatory function by pre-
cluding agency choice of which hazards to regulate and thereby
make any use of comparative risk methods moot. This is done by
those statutes which require agencies to review all new product
applications, as in the case of new drugs, food additives, pesti-
cides and toxic chemicals. By eliminating agency discretion to
choose which products to review for risks and possible standard-
setting, these laws preclude use of regulatory prioritization meth-
ods such as the comparing of risks. 22

Other types of statutes also restrict the first regulatory function
by setting the agency's priorities, either mandating or emphasiz-
ing agency action on several specific hazards. For example, the
Clean Air Act requires the Environmental Protection Agency
(EPA) to list and regulate "ambient air pollutants" and "mobile
source" pollutants with strong legislative history and statutory
language emphasizing EPA action on several specific pollutants
designated by Congress. 23

Nevertheless, most statutes do not restrict the first regulatory
function and afford agencies broad discretion to prioritize risks
for regulatory actions, whether by comparison of risk or other
methods. For example, other sections of the same Clean Air Act
call for EPA listing and regulation of "hazardous air pollutants"
and "stationary sources" of air pollution, without narrowing EPA
discretion and choice and without providing any statutory obsta-
cle to EPA's use of the comparative risk concept. 24

Given the discretion afforded by most statutes for choosing
among a universe of hazards for regulatory action, agencies have

22. For example, the Federal Insecticide, Fungicide and Rodenticide Act (FIFRA) provi-
sions on pesticide registration at 7 U.S.C. § 136a (1982 & Supp. IV 1986).

23. The Senate subcommittee report accompanying the 1970 amendments to the Clean
Air Act, 42 U.S.C. §§ 7401-7462 (1982 & Supp. III 1985), identified several pollutants as
broad threats to national health and recommended EPA regulation by means of national
ambient air quality standards under § 108, 42 U.S.C. § 7408 (sulfur oxides, carbon mon-
oxide, hydrocarbons, photochemical oxidants, fluorides, nitrogen oxides, polynuclear or-
ganic matter, lead and odors). I W. RODGERS, ENVIRONMEN rAL LAw: AIR AND WATER § 3.7
at 240-41 (1986). Section 7422 of the Act designates radioactive pollutants, cadmium,
arsenic and polycyclic organic matter for EPA to consider regulating under section 7408
(national ambient air quality standards) or section 7412 (hazardous air pollutant stan-
dards). Section 7521 designates carbon monoxide, hydrocarbons and nitrogen oxides for
EPA regulation as moving source pollutants.

24. See 42 U.S.C. §§ 7411(b) (stationary sources), 7412 (hazardous air pollutants)
(1982).

19871



10 COLUMBIA JOURNAL OF ENVIRONMENTAL LAW [Vol. 13:1

devised and used various types of screening or prioritization
methods. Some of these methods involve risk comparisons.

In Congressional hearings on Congressman Don Ritter's pro-
posed Comparative Risk Assessment Bill, 2 5 representatives of
several agencies were asked the same questions: "Considering
that your resources are limited, how do you determine the prior-
ity by which issues are addressed in the regulatory decision pro-
cess? Is risk assessment or any other method used in this ranking
process?"

26

The Occupational Safety and Health Administration (OSHA)
set forth its framework for comparing cancer risks and setting its
regulatory priorities, in which the following factors are consid-
ered in ranking the cancer risks:

(1) The estimated number of workers exposed;
(2) The estimated levels of human exposure;
(3) The levels of exposures to the substance which have

been reported to cause an increased incidence of neoplasms in
humans, animals, or both;

(4) The extent to which regulatory action could reduce not
only risks of contracting cancer but also other occupational and
environmental health hazards .... 27

Dr. Bailus Walker, for OSHA, explained that the resultant "pri-

ority list will obviously determine only the approximate order in
which rulemaking shall proceed," and asserted that given techni-
cal uncertainties, "any priority list ... cannot involve a strict nu-
merical ranking scheme." 28 He explicitly rejected the view of
several other witnesses before the hearing committee-that cost
factors (instead of health risk) be compared in prioritizing for reg-
ulatory action-as being impractical given data and resource con-
straints and the controversial, inaccurate nature of cost estimates
and comparisons which "would produce numbers that are so
highly speculative as to be essentially meaningless for comparison
purposes." 2 9 Walker also expressed OSHA's opposition to using

cost factors for prioritization on other grounds as well: that

25. H.R. 4939, 96th Cong., 2d Sess. (1980).
26. See, e.g., Comparative Risk Assessment: Hearings Before the Subcomm. on Science, Research and

Technology of the House Comm. on Science and Technology, 96th Cong., 2d Sess. at 437, 506, 519,

551 (1980) (responses by representatives of Food and Drug Administration, Occupational
Safety and Health Administration, Consumer Product Safety Commission and Environ-
mental Protection Agency) [hereinafter Comparative Risk Hearings].

27. Id. at 506.
28. Id.
29. Id. at 507.



Comparative Risk

OSHA must deal with many types of health risks besides cancer,
and that such ranking methods leave unanswered how one quan-
titatively evaluates different types of health risks (e.g., mortality,
morbidity) on a relative basis. 30

The Consumer Product Safety Commission (CPSC) system was
described by Dr. Peter Preuss.

We consider several factors in setting priorities: frequency and
severity of injuries; causes of these injuries; the relative cost
and consumer benefits of [CPSC] actions; the vulnerability of
those likely to be injured; the possibility of chronic illness or
future injuries; and probability of exposure to the risk.

These criteria are used to prepare our annual priority list

There are many questions that remain unanswered, which
make comparative assessment of risk extremely difficult. Even
if we are able to assess different types of risks accurately, we
would still have difficulty deciding among them. The choices to
be made could be, for example, among: (a) 0.1 percent
probability of electric shock; (b) 75 percent probability of sev-
ered fingers; (c) 0.001 percent probability of cancer; (d) 50 per-
cent probability of fire . . .(e) 0.00001 percent probability of
birth defects . . .or (f) 100 percent probability of a bruised
knee.31

Preuss also pointed to other problems with risk comparisons,
including the uncertainties, variable assumptions used and the
agency's view that consumer risks which arise from convenience
and luxury products should be treated differently than risks aris-
ing from other types of products. He concluded that "we see risk
assessment and comparative risk assessment as tools which can be
used along with many others to reach regulatory decisions. We
do not see them as ultimate determinants, nor even as tools that
are useful in all cases at all times." 3 2

Other agencies also presented their prioritization methods, in-
cluding the Food and Drug Administration (FDA) and the Envi-
ronmental Protection Agency (EPA). FDA's elaborate scheme for
comparing risks to set priority rankings was notably different
from that of other agencies in that it incorporated the views of
consumer and other public interest groups, thereby expanding
the concept of comparative risk assessment to include public per-

30. Id. at 507-08.
31. Id. at 511-12.
32. Id. at 512.

1987]



12 COLUMBIA JOURNAL OF ENVIRONMENTAL LAW [Vol. 13:1

ceptions and attitudes about risk. 33 In stark contrast to FDA's de-
tailed and sensitive approach was EPA's evasiveness and failure to
respond to the Committee's questions on how it set its regulatory
priorities, evincing its lack of any policy for establishing priori-
ties.3 4 EPA is now attempting to remedy this situation by con-
ducting a "Comparative Risk Project" for ranking risks subject to
its regulatory jurisdiction, with the intention of using the results
to set its funding and resource priorities.3 5

Although these agency systems have undoubtedly been modi-
fied since 1980, when they were described to the House Subcom-
mittee, the major issues raised at that time about the usefulness of
comparing risks to set priorities have not been publicly resolved
and presumably persist today.

These issues include (1) technical uncertainty about the health
risk attributes, since in-depth regulatory research and risk analysis
(with public comments and industry inputs) have not been done
at this early stage of regulation-thereby militating against any
quantitative comparison and rankings which would be relatively
conclusive or defensible; (2) inability to establish a common valu-
ation scale or system for priority-ranking different types of risks
with their different probabilities of manifestation and their differ-
ing health significance attributes, such as pain and suffering, job
disability, hospitalization, impairment of function, death, etc.;
(3) societal values and attitudes about risks, which may confound
purely quantitative approaches for ranking risk by incidence and
severity, such as the voluntariness of exposure, the importance of
the source of the risk to the person at risk, the negligent or willful
behavior of the risk generating organizations, and whether the
availability of self-help or compensatory options to persons at risk
should diminish agency concern for the risk; and (4) the technol-
ogy-forcing functions required by certain statutes which are not
related to any ranking of health risks.

33. Id. at 437. FDA's elaborate project priorities and methods are detailed at 438-89.
34. See id. at 542-51.
35. [Current Developments] Env't Rep. (BNA) 2203 (April 11, 1986). The project pro-

duced a report in February 1987. See EPA, UNFINISHED BUSINESS: A COMPARATIVE ASSESS-
MENT OF ENVIRONMENTAL PROBLEMS OVERVIEW REPORT (1987); E.P.A. Report Says Agency

Focuses on Lesser Problems, N.Y. Times, Feb. 19, 1987, at B6, col. 5.
The agency also began re-examining the hazard ranking system used to develop the

National Priorities List under the Comprehensive Environmental Response Cleanup and
Liability Act. [Current Developments] Env't Rep. (BNA) 707 (Sept. 12, 1986).



Comparative Risk

Finally, as the testimony of industrial officials and economists
ardently advocated, risks could also be compared and prioritized
by their economic attributes. Although this approach would pro-
vide a common metric (dollars) for valuing each risk and for esti-
mating the cost-effectiveness of regulatory actions, it has been
repudiated by OSHA, FDA and CPSC spokesmen on several
grounds: it would load arbitrary dollar valuations of different
types of risks onto uncertain health risk estimates; and it would
transform Congressional mandates for health risk reduction and
technology-forcing into mandates for minimizing the costs of risk
regulation.

From the foregoing, it appears that agencies not constrained by
statutory language develop several methods and criteria, includ-
ing risk comparison, to establish regulatory priorities but are re-
luctant to rely on risk comparison as a conclusive criterion, using
it only as one of several methods for eventually setting their regu-
latory priorities, or for later defending their choices of risks.

V. SEITING RISK LIMITS IN REGULATORY ACTIONS

A. Background

Agencies face the risk limit issue in two decisionmaking con-
texts. In the standard setting context, the agency has selected a
risk for regulatory action, measured it, considered several regula-
tory options and their economic and other implications, and fi-
nally confronts the issue of the risk limit-the stopping point for
its risk reduction standard.

In the permit or product approval context, the agency has eval-
uated a petition or application for approval of an activity (e.g.,
construction of a power plant) or product (e.g., a pesticide) and is
deliberating about the terms and conditions to be imposed on the
applicant. Some of these terms will be designed to control the
risk level of the proposed activity or product and, in some in-
stances, the agency can do this by simply incorporating existing
standards it has already set in other proceedings. Often, how-
ever, no generic standards are suitable or available for incorpora-
tion in the permit as limiting conditions, and the agency must
then grapple with the problem of setting ad hoc or special risk-
limiting terms to conclude its decision process.

Surprisingly little is known about how agencies actually set risk
limits in these two regulatory contexts, despite the volumes of

1987]



14 COLUMBIA JOURNAL OF ENVIRONMENTAL LAW [Vol. 13:1

published materials on uses of cost-benefit and risk-benefit analy-
sis in regulatory decisionmaking. According to Rodricks:

Although there have been numerous studies of... most ele-
ments of the risk assessment-risk management process, at least
one element appears to have escaped detailed analysis: the de-
termination of whether a given predicted risk poses a signifi-
cant threat to the public health and of the extent to which risk
reduction is needed to achieve public health protection.3 6

But this void is now being filled. Several policy and legal analy-
ses of agency methods for establishing what constitutes a "signifi-
cant risk" have been published recently.3 7 These studies have
been stimulated by the Supreme Court's 1980 decision which in-
validated OSHA's revised benzene standard because the agency
had not established that benzene, already subject to an earlier
OSHA standard, still posed a "significant risk" to workers and
merited the more stringent revised standard.3 8

Several studies have also evaluated the "de minimis risk" con-
cept and its proposed and actual uses in agency decisionmaking
to set risk limits.3 9 "De minimis risk" has been defined as a risk
level so trivial that it does not merit regulatory attention. 40 The
concept has therefore been proposed as a means for determining
what is not a significant risk and has been vigorously promoted by
the FDA as a rationale for approving the use of carcinogenic food
and cosmetic additives (e.g., use of methylene chloride for coffee

36. J. Rodricks, S. Brett & G. Wrenn, Significant Risk Decisions in Federal Regulatory
Agencies I (unpublished paper) [hereinafter Rodricks] (copy may be requested in writing

from ENVIRON Corporation, 1000 Potomac Street, Washington, D.C. 20007).

37. Cross, Beyond Benzene: Establishing Principles for a Significance Threshold on Regulatable

Risks of Cancer, 35 EMORY L.J. 1 (1986). See also Latin, The "Significance"of Toxic Health Risks:
An Essay on Legal Decisionmaking under Uncertainty, 10 EcOLOGY L.Q. 339 (1982).

38. Industrial Union Dept., AFL-CIO v. American Petroleum Institute, 448 U.S. 607
(1980).

39. See FIKSEL & BARAM, supra note 1; Mumpower, supra note 11; Cross, supra note 37.
See also Spangler, The Need for De Minimis Risk Standards in Regulatory Decisionmaking: An Indi-

vidual or Societal Risk Concept? in ENVIRONMENTAL HEALTH RISKS: ASSESSMENT AND MANAGE-
MENT 205 (R.S. McColl ed. 1987); M. Spangler, A Summary Perspective on NRC's Implicit
and Explicit Use of De Minimis Risk Concepts in Regulating for Radiological Protection in
the Nuclear Fuel Cycle (unpublished paper) (copy may be requested in writing from Miller
B. Spangler, Special Assistant for Policy Development, United States Nuclear Regulatory
Commission, Washington, D.C. 20555).

40. Alabama Power Co. v. Costle, 636 F.2d 323, 360 (D.C. Cir. 1968). See infra note 56
and accompanying text.



Comparative Risk

decaffeination) 4 1 which pose a lifetime risk of one in a million or
less for cancer.4 2 This FDA attempt to avoid the zero cancer risk
mandate of its governing statute's "Delaney Clause" by formula-
tion of a de minimis risk approach has recently been denied by a
federal Court of Appeals as being contrary to the strict statutory
language of the Delaney provision. 43

B. Agency Efforts

"Significant risk" and "de minimis risk" concepts essentially
represent the two sectors of any particular risk, the former being
the sector to be regulated or prevented, the latter being the sec-
tor to be unregulated or permitted. Because of differing statutory
nomenclature, some analysts and agencies use the terms "unac-
ceptable risk" and "acceptable risk" as synonyms for "significant
risk" and "de minimis risk," respectively.44 Thus, any particular
risk would accordingly be divided into two sectors, as follows:

Regulatory
Risk Limit

to permit by to prevent by
regulatory inaction regulatory action

0 --------------------------------- >
De Minimis or Risk Level Significant or
Acceptable Risk Unacceptable Risk

41. See Methylene Chloride: FDA ' De Minimis Tested, ENVIRON REPORT 2 (Fall 1986) (copy
may be requested in writing from ENVIRON Corporation, 1000 Potomac Street, Washing-
ton, D.C. 20007).

42. See Cooper, Stretching Delaney Till It Breaks, 9 REGULATION 11 (Nov./Dec. 1985).
43. Public Citizen v. Young, 831 F.2d 1108 (D.C. Cir. 1987). Petitioners challenged the

FDA's decision to list two color additives, Orange No. 17 and Red No. 19, based on quan-
titative risk assessments indicating that cancer risks presented by these dyes were trivial.
The decision hinged on construction of the Delaney Clause for color additives, 21 U.S.C.
§ 376(b)(5)(B), one of three Delaney Clauses in the Food, Drug and Cosmetic Act.

44. The term "unreasonable risk" is used in section 2605(a) of the Toxic Substances
Control Act, 15 U.S.C. §§ 2601-2654 (1982 & Supp. IV 1986), and is yet another synonym
for significant or unacceptable risk. For a discussion of the factors considered in deciding
whether a chemical poses an unreasonable risk, see J.C. DAVIES, S. GUSMAN & F. IRWIN,

DETERMINING UNREASONABLE RISK UNDER THE Toxic SUBSTANCES CONTROL ACT (1979)
(The Conservation Foundation, Washington, D.C.).

1987]



16 COLUMBIA JOURNAL OF ENVIRONMENTAL LAW [Vol. 13:1

Agencies have not coordinated their respective approaches, nor
have they acted with internal consistency, in defining these two
risk sectors and determining where the risk limit should lie. In
defining the significant or unacceptable risk sector, agencies are
directed by differing statutory criteria and therefore reach differ-
ing outcomes. But in defining the acceptable (or de minimis) risk
sector, agencies have recently adopted various simplistic formula-
tions, such as a 1 X 106 lifetime risk of cancer, to establish the
risk limit. These formulations have been made without the gui-
dance of any statutory criteria. No statutes directly define de
minimis risk, probably because Congress assumed that this risk
sector would be defined as a consequence of agency determina-
tion of significant risk in accordance with statutory language.

This results in a confusing set of outcomes across the several
agencies and within each agency. EPA, for example, operates
under the inconsistent demands of numerous regulatory pro-
grams. According to Cross:

Unfortunately, to date the various federal health and safety
agencies have adopted only a haphazard, disorganized re-
sponse to both the Benzene decision and the Court's rationale.
The result has been an inconsistent pattern of regulation which
may be questioned on both legal and public policy grounds.
When agencies have used a significant risk cutoff to guide regu-
lation, they have been unable to select even a consistent mea-
sure of risk. 45

For example, Rodricks reports de minimis determinations for
lifetime cancer risk by the FDA of one in 14,000 (PCBs in fish)46

and one in one million (1 X 10-6) (animal drug residue, methylene
chloride decaffeination under Delaney clause);47 EPA pesticide
determinations usually at 1 X 10-6 but running to 1 X 10- when
benefits were large, and to levels beyond 1 X 106 when benefits
were seen as negligible;48 EPA carcinogenic air pollutant determi-
nations usually at 1XIO-' but running to 1XI - ' for ra-
dionuclides and benzene; 49 EPA drinking water pollutant

45. Cross, supra note 37, at 3-4 (footnotes omitted).

46. Rodricks, supra note 36, at 6.
47. Id. at 4.
48. Id. at 6-7.
49. Id. at 10-12; see also [Current Developments] Env't Rep. (BNA) 616 (Aug. 10, 1984)

(EPA table of cancer risks for airborne toxics).



Comparative Risk

determinations at zero exposure;50 and EPA hazardous waste
cleanup determinations at Superfund sites ranging from 1 X 10-7

to I X 10-4.51

This disarray has been independently confirmed by Cross:

The preceding discussion demonstrates the tremendous vari-
ance in agencies' determinations concerning de minimis risk
levels that do not warrant regulation. The Consumer Product
Safety Commission, for example, will regulate a risk of three
chances in one hundred thousand, while OSHA will accept
risks as high as eight chances in one thousand, or almost three
hundred times the threshold level regulated by the CPSC. Sim-
ilarly, the Nuclear Regulatory Commission will accept a risk
more than one hundred times greater than will the Food and
Drug Administration.... The EPA permits a residual risk from
uranium tailings that is more than a thousand times the risk it
allows under the Clean Water Act. The risk level regulated by
the EPA under a single statute, FIFRA, has varied by two or-
ders of magnitude in recent years ...

... [T]he differences observed in practice are not based upon
any explained distinction in statutory mandates, but rather ap-
pear haphazard .... In sum, at the present time, there appears
to be no coherent significant risk policy in federal regulatory
agencies. 52

This haphazard or ad hoc approach to defining significant and
de minimis risk sectors, or risk limits, damages agency credibility
and makes the whole enterprise look as arbitrary as earlier agency
uses of cost-benefit analysis on an ad hoc basis. 53 Clearly, a prin-
cipled approach, embodied in a formal, generic policy, for making
both significant and de minimis risk determinations is needed.
Such an approach would provide for internal consistency within

50. Rodricks, supra note 36 at 12; see also [Current Developments) Env't Rep. (BNA) at
976 (Oct. 24, 1986) (risk from drinking water set at zero).

51. See Rodricks, supra note 36, at 12.
52. See Cross, supra note 37, at 43-44. Contra Travis, Richter, Crouch, Wilson & Klema,

Cancer Risk Management, 21 ENVr'L Sc. & TECH. 415, 420 (1987), who find "the history of
federal decision making indicates that all agencies are fairly consistent in their implicit
definitions of ... de minimis risk levels." They note that agencies always regulate chemi-
cals posing an individual risk of death greater than 4 X 10' and, with the exception of one
FDA decision, never regulate chemicals posing a risk of less than I X 106. Id. at 418. They
also find some correlation between the size of the population exposed and the de minimis
risk limit. See id. For example, for small exposed populations, regulatory action is never
taken for risks below I X 10" , but for exposures affecting the entire U.S. population, the de
minimis risk level is I X 10"6. Id.

53. See, e.g., Baram, supra note 8, at 519-20 (both EPA and NRC have applied cost-bene-
fit analysis inconsistently).

1987]



18 COLUMBIA JOURNAL OF ENVIRONMENTAL LAw [Vol. 13:1

each agency and foster inter-agency consistency to the extent per-
mitted by the differing statutory mandates of each agency. 54

Several federal court decisions have promoted agency develop-
ment and use of methods for setting risk limits in regulatory deci-
sions. For example, in Alabama Power Co. v. Costle,55 Judge
Leventhal considered EPA's prospective exemption of certain cat-
egories of stationary sources of air pollution from its regulatory
requirements aimed at prevention of significant deterioration of
air quality. Judge Leventhal, a noted proponent of agency rules
of reason, stated his view that:

Categorical exemptions may also be permissible as an exercise
of agency power, inherent in most statutory schemes, to over-
look circumstances that in context may fairly be considered de
minimis.... [T]he law does not concern itself with trifling mat-
ters, and this principle has often found application in the ad-
ministrative context. Courts should be reluctant to apply the
literal terms of a statute to mandate pointless expenditures of
effort .... The ability.., to exempt de minimis situations from a
statutory command is not an ability to depart from the statute,
but rather a tool to be used in implementing the legislative
design.

Determination of when matters are truly de minimis naturally
will turn on the assessment of particular circumstances, and the
agency will bear the burden of making the required showing.
But we think most regulatory statutes, including the Clean Air
Act, permit such agency showings in appropriate cases. 56

The court thus accepted the de minimis risk rationale, although it
found the particular exemptions made by EPA in the matter being
reviewed to be invalid.

54. For recent recommendations as to a generic approach, see FIKSEL & BARAM, supra
note 1; Cross, supra note 37. Traditionally, the Nuclear Regulatory Commission has sur-
passed all other agencies in setting forth generic principles and policies. Statutory lan-
guage may preclude agency adoption of a de minimis risk approach. Public Citizen v.
Young, 831 F.2d at 1108 (Delaney Clause for color additives does not permit FDA adop-
tion of de minimis risk method).

55. 636 F.2d 323 (D.C. Cir. 1979).
56. 636 F.2d at 360-61 (footnotes omitted). See also Monsanto v. Kennedy, 613 F.2d

947, 954 (D.C. Cir. 1979) (FDA has power under de minimis doctrine to exempt from
regulation acrylonitriles that diffuse in trivial amounts into foods). Cf Environmental De-
fense Fund v. EPA, 636 F.2d 1267, 1283 (D.C. Cir. 1980) (disallowing de minimis excep-
tion under Toxic Substances Control Act for materials containing less than 50 parts per
million PCB when far smaller concentrations shown to be toxic); Aqua Slide'n Dive v.
Consumer Product Safety Commission, 569 F.2d 831, 842 (5th Cir. 1978) (argument that
extremely remote risk of paraplegia from swimming pool slides is a "reasonable risk;"
court invalidates CPSC regulation requiring warning signs on slides as not "reasonably
necessary" to reduce the risk).



Comparative Risk

One year after Alabama Power Co. v. Costle, the Supreme Court
drove the point home by invalidating OSHA's revised benzene
standard because the agency had not demonstrated that the stan-
dard would prevent a significant risk. The plurality opinion in
this case, Industrial Union Department, AFL-CIO v. American Petroleum
Institute,57 thereby stands for the proposition that OSHA is re-
quired to make a threshold finding of fact that a significant risk is
present in the work place before the agency can promulgate a
valid standard for reducing this risk. 58 In accordance with the sig-
nificant risk/de minimis risk dichotomy discussed earlier, the
Supreme Court has essentially required OSHA to factually estab-
lish that any risk it intends to regulate is not a de minimis risk. As
a practical matter, other agencies must now heed this proposition
unless their authorizing statute requires zero risk regulation or
otherwise precludes agency determination of a significant risk
level on a discretionary basis, as in the case of the FDA's Delaney
Clause.

No court has yet offered clear guidance to the agencies as to
any quantitative measure for marking the division between de
minimis and significant risks, although in the benzene case, the
Supreme Court did go so far as to opine that the dividing line or
risk limit for a lifetime occupational cancer risk lies somewhere
between 1 X 10- and 1 X 10'. 59 Rodricks has found that since
the date of this decision, "OSHA has notjudged any occupational
carcinogenic risk to be clearly insignificant, but has not sought to
force predicted lifetime risks below ca. 10- 3. It appears that, in
least in principle [sic], OSHA is prepared to find some level of
occupational risk insignificant." 60

57. 448 U.S. 607, 642 (1980).
58. Cf Public Citizen v. Young, 831 F.2d 1108, discussed supra note 43.
59. Id. at 655. The Court of Appeals in Public Citizen v. Young also deliberately

avoided setting a "dividing point between de minimis and other risks." 831 F.2d at 1112
n.4.

60. Rodricks, supra note 36, at 19. Rodricks believes that this risk limit may be due to
several factors in addition to the Supreme Court's ambiguous placement of the risk limit as
being somewhere between I X 10-' and 1 X 10 -'; namely that a risk of 1 X 10' is "low
compared to other fatality hazards in jobs commonly thought of as 'safe,' " id. at 14, and
that the "feasibility criteria" in OSHA's statute for reducing risk militate against setting
tougher standards. Id. at 17. See Public Citizen Health Research Group v. Tyson, 796
F.2d 1479, 1502 (D.C. Cir. 1986): "OSHA... found that EtO exposure at I ppm (12-23
excess deaths estimated [per 10,000 workers]) poses a significant risk itself. The agency
set the PEL [standard] at I ppm, however, not because no excess deaths would occur at
that level, but because it could not show that any lower ... limit would be feasible."

19871



COLUMBIA JOURNAL OF ENVIRONMENTAL LAW [Vol. 13:1

Thus, the problem of developing a principled approach to de-
fining the risk limit remains a challenge for the federal agencies-
whether in the context of a specific decision process or in the con-
text of developing a consistent agency or inter-agency policy.

The comparative risk concept has recently been proposed, and
in some instances used, by agencies to support risk limit determi-
nations in various regulatory proceedings. This concept, and the
opportunities and problems it presents, will now be discussed.

C. Comparative Risk Concepts in Use

At this time, no federal agency has set forth a formal regulatory
policy for generic use of a comparative risk method to set consis-
tent limits in its regulatory proceedings. However, EPA has de-
veloped risk comparison data for several of its regulatory
programs, 6' discussed its potential uses and used risk compari-
sons to communicate with the public about new problems such as
radon.62 Two agencies, OSHA and the NRC, now regularly use
risk comparisons as one of several supports for the risk limits they
set in regulatory proceedings.

OSHA now considers five factors in establishing whether a risk
is significant, according to a recent report: the quality of the un-
derlying data, the "reasonableness" of the risk assessment, the
statistical significance of the findings, whether the risk at issue is
material and the numerical significance of the risk relative to
other occupational risks. 63

Obviously, the fifth factor involves risk comparisons, which, ac-
cording to Rodricks, use as "benchmarks" the fatality rates in cer-
tain occupations compiled by the Bureau of Labor Statistics. 64

The fatality rates arise from many different types of risks (e.g.,
mining fatalities from explosions, gas, structural failures, flood-
ing, accidents, etc.). Thus, OSHA's consideration of the risk limit
to set for cancer from toxic chemicals in a particular type of work
place will be based on five factors, one of which will be a numeri-

61. See Summary of Data on Specific Pollutants and Cancer Risk Estimates Excerpted from EPA

Draft Study on Air Toxics Problem in United States, [Current Developments] Env't Rep. (BNA)

616-18 (August 10, 1984)
62. See Radon: Threat Is Real, but Scientists Argue Over Its Severity, N.Y. Times, Sept. 2, 1986,

at C1, col. 1; Environmental Authorities Must Increase Focus on Individuals to Bring Progress, Rus-

sell Says, [Current Developments] Env't Rep. (BNA) 917 (Oct. 17, 1986).
63. OSHA Official Outlines Agency's Steps in Deciding if Significant Risk Exists, 16 Occupational

Safety and Health Reporter (BNA) 599 (Nov. 5, 1986).
64. See Rodricks, supra note 36, at 15-16.



Comparative Risk

cal comparison of the cancer fatality rates associated with the
chemical with death rates arising in very different occupational
contexts from very different types of risks.

Rodricks describes the growing reliance of EPA and NRC on
these same occupational fatality rates for evaluating, by means of
comparisons, whether certain risks being considered for regula-
tion by EPA and NRC are significant.

EPA and the Nuclear Regulatory Commission (NRC), for ex-
ample, have proposed to set federal radiation standards using
as a yardstick the fatality rates prevalent in industries com-
monly considered to be relatively safe.

In its radiation protection proposal, EPA noted that "the risk
ofjob-related accidental death in the safest of all major occupa-
tional categories, retail trades, [was] an annual death rate [of]
60 per million workers in 1975." This risk equates to a 45-year
worklife risk of 2.7 in 1,000. The Agency based its proposed
radiation protection guidelines on its finding that radiation
risks of a magnitude similar to 3 in 1,000 "do not appear unrea-
sonably high" because "[tihey are comparable to risk of acci-
dental death in the least hazardous occupations."

In a similar vein, NRC's recent radiation protection proposal
follows the approach recommended by the International Com-
mission on Radiological Protection . . .which developed its
guidelines by "comparing [radiation] risk with that of workers
in industries ... which are recognized as having high standards
of safety." As NRC pointed out, in such " '[s]afe' industries...
average annual mortality due to occupational hazards does not
exceed 10- .... ." This annual rate amounts to a 45-year life-
time risk in excess of 4 in 1,000. Like EPA, NRC proposed
standards on the basis that occupational mortality risks due to
radiation are "acceptable" if kept at or below this "safe indus-
try" risk level. 6 5

The NRC has also published proposed rules and a policy state-
ment in 1986 which express its reliance on risk comparisons for
setting limits on its regulatory actions. 66 For example, in its pol-
icy statement on how it will make decisions on the exemption of
certain wastes from disposal in licensed low level radioactive
waste disposal facilities, NRC has set forth several criteria, includ-
ing one that provides "the maximum expected effective dose

65. Id. at 15, 17.
66. Radioactive Waste Below Regulatory Concern; Policy Statement, 51 Fed. Reg.

30839 (Aug. 29, 1986) (to be codified in 10 C.F.R. § 2 app. B.); Safety Goals for the Oper-
ations of Nuclear Power Plants; Correction and Republication of Policy Statement, 51 Fed.
Reg. 30028 (Aug. 21, 1986).

1987]



22 COLUMBIA JOURNAL OF ENVIRONMENTAL LAw [Vol. 13:1

equivalent to an individual member of the public does not exceed
a few millirem per year for normal operations and anticipated
events." 67 Discussion of this criterion is focused on a one mil-
lirem per year risk limit, which in turn is justified by a comparison
of this risk limit to various EPA standards (which establish a 25
millirem per year risk limit for radioactive air pollution and the
uranium fuel cycle), to natural background levels of human expo-
sure (100 to 120 millirem per year), to Federal Radiation Council
guidance (500 millirem per year) and to various British and Cana-
dian standards. 68 Thus, comparative risk will be used to support
NRC determinations of radioactive wastes which fall "below regu-
latory concern" because they pose insignificant or de minimis
risks.

NRC's policy statement on Safety Goals for the Operation of
Nuclear Power Plants69 is also replete with uses of comparative
risk to set risk reduction controls on reactor operation. After as-
suring the public "that no death attributable to nuclear power
plant operation will ever be 'acceptable' " and drawing the dis-
tinction that it "is discussing acceptable risks, not acceptable
deaths," NRC sets forth its "qualitative safety goals," including
one that "societal risks to life and health from nuclear power
plant operation should be comparable to or less than the risks of
generating electricity by viable competing technologies ...."70

NRC also provides its "quantitative objectives," which "are to
be used in determining achievement of [its] safety goals:"

-The risk to an average individual in the vicinity of a nuclear
power plant of prompt fatalities that might result from reactor
accidents should not exceed one-tenth of one percent... of the
sum of prompt fatality risks resulting from other accidents to
which members of the U.S. population are generally exposed.
-The risk to the population in the area near a nuclear power
plant of cancer fatalities that might result from nuclear power
plant operation should not exceed one-tenth of one percent...
of the sum of cancer fatality risks resulting from all other
causes.

71

The Commission believes that this ratio of 0.1 percent ap-
propriately reflects both of the qualitative goals-to provide

67. 51 Fed. Reg. at 30844.
68. Id. at 30845.
69. 51 Fed. Reg. 30028.
70. Id.
71. Id. at 30028-29.



Comparative Risk

that individuals and society bear no significant additional
risk.

72

In separate views, Commissioners Asselstine and Bernthal took
issue with various aspects of the NRC policy statement. Bernthal
focused on whether the NRC approach would provide the public
with "clear, unambiguous, practical safety objectives" that would
answer the question "how safe is safe enough."' 73 He concluded
that "the question remains unanswered" despite the comparative
risk approach used, because the risks are not viewed as compara-
ble by the public 74 and because the 0.1 percent goals do not in-
clude population density considerations. 75

D. Judicial Review

Only a few agency decisions based in part on risk comparisons
have been tested in the courts. Despite vigorous attacks on
agency uses of comparative approaches to establish risk limits in
these regulatory decisions, the courts have approved such uses of
risk comparison.

In San Luis Obispo Mothers for Peace v. NRC,7 6 petitioners argued
that NRC had acted arbitrarily in issuing low and full power oper-
ating licenses to the Diablo Canyon power plant. Petitioners fo-
cused in particular on NRC's refusal to hold a hearing to consider
earthquake events and their potential effects on emergency re-
sponse plans for public safety, 7 7 and argued that this refusal was
arbitrary and irrational in that NRC had considered other natural
phenomena, such as volcanos, tornados, hurricanes, fog and
heavy rain, which could also have potential effects on emergency

72. Id. at 30030.
73. Id at 30033.
74. "It is unrealistic for the Commission to expect that society, for the foreseeable fu-

ture, will judge nuclear power by the same standard as it does all other risks." Id.

75. [A] power plant could be located in Central Park and still meet the Commis-
sion's quantitative offsite release standard....

... [T]he Commission's standards should preserve the important principle
that site-specific population density be quantitatively considered ...e.g., by
requiring that for the entire U.S. population, the risk of fatal injury . . .should
not exceed some appropriate specified fraction of the sum of the expected risk
of fatality from all other hazards to which members of the U.S. population are
generally exposed.

Id
76. 789 F.2d 26 (D.C. Cir. 1986), cert denied, 107 S. Ct. 330 (1986).
77. Id. at 29.

1987]



24 COLUMBIA JOURNAL OF ENVIRONMENTAL LAw [Vol. 13:1

plans and public safety. 78 NRC's defense rested on the very low
probability of an earthquake initiating or occurring contempora-
neously with a plant accident so as to interfere with an emergency

response. 79 Earthquakes as a risk factor were therefore essen-
tially insignificant or de minimis.

A majority of the court held that petitioners failed to establish
that NRC's refusal to require emergency response plans to con-

sider earthquakes was arbitrary or irrational. The majority ac-
cepted NRC's logic that the agency should consider only those
natural phenomena that "frequently occur" at a particular site
and that the phenomena to be considered in licensing a particular
plant will accordingly vary from site to site (e.g., snow to be con-
sidered for a plant in Pennsylvania, but not for a plant in Flor-
ida).8 0 It then found sufficient support in the record of the
licensing proceeding for NRC's exclusion of earthquakes:

[I]t is sufficient that the record establishes that fog is 24,200
times more likely to occur, and rain 6,875 times more likely to
occur, at Diablo Canyon than is a major earthquake.... Under
these circumstances, the Commission certainly drew a rational
distinction between rain and fog, on the one hand, and earth-
quakes, on the other. Given the relative probabilities, this
court cannot conclude that the Commission's decision was arbi-
trary and capricious. 8 '

Fourjudges joined in vigorous dissent on several grounds: first,
that the NRC had not " 'articulated a satisfactory explanation for
its action including a "rational connection between the facts
found and the choice made,"' " particularly since the agency's
earlier "design proceedings centered on the plant's proximity to

an active earthquake fault;"8 2 second, that the majority had effec-
tively gutted the emergency planning regulations by "downplay-

ing the frequency of earthquakes by multiplying their probability
by the probability of a radiological accident" to arrive at a de
minimis frequency finding for earthquakes, instead of dealing
with the probability of the "complicating event" (the earthquake)

alone;8 3 and finally, that the NRC had not adequately defined

78. Id. at 40-42.

79. Id. at 37-40.

80. Id. at 42.

81. Id.
82. Id. at 47.

83. Id. at 50. See also discussion at 55 (citation omitted): "By automatically multiplying

the 1 in 100,000 chance of a nuclear accident by the likelihood of any natural hazard oc-



Comparative Risk

"frequently occurring" natural hazards, that "by failing to... de-
scribe where the cut-off point between frequently and infre-
quently occurring phenomena might lie-the Commission has
made it impossible to apply the standard to hazards other than
those specifically listed." 84

Similar results, affirming agency reliance on comparative risk
methods to make risk limit decisions, have been reached in Mas-
sachusetts courts. In Town of Brookline v. Commissioner of the Depart-
ment of Environmental Quality Engineering (DEQE),85 Brookline
challenged DEQE's decision approving operation of Harvard
University's "MATEP" power plant.

DEQE had made this decision after ten years of regulatory
hearings and litigation, during which DEQE and state courts had
ordered some thirty-five design and operational changes in the
MATEP proposal. These prior changes had considerably reduced
MATEP risks to human health in the Metropolitan Boston area;
but when Harvard submitted its revised plan to DEQE for final
approval, DEQE conducted a risk analysis and determined that
the plant's operation would still cause up to four cancer deaths in
the exposed population of 1.6 million people over the next forty
years. Thus, DEQE faced the ultimate problem of determining
whether this remaining risk of four lung cancer deaths was too
great or significant to permit its final approval of the modified
plant proposal.

The agency dealt with the risk limit issue in its final decision
process by comparing the MATEP risks with the magnitude of
other risks common in society. Using rough comparisons, DEQE
determined that the four death risk from the MATEP operation
was not unreasonable and approved the plant's operation on this
basis, prompting the Brookline appeal. Specifically, the DEQE
administrator's final decision addressed the issue: "is the risk
posed by MATEP after application of the best available control
technology reasonable?" and concluded:

I adopt the Department staff's approach of determining rela-
tive risk as the most sensible and appropriate method of ad-
dressing the issue. The Department in all of its regulatory
activities . . . is faced with making judgements [sic] based on

curring, the majority reduces all simultaneous occurrences to a 'never-never' land beyond
rational planning."

84. Id. at 54.
85. 398 Mass. 404, 497 N.E.2d 9 (1986).

1987]



26 COLUMBIA JOURNAL OF ENVIRONMENTAL LAW [Vol. 13:1

relative risks with or without the benefit of clear numerical
standards....

Using the Department staff's relative risk method .. .I con-
clude that a maximum increased risk of lung cancer of zero to
four excess cases in a population of 1.66 million persons over a
period of 40 years . . . is not unreasonable. 8 6

In its appeal to the state's highest court, Brookline argued,
among other things, that the agency had exceeded its statutory
authority in using a "reasonable risk" criterion and that DEQE
had also acted impermissibly under its statutory mandate by com-
paring MATEP risks to various risks due to other sources and ex-
posure circumstances in order to determine whether the plant's
operational risk was reasonable and whether the facility should be
approved.

87

Brookline's argument that DEQE had exceeded its statutory au-
thority by devising and applying a reasonableness test was based
on provisions of the state's law and regulations for implementing
the federal Clean Air Act. The state law and regulations provided
that DEQE disapprove a "new source of air pollution," if it was to
(a) cause a nuisance, (b) be potentially injurious to human life or
(c) unreasonably interfere with the comfortable enjoyment of
life. 88 The court denied this argument:

The Legislature has granted DEQE broad authority ...
DEQE thus is charged with evaluating the technical evidence
and reaching a decision on the risk attributable to the new
source. That decision includes a determination of the bound-
ary within which the risk will be acceptable .... "[wihat may be
injurious to life ... is best left to the DEQE to determine on a
case-by-case basis in light of the most current scientific
evidence.'"89

The court then proceeded to deal with the other issues "mindful
of [its] limited role of review .... 90

Brookline's argument that DEQE had acted impermissibly in
using risk comparisons to reach its final decision was among the
several issues then considered. According to Brookline:

86. Medical Area Total Energy Plant, Inc., Application 77-51 at 15-16 (DEQEJanuary 3,
1985) (final decision on remand after hearing).

87. 398 Mass. at 410, 497 N.E.2d at 12.
88. Id. at 413 n.12, 497 N.E.2d at 14 n.12.

89. Id. at 411, 497 N.E.2d at 13 (citation omitted).

90. Id.



Comparative Risk

DEQE's health expert, Dr. Halina Brown, brought into the pro-
ceeding [DEQE final hearing on MATEP] a comparative risk
assessment which was not suggested in her pre-filed direct tes-
timony. She sought to justify the four lung cancer deaths.., by
comparing these with existing environmental risks, namely, the
risk of contracting cancer from polluted public drinking water
and contaminated urban air. Although Dr. Brown conceded on
cross-examination that such a risk assessment is not permitted
under any environmental laws or programs, nonetheless, in
concluding that four lung cancer deaths are acceptable, she
compared these deaths with the number of potential deaths
which could result from ingesting pollutants in public drinking
water at levels allowable under EPA's standards.... Dr. Brown
recognized that providing drinking water is an activity signifi-
cantly different from MATEP's electricity generation, in that
providing water is an essential public service and standards...
for public water supply must take account of existing pollution
conditions, the technical feasibility of reducing the pollutants
... and associated costs, in order to [enable] a municipality to

continue to provide drinking water to its citizenry. In contrast,
the electricity from MATEP is unnecessary since its intended
customers can continue to obtain their electrical needs from
Boston Electric Company; hence the carcinogenic emissions
from MATEP can be avoided without any interruption in essen-
tial services.

... The Hearing Officer also cast doubt upon the propriety
of comparing such risks when she concluded:

"it is also true that one can certainly not conclude that
the toleration of certain level[s] of cancer risk from
drinking water, for example, implies that society has
tacitly accepted an equivalent risk from other sources
or activities."

The Hearing Officer pointed out to Dr. Brown that a prob-
lem with her comparative risk analysis is that "the dirtier the
area is to begin with, the greater the new sources of pollution
that would be allowed," a result at odds with both the federal
and state clean air acts ... [and] directly contrary to the statu-
tory mandate. Dr. Brown appeared to recognize the fallacy of
her approach by conceding that her comparative risk assess-
ment probably was a mistake, and was confusing because she
did not make decisions by comparing risks anyway. However,
she offered no other basis for her "decision" that the four lung
cancer deaths attributable to MATEP are acceptable ....

Thus, the Deputy Commissioner [of DEQE] followed unlaw-
ful procedures in determining that MATEP's predicted deaths

1987]



28 COLUMBIA JOURNAL OF ENVIRONMENTAL LAW [Vol. 13:1

are acceptable by comparing them to the number of deaths
caused by existing sources of pollution. 9 '

Brookline's lengthy argument against comparative risk also re-

lied on prior decisions of the Massachusetts courts which had de-
nied arguments that comparative risk should be used to establish
the risk limit in regulatory proceedings:

In reviewing the Massachusetts Department of Public Health's
promulgation of an emergency regulation banning the sale of
food products containing EDB in amounts greater than a desig-
nated level, this Court found that the Department properly re-
lied on its staff's view that "there is no safe level of exposure
for a cancer-causing agent" and that genetic changes which re-
sult in cancer can occur from only one "hit" of carcinogenic
substances. American Grain Products Processing Institute v. Depart-
ment of Public Health, 392 Mass. 309, 324-325 (1984). Nothing
in the record of the EDB rule-making proceeding showed that
actual deaths would occur if EDB existed in foods in quantities
above the proscribed level, rather the record showed that the
potential for such deaths existed. An expert for the opponents
of the EDB regulation attempted to trivialize the 1 ppb maxi-
mum allowable level set by the Department by asserting that
such a quantity is:

".... one inch in 16,000 miles, or roughly two-thirds of
the distance around the earth. One ppb is one second
in 33 years, or one minute in the time elapsed since
the birth of Christ. One ppb is one penny in ten mil-
lion dollars. One ppb is one teaspoon of fertilizer
spread evenly over a garden of 5,000 acres. One ppb
is equivalent to four drops of water in a filled Olympic-
sized pool (64,000 gallons). Or for those of you on a
low-carbohydrate diet, one ppb is one crouton in a
salad weighing 500 tons." 92

Brookline then proposed an alternative approach, based on tes-

timony from its own experts which would deny both comparative

risk and de minimis approaches:

MATEP's expert health witness, Dr. Jeffrey Harris, used the
same tactic to attempt to trivialize the risk of 4 lung cancer
deaths attributable to MATEP by asserting that the risk is
equivalent to the risk of cancer to a 40-year-old man who
smokes 1.4 cigarettes per year for twenty years. Although Dr.
Harris did not offer any evidence to substantiate his compari-

91. Brief for Appellants at 29-32, Town of Brookline v. DEQE, 398 Mass. 404, 497

N.E.2d 9 (1986) (SJC-4135) (citations omitted), filed by Bracken and Baram.
92. Id. at 38-39 (quoting American Grain Products Processing Institute v. Department

of Public Health, 392 Mass. 309, 330-331, 467 N.E.2d 455, 470 (1984)).



Comparative Risk

son, such a comparison is wholly inappropriate since smoking
is an avoidable risk which the residents of the MATEP impact
area can avoid if they choose not to smoke. The Hearing Of-
ficer appeared to recognize that comparison of an avoidable
risk with an unavoidable risk is inappropriate. Brookline sug-
gests that their expert witness, Dr. David Ozonoff, had the right
approach. He testified:

"If I have a preventable risk, then my wish is to prevent it
and not to compare it"

"If 99.5 percent of all cancers are caused by something
that I couldn't avoid, that wouldn't excuse my (sic) dealing
with the half a percent that I could avoid ..

"I think that our society has been fairly clear that there are
stricter standards for imposed risks than chosen risks." 9 3

Dr. Ozonoff's view was supported by Brookline's medical ex-
pert, Dr. Marvin Schneiderman, who, when asked on cross-exami-
nation by DEQE's General Counsel whether the plant should be
approved if the price for operation was the "human sacrifice" of
one person each ten years, replied:

So, four deaths is not so trivial. Is this hypothetical; are these
hypothetical deaths? If we knew specifically who was going to
die, we absolutely, certainly, would not permit it to happen....
It's only because we don't know who it is and we're fairly cer-
tain that it's not going to be ourselves-for what reason I'm not
so sure. But we know it's not going to be ourselves. We then
sort of say these are acceptable deaths. I don't know who they
are acceptable to. They're certainly not acceptable to the per-
son who dies.
. . . [L]ung cancer is a very nasty disease and the survival

rates rather poor.94

During oral argument, Justices Liacos and Hennessy sharply
questioned the comparative risk basis for DEQE's decision on
grounds that its use of risk comparisons was ad hoc and not struc-
tured or preceded by a generic DEQE regulation and that such
comparisons would facilitate the siting of polluting facilities,
"stand the Clean Air Act on its head," and lead to giving less reg-
ulatory protection to already polluted neighborhoods.

Nevertheless, the court ultimately decided to affirm the DEQE
decision because of the "reasonable determination by DEQE that

93. Id. at 40 (citations omitted).
94. Id. at 45.

19871



30 COLUMBIA JOURNAL OF ENVIRONMENTAL LAW [Vol. 13:1

the potential risk due to MATEP is within acceptable limits." 95

As for the disputed comparative risk approach, the court found
that "It]his is a decision to be made by the agency within its area
of expertise," which it would not scrutinize "except to assure that
it is rational and conforms to the law." 96 In so deferring to
agency expertise, the court also lent its seal of approval to the
general proposition that risk comparisons were a reasonable aid
to agency decisionmaking:

The meaning of an increase in risk of 0.005% is difficult to
grasp.... DEQE used the comparisons of the MATEP... risk
and other risks as an aid to comprehension and as a means to
understand better the magnitude of risks society generally
chooses to accept. Such comparisons are part of a rational
method of determining the acceptability of a risk.

We do not share the sense of impending doom that risk com-
parisons seem to evoke in the Brookline challengers. Brook-
line's fears that such comparison will lead to approval of ever
dirtier emissions are unfounded. We see no reason to presume
that DEQE will abandon its statutory role of rationally deter-
mining what risks are acceptable to our society under its statu-
tory mandate. 97

Such federal and state court decisions evidence judicial willing-
ness to accept agency use of risk comparisons as a reasonable and
rational basis for setting risk limits in regulatory decisionmaking.
Each case involved judicial review of an agency permit decision
made after extensive quasi-judicial proceedings by administra-
tors, and each was marked by a dramatic underlying condition-
whether a facility already built at substantial cost should be al-
lowed to operate.

Whether similar judicial approval of risk comparisons in stan-
dard-setting and other types of regulatory decisions would follow,
where the economic stake is not as clearly or dramatically
presented, is an open question at this time. Since judicial review
of agency quasi-legislative or standard-setting actions is usually
less rigorous than that applied to permit and licensing decisions
which arise from agency adjudicatory functions, one would expect
that risk comparison in standards decisions would also receive
favorable judicial responses, unless excluded by statutory lan-
guage such as that of the Delaney Clause, which has been con-

95. 398 Mass. at 416, 497 N.E.2d at 16.
96. Id. at 415, 497 N.E.2d at 15.
97. Id. at 415-16, 497 N.E.2d at 15-16.



Comparative Risk

strued to bar use of a de minimis approach. 98 In an early opinion
under the Clean Air Act on EPA's new source performance stan-
dard for Portland cement plants as stationary sources of air pollu-
tants, 99 Judge Leventhal set forth a fundamental objection to the
use of risk comparisons:

Petitioners also challenge the cement standards as unfair in
light of lower standards mandated for fossil-fuel-fired steam
generating power plants and incinerators. They claim that
while the cement standard, as expressed in grains of particu-
lates allowed per standard cubic foot of gas (g/scf), requires a
reduction to .03, power plants are permitted to reach .12 and
incinerators to be at .10....

EPA, in response to comments. . . stated... "The difference
.. is attributable to the superior technology available

therefor."
This statement seems to be supported by the EPA Back-

ground Document....

The core of our response to petitioners is that the Adminis-
trator is not required to present affirmative justifications for
different standards in different industries. Inter-industry com-
parisons of this kind are not generally required, or even pro-
ductive;. and they were not contemplated by Congress in this
Act. . . . [T]here is no requirement of uniformity of specific
standards for all industries. The Administrator applied the
same general approach, of ascertaining for each industry, what
was feasible in that industry.' 00

Thus, statutory language of the Clean Air Act has been con-
strued as excluding certain EPA decisions based on inter-industry
risk limit comparisons. The Leventhal view may therefore pres-
ent a formidable obstacle for future advocates of setting stan-
dards on the basis of comparison methods, particularly since it
has been reaffirmed by the influential D.C. Circuit in rejecting the
FDA's use of a de minimis approach some fourteen years later.' 0 '

Legislative authorization for agency use of risk comparisons
could overcome this obstacle and provide more assurance of
favorable judicial response. Although no statutes on risk explic-
itly authorize use of comparative risk methods, some statutes ap-

98. See supra note 43.
99. Clean Air Act § 111, 42 U.S.C. § 7411 (1982) (new source performance standards).
100. Portland Cement Association v. Ruckelshaus, 486 F.2d 375, 388 (D.C. Cir. 1973)

(standard remanded for other reasons).
101. 831 F.2d 1108. See supra note 43.

1987]



32 COLUMBIA JOURNAL OF ENVIRONMENTAL LAW [Vol. 13:1

proach this condition. A Massachusetts statute provides that site
selection by a special state board for a hazardous waste facility:

shall be subject to such limitation with respect to the extent,
character and nature of operation thereof as will insure that the
facility imposes no significantly greater danger to the public
health or public safety from fire, explosion, pollution, dis-
charge of hazardous substances, or other construction or oper-
ational factors than the dangers that currently exist in the
conduct and operation of other industrial and commercial en-
terprises in the Commonwealth not engaged in the treatment,
processing or disposal of hazardous waste, but utilizing
processes that are comparable. 0 2

The other statute, the Superfund Amendments and
Reauthorization Act of 1986 (SARA),1 0 3 also deals with hazardous
waste regulation and provides EPA with a comparative measure
for determining "how clean is clean enough" in conducting its
remedial actions at hazardous waste sites designated for cleanup.

[T]he remedial action selected under section 9604 of this title
or secured under section 9606 of this title shall require, at the
completion of the remedial action, a level or standard of con-
trol for such hazardous substance ... which at least attains...
legally applicable or relevant and appropriate standard[s] ...
[including] Maximum Contaminant Level Goals established
under the Safe Drinking Water Act and water quality criteria
established under section 304 or 303 of the Clean Water
Act .... 104

102. MASS. GEN. L. ANN., ch. 111, § 150B (1983).

103. Pub. L. No. 99-499, 1986 U.S. CODE CONG. & AOMIN. NEWS (100 Stat.) 1613.

104. SARA § 121(d)(2)(A), 42 U.S.C.A. § 9621(d)(2)(A) (West Supp. 1987) ("Degree of
cleanup") (citations omitted). "Legally applicable standards" are defined to include stan-
dards and criteria under various federal laws administered by EPA, more stringent stan-
dards set by state laws, and standards contained in state programs which have been
approved by EPA.

In the full, section 121 (d) (2) (A) provides:
With respect to any hazardous substance, pollutant or contaminant that will remain

onsite, if-
(i) any standard, requirement, criteria, or limitation under any Federal environ-

mental law, including, but not limited to, the Toxic Substances Control Act, the
Safe Drinking Water Act, the Clean Water Act, the Marine Protection, Research and
Sanctuaries Act, or the Solid Waste Disposal Act; or.

(ii) any promulgated standard, requirement, criteria, or limitation under a State
environmental or facility siting law that is more stringent than any Federal standard,
requirement, criteria, or limitation, including each such State standard, require-
ment, criteria, or limitation contained in a program approved, authorized or dele-
gated by the Administrator under a statute cited in subparagraph (A), and that has
been identified to the President by the State in a timely manner,



Comparative Risk

Thus, use of comparative risk methods for making regulatory
decisions on risk limits has been favorably viewed thus far by the
judiciary and has been authorized by several legislative actions.
Nevertheless, several analytic problems and legal issues are ap-
parent and will undoubtedly be raised in future litigation. Many
of these issues were raised in the MATEP case but were not fully
considered by the Massachusetts court because of its long stand-
ing deference to agency expertise, a tradition which does not pre-
vail in other courts.

E. Problems to Solve

The analytic problems involve, for example, the quality of the
estimates of other societal risks which are being compared to the
particular risk before the regulatory agencies. These estimates,
of variable conclusiveness, arise from very different types of stud-
ies-different in terms of the science involved (e.g., toxicology,
epidemiology, pharmacokinetics, etc.), the assumptions and
causal models used (e.g., as to dose-response relationships), the
confidence levels in the results, overall rigor and documentation
of the efforts, resources expended and objectivity (e.g., many
studies are done for advocacy use). As the San Luis Obispo Mothers
of Peace v. NRC case demonstrated, even studies arising from the
same discipline are hotly contested because of methods and as-
sumptions used.' 0 5 Thus, the risk estimates being compared arise

is legally applicable to the hazardous substance or pollutant or contaminant con-
cerned or is relevant and appropriate under the circumstances of the release or
threatened release of such hazardous substance or pollutant or contaminant, the re-
medial action selected under section 9604 of this title or secured under section 9606
of this title shall require, at the completion of the remedial action, a level or standard
of control for such hazardous substance or pollutant or contaminant which at least
attains such legally applicable or relevant and appropriate standard, requirement, cri-
teria, or limitation. Such remedial action shall require a level or standard of control
which at least attains Maximum Contaminant Level Goals established under the Safe
Drinking Water Act and the water quality criteria established under section 304 or 303
of the Clean Water Act, where such goals or criteria are relevant and appropriate
under the circumstances of the release or threatened release.
Note that SARA § 122(g), 42 U.S.C.A. § 9 62 2 (g), provides for "de minimis settlements"

by EPA with "potentially responsible parties" which contributed hazardous substances
which are "minimal" in terms of amount and hazardous effects.

105. See discussion of San Luis Obispo Mothers for Peace v. NRC, supra notes 76-84 and
accompanying text. A further example is the lack of consensus among epidemiologists
regarding the use of basic statistical tools, such as significance tests and confidence inter-
vals, in assessing results. Thompson, Statistical Criteria in the Interpretation of Epidemiologic
Data, 77 AM. J. PUBLIc HEALTH 191 (1987). Some authors have been criticized for using
confidence intervals to over-interpret skimpy data. Id.

1987]



34 COLUMBIA JOURNAL OF ENVIRONMENTAL LAw [Vol. 13:1

from studies which are often not comparable in quality, nor even
evaluated as to their merits.

Another analytic problem pertains to whether risks to individu-
als, populations or both are to be considered and the comparabil-
ity of the individuals and the populations at risk. In the Brookline
case, the entire population of Metropolitan Boston would be at
risk from breathing the air to be polluted by MATEP. But the
other risks used by DEQE for comparison purposes applied only
to subsets of this population; for example, the smoking risk per-
tains only to smokers (and perhaps to those who inhale sufficient
"side-stream" smoke), and the risk of drinking polluted water var-
ies from person to person and from town to town in Metropolitan
Boston, depending on individual consumption and the various
water supply systems being used.

Analytic problems with risk comparison extend further to the
types of risks involved and their significance. Comparing deaths
from air pollutants that put lung function at risk with deaths from
water pollutants that put kidneys at risk involves comparing
deaths which arise from different biological processes over differ-
ent time frames and comparing deaths involving different eco-
nomic burdens, levels of anxiety, pain, suffering and life support
needs.

Finally, there is the issue of whether comparing risks from vol-
untary activity (e.g., smoking) with risks from imposed activity
(e.g., industrial air pollution) is valid from an analytic standpoint,
and convincing or acceptable to the public. 106

Slipshod, ad hoc practices of risk comparison prevail at this
time and are likely to increase because most agency proceedings
are not governed by rules of evidence which would screen out
hearsay and other unreliable types of evidence and because courts
which review agency decisions no longer require that the agency
decisions be supported by "competent evidence" (or even a "re-
siduum" of competent evidence), the type of evidence required

106. See Rodricks, supra note 36, at 21.

It is far from clear how to choose the appropriate background of risk against which

to make comparisons. Most analysts, for example, would not compare voluntarily as-

sumed risks to involuntarily assumed risks .... [Elven more difficult is the issue of the

relative degrees of reliability in the risk figures being compared .... [F]urther atten-

tion needs to be devoted to the appropriateness of various risk comparison

procedures.



Comparative Risk

under the rules of evidence which apply to civil actions in the
courts.

These analytic problems raise several legal and policy issues. It
could be argued that use of sloppy comparisons to make decisions
constitutes a denial of due process, in that persons who are likely
to be adversely affected have little opportunity, in a regulatory
proceeding on a particular risk, to challenge the estimates of
other societal risks or regulatory determinations being used as
comparisons. Sloppy, ad hoc comparisons also provide a basis
for arguing on judicial review that an agency's ultimate decision is
arbitrary or is not based on substantial evidence and is therefore
invalid-despite the cases discussed earlier which were decided
by courts under the pressure of extraordinary economic circum-
stances (shut down of a costly facility leading to huge economic
losses for the facility owners). Repeated use of sloppy compari-
sons by an agency to decide in favor of polluting and other risk-
creating activities could eventually lead to public and judicial rec-
ognition that legislated goals are being subverted, erode agency
credibility and result in reversals of agency decisions. There is
ample support for these dire predictions in the troubled history of
agency uses of cost-benefit analysis and environmental impact
analysis.

Agencies should carefully evaluate their proposed uses of com-
parative risk and the analytic and legal issues likely to arise, such
as those discussed, and develop generic rules to provide for a
principled framework which will govern their future uses of com-
parative risk methods and assure consistent and predictable
practices.

The generic rule for an agency should provide criteria for as-
suring that only qualitatively comparable studies be relied on for
the risk estimates to be compared, that the estimates deal with
comparable populations at risk, and that the attributes of the risks
being compared (e.g., economic impacts, pain and suffering, mor-
tality and morbidity, etc.) be sufficiently similar to justify com-
parison.

The generic rule should also provide a set of evidentiary rules
to govern the admissability and supporting documentation for the
risk estimates, and procedures to assure that interested parties
have the opportunity to evaluate and challenge the estimates be-
ing used for comparative purposes. Finally, the generic rule
should include a policy statement and rationale for assuring that

1987]



36 COLUMBIA JOURNAL OF ENVIRONMENTAL LAW [Vol. 13:1

future uses of comparative risk methods by the agency will be
consistent with the goals set forth in the statute authorizing its
particular regulatory program.10 7

Adopting an appropriate generic rule will not provide a com-
plete solution, however. Comparative risk faces an analytic di-
lemma with possible policy consequences when the other risks
being compared are those which have been determined to be de
minimis, insignificant or otherwise unworthy of regulation by
other agencies. Earlier discussion described the ad hoc and in-
consistent approaches used by certain agencies such as EPA in
establishing de minimis risk levels as limits for their regulatory
actions. When a comparative risk approach is based on consider-
ation of such risk limits, it clearly lacks convincing or consistent
support.

Further, by relying on these de minimis risk limits set by other
agencies under other statutes with their different provisions, an
agency is vulnerable to the argument that its decisions are ultra
vires and therefore invalid, in that they are based not on the pro-
visions and criteria of its own statute, but on the criteria of the
other statutes which governed the setting of risk limits by the
other agencies.

Even if a standard or other regulatory decision is based on in-
ter-industry comparisons of risk limits set by the same agency
under the same statutory mandate, it may be found to be arbitrary
under Judge Leventhal's rationale in Portland Cement Association v.
Ruckelshaus,10 8 or a deviation from statutory language, as in Public
Citizen v. Young.' 09

Thus, an agency should use care in its choice of risk limits set
by other agencies which it intends to use for comparison pur-
poses. It must also establish that the outcome of its comparative
risk exercise, its risk limit, is consistent with the criteria of its own
enabling statute, irrespective of its consistency with the determi-
nation of other agencies.

107. The generic rule, suggested above, provides a process approach without recom-
mending any particular risk limit outcome to permit agency flexibility in choosing the risk
limit for each particular case. For an "outcome" approach proposing specific risk limits,
see Cross, supra note 37, at 44-56. For example, Cross suggests that a lifetime mortality risk
of one chance in one hundred thousand be set as the de minimis threshold of regulation
for "average environmental risks." Id. at 51.

108. 486 F.2d 375 (D.C. Cir. 1973). See supra notes 100, 101 and accompanying text.

109. 831 F.2d 1108. See supra note 43.



Comparative Risk

VI. EVIDENTIARY USE IN COMMON LAW LITIGATION

A. Background

Thousands of toxic tort actions now crowd the dockets of state
and federal courts. The vast majority of these cases involve
claims of personal injury from exposure to asbestos. Estimates of
potential liability for the asbestos industry and their insurers
range from $7.6 billion to $87.1 billion.

A new category of asbestos-related claims is now emerging
which poses even greater liability potential. These claims allege
injury to property caused by asbestos installed in buildings, par-
ticularly in schools. As a result:

Thousands of school districts all over the country have sued
asbestos product manufacturers, miners, and distributors, al-
leging damages caused by the presence of asbestos-containing
materials in their school buildings. Claiming that the presence
of asbestos in the buildings is a health hazard that must be re-
medied, the school districts seek indemnification for costs asso-
ciated with inspection, removal, replacement, and
encapsulation of asbestos-containing products, loss of the
buildings' use during this process, and reduction in the build-
ings' market value .... Such claims ... may even outdistance
... bodily injury claims in dollar terms. 110

The school district plaintiffs and others with similar property
damage claims involving asbestos rely on the most favorable the-
ories of tort and contract liability available to them. In general,
the theories that are the most favorable are strict products liabil-
ity and breach of warranty. The plaintiff can argue both in the
same action but the formulation and availability of these theories
varies from state to state.

Generally, in strict products liability:

the plaintiff is not required to impugn the conduct of the maker
or other seller.... [T]he product must be in 'a defective condi-
tion unreasonably dangerous.' This simply means that the
product must be defective in the kind of way that subjects per-
sons or tangible property to an unreasonable risk of harm."'

110. Arness & Eliason, Insurance Coverage for Property Damage in Asbestos and Other Toxic Tort
Cases, 72 VA. L. REV. 943, 945 (1986) (footnote omitted). "About 15,000 school districts
are participating in a class action against 55 defendants .... In addition, about 150 school
districts are proceeding individually." Id. at 945 n.9.

111. PROSSER AND KEETON ON TORTS, supra note 6, § 99 at 695 (citing § 402A of the
RESTATEMENT (SECOND) OF TORTS).

1987]



38 COLUMBIA JOURNAL OF ENVIRONMENTAL LAw [Vol. 13:1

In breach of warranty actions, the plaintiff must prove that the
asbestos product does not conform to the defendant's express or
implied representations. Most plaintiffs in the asbestos cases rely
on the more favorable implied warranty:

Recovery under an implied warranty theory . . . does not rest
on the written or spoken words of the seller. Rather, it rests on
the implied warranty of merchantability, which the Uniform
Commercial Code has codified. This implied warranty requires
the manufacturer to produce goods that are of merchantable
quality and that are reasonably fit for their intended purpose.
To recover... the plaintiff must prove that ... (4) the defective
nature of the goods caused the injury both proximately and in
fact .... 112

Thus, under both theories, litigants in such property damage
cases involving asbestos in buildings must address the issue of
whether the asbestos product as presently situated in a particular
building is defective, i.e., whether it poses an unreasonable health
risk to exposed persons (e.g., school children, teachers, and other
employees), thereby renders the structure unfit for continued use
and consequently injures the property interest of the plaintiff.

In addressing this issue, both parties must deal with factual
questions about prospective causation of harm to exposed per-
sons, and this inevitably leads to expert testimony on health risk
estimates. The expert views on health estimates may converge,
but more likely will conflict. In either case, the ultimate question
is then raised, namely, what level of health risk is unreasonable.

B. Expert Testimony

In several cases, defendants have sought to use comparative
risk evidence to convince the jury that the asbestos products as
installed, and the health risks as estimated, do not constitute an
unreasonable risk to exposed persons and thereby do not make
the structure unfit nor impair the plaintiff's property interest.

In County of Anderson, Tennessee v. United States Gypsum Co., 113 de-
fendants' expert witness, Dr. Kenny Crump, a biostatistician and
consultant, testified under direct examination by defendant's at-
torney as to the nature and value of probabilistic risk assessments
and then sought to explain the difference between his estimate of

112. Special Project-An Analysis of the Legal, Social and Political Issues Raised by Asbestos Liti-
gation, 36 VAND. L. REV. 573, 585 (1983) (footnotes omitted).

113. Civ. Action No. 3-83-511 (E.D. Tenn. 1985).



Comparative Risk

.001 asbestos fibers per c.c. in the Anderson schools and 5.00 fi-
bers per c.c., which he offered as the level "that can cause
disease."

Well, five fibers per cc is five thousand times larger than the
.001 fibers per cc. A difference of five thousand ig somewhat
difficult to grasp. If you think of a tooth pick which is a little
less than three inches long, it would take three football fields
stretched end to end to be five thousand times as long as a
toothpick .... [or] four Norris Dams stretched end to end to
equal five thousand times the length of a toothpick." 14

If a person is exposed to .001 fibers per c.c., he would have
to be exposed for 5,000 years in order to receive exposure
equal to five fibers for one year.' '5

Crump then began his comparative risk testimony with a table
of risk estimates:

The first number here is the risk of cancer from smoking cig-
arettes for 20 years. Assume one pack per day .... The risk
from smoking cigarettes I estimate to be 88,000 per million
persons. That's a risk of almost one in ten. Now I personally
consider that to be a high risk .... 116

He then proceeded to discuss other risks in the table in de-
scending order of magnitude: drinking diet soft drinks containing
saccharin (170 per million), inhaling side-stream smoke (110 per
million), drinking chlorinated water in Anderson County (26 to
17 per million), riding in a school bus (50 per million), eating
peanut butter which contains the naturally-occurring carcinogen
aflatoxin (11 per million), living in a brick house with radon gas (4
per million)." 7 He then compared these numerical estimates
with his estimates of .8 to .7 per million risk from attending the
Anderson county junior and senior high schools and concluded
that for the senior high school, "one would have to wait 4,000
years before you would expect to see a single cancer," and that
for the junior high school, "you would expect to wait for 12,000
years .... 118

114. Transcript of Proceedings March 7, 1985 at 2467, County of Anderson, Tennessee
v. United States Gypsum Co., Civ. Action No. 3-83-511 (E.D. Tenn. 1985).

115. Id at 2470.

116. Id. at 2475.

117. Id. at 2475-79.

118. Id. at 2479.

19871



40 COLUMBIA JOURNAL OF ENVIRONMENTAL LAW [Vol. 13:1

He then offered several variations on the comparative risk
theme. These included a bar graph illustrating the relative mag-
nitudes of the tabulated risks previously discussed:

We have had to take smoking off the chart, we would have to
make the chart a quarter of a mile long to include smoking. If
you want to, imagine in your mind a line here [that] goes ten
blocks or so away.... The longest one here is saccharin which
is about.., three feet long.., two small bars here are the risk
from being exposed . .. in these schools .... 119

Crump also used a table showing his estimates of loss of life ex-
pectancy for each of the risks:

If you smoke cigarettes for twenty years, that activity will cause
you to lose between two and five years off of your life ....
Drinking one diet soft drink.., throughout your whole life...
would reduce your life expectancy by 22 hours .... Drinking
the water in Anderson County or Knoxville throughout your
entire life [and life] expectancy would be reduced by two to
three hours....

Using the exact same methods for asbestos risks and expo-
sure to two schools for three years I estimate the loss of life
expectancy . . . [for] Norwood Junior High ... would be six
minutes .... The loss ... for attending Clinton Senior would
be five minutes.

120

Direct examination concluded with the following question and
answer:

Q Dr. Crump, based on your professional experiences, do
you have an opinion as to whether asbestos at the level to exist
in the Anderson County schools created any unreasonable risk
of harm to the employees or students?

A: I have compared the risk and found them to be less than
[the] risk of other activities we consider to be safe and my con-
clusion is there was not an unreasonable risk of harm from ex-
posures in those schools. 12 1

Plaintiff's attorney then cross-examined Crump on several key
points to undermine his testimony. The initial line of questioning
sought to get Crump to admit "that there is a statistical risk.., of
developing an asbestos related disease and dying from it .. .[as]
depicted in that chart at Norwood ... and Clinton."' 22 Despite
repeated denials, Crump finally admitted under continuing exam-

119. Id. at 2481.

120. Id. at 2482-83.
121. Id. at 2483-84.
122. Id. at 2485.



Comparative Risk

ination that his chart showed an "upper limit" for statistical risk
for each activity including attending the two schools.' 23 The next
line of questioning brought Crump to disclose his extensive work
as consultant and expert witness for numerous industrial organi-
zations over several years, including the industrial Asbestos Infor-
mation Association, Johns-Manville, W.R. Grace & Co. and other
asbestos producers, and the fact that he had never consulted for a
school board. 124 This was followed by cross-examining Crump on
the uncertainties inherent in quantitative risk assessment. Plain-
tiffs' counsel elicited Crump's admission as to these uncertainties
and acknowledgement of his pre-trial published statements to this
effect. '

25

Cross-examination was then targeted on the quality of Crump's
risk estimates for the two schools in question and secured admis-
sions by Crump that he had not done such studies before; had not
visited the two schools in question; had relied on air sampling
data on the two schools to develop his risk estimates, but had not
heard of various personal observations from persons at the school
as to visibly heavier concentrations of fibers during air condition-
ing repairs and other special circumstances; had not considered
the synergistic effects of smoking in an asbestos environment; had
not considered employee exposure over a thirty year job period,
but only student exposure over three years; and had not consid-
ered the cumulative risk for the same students attending both
schools over a period of six years. These questions forced Crump
to offer various revised estimates as to student risk, up from .7
and .8 per million to 1.5 per million for children attending both
schools, and to 2 per million for student smokers attending one
of the schools. Thus, plaintiff's attorney worked at impugning
Crump's objectivity and expertise, the quality of his risk analysis
and resultant estimates.126

Crump has offered similar risk estimates, comparisons and
opinions about the reasonableness of asbestos risks in other cases
as an expert for defendants and added risks from taking aspirin,
having chest X-rays and inhaling radon gas in masonry school
buildings to his repertoire of risk comparisons. He faced increas-
ingly intensive cross-examinations in these cases. In some of

123. Id. at 2488.
124. Id. at 2488-96.
125. Id. at 2489-91.
126. Id. at 2496-509.

1987]



42 COLUMBIA JOURNAL OF ENVIRONMENTAL LAW [Vol. 13:1

these cases, cross-examination has exposed important disparities
between the methods used by Crump to develop the risk esti-
mates used in his comparative risk tables and charts, such as his
use of a threshold for asbestos risk (5 fibers per c.c.), in contrast to
his use of a linear dose-response or no threshold approach for other risks
being compared, a disparity of method which enlarges the differ-
ence between the asbestos risks and the other risks. 127

At this time, one cannot fully evaluate how comparative risk evi-
dence influences outcomes in these jury trials. Some cases have
not been fully resolved or have been settled out of court. Many
trial court decisions are not reported. Other considerations may
influence outcomes (e.g., evidence of defendant's failure to warn
or deliberate non-disclosure). Additionally some courts have ex-
cluded the comparative risk evidence offered by the defendants.

For example, City of Greenville v. WR. Grace & Co. 128 marked the
first verdict against a defendant in cases involving property dam-
age from installed asbestos products. Grace was ordered to pay
$6.4 million in actual damages and $2 million in punitive dam-
ages for its sale of asbestos fire-proofing materials to the city in
1971 without disclosing that the materials contained asbestos. In
this case, district court excluded Grace's comparative risk evi-
dence, under Federal Rule of Evidence 403, on the ground that
the probative value of such evidence was substantially outweighed
by the risk of jury confusion and undue delay. 12 9

On appeal to the Court of Appeals for the Fourth Circuit,
Grace contended that "the evidence presented at trial was insuffi-
cient to support a finding that the levels of asbestos contamina-
tion ... posed a serious risk to the building's occupants,"'' 30 and
that the district court had erred in refusing to admit its compara-
tive risk evidence which included data on risks posed by peanut
butter, diet soft drinks and other products.' 3 1 In affirming the
district court decision, the court of appeals noted that Green-
ville's experts had used techniques accepted in the scientific com-

127. See Trial Transcript at 1429, Spartanburg School District Seven v. National Gyp-
sum Co., Civ. Action No. 83-1744-14 (D.S.C.July 29, 1985); Trial Transcript at 585, City
of Greenville v. W.R. Grace & Co., 640 F. Supp. 559 (D.S.C. 1986) (No. 85-1693-3).

128. City of Greenville v. W.R. Grace & Co., 640 F. Supp. 559 (1986).
129. Id. at 571-72. See also Asbestos Litigation Reporter (Andrews Publications) at

11,735-36 (Feb. 21, 1986) and discussion in City of Greenville v. W.R. Grace & Co., 827
F.2d 975, 981 n.3 (4th Cir. 1987).

130. 827 F.2d at 980 n.2.
131. Id. at 981 n.3.



Comparative Risk

munity to estimate the risk, and that their testimony therefore
"could reasonably support a jury finding that the levels of asbes-
tos contamination.., posed significant health risks."1 32 Further,
it noted that Grace's claim of district court error in excluding
comparative risk evidence was without merit, as Grace's counsel
had conceded during oral argument, and "the District Court
acted within its discretion in excluding the 'comparative risk'
evidence."

33

Since Greenville, several other verdicts have been reached
against defendants.134 In one case in which Crump's comparative
risk testimony was admitted and verdict returned for the defend-
ant, a new trial has been ordered because of a faulty jury
instruction. 135

The admissability of comparative risk evidence in personal in-
jury and property damage actions has long troubled the courts.
The reasons were cogently stated in a recent decision of the
Supreme Judicial Court of Massachusetts, Kromhout v. Common-
wealth of Massachusetts.'36 In this case, plaintiff sought to establish
that the state was liable in tort for the accidental death in 1981 of
her husband, a motorist, due to a drainage defect in a state high-
way. During the trial, the plaintiff introduced evidence of twenty-
one accidents at the same site, during the six years preceding her
husband's accident, and plaintiff's expert on transportation testi-
fied that, in his opinion, this number of accidents at the same site
in a six year period was significant. The state's expert testified
that the total of twenty-one accidents was not significant. The
jury returned a verdict for the plaintiff.

Following appeal to the Supreme Judicial Court by the state,
the Court held that plaintiff's evidence was not admissible at trial,
reversed the trial court judgment and remanded the case for a
new trial. In finding the admission of such evidence to be revers-
ible error, the Supreme Judicial Court provided the following
rationale:

132. Id. at 980 n.2.
133. Id. at 981 n.3.
134. See New Trial Ordered in S.C. Asbestos Case, Nat'l L.J., December 15, 1986, at 3, col. 1.
135. Spartanburg County School District Seven v. National Gypsum Co., 805 F.2d 1148

(4th Cir. 1986). The jury was erroneously instructed that the state of the art was a defense
to a breach of warranty action. The court did not address plaintiffs' allegation that defend-
ants had concealed certain documents, as reported by the National Law Journal article,
supra note 134.

136. 398 Mass. 687, 500 N.E.2d 789 (1986).

1987]



44 COLUMBIA JOURNAL OF ENVIRONMENTAL LAw [Vol. 13:1

"The admissability of evidence of injury to others at other
times by reason of the same thing that caused the plaintiff's
injury, for the purpose of showing that thing to be dangerous,
has often come before this court. Such evidence is open to
grave objections. Its persuasive force depends upon similarity
in the circumstances of different injuries, of which it is hard to
be certain. Substantial identity in the alleged defective condi-
tion is only the first essential. The person who was injured at
the time to which the offered evidence relates may have been
defective in eyesight, feeble, or careless. The fact that he was
injured may have little or no bearing upon the danger to a nor-
mal traveller. Moreover, though the same defective condition
may have been present at both times, the actual causes of the
two injuries may have been different. Unless a comparison of
the circumstances and causes of the two injuries is made, the
injury to another is without significance. But if such a compari-
son is undertaken, the minds of the jurors must be diverted
from the injury on trial into a detailed and possibly protracted
inquiry as to injuries received by others at various times. Those
injuries have only a collateral and often minor bearing upon
the case. As to them the opposing party will often be ill pre-
pared to present evidence. There is danger that a jury may dis-
regard the real differences in the circumstances of the two
incidents, and find upon mere superficial similarity that a dan-
gerous condition existed. Similar considerations apply where
evidence that other people, confronted at other times with the
same alleged danger, suffered no injury is offered to prove the
want of a dangerous condition."

It is true that where substantial identity in the circumstances
appears, and the danger of unfairness, confusion or undue ex-
penditure of time in the trial of collateral issues seems small,
the admission of such evidence has resided in the judge's
sound discretion. That was clearly not the case here. No such
identity of circumstances was shown .... 137

Six months earlier, the same court had affirmed the state
agency decision based on comparative risk evidence in Town of
Brookline v. Commissioner Department of Environmental Quality Engineer-

ing, 138 discussed previously, demonstrating the judiciary's differ-

137. 398 Mass. at 693, 500 N.E.2d at 793 (quoting Robitaille v. Neteco Comm. Theatre
of North Attleboro, 305 Mass. 265, 266-67, 25 N.E.2d 749-50 (1940)). See discussion of
the admissability of evidence in negligence cases as to prior accidents at a particular site,
provided that the conditions at the time of the earlier accidents (or lack of accidents) were
"substantially the same" and there is a "showing of the relevant conditions ... prevailing
at the time of the earlier accidents . C. KRAMER & D. KRAMER, EVIDENCE IN NEGLI-

GENCE CASES 58-61 (8th ed. 1987).

138. 398 Mass. 404, 497 N.E.2d 9 (1986). See supra notes 85-97.



Comparative Risk

ential standards for private civil actions and for regulatory
decisionmaking.

C. Rules of Evidence and Judicial Concerns

Rules of evidence govern the admissability of all evidence at
trial, including expert testimony as to comparative risk and expert
opinion as to the ultimate factual question before the jury, such as
whether a product is unreasonably dangerous.

Traditionally, these rules have varied from state to state, but
many states have recently adopted the Federal Rules of Evi-
dence,' 3 9 which apply in federal courts, leading to greater uni-
formity of rules among these states. Nevertheless, complete
uniformity in application is unachievable, since rules of evidence
provide that issues regarding the admissability of expert testi-
mony and opinions are left to the discretion of the trial courts.' 40

Such discretion, which is exercised on a case-by-case basis by dif-
ferent judges in different states, inevitably leads to variability in
the treatment of expert testimony and opinions.' 4 '

In general, rules of evidence seek to guarantee that evidence by
witnesses, expert or non-expert, will be "relevant" and "mate-
rial" as to the issues and "competent" in the sense that the evi-
dence is not barred by any specific exclusionary rules.' 42

Comparative risk evidence, such as the Crump evidence in the
school cases discussed earlier, would have run a considerable risk
of exclusion under rules of evidence in effect in many states

139. FED. R. EVID., Pub. L. No. 93-595, 88 Stat. 1926 (1975), codified at 28 U.S.C. app.
678 (1982).

140. See FED. R. EVID. 104(a) (qualifications of witnesses and admissability of evidence
shall be determined by the court), and state counterparts, ALASKA R. EVID. 104(a); ARIz.
REV. STAT. ANN., Rules of Evidence, Rule 104(a); ARK. STAT. ANN. § 28-1001, Uniform
Rules of Evidence, Rule 104(2); COLO. R. EViD. 104(a); FLA. STAT. ANN. § 90.105; HAW. R.
STAT. § 626-1, Rules of Evidence, Rule 104(a); IDAHO R. EviD. 104(a); IOWA R. Evio.
104(a); ME. R. EVID. 104(a); MICH. R. EVID. 104(a); MINN. R. EVID. 104(a); Miss. R. EVID.
104(a); MONT. CODE ANN. tit. 26, ch. 10, Rules of Evidence, Rule 104(a); NEa. REV. STAT.
§ 27-104(a); NEv. REV. STAT. § 47.060; N.H. R. EVID. 104(a); N.M. R. EViD. 104(a); N.C. R.
EVID. 104(a); N.D. R. EVID. 104(a); OHIO R. Evw. 104(a); OKLA. STAT. ANN., tit. 12, § 2105;
OR. EVID. CODE 104(a); S.D. CODIFIED LAWS ANN. § 19-9-7 to -11; TEX. R. EVID. 104(a);

UTAH R. EVID. 104(a); VT. R. EVID. 104(a); WASH. R. EvID. 104(a); W. VA. R. EVID. 104(a);
Wis. STAT. ANN. § 901.04(1); Wyo. R. EVID. 104(a). See also P.R. R. Evid. 9.

141. In addition, states have re-drafted and amended the Federal Rules of Evidence
during adoption, creating another source of variation among states. State adoptions and

variable results are discussed in Wroth, The Federal Rules of Evidence in the States: A Ten Year
Perspective, 30 VIL. L. REV. 1315 (1985).

142. See C. KRAMER & D. KRAMER, supra note 137, at 3, 17-30.

1987]



COLUMBIA JOURNAL OF ENVIRONMENTAL LAW [Vol. 13:1

before their adoption of the more liberal Federal Rules. The
Crump evidence as to various risks in society based on studies
done by others would most likely have been excluded under rules
barring "hearsay" evidence, hearsay being "what someone other
than the witness said or did . . . what the witness heard X say,
rather than that which is based on his personal knowledge of
observation."1

43

Similarly, expert opinion as to the ultimate factual issue, for ex-
ample, whether a product is unreasonably dangerous, would not
have been admissible under the rules and common law in effect in
many states before their adoption of the Federal Rules. Such ex-
pert opinion evidence would have been subject to exclusion on
several grounds: it would confuse the jury, cause undue delay, or
displace the jury's function of resolving the ultimate factual issue;
it would be based on studies and technical reports not before the
court and therefore would not be subject to critical review or
cross-examination by the opposing party; it would permit an ex-
pert to speculate and subjectively rely on unaccepted or unproven
experiments. 1

44

However, in federal and state courts now following the new
Federal Rules, expert opinions are more readily admissible. Rule
702 provides for judicial admissability of expert testimony when it'
is "helpful," i.e., "if scientific, technical, or other specialized
knowledge will assist the trier of fact to understand the evidence
or to determine a fact in issue." Rule 704 provides that "testi-
mony in the form of an opinion or inference otherwise admissible
is not objectionable because it embraces an ultimate issue to be
decided by the trier of fact." Further, rule 703 provides that ex-
pert opinions are judicially admissible when based on facts which
either are introduced as evidence, or are of a kind "reasonably
relied on by experts in the same field." Thus expert opinions and
inferences can now be admitted even if the underlying facts or
data are not admitted or are not admissible as hearsay. These and

143. Id. at 25 (footnote omitted) (quoting FED. R. EviD. 801(c), 802).
144. Evidence law on expert opinions became particularly restrictive in courts following

the rule of Frye v. United States, 293 F. 1013 (D.C. Cir. 1923). The Frye rule, set forth in a
case involving the exclusion of the results of a lie detector case, provides that "[w]hile
courts will go a long way in admitting expert testimony deduced from a well-recognized
scientific principle . . . the thing from which the deduction is made must be sufficiently
established to have gained general acceptance in the particular field in which it belongs."
Id. at 1014. See discussion of this general acceptance test in Weinstein, Improving Expert
Testimony, 20 U. RICH. L. REV. 473, 476 (1986).



Comparative Risk

other provisions of the new Federal Rules have led to a consider-
able liberalization of evidence law. 14 5

But new problems have arisen, according to Judge Jack
Weinstein:

The Federal Rules of Evidence and their state counterparts
have thus produced an enormous loosening up of the restric-
tions on the admission of expert testimony .... This relaxation
was needed to give the trier of fact convenient access to the
reliable technical knowledge that is available in our modem so-
ciety. As might be expected, however, the modification of the
old rules has led to new difficulties.

An expert can be found to testify to the truth of almost any
factual theory, no matter how frivolous, thus validating the case
sufficiently to avoid summary judgment and force the matter to
trial. At the trial itself an expert's testimony can be used to
obfuscate what would otherwise be a simple case. . . . Juries
and judges can be, and sometimes are, misled by the expert-
for-hire.

New discretionary controls .. .may be needed to curb the
potential abuses that have appeared .... But how can discre-
tion be bridled in a manner predictable and fair? How can the
nonexperts control the experts? 14 6

Introduction of comparative risk testimony and expert opinions
on the reasonableness of risk, as in the asbestos in schools cases
previously discussed, illustrate the validity of these judicial con-
cerns. Weinstein and others have proposed various measures,
some of which bear on the use of comparative risk evidence.
Their proposals include greater judicial supervision of expert tes-
timony before trial, including pre-trial provision of all underlying
records from which data were collected to the opposing party, ' 47

pre-trial conferences for the opposing parties with the persons
who compiled the "underlying records,"' 148 pre-trial agreement
on the database, 49 and the taking of depositions from non-testi-

145. See e.g., FED. R. EvID. 803(6), (8) and (18) (exceptions to hearsay rule for business
records, public records and learned treatises). See also Weinstein, supra note 144, at 481-
82, for a discussion regarding new exceptions to the hearsay rule for learned treatises,
business records and government reports, which reinforce the liberalizing influence of the
Federal Rules.

146. Weinstein, supra note 144, at 481-82.
147. See PANEL ON STATISTICAL ASSESSMENTS AS EVIDENCE IN THE COURTS, THE EVOLVING

ROLE OF STATISTICAL ASSESSMENTS AS EVIDENCE IN THE COURTS 360.
148. Id.
149. Id. at 351.

1987]



48 COLUMBIA JOURNAL OF ENVIRONMENTAL LAw [Vol. 13:1

fying experts. 150 These pre-trial procedures available to the judi-
ciary would "subject expert testimony to more informed scrutiny
by the opposition's experts and lawyers. 151

Other reforms, cited by Weinstein, include professional licens-
ing of experts as qualified for testimony in certain fields with vio-
lations subject to disciplinary action and the development and
application of new ethical standards for expert witnesses.1 52

Standards now being considered by a National Academy of Sci-
ences' Panel would pertain to the reliability of statistical tech-
niques, the disclosure of methodologies used, and other "aspects
that may raise ethical considerations," and a requirement "that
statistical experts who consult or testify in litigation maintain the
degree of professional autonomy required by independent scien-
tific research."1 53

Finally, stronger judicial supervision may be needed. According
to Weinstein, who has presided over the Agent Orange litigation,
this may take various forms, ranging from changes in allocating
the burden of proof to greater use of court-appointed expert wit-
nesses as permitted by the new Federal Rules (Rule 706), to
court-commissioned studies by government agencies or impartial
experts.154

This need for judicial activism may extend to greater use .of
summary judgment "to prevent the enormous waste of resources
caused by taking baseless or overwhelmingly strong cases to
trial," 55 as in cases where "examination of the basis of an ex-
pert's opinion reveals that it is supported by no reliable evidence
at all .... In other cases, an expert's opinion is supported by some
credible evidence, but . . . there is other, much more persuasive
evidence available which undermines the expert's opinion and
which the expert is ignoring."' 156 True to his words, Weinstein
has granted summary judgment and dismissed a complaint filed
by an alleged victim of Agent Orange, because the plaintiff's
medical expert relied on a causation hypothesis without any scien-

150. Id. at 241.
151. Weinstein, supra note 144, at 484.

152. Id. at 485-86.
153. Id. (discussing study of PANEL ON STATISTICAL ASSESSMENTS AS EVIDENCE IN THE

COURTS, supra note 147).

154. Id. at 486-91.

155. Id. at 492.

156. Id. at 493.



Comparative Risk

tific support and excluded other potential causes without any fac-
tual basis. 157

Thus, use of expert testimony about comparative risk and ex-
pert opinion as to unreasonable risk is now more readily permit-
ted by new evidentiary rules, but early experience has already led
to judicial concern and action over abuses, including its unrelia-
bility and misleading influences. Experts therefore face a sensi-
tized and increasingly critical judiciary, despite new evidentiary
rules, and this in itself may work to promote more responsible
testimony and opinions pertaining to risk comparisons. But a
more systemic reform is needed to guide case-by-case decisions
by the judiciary. Principles of responsibility and reliability in the
use of comparative risk evidence should be developed by the sci-
entific community to meet this need.'5 8

VII. CONCLUSIONS

Now that risk analysts produce quantitative estimates of risk
and these estimates can be statistically compared, many have pro-
posed, and some have used, risk comparisons to resolve three risk
decisionmaking problems: prioritizing risks to allocate regulatory
resources, determining the risk limit in licensing and standard-
setting actions and the common law or toxic tort problem of de-
termining whether a risk at issue is unreasonable.

Use of risk comparisons for setting regulatory priorities has
raised several issues that have yet to be resolved and that there-
fore obstruct full reliance on this approach. Some of the issues
are legal in that several statutes that mandate risk regulation pro-
vide for it in a manner that militates against agency use of com-
parisons for setting priorities. But most of the issues pertain to
the uncertainty of the risk estimates being compared and, more
fundamentally, to their lack of comparability due to the absence

157. See In re Agent Orange Product Liability Litigation (Lilley), 611 F. Supp. 1267,
1280-83 (E.D.N.Y. 1985).

158. For recent general background as to evidentiary law relevant to experts and com-
parative risk testimony, see Weinstein, supra note 144; Wroth, supra note 141; C. KRAMER &
D. KRAMER, supra note 137; PANEL ON STATISTICAL ASSESSMENTS As EVIDENCE IN THE

COURTS, supra note 147. See also Carlson, Policing the Bases of Modern Expert Testimony, 39
VAND. L. REV. 587 (1986); Reisel, Discovery and Examination of Scientific Experts, 32 PRAc.

LAw. 59 (September 1986); Field & Baram, Screening and Monitoring Data as Evidence in Legal

Proceedings, 28 J. OCCUPATIONAL MED. 946 (1986); Durst, Evidentiary Use of OSHA Regula-

tions, 17 TRIAL LAW. Q., No. 3, at 5 (1986); Faulk, Strategic and Scientific Considerations in Toxic
Tort Defense, 26 S. TEX. L.J. 513 (1985).

1987]



50 COLUMBIA JOURNAL OF ENVIRONMENTAL LAW [Vol. 13:1

of any accepted generic criteria for valuing the disparate types of
risks in order to rank order them on a common scale. Neverthe-
less, some agencies have proceeded to use risk comparisons as
one of several methods for structuring their regulatory priorities,
particularly when the types of risks involved are essentially com-
parable, as in the case of the risk of death due to cancer.

Agency use of risk comparisons for setting the risk limits in
their licensing and standard-setting decision processes is now oc-
curring on an ad hoc basis, although the Nuclear Regulatory
Commission has recently proposed to formally adopt a generic
approach.1 59 Several federal and state agency decisions in license
cases based on risk comparisons have been affirmed in the courts
as being reasonable and rational, despite revelations that such
comparisons used risk estimates derived from studies of variable
quality and studies which used different assumptions and analytic
methods.16

0

Nevertheless, future judicial review may not so readily defer to
agency expertise in making decisions on the basis of such ad hoc
comparisons. Many of these comparisons have obviously been
flawed in several material respects: for example, the lack of com-
parability of the risks, or of the exposed populations studied, and
the ad hoc or opportunistic use of available risk estimates without
a properly structured set of generic principles to guide agency use
of risk comparisons. Appropriate generic rules could be devel-
oped by the agencies to avoid these problems and to assure that
any use of comparative risk methods will be consistent with statu-
tory mandates.

Finally, comparative risk evidence is also being used in toxic
tort and related common law litigation on unreasonable risk is-
sues, most prominently in the asbestos in schools cases. Expert
testimony and opinions, based on risk comparisons, have been
subject to extensive cross-examination, revealing various inade-
quacies including the expert's use of risk estimates derived from
studies of variable quality, studies which use differing assump-
tions and analytic methods. The effectiveness of such testimony
is an open question at this time. What is clear is that new, liberal
rules of evidence used in federal and many state courts now per-

159. See supra notes 66-75 and accompanying text.
160. See supra notes 76-97 and accompanying text.



Comparative Risk

mit such testimony, a sharp departure from earlier rules restrict-
ing hearsay testimony and opinions in these courts.

But growing judicial dissatisfaction with the use of experts on
risk is leading to the development of new means for curbing
abuses in the use of expert testimony under the new rules of evi-
dence. Some of these means involve greater judicial activism
before trial and during trial, such as the summary dismissal of an
action when a plaintiff's expert opinion has not been supported
by reliable scientific evidence. 161 Other reforms are being devel-
oped within the scientific community and could provide a consis-
tent set of principles for judicial use in determining the
admissability of expert testimony and opinions, as well as for use
by experts themselves to guarantee the quality of their testimony.

Thus, future use of comparative risk to set priorities and risk
limits in the agencies, and to determine whether a risk is unrea-
sonable in common law litigation, will depend to a considerable
extent on the responsibility of its proponents. As agencies and
courts grapple with the need to develop procedures for assuring
the appropriate use of scientific expertise and risk comparisons,
there is a concomitant need for the risk analysis community to
deal with the substantive issues and put its own house in order.
To do this, risk analysts must work to assure that professional at-
tributes of objectivity and reliability, and humanistic considera-
tions about the appropriateness of comparisons in particular
circumstances, govern the use of comparative risk methods in the
agencies and courts.

161. See supra notes 146-157 and accompanying text.

1987]