Risk Exposure as Injury: Alleviating the Injustice of
Tort Causation Rules
David Gerecke*
Increasingly, courts are faced with cases
involving injuries stemming from exposure to
risk. Our scientific knowledge of sources of
disease, and our ability to correlate exposure
and injury, are increasing. However, standard
causation rules are inadequate in situations
involving the indeterminate defendant, the
indeterminate plaintiff, and loss of a chance
patterns.
The author argues that a “risk as injury”
approach will resolve problems associated
with causal indeterminacy. The “all or noth-
ing” requirement of current tort principles fre-
quently causes injustice. However, a risk-
based theory is compatible with general
objectives of tort liability. Moreover, probabi-
listic evidence may be as adequate as particu-
laristic evidence.
The “risk as injury” theory has been applied by
American courts in cases involving DES. The
author argues that the approach taken in deci-
sions such as Sindell may be extended to other
fact patterns involving indeterminacy. The
case of Janiak shows that Canadian courts
may be ready to accept risk-based liability as
a method which resolves causal indeteminacy
yet maintains the integrity of the causation
requirement.
De plus en plus, les tribunaux doivent traiter
de cas oa un dommage corporel rdsulte de
l’exposition d’une personne A un risque. Nos
connaissances scientifiques des origines des
maladies et notre aptitude h mettre en corrdla-
tion exposition et risque ne font qu’augmenter.
Toutefois, il s’av~re que les r~gles tradition-
nelles en matibre de causalit6 sont devenues
caduques aux fins de rdgler les causes impli-
quant des demandeurs ou des ddfendeurs inde-
termines ou encore des 6valuations de pertes
de chance.
Selon l’auteur, l’approche pr6nant l’quiva-
lence entre le risque et le pr6judice devrait per-
mettre de rdsoudre les probl~mes relatifs A
l’imprdcision causale. Le critre actuel, en
d~lits, du << tout ou rien >> est souvent la cause
d’inustice. Pourtant, une throrie basre sur le
risque est compatible avec les objectifs de la
responsabilit6 delictuelle. De plus, une preuve
de probabilit6 peut 8tre tout aussi probante
qu’une preuve particuli~re.
La throrie du < risque 6quivaut prejudice >> a
6t6 appliqure par les tribunaux arn6ricains
dans des cas relatifs au DES. L’auteur suggre
que l’approche adoptee dans des decisions
comme telle que Sindell peut 6tre transposee k
d’autres situations de faits en mati6re d’impr-
cision causale. L’affaire Janiak indique que les
tribunaux canadiens semblent disposes k
accepter la responsabilit6 basee sur le risque.
Cette approche est la seule qui puisse resoudre
les probl~mes de causalit6 tout en maintenant
l’integrit6 de cet imperatif.
* Macleod Dixon, Calgary, Alberta. The author would like to thank Professor Ken Cooper-
Stephenson of the University of Saskatchewan, College of Law, for his invaluable direction and
insights.
McGill Law Journal 1990
Revue de droit de McGill
McGILL LAW JOURNAL
[Vol. 35
Synopsis
Introduction
I.
H. Tort Objectives and Causal Indeterminacy
The Source of Injustice – The All or Nothing Approach
A. Compensation
B. Deterrence
C. Other Objectives: Knowledge and Justice
I. The Problem of Statistics: Probabilistic v. Particularistic Evidence
IV. Risk Factors: Binding the Patterns Together
V. The Indeterminate Defendant
A. The DES Dilemma
B. Sindell: Risk as Injury in its Infancy
C. The Evolution of the Theory
D. Extending Sindell
E. The Significance of Breach in Risk-based Liability
F. The Operation of Risk-based Liability
VI. The Indeterminate Plaintiff
A. Redress for the Mass Exposure Victim
B. McGhee Overturned
C. Epidemiological Evidence
D. Calculating Awards
E. The Unavailability of Class Actions
F. No Manifested Injury: Is Accurate Compensation Possible?
VII. Loss of a Chance
A. Janiak: The Foundation for Risk-based Liability
B. Valuing the Loss of a Chance
VIII. Judicial or Legislative Reform?
Conclusion
.~: *
*
19901
RISK EXPOSURE AS INJURY
Introduction
Scientific developments in this half century have had an impact on tort
causation in two unique but interrelated ways. First, populations are being
exposed to hazardous substances more often and in greater intensity than ever
before. Second, knowledge is constantly increasing as to the sources of disease
and other mysterious injuries. This new knowledge, however, is frequently
insufficient to link a particular defendant’s conduct with a particular plaintiff’s
injury in the eyes of the law.’ The difficulty is that there is a substantial differ-
ence between medical causation and legal causation. While scientists make pre-
dictive statements based on observations of repetitions of events, our courts
treat a submission which is probably true as if it were a certain fact. In the result,
the standard causation inquiry is inadequate to deal with the increasingly com-
plex circumstances which are confronting it. This article postulates a “risk as
injury” thesis as an answer to this inadequacy.
The problem arises primarily in three contexts. They are known as the
“indeterminate defendant”, “indeterminate plaintiff”, and “loss of a chance” pat-
terns.’ In each case the plaintiff generally will be denied recovery because of his
or her inability to show a connection (in the eyes of the law) between the defen-
dant’s tortious conduct and the injury he or she has suffered.
1There are exceptions. For example, scientists are able to point to asbestos as the “cause” of var-
ious forms of cancer, because when asbestos causes an injury it carries a “signature.” D.A. Farber,
“Toxic Causation” (1988) 19 Land Use and Environment L. Rev. 557 at 589-90.
2
1t will be argued that these patterns are more accurately described as “sub-patterns” and they
will be referred to as such.
REVUE DE DROIT DE McGILL
[Vol. 35
There has been some litigation3 and considerable commentary4 in each of
these contexts. The American courts and academics have been the most active,
and in fact have dominated the dialogue. One of the goals of this article is to
bring a Canadian perspective to the discussion, although many of the principles
and precedents involved are common to the American and Commonwealth sys-
tems. The major difference is found more in judicial attitude; the U.S. courts
have taken the lead in resolving the injustice of causal indeterminacy.
More important than relating the issue to Canadian law is the task of pro-
viding a unified approach to the indeterminate defendant, indeterminate plain-
3See, for example, the following cases involving indeterminate defendents: Sindell v. Abbott
Laboratories, 607 P.2d 924 (Cal. 1980) (DES) [hereinafter Sindell]; Collins v. Eli Lilly & Co., 342
N.W.2d 37 (Wis. 1984) (DES) [hereinafter Collins]; Martin v. Abbott Laboratories, 689 P.2d 368
(Wash. 1984) (DES) [hereinafter Martin]; Hardy v. Johns-Manville Sales Corp., 509 F.Supp. 1353
(1981) (rev’d on other grounds) (asbestos); but see, Payton v. Abbott Labs, 437 N.E.2d 171 (Mass.
1982) (DES) [hereinafter Payton]; Starling v. Seaboard CoastLine R. Co., 533 F.Supp. 183 (1982)
(asbestos) [hereinafter Starling]. For cases involving an indeterminate plaintiff, see In re “Agent
Orange” Product Liability Litigation, 565 F.Supp. 1263 (1983) [hereinafter Agent Orange]; Ayers
v. Jackson Township, 525 A.2d 287 (NJ. 1987) (various toxins contaminating region’s well water)
[hereinafter Ayers]. Cases involving loss of a chance include Herskovits v. Group Health
Co-operative of Puget Sound, 664 P.2d 474 (Wash. 1983) (reduction in patient’s chance of survival
due to physician’s failure to make timely diagnosis) [hereinafter Herskovits]; Chaplin v. Hicks,
[1911] 2 K.B. 786 (C.A.) (loss of chance to compete for prize in beauty contest) [hereinafter
Chaplin]; but see also Hotson v. East Berkshire Health Authority, [1987] 1 A.C. 750 (H.L.) (delay
in treating injury) [hereinafter Hotson].
4See, for example, N. Sheiner, “DES and a Proposed Theory of Enterprise Liability” (1978) 46
Fordham L. Rev. 963 [hereinafter cited as Fordham Comment]; G.O. Robinson, “Multiple
Causation in Tort Law: Reflections on the DES cases” (1982) 68 Virg. L. Rev. 713; T.T. Currie,
“Risk Contribution: An Undesirable New Method for Apportioning Damages in the DES Cases”
(1985) 10 J.Corp. L. 743; R.W. Wright, “Causation in Tort Law” (1985) 73 Calif. L. Rev. 1737
at 1813-1826; J.G. Fleming, “Probabilistic Causation in Tort Law” (1989) 68 Can. Bar Rev. 661.
See also D. Rosenberg, “The Causal Connection in Mass Exposure Cases: A ‘Public Law’
Vision of the Tort System” (1984) 97 Harv. L. Rev. 851; R. Delgado, “Beyond Sindell: Relaxation
of Cause-In-Fact Rules for Indeterminate Plaintiffs” (1980) 70 Calif L. Rev. 881; S. Gold,
“Causation in Toxic Torts: Burdens of Proof, Standards of Persuasion, and Statistical Evidence”
(1986) 96 Yale L.J. 376; G.O. Robinson, “Probabilistic Causation and Compensation for Tortious
Risk” (1985) 14 J.Legal Stud. 779; B. Pardy, “Risk, Cause, and Toxic Torts: A Theory for a
Standard of Proof’ (1989) 10 Adv. Q. 277; M.C. Andrues, “Proof of Cancer Causation in Toxic
Waste Litigation: The Case of Determinacy Versus Indeterminacy” (1988) 61 S. Calif L. Rev.
2075; Farber, “Toxic Causation”, supra, note 1.
See also J.H. King, “Causation, Valuation, and Chance in Personal Injury Torts Involving
Preexisting Conditions and Future Consequences” (1981) 90 Yale L.J. 1353; S.F. Brennwald,
“Proving Causation in ‘Loss of a Chance’ Cases: A Proportional Approach” (1985) 34 Cath. U.L.
Rev. 747; K. Cooper-Stephenson, “Assessing Possibilities in Damage Awards; The Loss of a
Chance or the Chance of a Loss” (1973) 37 Sask. L. Rev. 193; B. Coote, “Chance and the Burden
of Proof in Contract and Tort” (1988) 62 Austral. L.J. 761; R.W. Wright, “Causation,
Responsibility, Risk, Probability, Naked Statistics, and Proof: Pruning the Bramble Bush by
Clarifying the Concepts” (1988) 73 Iowa L. Rev. 1001 at 1067-77; R.A.B. Bush, “Between Two
Worlds: The Shift From Individual To Group Responsibility in the Law of Causation of Injury”
(1986) 33 UCLA L. Rev. 1473.
1990]
RISK EXPOSURE AS INJURY
tiff, and loss of a chance cases. That unified approach may be described as a
“risk as injury” thesis. This involves application of a “probabilistic increased-
risk concept”,5 under which exposure to risk is classified as an actual injury
worthy of redress. Liability for risk exposure will permit the plaintiff to recover
in some circumstances when he or she previously would have been barred; how-
ever, the award will be discounted to the extent that it is likely that the plaintiff’s
injury was caused by risk factors other than the defendant’s negligence. Thus,
it is also referred to as “proportional recovery”. A simple example of propor-
tional recovery might run as follows. It is established that there is a thirty per-
cent chance that the defendant caused the plaintiff’s injury (but a seventy per-
cent chance that the defendant had nothing to do with the injury). As will be
explained, traditional causation law would deny any recovery to the plaintiff.
Proportional recovery would award thirty percent of the plaintiff’s damages
against the defendant. Such an award most precisely reflects the risk created by
the defendant and the damage suffered by the plaintiff through that risk.
The need for a unified approach arises from two distinct sources of confu-
sion. First, commentators have generally not attempted to resolve all three pat-
terns at once; most have chosen instead to tackle only one or at most two. The
result has been a patchwork of theories and proposals. In one instance, for
example, the consequence has been that the writer, while calling for some form
of proportional recovery, simply failed to realize the potential breadth of appli-
cation of his theory;6 at other times commentators have been unable to get
beyond the technical aspects of the particular approaches they were criticizing
in order to apprehend the existence of any underlying principle.7 Second, advo-
cates of proportional recovery have travelled differing theoretical routes in
arriving at their conclusions! At times this leads to differences in practical
application.9
5Wright, “Causation in Tort Law”, ibid. at 1814.
6See Delgado, supra, note 4 at 889; see also, Rosenberg, supra, note 4 at 876; Robinson,
“Probalistic Causation”, supra, note 4 at 796-97, where it is argued that proportional liability will
have only limited application to sporadic accidents.
7Much of the early criticism of attempts to introduce some form of proportional liability failed
to realize that market share liability was simply the manifestation of a higher principle; and more-
over that the form utilized to solve, i.e., the DES problem should be considered as nothing more
than a structure for applying that principle to a particular fact pattern. See, for example, J.B.
Newcomb, “Market Share Liability for Defective Products: An Ill-Advised Remedy for the
Problem of Identification” (1981) 76 Northwestern L. Rev. 300.
8See Bush, supra, note 4 at 1487-91 for a detailed discussion of how the commentators differ
in their approaches. Bush’s own theory is unique as well. Yet another proposal (awarding the bulk
of the compensation to the “most likely victim” of the defendant’s negligence) can be found in
Farber, supra, note 1.
9Perhaps the most marked division among commentators is in reference to whether to permit
individual plaintiff suits. See Bush, supra note 4, at 1487-91.
McGILL LAW JOURNAL
[Vol. 35
This article argues that risk-based liability is necessary to resolve the
dilemma created by causal indeterminacy. Part I establishes that injustice inev-
itably flows from the “all or nothing” requirement of current tort principles. In
Part II, the all or nothing and risk-based liability approaches are examined in the
context of the underpinnings of tort liability. Part HI discusses statistical meth-
ods of proof. Part IV demonstrates that a single integrated thesis underlies all
three sub-patterns such that the sub-patterns are merely extensions of each other.
Parts V through VII discuss the sub-patterns, and their application to risk-based
liability. In trying to solve the indeterminate defendant problem, the important
case of Sindell v. Abbott Laboratories0 provided the first explicit recognition of
a form of risk-based liability. Part V examines the principles embodied in that
case and how they evolved in subsequent decisions. Part VI illustrates how such
analysis is properly applied to the indeterminate plaintiff sub-pattern. Part VII
begins with a discussion of loss of a chance paradigms, and suggests that the
future foundation for risk-based liability rests on two decisions, Herskovits v.
Group Health Co-op” and Janiak v. Ippolito.2 Herskovits provides firm support
for the loss of a chance approach in the United States, while the unanimous
decision of the Supreme Court of Canada in Janiak is capable of significant and
exciting expansion in Canada. Part VIII very briefly examines whether the court
system is .the proper forum for the undertaking of such potentially momentous
reforms. Finally, this article concludes that risk-based liability is the only
method which can fairly resolve causal indeterminacy and yet maintain the
integrity of the causation requirement.
Before embarking on this discussion, three points must be made. First, the
consequence of the risk-based liability approach is to make the causation
requirement relatively easy to satisfy. The integrity of the causation inquiry is
maintained because risk is injury, and there is a resulting shift in the focus of
the analysis to damages assessment.’ 3 Cases such as McGhee v. National Coal
Board4 illustrate that the causation inquiry is unsuited to the task of unravelling
the riddle of causal indeterminacy. Our law has long required a “yes or no”
answer to the issue of factual causation, and seems incapable of moving from
that. Where there is causal indeterminacy, the result is ineffective, muddled, and
unfair law, which has finally led to a severe backlash. 5 The distinction made in
focusing on damages instead of causation involves more than semantics; put
simply, the courts need to use tools appropriate to the task at hand. The damages
‘0Supra, note 3.
“Supra, note 3.
12[1985] 1 S.C.R. 146, 16 D.L.R. (4th) 1 [hereinafter Janiak cited to S.C.R.].
13Nonetheless, it is the obstacle of causation which needs to be overcome, so the terminology
of this article will generally be causation-oriented.
14[1972] 3 All E.R. 1008 (H.L.) [hereinafter McGhee].
15See Wilsher v. Essex Area Health Authority, [1988] 1 All E.R. 871 (H.L.) [hereinafter Wilsher].
For a discussion of this case and McGhee, see infra, notes 182-93.
19901
RISK EXPOSURE AS INJURY
inquiry does not suffer from some of the logical restraints which bind causation,
and has the theoretical flexibility to master the complexity of risk-based liability
cases. The shift to a focus on damages assessment is thus an integral feature of
risk-based liability.
Second, the “indeterminate defendant”, “indeterminate plaintiff”, and “loss
of a chance” sub-patterns are truly only labels for general fact patterns which
parallel one another on a theoretical level. They are “three sides of the same
coin”, and each should generally be treated in an identical fashion. 6
Finally, a brief comment should be made on the proper scope of the risk
as injury thesis. Richard Wright argues that risk exposure has its place on the
“frontiers of tort liability”, 7 but his view of the theory’s potential breadth may
be somewhat narrow. For example, it may have application in some sporadic
accident circumstances.”8 But while it might be possible for risk exposure theory
to move to the mainstream of tort litigation, such a development is highly
improbable for the foreseeable future. More likely, the theory will find use
where ambiguous causal relationships exist. In other words, it will be used
where it is useful, and those circumstances are limited at present.
I. The Source of Injustice – The All or Nothing Approach
The specific obstacle to recovery which critics of the traditional causation
requirements are attempting to surmount is the “all or nothing” standard for
recovery. Under this standard, if the plaintiff meets the burden of proof (gener-
ally the balance of probabilities) regarding all the elements of the tort (includ-
ing, specifically, causation), then the court will treat the defendant’s causal
responsibility as a certainty. 9 Conversely, if the plaintiff fails to meet the bur-
den, it will be considered a certainty that the defendant’s actions imposed no
harm upon the plaintiff.
The flaws of the all or nothing approach are exposed in the increased risk
cases. In practical terms, the approach’s most glaring weakness is its denial of
recovery to almost all increased risk plaintiffs. For example, in the industrial
waste context, it is virtually impossible for an individual to show that her inju-
ries have been caused by the defendant. In many cases she can prove that the
defendant negligently disposed of the dangerous substance, that she was within
the area of risk, and that she suffered some injury. But under traditional
Canadian tort law she must also be able to show, on the balance of probabilities,
16See infra, notes 87-92 (Part IV).
17Wright, “Causation in Tort Law”, supra, note 4 at 1813.
‘8See infra, notes 143-44.
19Courts have, however, found ways to discount awards on other grounds –
in essence applying
the all or nothing standard to causation in all cases, but not always to the issue of damages. In
awards for loss of sight, for example, life expectancy will be calculated into the measure.
REVUE DE DROIT DE McGILL
[Vol. 35
that the defendant’s negligent conduct was the legal cause of her injuries.
Considerable controversy exists as to what is the threshold for the standard of
proof,’ but assume for argument’s sake that the plaintiff must show that it is
more likely than not (ie., greater than fifty percent chance) that the defendant
caused her injury.2 While the plaintiff may be able to show some connection
between the dangerous substance and the cancer through epidemiological evi-
dence, she will rarely be able to recover under traditional tort principles. Unless
she can show that there is a greater than fifty percent chance that the defendant’s
negligence led to her injuries, she will fail.’ This fact pattern is known as the
“indeterminate plaintiff” problem: although it is certain the defendant must have
caused some injuries, the plaintiff is unable to show under traditional causation
principles that she was one of the defendant’s victims. While a representative
action might be of assistance, access to such procedures is severely limited in
Canada.’
The industrial waste plaintiff’s problems are mirrored in other contexts.
Consider the injured person who goes to a doctor for treatment on her leg.
Assume that if the doctor correctly diagnoses and treats the injury, there is a
thirty percent chance the leg can be saved.’ However, the doctor is negligent,
and the patient loses her leg. Under the analysis applied in the industrial waste
context, she can not recover from the doctor because it was more likely than not
(seventy percent) that the leg would have been lost anyway. This is a case of
“loss of a chance.” It might also be referred to as a problem of “indeterminate
harm,” because it is unknown whether the plaintiff has been injured at all by the
defendant’s negligence.’
2See infra, notes 67-70.
21The plaintiff’s case will likely rest on epidemiological studies which compare the rates of can-
cer, for example, of the group exposed to the dangerous substance (of which the plaintiff is one)
to those of the larger population base which was not exposed. The increase in the incidence of can-
cer will be attributable to the defendant’s negligence.
22It is impossible for most plaintiffs to demonstrate this, as they would be required to produce
evidence to the effect that the defendant’s negligence more than doubled the cancer rate in the area
of risk.
23See infra, notes 205-12.
24In such a case, the chance the patient will heal fully is calculated in much the same manner
thirty out of one hundred people with a similar injury could
as is used in epidemiological studies –
be expected to heal fully.
25The plaintiff is trying to establish that a beneficial future event would have occurred if not for
the defendant’s negligence. Assuming that such an occurrence was not a certainty, there exists the
possibility that the event would not have occurred and that the defendant’s negligence has not actu-
ally cost the plaintiff any benefit.
This situation is paralleled where the plaintiff has been exposed, for example, to toxic waste, but
has yet to manifest any injury; epidemiological studies may be able to accurately determine the
plaintiff’s chance of contracting disease in the future as a result of that exposure. See infra, notes
194-201.
19901
RISK EXPOSURE AS INJURY
The third context in which the plaintiff might encounter similar difficulties
with the causation requirement also involves mass torts. United States courts
have already considered the “indeterminate defendant” question in a number of
jurisdictions. The diethylstilbestrol (DES) drug litigation poses the problem of
identifying which defendant company manufactured the particular drug taken
by the plaintiff’s mother.26 The drug was distributed under a generic label.’
Again, the plaintiff may be unable to prove on a balance of probabilities that any
single defendant caused her injury.
The all or nothing approach works reasonably well in the context of the
ordinary sporadic accident case “in which causal indeterminacy arises randomly
and always signifies a substantial chance that the defendant in fact harmed no
one.”‘
In adjudication of sporadic accidents, all or nothing is possessed of a
number of virtues, not the least of which is efficiency. The all or nothing rule
interacts well with the traditional presumptions of tort law –
for example, that
the plaintiff carries the burden of proof. The combination of the placing of the
burden and the all or nothing rule are the law’s solution to fact indeterminacy.29
Unless the plaintiff can overcome this obstacle, the court will refuse to inter-
vene; but because of the nature of causal indeterminacy, the plaintiff will virtu-
neither the plaintiff nor the defendant possesses sufficient
ally never succeed –
information to satisfy the burden of proof. At issue is a policy question con-
cerning the kinds of interests the law should be protecting; and despite that it
appears that maintenance of the all or nothing rule may be a display of “healthy
judicial conservatism”,31 a choice in favor of doing nothing to change the rule
26See infra, notes 94-136.
27For a detailed description, see Collins, supra, note 3 at 44.
28Rosenberg, supra, note 4 at 858. But circumstances may arise where the all or nothing
approach is inadequate to deal with sporadic accidents. See the hypothetical infra, note 141.
29Cooper, supra, note 4 at 215.
30Early commentary on Sindell focused on that case’s roots in Summers v. Tice, 199 P.2d 1 (Cal.
1948), particularly with respect to the shifting of the burden of proof of causation to the defendants.
Robinson, “Multiple Causation in Tort Law”, supra, note 4 at 729.
that in these cases neither party will be able to satisfy the burden –
matter, is dispositive.” Ibid. at 728-36.
The danger of the emphasis on the placement of the burden in cases of causal indeterminacy is
its placement, “as a practical
31J. E. Coons, “Approaches to Court Imposed Compromise- The Uses of Doubt and Reason”
(1964) 58 Northwestern U.L. Rev. 750 at 756.
McGILL LAW JOURNAL
[Vol. 35
is in fact a choice.32 The issues need to be confronted openly.33
,Efficiency and consistency are the primary virtues of the all or nothing
rule. Neither, however, provides sufficient justification for its application across
the board. They are desirable and necessary for dispensing justice, but expedi-
ence cannot be substituted for fairness on a scale such as the one involved in
this problem. 4 The increased risk cases, in all forms, represent substantial num-
bers of claims for substantial awards; and the cost of errors against plaintiffs,
many of whom have suffered catastrophic injuries, is high.
I. Tort Objectives and Causal Indeterminacy
It is thus necessary to consider the all or nothing rule and the probabilistic
increased risk approach in the context of the objectives of tort law. The most
important of these are compensation and deterrence. Each of these, and some
objectives of lesser importance, favors the classification of risk as an injury.
A. Compensation
Compensation is fundamental to tort law’s operation.35 In the indeterminate
defendant cases there is no doubt that the plaintiffs’ injuries resulted from neg-
ligence, so an approach based upon probability can comfortably award full com-
pensation; the all or nothing rule would usually leave the plaintiff without a
remedy (at least in the DES and asbestos cases). If applied in the form of a mod-
ification of the causation rule, proportional recovery in an indeterminate plain-
tiff case (i.e., toxic tort) operates in a rough and ready fashion. Critics charge
that proportional liability, in subverting the causation requirement, gives a
32Coons argues that where both parties offer equally strong arguments and the case is decided
on the basis of which party carries the burden of persuasion, “the all-or-nothing judgment appears
an arbitrary preference of one litigant over another.” Ibid. at 758. This needs to be taken one step
farther. In cases of causal indeterminacy, irrespective of how strong the respective parties’ argu-
ments are, the all or nothing rule constitutes an arbitrary preference for the party which is able to
assemble its case in the traditional manner (rather than relying upon probabilities). For a discussion
of the virtues of probabilistic and particularistic types of evidence, see infra, notes 65-86.
Robinson argues in the DES context that where the defendants’ conduct has been negligent and
the plaintiff has suffered injury, the law may recognise a policy favoring liability even though the
plaintiff cannot prove which defendant caused the injury. See “Multiple Causation in Tort Law”,
supra, note 4 at 736.
33Wright, “Causation in Tort Law”, supra, note 4 at 1815.
34″M]ere ease of administration and simplicity are no excuse for injustice.” Cooper, supra, note
4 at 217, referring to Chaplin, supra, note 3, and to the principle that difficulty in assessing dam-
ages is no bar to an award.
35Compensation is particularly important considering the nature and severity of the injuries suf-
fered by many of the plaintiffs who are prejudiced by the all or nothing rule. For a detailed descrip-
tion of a DES daughter’s injuries, see H.S. Abrahams and B.J. Musgrave, “The DES Labyrinth”
(1981-82) 33 S.Carolina L. Rev. 663 at 664-9.
19901
RISK EXPOSURE AS INJURY
windfall to the plaintiff.36 This fear is strongest in the context of loss of a chance
and indeterminate plaintiff cases, and is particularly acute when recovery based
on future risk of contracting disease is contemplated.17 The argument is that
some members of the plaintiff class (those who were in fact injured by the
defendant) will be undercompensated, while the rest (those who suffered no
injury at the defendant’s hands) will be overcompensated.”
This criticism is accurate, however, only if recovery is granted with resort
to a blurring of the causation requirements (in order, for example, to compensate
a future injury).39 The risk exposure thesis preserves the causation requirement
rather than relaxing or subverting it. The crucial point to understand is that the
risk as injury approach reformulates or expands the concept of damage to
embrace tortious exposure to risk as a legal injury. A new legal interest worthy
of protection is recognised. The plaintiff must always show on the balance of
probabilities that the defendant caused her exposure to the risk. There is never
a windfall, because in each case where compensation is awarded the plaintiff
will have suffered a legal injury. The fairness of a particular award will not
depend upon the ultimate development of disease, which, in the risk exposure
cases, is not the injury being litigated.4 This approach is perfectly consistent
with the desire to conduct an overt (rather than covert) analysis of policy and
principle to resolve the causation dilemma. There is no sleight of hand here.
360f course, the question of windfalls depends upon which party has the stronger case. Where
the defendant is likely to succeed under the all or nothing analysis, he or she will object to the intro-
duction of proportional liability because it will confer a windfall.
37See “Proof of Cancer Causation”, supra, note 4 at 2107. Andrues writes:
“[]nherent in the use of statistical analyses is the very real possibility that a plaintiff
will never develop cancer. Any present recovery of damages for an injury which never
materializes creates a windfall recovery for the plaintiff.”
Ibid. at 2107. See also Wright, “Causation, Responsibility, Risk”, supra, note 4 at 1072.
38Delgado, supra, note 4 at 892. See also D.S Pegno, “An Analysis of the Enhanced Risk Cause
of Action (Or How I Learned to Stop Worrying and Love Toxic Waste)” (1988) 33 Villanova L.
Rev. 437 at 461. See also Anderson v. W.R. Grace & Co., 628 F. Supp. 1291 (D. Mass. 1986). The
Anderson court cites at 1232 a passage from Arnett v. Dow Chemical Corp., No. 729586 (Cal. S.C.,
Mar. 21 1983), which states that “[t]o award damages based on a mere mathematical probability
would significantly undercompensate those who actually do develop cancer and would be a wind-
fall to those who do not.”
39Pegno, ibid. at 461 n.146.
4As well, the choice of remedy can help to prevent over-compensation. Perfect compensation
(or as near it as can ever be possible) can be ensured by awarding to the risk exposure plaintiff
who is claiming for future injury the cost of premiums for insuring against the eventual occurrence
of that injury. This argument will be developed further at a later point: see infra, notes 213-23.
410f course, in assessing damages the award is discounted by the chance that the defendant’s
negligence did not cause the plaintiff’s physical or economic loss. This is the only point in the
inquiry at which the physical or economic loss is relevant; the plaintiff’s liability has already been
established as well as her legal injury –
exposure to a tortious risk.
REVUE DE DROIT DE McGILL
[Vol. 35
The real danger of windfalls exists under the traditional causation
approach. Suppose, in a toxic tort case, that each of ten plaintiffs can show a
seventy percent likelihood that the defendant caused their injuries. The defend-
ant will be liable for the full damages suffered by each plaintiff; three of the
plaintiffs have probably recovered a windfall. In an opposite case, none of the
plaintiffs will be able to prove their cases, and the defendant will receive a
windfall.
B. Deterrence
Concomitant with the fear of windfalls is a belief that proportional recov-
ery will over-deter defendants and discourage socially desirable activity.42
Optimal deterrence is achieved when the precise amount of accident costs cre-
ated by the defendant is passed on to him or her.43 As David Rosenberg states:
The threat of tort liability should induce rational actors to take ‘optimal care’ –
that is, to reduce the chance of accidents to the point at which the cost of any fur-
ther accident prevention measures would exceed the injury losses they would pre-
vent. Optimal care thus minimizes the sum of accident costs.44
Liability for risk creation provides a perfect fit from the defendant’s perspective,
because it is the injury to the class of plaintiffs which is important.45 That injury
reflects the whole of the risk created by the defendant, so liability for risk cre-
ation indeed achieves optimal deterrence.46 The key is that liability is attached
to the defendant’s conduct (the creation of the risk) rather than to its impact (the
resultant injury).47
In the context of causal indeterminacy, the all or nothing approach fails to
deter the creation of unreasonable risks.4″ Glen Robinson writes that:
Normally, of course, deterring the risk will deter the harm; nevertheless
legal rules that are forced on only the latter may not deter the former effec-
tively. The DES cases illustrate this point. To insist that a particular injury
be linked to a particular manufacturer’s product is to invite undeterrence of
the risk in every case where there is no proof of specific causation.49
42D.A. Fischer, “Products Liability – An Analysis of Market Share Liability” (1981) 34 Vand.
L. Rev. 1623 at 1629. See also L.J Chastain, “Market Share Liability and Asbestos Litigation: No
Causation, No Cause” (1986) 37 Mercer L. Rev. 1115 at 1129.
430f course, the defendant will in turn transfer these costs to the public through insurance or
higher prices. This sort of loss-spreading is also a function of tort law. See Delgado, supra, note
4 at 893; Robinson, “Probabilistic Causation”, supra, note 4 at 785.
“Rosenberg, supra, note 4 at 861-62.
45Delgado, supra, note 4 at 893.
“As it will even if risk exposure is not treated as an injury. For deterrence purposes it is only
necessary that liability equal the total of injuries caused.
47Delgado, supra, note 4 at 893.
4SRobinson, “Multiple Causation in Tort Law”, supra, note 4 at 740.
491bid. at 740. Parallels exist in the loss of a chance and indeterminate plaintiff contexts. A doc-
tor treating a patient who has only a thirty percent chance of survival is not deterred from negli-
gence by the all or nothing rule because he or she can never be made liable. Likewise, a company
19901
RISK EXPOSURE AS INJURY
Of course, not all risk creation will render the defendant liable. The law is
only concerned with deterring unreasonable risks. Defendants must be negligent
for the causation inquiry to have any relevance; and while this seems an obvious
prerequisite for all tort liability, the point seems to have escaped some commen-
tators and courts.50 That only tortious conduct will attract liablity is particularly
significant in light of the complaint that acceptance of liability for risk exposure
will have the effect of discouraging socially desirable activity.”1
Is tort liability an effective deterrent? Detractors of liability for tortious risk
creation are clearly convinced that it operates as a deterrent –
otherwise there
would be no reason for them to oppose it in fear of overdeterrence. Rosenberg
also believes that the threat of liability has deterrence value: “[b]oth the theory
of profit maximization and recent (though tentative) empirical data indicate that
firms are becoming increasingly sensitive to the prospect of tort liability. 52
Moreover, liability for risk creation will provide for defendants the motivation
and the ability to determine and minimize “the degree of excess risk attributable
to their activities.”53
negligently disposing of toxic waste need not fear liability as long as the background risk for the
plaintiff class exceeds the risk it is creating.
50See the discussion in Robinson, “Multiple Causation in Tort Law”, supra, note 4 at 741 n.106.
See also, Rosenberg, supra, note 4 at 867 n.68 (although he discusses this in the context of strict
liability). In tort law, a firm taking optimal care will be relieved of liability because it has met the
requisite standard of care.
51This argument is advanced in Fischer, supra, note 42 at 1629 & 1653-54 and in Chastain,
supra, note 42 at 1140. Fischer argues at 1654 that the deterrent effect of potential mass liability
will influence manufacturers to forego marketing risky new products, and that this may result in
the withholding of ultimately harmless products with great social utility.
The response is that a burden is properly placed upon the manufacturer to carefully test the pro-
duct and to balance the potential risks to the public against the perceived benefits. If, at the time
the decision to market is made, that decision is reasonable under this formula (which is no more
than the equivalent of the Learned Hand formula), then the defendant will not be found negligent.
Thus, if proper care is taken in the testing and the decision to market, the risk exposure approach
will not lead to overdeterrence; rather the effect is to encourage the defendant to exercise precisely
the care necessary to avoid liability.
Finally, while Fischer fears the imposition of crushing liability on defendants, it must be remem-
bered that they have the ability to spread their losses; plaintiffs who have suffered crushing injuries
will often have no other recourse. I do not raise this to promote deep pocket liability, but to show
that the defendant’s interest in avoiding a “crushing” loss is no greater than that of the plaintiff.
52Rosenberg, supra, note 4 at 862 n.51.
53Ibid. at 877. This statement holds true only for cases where the defendant possesses consid-
erable resources and/or knowledge, such as, for example, where the defendant is the manufacturer
of a drug causing birth defects or a doctor treating a patient. There is no logical reason, however,
why liability for tortious risk exposure should not extend to cases in the sporadic accident field
(see infra, notes 143-44); it seems unlikely that any theory of causation will have a considerable
practical deterrent effect in such circumstances.
McGILL LAW JOURNAL
[Vol. 35
C. Other Objectives: Knowledge and Justice
Risk-based liability should have the effect of generating knowledge. Under
the all or nothing approach no incentive exists for defendants to conduct
research into injury causation;54 liability for risk exposure will encourage
defendants to examine their own activities afresh in order to discover exculpa-
tory evidence.5 Such research may uncover more efficient methods of reducing
risk, and should reduce injuries over time.
Which approach will provide more accurate results? If one considers man-
ifested injuries to be the interest to be protected, proportional recovery will min-
imize errors in the long run. 6 It is argued, however, that our justice system
requires that the interests of individuals not be undermined in favor of fewer
errors over the long term. 7 In individual cases, the all or nothing rule will prod-
uce fewer numerical errors and pay out a smaller amount of money damages in
error. 8 But the types of errors created must also be considered –
in particular,
the number of large errors promulgated under each approach. Arguably, large
errors are more dangerous because they are more likely to inflict a crushing
result upon a party. Proportional liability will create significantly fewer large
errors.59
The criticisms directed at proportional recovery in the context of minimi-
zing errors have been predicated on the assumption that the interest protected
is the actual physical injury suffered by the plaintiff.’ The approach proposed
by this article dispenses with such analysis, and advocates the protection of the
plaintiff against risk creation. The injury is exposure to the risk. As such there
are no errors created, because in all cases the plaintiff is compensated for pre-
cisely that injury. Thus, in the sense of minimizing errors, liability for risk expo-
sure is far superior to the all or nothing rule.
Risk-based liability should carry the benefit of encouraging fairer settle-
ments for plaintiffs. Under the all or nothing rule, defendants have no incentive
to settle for more than a nominal sum. Negative publicity is the only risk to
54Delgado, supra, note 4 at 894.
55Ibid. This point also rebuts arguments that liability under risk exposure should be imposed in
the indeterminate defendant context where a defendant can show that he was not or could not have
been responsible for the plaintiff’s injuries. To impose liability in these situations would clearly
destroy the incentive to generate knowledge.
56M. Dant, “Gambling on the Truth: The Use of Purely Statistical Evidence as a Basis for Civil
Liability” (1988) 22 Columbia J. Law & Soc. Problems 31 at 49.
571bid. at 49-50. Bush, supra, note 4 is devoted almost entirely to refuting arguments such as
this.58Brennwald, supra, note 4 at 779.
591bid. at 781 (writing in the context of loss of a chance cases).
6See, for example, Fischer, supra, note 42.
1990)
RISK EXPOSURE AS INJURY
which they are exposed in proceeding to trial: there is virtually no chance of an
adverse judgment in most cases. The new theory will grant some recovery to
many more plaintiffs, so it enhances their bargaining positions for settlement
negotiations. Where plaintiffs and defendants make similar estimates of the
probability of causation, settlements will be encouraged; where estimates vary
widely, the question will likely end up in the courts.6″ In either instance, the
plaintiff is no longer being forced to accept a settlement imposed on the defen-
dant’s terms.
It is also argued that proportional recovery is unfair to defendants, because
it is no more than an application of a “‘deep pocket’ theory of liability, fastening
liability on defendants, presumably because they are rich”. 62 No doubt, plaintiffs
will be the primary beneficiaries of risk-based liability. It is fallacious, however,
to presume that the relative wealth of the parties bears on the theory’s validity.
Arguably, risk-based liability is capable of application in sporadic accident
cases, where defendants may be average individuals no better equipped to bear
the loss than a corporate defendant. Moreover, “rich” defendants may invoke
the theory to reduce their liability. In Hardy v. Johns-Manville Sales Corp.,63 the
defendant asbestos manufacturer successfully applied to court for leave to file
cross-actions against other manufacturers for contribution under market share
theory.
In summary, it is clear that the tort objectives of compensation and deter-
rence are satisfied through the imposition of liability for risk exposure. In addi-
tion, the generation and accuracy of knowledge should be enhanced. The pres-
ent all or nothing causation requirements fail to achieve these objectives.
Defendants are permitted to avoid liability in circumstances where their conduct
has caused statistically demonstrable losses.’ Justice demands the acceptance of
liability for risk exposure, which permits plaintiff recovery without unduly prej-
udicing defendants or deterring valuable activity.
II. The Problem of Statistics: Probabilistic v. Particularistic Evidence
The remaining objection to the acceptance of proportional recovery is
based on hesitancy regarding statistical evidence. It should be readily apparent
that “causation ‘ arguments in risk-based liability cases depend heavily on pro-
61Rosenberg, supra, note 4 at 897-98. But the court’s attention should be focused on this issue
of valuation, which may increase the efficiency of trials.
62Sindell, supra, note 3 at 941 (in Richardson J.’s dissent).
63509 ESupp 1353 (1981) (rev’d on other grounds).
64King, supra, note 4 at 1377.
65The evidence led to show “causation” does not really pertain to causation at all; in truth, to
show causation under risk-based liability it is only necessary to establish that the defendant created
a tortious risk to which the plaintiff was exposed, without reference to any actual injury. The bulk
of the evidence led will pertain to valuation, or assessment of damages.
REVUE DE DROIT DE McGILL
[Vol. 35
babilistic and statistical evidence, particularly epidemioligical studies. The
majority of these plaintiffs’ cases will fail, even with risk as a protected interest,
if probabilistic evidence is excluded. Whether to admit such evidence for the
purpose of demonstrating causation has been the source of a heated academic
debate.6
Unfortunately, the participants in the debate are so polarized that it is
nearly impossible for an unbiased newcomer to choose a position without resort
to simple intuition. This is perhaps appropriate, because at the core of the con-
troversy lies a question which begs an intuitive response: what is the probative
value of statistics?
The question arises because of uncertainty as to what constitutes the “bal-
ance of probabilities” or “more likely than not” threshold. Some argue that the
plaintiff must establish that there is a greater than fifty percent probability that
her submissions are true. 7 Others have argued for an “actual belief’6 standard;
this has been criticized as demanding too much from the plaintiff.69 In its place
has been suggested a non-probabilistic requirement that jurors be “inclined” to
believe the truth of the plaintiff’s allegations.’
Critics of statistical evidence have attempted to demonstrate its unreliabil-
ity through the use of various hypotheticals.7′ One of the most famous of these
is the “Blue Bus” case.’ Suppose Helen is struck by a bus driven negligently.
66Among those favoring the use of statistical evidence are J. Brook, “The Use of Statistical
Evidence of Identification in Civil Litigation: Well-worn Hypotheticals, Real Cases, and
Controversy” (1985) 29 St. L. U.LJ. 293; Rosenberg, supra, note 4 at 869-74; Gold, supra, note
4 at 379-92; D. Kaye, “The Limits of the Preponderance of the Evidence Standard: Justifiably
Naked Statistical Evidence and Multiple Causation” [1982] Am. B. Found. Research J. 487.
Others argue that liability should not be based solely on statistical evidence. See Dant, supra,
note 56; Wright, “Causation in Tort Law”, supra, note 4 at 1821-25; L.H. Tribe, “Trial by
Mathematics: Precision and Ritual in the Legal Process” (1971) 84 Harv. L. Rev. 1329; L.R. Jaffee,
“Of Probativity and Probability: Statistics, Scientific Evidence, and the Calculus of Chance at
Trial” (1985) 46 U. Pitt. L. Rev. 925.
67Dant, ibid. at 33.
68Wright, “Causation in Tort Law”, supra, note 4 at 1823-26.
69The requirement that the trier of fact be “convinced” or have an “actual belief” that the alleged
facts are true “is more than the preponderance standard should demand if it is to be distinct from
the competing standards of…’beyond a reasonable doubt.”‘ Dant, supra, note 56 at 59.
70Ibid.
71See, for example, the Gatecrasher case: L. Cohen, “Subjective Probability and the Paradox of
the Gatecrasher” [1981] Ariz St. L.J. 627. Only four hundred of one thousand attendees at a rodeo
have paid. If there is no other evidence, can the management pick one person, A, and sue for non-
payment on the basis that there is a 60% probability that A did not pay? See Dant, supra, note 56
generally and at 40, 52 & 54-58; Brook, supra, note 66 at 310-23 & 330-31.
72This hypothetical is presented by commentators with different variations in the facts, but the
principal at issue is always the same. See Brook, ibid. generally and at 298, 324-29 & 345-49;
Dant, supra, note 56 at 34, 40 & 42 (red cabs). See also Smith v. Rapid Transit, Inc., 58 N.E.2d
754 (Mass. 1945).
1990]
RISK EXPOSURE AS INJURY
She sues the Blue Bus Company, whose buses are blue in color, for the injuries
she suffers. At trial she offers no evidence as to the color of bus that struck her
(nor any other evidence regarding the particular bus or driver). She bases her
claim on the fact that eighty percent of the buses which use that street are blue
buses belonging to the Blue Bus Company, believing she has proven causation
on the balance of probabilities. The defendant complains that there is no evi-
dence to link its buses to the accident.73
Many commentators would require the presentation of “particularistic”
evidence that a blue bus injured Helen, ie., an eyewitness account. Statistical
evidence provides no information about the buses in particular, but merely the
proportion of blue buses to grey buses on the road. 4 On the other hand, if a wit-
ness were to testify to having seen a blue bus cause the accident,75 this partic-
ularistic evidence would identify blue buses specifically. While probabilities
assist in predicting the statistical likelihood of an event occurring over the long
run, only particularistic evidence provides a “rational basis for forming beliefs
about individual events.” 76
While it cannot be denied that probabilistic evidence fails to focus on the
particular defendant, the necessity of such a requirement can be questioned.77
According to Rosenberg:
[T]he entire notion that “particularistic” evidence differs in some significant qual-
itative way from statistical evidence must be questioned. “Particularistic” eviden-
in fact no less probabilistic than is the statistical evidence that courts purport
ce…is
to shun. All knowledge of past as well as future events is probabilistic. Inevitably
it rests on intuitive or more rigorously acquired impressions of the frequency with
which similar events have occurred in like circumstances.78
For example, if a person is witnessed slipping while walking on a patch of ice,
one naturally concludes that the ice caused the fall. That conclusion, however,
73Brook, supra, note 66 at 298.
74Dant, supra, note 56 at 40. See also L.L. Jaffee, “Res Ipsa Loquitor Vindicated” (1951) 1
Buffalo L.Rev. 1 at 3-4.
ness account…is among the most unreliable forms of proof.” Rosenberg, supra, note 4 at 872.
75Probabilists question the perceived reliability of such particularistic evidence. “[T]he eyewit-
76Dant, supra, note 56 at 46.
Brook sets out to discover why most people are made uncomfortable by the use of statistics to
impose liability in the Blue Bus case. He argues that it is the very “nakedness” of the statistical
evidence which militates against its use: that no other evidence (ie., of a particularistic nature, such
as an eyewitness account) is being presented. He writes that:
If the situation is such that we could reasonably expect the plaintiff to have other, more
traditional types of evidence if his factual contentions are correct, then his failure to
bring such other evidence into the courtroom understandably adds to our doubt that the
facts are actually as the plaintiff would have us believe.
Supra, note 66 at 324.
77Brook, supra, note 66 at 333-40; Rosenberg, supra, note 4 at 869-73.
78Rosenberg, ibid. at 870.
McGILL LAW JOURNAL
[Vol. 35
is arrived at from past experience which tells us that the frequency of falls
increases under icy conditions.79 It is a probabilistic inference based on obser-
vations of the frequency of falls under different conditions.” So, it is argued, if
all evidence is ultimately probabilistic, why should explicitly probabilistic evi-
dence be excluded?
If the commentators are divided on the general use of statistical data, they
are agreed that it is appropriate for use in a toxic tort or Sindell-type situation.8
There, every defendant has engaged in tortious conduct, 2 and each will have to
pay out the equivalent of the total damages it caused. 3 In contrast, there is a dis-
tinct possibility that the Blue Bus Company has done nothing wrong. 4
Additionally, the preceding discussion is largely moot in the context of
risk-based liability. Mary Dant suggests that, in arriving at a decision, the trier
of fact is searching for the best explanation, and as such needs only to believe
the plaintiff’s explanation over that of the defendant in order to find for the
plaintiff (or vice versa).8 5 In risk as injury cases, that explanation relates not to
whether the defendant actually caused the plaintiff’s injury, but to whether the
defendant imposed a tortious risk which could have led to the injury. Once tor-
tious risk creation is shown, the statistics are used to measure the plaintiff’s
compensable loss.86 Regarding the causation issue, the statistics merely assist in
making an inference to causation. In a toxic tort case, for example, there will
be additional evidence as to the defendant’s negligence, and the type and mag-
nitude of risk the release of toxic substances is likely to create. Thus, in risk as
injury cases, the statistics will not be used to prove that the defendant was
negligent.
791bid. at 870 n.77.
SThe response is that the trier of fact does not reach a conclusion based on inferences from sta-
tistics, but rather is searching for the best or most plausible explanation. Thus if more evidence
(such as eyewitness testimony, or evidence that no other company’s buses were in the area at the
relevant time) were provided in the Blue Bus case, the best explanation might be that a blue bus
did indeed strike Helen. The proper role of statistics is to assist in inferring to the best explanation.
Dant, supra, note 56 at 53-56.
81Ibid. at 61-68; Brook, supra, note 66 at 341-51; Rosenberg, supra, note 4 at 869-73; Wright,
82Brook, ibid. at 346-48.
83Dant, supra, note 56 at 68. This merely restates the point that risk-based liability creates per-
“Causation in Tort Law”, supra, note 4 at 1825.
fect deterrence for each defendant.
84This unwillingness to impose liability upon a potentially innocent defendant is reflected in
Sheffield v. Eli Lilly and Co., 192 Cal.Rptr. 870 (App. 1983) [hereinafter Sheffield]. See infra, notes
142-44.
85See supra, notes 70 & 80.
86Brennwald, supra, note 4 at 786.
19901
RISK EXPOSURE AS INJURY
IV. Risk Factors: Binding the Patterns Together
As stated earlier, “indeterminate defendant”, “indeterminate plaintiff”, and
“loss of a chance” are no more than convenient labels used to designate classes
of circumstances which fall within one larger pattern. Each of the classes, or
sub-patterns, possesses distinct features the details of which require individual
treatment; nevertheless, the overall approach to each must be identical.
Thus, it is necessary to examine how they interact. In the past, where there
has been any attempt to link the subpatterns, the relationships between them
have been largely misunderstood. A typical approach is to treat the indetermi-
nate defendant sub-pattern as the direct inverse of the indeterminate plaintiff
sub-pattern. In the former pattern, a single plaintiff knows she has suffered
injury at the hands of one of a number of potential defendants, but does not
know which caused her actual injury. For the indeterminate plaintiff, the scene
is reversed. Now, the plaintiff is one of a number of “victims”, some of whom
have been injured by a defendant’s negligence, but “who are unable to deter-
mine which among them has suffered injury at the defendant’s hands.”‘
Yet, such an approach is misleading. Richard Delgado states that the uncer-
tainty in the indeterminate defendant problem occurs at the “origin or the start-
ing point of causation”.88 Conversely, the uncertainty for the indeterminate
plaintiff occurs at the “terminus”89 of causation. While this characterization
appears correct on the surface, the statement is an incorrect description of the
patterns’ interrelationship.
To illustrate that the sub-patterns operate within the same parameters, con-
sider the following hypothetical. If a worker in a factory containing asbestos
products develops cancer, but cannot prove which manufacturer produced the
asbestos, he has an indeterminate defendant problem. If he also cannot prove
conclusively that the cause of the cancer was asbestos exposure (ie., if he does
not develop a “signature disease”) he is an indeterminate plaintiff. If his disease
prevents him from pursuing an employment opportunity (which he might have
secured) then he is also a loss of a chance plaintiff. Thus, the sub-patterns are
simply extensions, rather than inversions, of one another.
Consider how one’s analysis of an indeterminate defendant problem would
be altered if it were shown that background risks existed (ie., that there was a
chance that the injury occurred irrespective of any negligence).9″ All non-
tortious risks must be accounted for. Each “victim’s” injury is the consequence
of exposure to risk factors. Those risks may be entirely natural, they may be the
87Delgado, supra, note 4 at 882.
“Ibid. at 883.
89Ibid.
90See infra, note 142.
REVUE DE DROIT DE McGILL
[Vol. 35
result of non-tortious human conduct, or they may arise from negligence. In the
DES cases (indeterminate defendant) it is clear (because of the unique nature of
the injuries suffered) that the injuries were caused by negligence. Thus it is
unnecessary to consider other risk factors. The plaintiff simply does not know
which defendant created the risk which caused the actual injury. In the indeter-
minate plaintiff sub-pattern, the plaintiff does not know which risk factor, neg-
ligent or natural, caused the actual injury. In the loss of a chance sub-pattern,
the plaintiff’s “condition”‘” might not have been bettered even if the defendant’s
negligence had not occurred. There were risk factors operating to prevent the
plaintiff from achieving success, some tortious and some not. No one knows
which of these was the operating factor.
The solution to all of these sub-patterns is to compensate for tortious expo-
sure to risk. The risk as injury inquiry consists of identifying and quantifying
as well as possible the sources of risk (risk factors), and compensating the plain-
tiff for exposure to those arising from negligent conduct;92 no compensation will
be forthcoming for non-tortious risk factors.
V. The Indeterminate Defendant
A. The DES Dilemma
The extension of liability for risk exposure has been largely misunderstood
in the indeterminate defendant context. Plaintiffs first made inroads toward
recovery in Sindell, where the “market share” theory was created.93 But while
that case represented a significant advance toward the recognition of exposure
to risk as a protected interest, it failed to account sufficiently for the existence
of a broader underlying principle. As a result, Sindell itself was fundamentally
flawed, and it spawned much criticism. Subsequent cases attempted to remedy
Sindell’s flaws, but again usually fell short of achieving a satisfactory solution.
The classic case of the indeterminate defendant problem arose in the DES
cases (of which Sindell was one). DES is a man-made estrogen synthesized by
British scientists in the 1930s.1 Originally, the drug was intended for use in
treating certain female disorders believed to stem from low estrogen levels.95 In
1947 the U.S. Food and Drug Administration (FDA) approved the use of DES
91In this context “condition” is used in the broadest sense to connote the plaintiff’s overall state
of well-being, rather than in reference to any particular medical condition.
92It will be remembered that once causation of some degree of risk is established (that being the
injury), the detailed portion of this inquiry is conducted under the measure of damages heading,
where risk as injury is an accepted notion. See K.D. Cooper-Stephenson & I.B. Saunders, Personal
Injuries Damages in Canada (Toronto: Carswell, 1981) c. 3. See also supra, notes 13-15.
93Supra, note 3.
94Collins, supra, note 3 at 43.
95Robinson, “Multiple Causation in Tort Law”, supra, note 4 at 718.
1990]
RISK EXPOSURE AS INJURY
to prevent miscarriages.96 From that time until 1971, DES was manufactured by
scores of drug companies,’ and prescribed to up to three million women. 9 In
1971 the drug was banned for use by pregnant women in response to studies
linking it to the development of a rare form of cancer (vaginal adenocarcinoma)
in the users’ daughters.99
DES daughters were confronted by a number of obstacles to recovery, pri-
marily that they were unable to prove which manufacturer produced the drug
that caused the cancer in any single case. The drug was marketed generically,”
meaning that the mother likely did not know whose DES she was consuming.
Moreover, the cancer does not manifest itself for at least ten to twelve years,01
and DES was likely not recognised as the cause of the condition until somewhat
later. The combination of these delays resulted in the destruction of records nec-
essary to prove the plaintiff’s case under traditional principles.”
The problems of proof presented by the latency periods were compounded
by the number of DES manufacturers. In the Collins case it was alleged that
between 1957 and 1958 at least one hundred and twenty companies marketed
DES in a particular dosage.’0 3 Hundreds of suits were filed against drug manu-
facturers as a consequence of DES use,” but only a few have been successful.1 5
The successes all resulted from relaxation of the causation requirement. The
most famous of these cases is Sindell, which imposed liability on defendants
according to their respective market shares.
Although the methods used to impose liability in the successful DES cases
are different in a number of ways from the theory proposed in this article, it is
instructive to examine those decisions. They reveal the weight of judicial con-
cern and the policy considerations underlying that concern, and they show the
evolution of market share theory toward liability for risk creation.
96Fordham Comment, supra, note 4 at 963.
97Estimates of the number of manufacturers range from ninety-four to three hundred. See ibid.
at 964 n.3.
981bid. at 965 n.6. The author notes that the number of DES daughters may be as low as half
a million, but this is considered a “conservative estimate”.
99Ibid. at 964-66 and n.5-12.
‘0’Collins, supra, note 3 at 44.
‘0 1Sindell, supra, note 3 at 925.
102Fordham Comment, supra, note 4 at 972 n.26.
’03Collins, supra, note 3 at 44. Five or six manufacturers, however, accounted for up to ninety
percent of the market for the drug. See Fordham Comment, ibid. at 977.
’14Robinson, “Multiple Causation in Tort Law”, supra, note 4 at 718-19.
’05Sindell, supra, note 3; Collins, supra, note 3; Martin, supra, note 3; McCormack v. Abbott
Laboratories, 617 F.Supp. 1521 (D.C. Mass. 1985) [hereinafter McCormack].
McGILL LAW JOURNAL
[Vol. 35
B. Sindell: Risk as Injury in its Infancy
The decision in Sindell was prompted by the persuasive Fordham
Comment. Market share theory modified the approach used in Summers v.
Tice,” a case virtually identical in facts and reasoning to Cook v. Lewis. 7
Summers and Cook shifted the burden of proof regarding causation to the
defendants to show that they were not responsible for the harm suffered by the
plaintiff. The shift was justified because both defendants were negligent toward
the plaintiff, their very negligence was the cause of the plaintiff’s inability to
identify who caused the injury, and if the court did not take action, it was likely
that both defendants would escape liability.”0 8
The fact patterns in Sindell and Summers bear striking similarities.”
Nevertheless, the Sindell court was unwilling to follow Summers because only
five of approximately two hundred companies manufacturing DES were joined
in the action; it was highly possible that none of the defendants joined made the
DES which caused the injury.”0 The court also declined to follow a number of
other approaches, such as enterprise liability, proposed by the plaintiff.”‘
Instead, the court created the market share theory, which required the plain-
tiff to join a “substantial share of the appropriate market”.”. Once the plaintiff
showed a complete cause of action in all other respects, and met the require-
ments of the exception,” 3 the direct causation requirement would be waived.
Liability would be apportioned to market share. The court anticipated that, with
1’6Supra, note 30.
101[1951] S.C.R. 830, [19521 1 D.L.R. 1. In each of these cases the plaintiff was injured when
two hunters shot negligently in his direction. The plaintiff was unable to show which hunter had
fired the shot that actually caused the injury, but in each case the court held both defendants jointly
and severally liable for the whole of the damages.
“8Sindell, supra, note 3 at 928.
l’9Fordham Comment, supra, note 3 at 987:
In each situation, all defendants are tortfeasors owing a duty of care to the injured
plaintiff. In both the DES cases and Summers, the tortious nature of each of the defen-
dants’ conduct was identical and created the same type of risk. Neither the plaintiff in
Summers hit by a bullet nor the DES daughter who developed cancer is at fault for
being unable to identify the one who caused his injury. In both cases the defendants
created the conditions which caused the plaintiff’s inability to identify – by shooting
simultaneously in Summers and by manufacturing a single drug under a variety of trade
names in the DES cases.
“”Sindell, supra, note 3 at 931.
“‘Ibid. at 928-935. See also Abrahams & Musgrave, supra, note 35 at 677-83; Currie, supra,
note 4 at 747-51 & 755-60.
” 2Sindell, supra, note 3 at 937.
” 3Supra, note 109. Regarding the requirement that the defendants’ conduct be identical and cre-
ate the same type of risk, courts have often required that the goods be “fungible”. See, for example,
Starling, supra, note 3 at 191.
19901
RISK EXPOSURE AS INJURY
a substantial share of the market represented, each defendant’s liability would
equal approximately the damages caused by the DES it had manufactured.”‘
The court justified its departure from traditional causation requirements in
a number of ways, 5 but most compelling was the need for flexibility:
In our contemporary complex industrialized society, advances in science and tech-
nology create fungible goods which may harm consumers and which cannot be
traced to any specific producer. The response of the courts can be either to adhere
rigidly to prior doctrine, denying recovery to those injured by such products, or to
fashion remedies to meet these changing needs.” 6
The case provided an ingeniously fair and simple solution; nevertheless, it was
not without its detractors, for the Sindell court failed to address many of the sub-
tle and complex difficulties raised by market share liability. Some courts flatly
rejected Sindell; 7 others saw the theory as unrefined but important, and capable
of (or receptive to) improvement.1
C. The Evolution of the Theory
The courts following Sindell showed an increased understanding of the
principles underlying market share theory. In Martin,”9 another DES case, the
Supreme Court of Washington found for the plaintiff for the same policy rea-
sons as the court in Sindell,” but declined to follow market share theory as for-
mulated in that case. The court had two primary concerns. First, the Sindell
court failed to define the term “substantial” share of the relevant market.
Compounding this omission, the court did not specify what would occur if the
plaintiff joined less than one hundred percent of the market. The implication is
that the defendants would have to pay one hundred percent of the damages
liability by forcing each of the joined
regardless.’ To do so would “distort”‘”
defendants to pay more than would be attributable to the risk it created.
The Martin court rejected the “substantial share” requirement altogether,
because it did not alter the probability that any particular defendant caused the
injury.”z Instead, upon proving the necessary elements, the burden would shift
“4Sindell, supra, note 3 at 937-38.
“5lbid. at 936.
6Ibid.
1
1 7See, for example, Payton, supra, note 3; Starling, supra, note 3.
118See Collins, supra, note 3; Martin, supra, note 3; McCormack, supra, note 105. In Sheffield,
supra, note 84, the court approved of the theory, but declined to use it because the circumstances
were not appropriate.
119lbid.
“2Ibid. at 382.
“-‘Ibid, at 380-81.
1-Ibid.
123Ibid. at 382.
REVUE DE DROIT DE McGILL
[Vol. 35
to each defendant to establish its market share.” Each defendant would have
the opportunity to exculpate itself by establishing that it could not have been
their DES which the mother ingested. 5 Any defendants unable to exculpate
themselves would be presumed initially to have equal shares of the market.
Each would be entitled to rebut this presumption by establishing its actual share
of the relevant market.” Defendants would be able to implead third party
defendants to reduce their presumptive shares, and if all defendants were able
to establish a share, any shortfall would be borne by the plaintiff.27 The plaintiff
is thus induced to account for as much of the market as possible, or risk being
unable to recover a potentially substantial portion of her damages.
The Martin court’s reasoning was wholeheartedly supported by the
Massachusetts District Court in McCormack,” which was particularly signifi-
cant because a Massachusetts court had previously rejected the pure Sindell
approach.’29 Garrity J. conceded it was inevitable that defendants would be held
liable to plaintiffs whose injuries they did not cause, but argued that:
[U]nder market-share theory, a plaintiff must first prove that a defendant acted tor-
tiously before any liability may be imposed. Consequently, a defendant who erro-
neously is held liable to a particular plaintiff can not be considered wholly inno-
cent of wrongdoing. Such defendant, by engaging in the conduct found to be
negligent, contributed to the risk of injury to the public in general and conse-
quently shares some degree of culpability in producing or marketing DES. 130
The modifications made in Martin and McCormack brought market share the-
ory closer to liability for risk exposure. No longer would fewer than one hun-
dred percent of potential defendants be held liable for the whole of a plaintiff’s
damages, and a plaintiff would not be required to name more than one defendant
in the action.13 1
14Ibid.
1″5Ibid. A defendant might argue, for example, that it did not distribute the drug during the time
it was prescribed to the mother, or that it did not market the drug in the relevant geographical area.
l261bid. at 383.
1271bid. The Martin approach would operate in the following manner. If X and Y are defendants
who are unable to exculpate themselves, and the plaintiff’s damages are $100,000, the initial pre-
sumption is that each is liable for $50,000. If X can show it held only twenty percent of the relevant
market and Y fails to prove its market share, X will be liable for twenty percent of the damages
($20,000) and Y will be liable for the remaining eighty percent ($80,000). If X establishes a twenty
percent market share and Y a forty percent market share, X will again be liable for $20,000 and
Y for $40,000. The plaintiff will be unable to recover the remaining $40,000 because she failed
to account for forty percent of the market.
reasons as offered in Martin. See text accompanying notes 120-23.
1-8Supra, note 105.
’29Payton, supra, note 3 at 189. The court chose not to follow Sindell’s lead for largely the same
130McCormack, supra, note 105 at 1527.
131A plaintiff is, of course, putting herself at risk in naming only one defendant. If the plaintiff
is unable to prove her case against that defendant, or the defendant turns out to be judgment-proof,
1990]
RISK EXPOSURE AS INJURY
Collins represented an attempt to understand the link between market share
liability and risk creation. In that case, since the market was fluid and many
companies did not have adequate records, 3 the court redefined the role of mar-
ket share in apportionment (it was to be considered a “relevant factor”‘ 33). There
was brief discussion of Robinson’s argument that “the critical point is the cre-
ation of a risk that society deems to be unreasonable, not whether anyone was
injured by it.” ’34 Nevertheless, the court was somewhat unsettled by that
approach and was not prepared to see risk creation as a protected interest. The
court in Collins wrote:
Although we find Robinson’s ‘risk contribution’ theory sound to the extent that it
recognizes that all DES drug companies contributed in some measure to the risk
of injury, we do not agree that this is a sufficient basis in itself for liability. We
still require it be shown that the defendant drug company reasonably could have
contributed in some way to the actual injury.’1
Despite its discomfort, the court proposed a mode of apportionment closely
approximating that which would result if risk creation were considered the key
to recovery. Liability would be apportioned according to comparative negli-
the liability of each defendant would mirror the percentage
gence principles –
of causal negligence attributable to it.136
D. Extending Sindell
Simple market share liability is limited. It is an adequate method of appor-
tionment in the DES cases, but has virtually no practical application otherwise.
It can only work in product liability cases where there is in fact a “market” and
the defendants have created the same type of risk. It works in the DES context
the statute of limitations may expire before the plaintiff can explore alternatives for recovery.
Collins, supra, note 3 at 51.
132Ibid. at 48.
133Ibid. at 49.
134Robinson, “Multiple Causation in Tort Law”, supra, note 4 at 739.
135Collins, supra, note 3 at 49 n.10.
1361bid. at 53. The court offered the following factors as important in determining each defen-
dant’s proportion of liability:
[W]hether the drug company conducted tests on DES for safety and efficacy in use for
pregnancies; to what degree the company took a role in gaining FDA approval of DES
for use in pregnancies; whether the company had a small or large market share in the
relevant area; whether the company took the lead or merely followed the lead of others
in producing or marketing DES; whether the company issued warnings about the dan-
gers of DES; whether produced or marketed DES after it knew or should have known
of the possible hazards DES presented to the public; and whether the company took
any affirmative steps to reduce the risk of injury to the public.
Ibid. While these factors represent an attempt to more accurately guage the quantum of risk created
by each defendant, many are not relevant to this task. See infra, notes 148-50.
McGILL LAW JOURNAL
[Vol. 35
because the resultant apportionment of liability generally reflects contribution to
risk.
Sindell-type liability cannot be extended without a broader theoretical
foundation. To date, courts have been unwilling to expand market share liability
beyond DES cases. When liability is tied to risk creation, and exposure to risk
treated as an actual injury, far more flexibility is permitted and the criticisms
directed at Sindell are easily answered. It becomes obvious that only one plain-
tiff need proceed at one time, and that a substantial proportion of potential
defendants need not be joined. Most importantly, a theory based on risk as
injury should assist plaintiff recovery in many more cases.
As noted, market share liability only functions when the defendants have
created the same type of risk. Liability for risk creation allows for much greater
flexibility as long as each defendant’s liability is in proportion to the magnitude
of risk created by his or her activity.’ Logically, the risk liability rule must be
considered to extend to “cases involving multiple and different risk-creating
activities.”‘ 39 Robinson examines this problem in detail and creates a hypothet-
ical in which Horace Tumor develops cancer, having been exposed to asbestos,
toxic waste, and negligently manufactured medication. 40 The source of each is
known, but it is not known which caused his cancer. However, it is possible to
estimate their respective contributions to the risk of cancer as forty percent,
thirty-five percent, and twenty-five percent respectively. 4’ Horace will likely
fail on any theory in arguing that any one of these defendants caused his injury.
Moreover, each risk is qualitatively distinct. But if the risk imposed by each
137Most of these instances have occurred in asbestos litigation. See, for example, Starling, supra,
note 3, in which the court denied recovery under market share for injuries allegedly caused by
asbestos. It concluded at 191:
[A]sbestos products are not fungible commodities. The injuries caused by asbestos
exposure are not restricted to asbestos products –
other products, such as cigarettes,
may have caused or contributed to the injury. Additionally, products containing asbes-
tos are not uniformly harmful – many products contain different degrees of asbestos
…. Thus ‘the total risk created by any manufacturer would be a function of both its share
of the market and the relative harmfulness of its products’; but a company’s market
share could not be adjusted for the latter relation….Neither could it be adjusted for the
possible harmful effect of nonasbestos products.
Risk-based liability answers all of these concerns.
It should be noted that asbestos litigation has not been a real issue in Canada. One reason for
this is that most asbestos exposure occurs at the workplace, and there is extensive Workman’s
Compensation legislation in place. See Ontario, Report of The Royal Commission on Matters of
Health and Safety Arising from the Use of Asbestos in Ontario (Toronto: Queen’s Printer, 1984)
(Chair: J. Stefan Dupre).
4 at 1498 (analyzing group responsibility).
13SRobinson, “Multiple Causation in Tort Law”, supra, note 4 at 750. See also Bush, supra, note
139Robinson, ibid. at 750.
14OIbid.
1411 have altered the percentages used in Robinson’s example.
1990]
RISK EXPOSURE AS INJURY
823
defendant is viewed as an actual legal injury, and it is possible to estimate the
contribution by each to the total risk,”4 there is no reason to deprive Horace of
recovery, irrespective of the fact that the risks are of different types.
Robinson has also argued that risk-based liability is not useful in the con-
text of sporadic accidents, such as automobile collisions.’43 In the average case
involving a collision he may be correct, but sporadic accidents may generate sit-
uations in which liability for risk creation is required to solve causal indetermi-
nacy. Imagine a situation where:
[A]n individual drove carelessly down a street where a pedestrian stood on the
curb waiting to cross. Just as the driver reached the point where the pedestrian
stood, a careless tree-trimmer working directly over the pedestrian’s position
dropped a tree limb. Simultaneously, a large, unleashed dog ran past the pedes-
trian. At that moment, something –
the pedestrian could not afterward say what
–
caused the pedestrian to fall, and he suffered a serious head injury.144
It will be difficult to estimate each causal agent’s respective contribution to risk,
but no more so than if a court were faced with one defendant (ie. the driver) and
forced to make an all or nothing determination. In any case, a court would be
forced to make such estimates if it were found that the plaintiff were contribu-
torily negligent. Thus, while the task of estimating each defendant’s contribu-
tion to risk may seem imposing, no court should refrain from use of risk-based
liability by reason of that obstacle alone.
E. The Significance of Breach in Risk-based Liability
Tort principles require that no liability be imposed unless it is established
that the defendant has acted tortiously. 45 Consequently, risk-based liability can-
not be used to resolve causal indeterminacy in cases of intermittent torts. Such
circumstances might arise where a defective unit of an otherwise safe generi-
cally marketed product is distributed and injures a person. Liability could be
142Later in the article it will become apparent that this example is simplistic: for example, what
are we to do if it is discovered that the drug, although a potential cause of the cancer, was not man-
ufactured negligently (the drug company met the standard of care)? Alternatively, how does the
theory deal with risks which are known to exist, but cannot be attributed to any defendant (ie.,
background risks)? With regard to the first question, no liability can be imposed against a defend-
ant who has not been shown to have acted tortiously. With regard to the second, the solution under
an integrated theory of liability for risk exposure is simple. These non-tortious risks must be
accounted for in apportionment, as would be done in a loss of a chance or indeterminate plaintiff
case; see Robinson, “Multiple Causation in Tort Law”, supra, note 4 at 759-64. Thus, Horace’s
recovery will be discounted to the extent to which it is possible that a non-tortious source of risk
could have caused his injury. This problem did not arise in Sindell because of experts’ ability to
link adenocarcinoma to DES.
143″Probabilistic Causation”, supra, note 4 at 797.
144Bush, supra, note 4 at 1499.
145See, for example, Wright, “Causation, Responsibility, Risk”, supra, note 4 at 1075.
REVUE DE DROIT DE McGILL
[Vol. 35
imposed on each manufacturer of the product “in proportion to the manufactu-
rer’s risk contribution as measured by its long-run accident (or defect) rate.”’46
To do so, however, would be to extend risk-based liability beyond its proper
bounds: risk would be measured as an average over a series of transactions,
rather than in terms of a single transaction. Defendants who were creating no
risk at the time the plaintiff was injured could be held liable; one necessary jus-
tification for recognising risk exposure as a category of injury is that innocent
defendants will never be exposed to liability.’47
The nature or existence of the defendant’s breach of duty is important in
yet another context. The Collins court listed factors which might be considered
in determining risk contribution in the DES cases. 4 s Among those factors were
the defendant’s role in gaining FDA approval for the drug, whether the defend-
ant was an industry leader in producing or marketing the drug, and whether it
took affirmative steps to reduce the risk to the public. The last of these factors
can have an impact in two ways. If the steps taken succeed in reducing the level
of risk created by the defendant, its liability should be reduced accordingly; if
those steps are sufficient to meet the standard of care imposed on the defendant,
it will be exculpated of any liability. This involves no more than application of
standard tort principle.
The first and second factors point to a different problem. Suppose a DES
manufacturer is exculpated because it was able to show that it did not manufac-
ture the DES taken by the plaintiff’s mother (for example, that she used a dif-
ferent dosage than that produced by the manufacturer). One commentator
argues that under the Collins formula this would not mean the manufacturer did
not contribute to the risk of injury, particularly if it had been an industry
leader.’49 If so, it would be wrong to exculpate a defendant simply on the basis
that he or she did not directly contribute to the plaintiff’s risk.
It seems a formidable obstacle, yet the solution is quite simple. Causation
must be relevant to the alleged tortious conduct. In the DES cases the plaintiffs
are claiming that the manufacturers negligently failed to adequately test the
drug and then negligently marketed it. 50 They can only be held responsible for
consequences attributable to these acts. Whether the defendant was an industry
leader is irrelevant in measuring the risk created, because the plaintiff’s claim
146Rosenberg, supra, note 4 at 868.
147The California Court of Appeal took this view in Sheffield, where recovery was denied to a
plaintiff injured by a defective polio vaccine which was made from a safe uniform formula. The
court was not prepared to use Sindell causation to “furnish a key to unlock a treasure chest of a
shared liability indiscriminately imposed on manufacturers of safe and defective products of the
same nature.” Supra, note 84 at 878.
148See supra, note 136.
149Currie, supra, note 4 at 763-64.
‘5Sindell, supra, note 3 at 932.
1990]
RISK EXPOSURE AS INJURY
does not involve negligence as an industry leader. Thus these factors do not
belong in the causation inquiry.
F. The Operation of Risk-based Liability
How will liability for risk exposure operate in the context of the indeter-
minate defendant? To begin with, the plaintiff must establish that the defen-
dant’s conduct was tortious. The plaintiff should not be required to join a sub-
stantial share of potential defendants,’ but he or she will be induced to join as
many as possible because each defendant’s liability will be limited to its share
of the entire risk. 52 In addition only factors relevant to the alleged breach of
duty may be considered in measuring each defendant’s risk contribution. A
more difficult issue arises in deciding how to determine a defendant’s contribu-
tion. It is important to understand how risk-based liability operates. Contrary to
what has been suggested, there is no relaxation of the causation requirement,’53
nor is the burden of proof shifted to the defendant. 54 The integrity of the causal
link requirement is maintained.’55 Consequently, the Martin method, which pre-
sumes that all defendants created an equal share of the risk and permits them
to rebut the presumption by establishing a lesser share, is unfair. 56 Undoubtedly
the plaintiff would benefit, and the process of measuring risk might be simpli-
fied. But risk is treated as an injury to relieve plaintiffs of the injustice of factual
indeterminacy. As such it is no more than another protected interest, and no fur-
ther departure from traditional tort principles is warranted. So where the plain-
tiff is faced with risk indeterminacy, the burden of proof should not be shifted
to defendants. 57 Instead, both plaintiffs and defendants should present evidence
relating to risk contribution,’ and the courts will determine defendants’ respec-
tive contributions in a manner similar to that used in cases of contributory
negligence.
151See Martin, supra, note 3.
152Collins, supra, note 3 at 57. See also Currie, supra, note 4 at 765.
153See Robinson, “Multiple Causation in Tort Law”, supra, note 4 at 726 n.54.
154Sindell, supra, note 3 at 937.
155Wright, “Causation in Tort Law”, supra, note 4 at 1815.
156See supra, notes 124-27.
15TWhile it may be tempting to do so on the basis that the defendants are in possession of supe-
rior resources and knowledge, one must remember that not all defendants have such an advantage.
For example, to place such a burden upon defendants in the DES cases might impose inordinately
large shares of risk on small companies possessed of lesser knowledge because they were unable
to reduce their presumed shares.
158lnfonnation should be available to plaintiffs through normal discovery procedures, and courts
should be vigilant to ensure that parties comply fully with procedures.
McGILL LAW JOURNAL
[Vol. 35
VI. The Indeterminate Plaintiff
A. Redress for the Mass Exposure Victim
On the surface, the indeterminate plaintiff and loss of a chance sub-patterns
might be thought to be more closely connected to each other than to that of the
indeterminate defendant. While that is not the case, they do occasionally share
issues not prevalent in indeterminate defendant problems. These relate primarily
to the types of evidence admissible to show causation.” 9 Still, they deal primar-
ily with different fact patterns, the causation difficulties arise in different ways,
and each raises unique technical obstacles. Thus, they warrant separate
treatment.
Toxic tort cases form the typical indeterminate plaintiff pattern. Agent
Orange,16 PCBs,16 1 radon, 62 and potentially hundreds of other substances, 63
often working in concert,1″pollute the air165 and our waters,’ 66 putting millions
at risk of disease at any one time. 67 It is certain that many more hazards, present
and future, have yet to be discovered. 16
The nature of the hazards makes it impossible for most victims to establish
that the toxic or radioactive waste was the cause of their injuries. 69 Most of the
resulting diseases (ie., cancer) leave no physical “trail” to the inducing agent,
so observation of symptoms is often of little help in identifying the cause;” The
injuries may remain latent for years.17 1 They can “occur at background levels
159Such as epidemiological studies. See infra, notes 194-201.
6O’See, for example, Agent Orange, supra, note 3.
161See Pardy, supra, note 4.
162L.L. Gonsalves, “Probability of Causation in Radiation Tort Litigation” (1989) 24 Tulsa L.J.
479 at 482-83.
163See Delgado, supra, note 4 at 884 n.16; M. Mellon etaL., The Regulation of Toxic and Oxidant
Air Pollution in North America, (A joint project of the Canadian Environmental Law Research
Foundation, Toronto and the Environmental Law Institute, Washington, D.C.) (Don Mills, Ont.:
CCH Canadian Ltd., 1986) at 16-28; Law Reform Commission of Canada, Workplace Pollution
(Working Paper No. 53) (Ottawa: 1986) at 91-93.
164See Ayers, supra, note 3; and Andrues, supra, note 4 at 2075.
165See Regulation of Toxic and Oxidant Air Pollution, supra, note 163.
166See Ayers, supra, note 3; Pardy, supra, note 4.
167By 1981, asbestos exposure had increased the risk of contracting cancer for as many as
twenty-seven million people. See Rosenberg, supra, note 4 at 853 n.7.
a detailed analysis of this process, see generally Andrues, supra, note 4 at 2078-98.
16Slbid. at 854 n.19.
169The plaintiff’s first step is to identify the toxic agent as capable of inducing her injury. For
170Gold, supra, note 4 at 379.
171The minimum latency period for most cancers arising from exposure to radioactivity is ten
years. See Gonsalves, supra, note 162 at 495.
1990]
RISK EXPOSURE AS INJURY
without any apparent cause.”‘”2 Moreover, the victim usually is unaware when
the toxic exposure is occurring:’73 the exposure event is inconspicuous,174 and
there will be no identifiable irritation or traumatic injury to signify exposure. 5
Given the widespread and crushing nature of the injuries toxic tort victims
can suffer, the practical need for deterrence in this particular context is intensi-
fied. “[M]ass exposure torts are frequently products of the deliberate policies of
businesses that tailor safety investments to profit margins.”’76 No threat of lia-
bility means no economic incentive to invest in safety.177 As a result, some pro-
ducers of industrial waste feel no compunction about dumping wastes where
those wastes will inevitably and permanently scar the environment and its abil-
ity to support healthy life. 178
There does not seem to have been any reported toxic tort litigation to date
in Canada. Bruce Pardy’s article’79 is one of few examples in Canadian literature
to indicate that anyone has even considered the possibility.’ Doubtless, plain-
tiffs have been deterred by obstacles such as establishing causation, running a
class action, and obtaining sufficient damages to make a suit economical.’
Obstacles to recovery also have the effect of placing defendants in superior bar-
gaining positions in settlement discussions, and victims have likely been forced
to accept reduced compensation because of the negligible chance of success at
trial.
B. McGhee Overturned
In McGhee v. National Coal Board, the House of Lords adopted an
approach which might have assisted plaintiff recovery in these circumstances.8 2
In general terms, Lord Reid83 and Lord Wilberforce” u were prepared to shift the
burden of disproving causation to the defendant in certain cases of factual inde-
‘7 2Gold, supra, note 4 at 376.
173Pegno, supra, note 38 at 443.
174Rosenberg, supra, note 4 at 856.
175B. Black & D.E. Lilienfeld, “Epidemiologic Proof in Toxic Tort Litigation” (1984) 52
Fordham L. Rev. 732 at 744.
176Rosenberg, supra, note 4 at 855.
177Gonsalves, supra, note 162 at 507.
178Ibid.
179Supra, note 4.
18See also, Fleming, supra, note 4; L. Nissen, “Class Actions in Canada: An Environmental
Perspective” (1984) 48 Sask. L. Rev. 29.
’81″Highly Toxic Chemicals: Detection and Protection Methods” in Proceedings of a
Symposium, H.B. Schiefer, ed. (Saskatoon: Toxicology Research Centre, University of
Saskatchewan, 1985) at 92-97.
’82Supra, note 14.
’83Ibid. at 1010-11.
’94Ibid. at 1011-13.
REVUE DE DROIT DE McGILL
[Vol. 35
terminacy. Lord Wilberforce, whose reasoning has garnered the most Canadian
support,”8 5 argued that the defendant’s negligence caused the factual indetermi-
nacy and that the defendant should not escape liability because of the plaintiff’s
inability to show cause. That interpretation has since been discredited by the
present House of Lords. Lord Bridge states in Wilsher v. Essex AHA that
McGhee “affirmed the principle that the onus of proving causation lies on the
pursuer or plaintiff.”‘ 86 His Lordship argues that Lord Wilberforce was the only
member of the House of Lords in McGhee to advocate a burden shift and that
the majority opinion was actually expressed by Lord Reid, who made a “legit-
imate inference of fact that the defenders’ negligence had materially contributed
to the pursuer’s injury.”‘” McGhee, therefore, laid down no new principle of
law.
One can easily understand Lord Bridge’s aversion to the common interpre-
tations of McGhee. Lord Wilberforce’s approach represents precisely the sort of
falsification of causation requirements which risk-based probability is intended
to replace. To shift the burden to the defendant is simply to decide for the plain-
tiff, for the defendant has no greater knowledge.’ On the other hand, Lord Reid
chose to relax the burden upon the plaintiff, arguing that there is “no substantial
difference between saying that what the [defendants] did materially increased
the risk of injury to the appellant and saying that what [they] did made a mate-
rial contribution to his injury.”‘ 9 With respect, such a difference clearly does
exist. Lord Reid’s approach severs the connection between conduct and liability
such that accurate compensation and deterrence can not be achieved.
While McGhee is flawed, Lord Bridge’s argument that it broke no new
ground is difficult to accept. For over fifteen years Canadian courts’ 90 and aca-
demics’91 have seen McGhee as precisely laying down a new principle of law.
Moreover, it can be questioned whether the facts of McGhee are susceptible to
the kind of analysis which Lord Reid is supposed (by Lord Bridge) to have con-
ducted.’ 9′
’85See Nowsco Well Service Ltd. v. Canadian Propane Gas and Oil Ltd. (1981), 7 Sask. R. 291,
122 D.L.R. (3d) 228 (C.A.); Letnik v. Metropolitan Toronto (Municipality), [1988] 2 F.C. 399, 49
D.L.R. (4th) 707 (C.A.).
186Supra, note 15 at 881.
1871bid. at 881-82.
188Pardy, supra, note 4 at 288. See also supra, note 30.
189Supra, note 14 at 1011.
19See supra, note 185.
19’See, for example, A. Linden, Canadian Tort Law, 4th ed. (Toronto: Butterworths, 1988) at
101-103; Cooper-Stephenson & Saunders, supra, note 92 at 653.
192Professor Fleming explains:
…Lord Bridge sought to justify McGhee as having been based on an inference that the
defendant’s fault not merely increased the risk of dermatitis but had actually cumula-
tively to the causation of dermatitis. This explanation is weakened, however, by the
1990]
RISK EXPOSURE AS INJURY
In any event, Wilsher ought not to have a significant impact upon the risk
exposure thesis. In terms of substantive law, Lord Bridge’s analysis is confined
to McGhee and its interpretations. Any broader influence is likely to result from
the court’s attitude as illustrated in the case, and that attitude is better articulated
in Hotson, where their Lordships were not prepared to compensate for loss of
a chance.’19 3
C. Epidemiological Evidence
Once again, the solution is to classify risk as a type of injury. The plaintiff
would be required to show that he or she was exposed to the toxic substance,
that exposure to the substance caused an increased risk of disease, and the extent
of the increased risk.194 The plaintiff will rely primarily on epidemiological evi-
dence to provide the link between exposure and disease. 9′ “[Elpidemiology
involves the study of human disease processes within population groups.”’96 “It
seeks to establish associations between alleged causes and effects by one of two
methods: either comparing the incidence of disease across exposed and unex-
posed populations, or comparing the incidence of exposure across sick and
healthy populations.”‘”
In the context of all or nothing causation, the probative value of epidemi-
ological studies is limited: they cannot be used to prove causation conclusively
in any individual case. 9 ‘ Under risk-based liability, however, epidemiology pro-
vides precisely the sort of evidence the plaintiff needs. Epidemiologists can esti-
mate the magnitude of the association between the risk factor and the disease.’99
They can describe in percentages both the proportion of disease incidence
attributable to background risks,” and the increase in incidence resulting from
the defendant’s negligence.”‘
absence of any reason for drawing an inference of cumulative rather than alternative
causation.
Supra, note 4 at 670.
193See infra, notes 235-38.
194Pegno, supra, note 38 at 456; see also Gonsalves, supra, note 162 at 504.
195See generally, Black & Lilienfeld, supra, note 175 at 755-61. Some commentators believe that
epidemiological studies are the only evidence adequate for this purpose. See, for example,
Rosenberg, supra, note 4 at 856; Andrues, supra, note 4 at 2089. But see Gold, supra, note 4 at
393-94, where he argues that the classes of evidence admitted in these cases need to be broadened
to include, for example, animal and in vitro experiments.
196Andrues, ibid. at 2088.
197Gold, supra, note 4 at 380. See also Andrues, ibid. at 2089-92.
19SGold, ibid. at 380; Andrues, ibid. at 2094; Rosenberg, supra, note 4 at 857.
199Black & Lilienfeld, supra, note 175 at 757.
20oThe admissibility of statistical evidence for these and similar purposes has been disputed. See
supra at notes 65-86.
20’See Black & Lilienfeld, supra, note 175 at 757-61; Rosenberg, supra, note 4 at 857; Pegno,
supra, note 38 at 457. For the study to have validity, the epidemiologist must show a “statistically
significant relationship” between the disease and the alleged cause; in effect, that the difference in
McGILL LAW JOURNAL
[Vol. 35
D. Calculating Awards
Treating risk as an injury necessarily involves discounting the plaintiff’s
recovery according to the chance that his or her injury was not caused by the
defendant’s negligence. One commentator has suggested that while the all or
nothing standard should be abandoned, awards should be discounted intuitively
without attempting to arrive at an exact probability estimate.”a To prefer a
purely scientific calculation, ie. to seek a “mystically precise probability that
would determine the proportional recovery”, is to “embrace the myth of the
magic number.” 3 The argument has some merit, but the dispute may ultimately
be one of semantics. Generally the defendant can be expected to present its own
evidence as to the magnitude of the risk created, and it will be the task of the
court to determine how the award should be discounted. The final award may
be the result of compromise where there is little to choose between the opposing
experts’ opinions. In all cases, that final award must be calculated with refer-
ence to the plaintiff’s total injuries, and as such will constitute a percentage of
those injuries. Thus, the award will always reflect the court’s legal determina-
tion of the proportion of total risk which was created by the defendant. Stating
that the final award is arrived at through intuitive, rather than scientific, reason-
ing cannot be said to enhance the validity of that award, even if the scientific
estimate cannot be one hundred percent accurate.24
E. The Unavailability of Class Actions
The risk as injury approach may be of particular importance to the indeter-
minate plaintiff in Canada, for whom the option of a class action appears to be
unavailable. While United States courts have attempted to resolve the difficul-
ties of ensuring just compensation for each class member,”5 the Canadian
response in Naken v. General Motors of Canada Ltd.2″ was to define restric-
incidence rates does not arise from chance. Black & Lilienfeld at 757. Even then, it may be difficult
to make exact percentage estimates. Pegno, at 457.
22Gold, supra, note 4 at 399.
203Ibid. at 397.
204Additionally, because Gold’s approach does not recognize risk as a type of injury (if the
defendant’s conduct is found to be a substantial potential cause of the plaintiff’s injury, the link
will be considered close enough to warrant liability: ibid. at 395-96), it may be vulnerable to the
charge that there is an insufficient connection between the conduct and injury in individual cases.
Another theory for apportionment has been suggested in Farber, supra, note 1. The most likely
victim (MLV) approach awards nothing to those plaintiffs whose injuries were least likely to have
been caused by the defendant, and fully compensates the injuries of those with the highest causa-
tion probabilities. See at 559 & 581-89.
25 See Delgado, supra, note 4 at 901 n.97. But courts have shown some reluctance to certify
class actions. See Rosenberg, supra, note 4 at 908 n.224.
206[1983] 1 S.C.R. 72, 144 D.L.R. (3d) 385 [hereinafter Naken cited to S.C.R.].
1990]
RISK EXPOSURE AS INJURY
tively the meaning of “same interest” in Rule 75 of the Ontario Rules of
Practice, which reads:
75. Where there are numerous persons having the same interest, one or more may
sue or be sued or may be authorized by the court to defend on behalf of, or for the
benefit of, all.
In Naken, the plaintiffs attempted to file a class action suit for breaches of war-
ranty involving up to four thousand and sixty-two Firenza automobiles. The
Supreme Court dismissed the class action on the basis that the evidentiary dif-
ficulties outweighed the expedience of allowing such an action.’ Estey J. was
unsettled by problems in identifying members of the class 8 and the conduct of
discovery procedures and ordering of costs (because members of the class were
not parties to the action).’ The Ontario Law Reform Commission’s proposed
Class Actions Act illustrates that these difficulties are not insurmountable.21 The
effect of Naken, however, is to preclude use of class actions where damage
claims of class members must be assessed individually.211
In a toxic tort action, indeterminate plaintiffs will always require such sep-
arate assessment. Under any theory of proportional recovery other than “risk as
injury”, plaintiffs, forced to proceed one at a time, are vulnerable to the argu-
ment that epidemiological studies cannot be used to prove causation in individ-
ual cases. 212 In fact, it appears that an indeterminate plaintiff relying upon epi-
demioligical evidence to show causation must argue a theory of risk-based
probability in order to succeed.
F. No Manifested Injury: Is Accurate Compensation Possible?
Up to this point this section has dealt with the case in which the indeter-
minate plaintiff has manifested a physical injury. Recovery appears to depend
on classifying risk exposure as a legal injury. But if exposure to risk constitutes
an injury, is it necessary to require the plaintiff to have developed the threatened
disease in order to justify compensation? Suppose, for example, that it is more
probable than not that the plaintiff will never contract the disease. If the plaintiff
is awarded compensation, he or she might then be said to enjoy a windfall. 2 3
Additionally, if the disease does manifest itself, the plaintiff may then have been
20Nissen, supra, note 180 at 45.
208Naken, supra, note 206 at 98-99.
2091bid. at 99-100.
21 See Ontario Law Reform Commission, Report on Class Actions (1982) at 861, ss 3, 4, 7, 21,
and 31. See generally the Report for the policies behind these provisions.
2 1Nissen, supra, note 180 at 46.
2 12See supra, note 197.
213Andrues, supra, note 4 at 2107. But “the windfall so created must be set against the windfall
to risk creators under the present system by escaping all liability for tortiously creating risks”.
Robinson, “Probabilistic Causation”, supra, note 4 at 786.
REVUE DE DROIT DE McGILL
[Vol. 35
undercompensated –
the recovery will have been insufficient to redress the
actual loss suffered.214 One commentator, moreover, argues that such an exten-
sion of liability would undermine recovery for indeterminate plaintiffs who
have suffered physical injuries by invalidating the standards of recovery.215
It cannot, however, be argued that the plaintiff who has yet to manifest
injuries has suffered no loss. 16 Exposure to the excess risk attributable to the
defendant’s conduct “devalues” the victim.217 Additionally, there are real costs
associated with exposure to a risk of disease. Victims of risk exposure may wish
to undergo medical treatment to prevent or monitor possible development of
disease; such treatment would constitute a real and justifiable expense.218
A victim may also choose to purchase insurance against the development
of disease. Why not ensure that any injury he or she suffers is fully compensated
by awarding the cost of premiums? Any discomfort created by awarding dam-
ages to plaintiffs who have yet to manifest physical injuries should be counter-
balanced by the consideration that the plaintiff’s purchase of insurance is no
more than an attempt to mitigate the risk exposure inflicted; it is a legitimate
expenditure for which the defendant should be liable to reimburse the plaintiff.
214Robinson, ibid. at 786.
215Andrues, supra, note 4 at 2104. The author suggests, at 2111-16, that other types of measures
would be more appropriate for protecting the rights of the victim who has yet to develop disease,
such as preventing the triggering of statutes of limitations until the manifestation of disease, and
permitting plaintiffs to separate components of their claims. The latter would entail “suing on some
injuries in the present while preserving the right to recover for future injuries later.” Ibid. at 2112.
2161s the present injury requirement fair? Consider the following hypothetical: suppose a drug
company negligently makes and distributes a carcinogenic drug.
One commentator argues that there is already precedent in Canada for the application of risk-
based liability to possible future losses. See Pardy, supra, note 4 at 291.
217The devaluation might be reflected if the victim were negligently killed in an auto accident
shortly after the toxic exposure. In awarding damages against the driver, a court might discount
the victim’s life prospects by the chance of death or disability created by the exposure. “The excess
risk would therefore devalue the entitlements of the automobile accident victim in a very real
sense.” Rosenberg, supra, note 4 at 886.
218See Ayers, supra, note 3 (allowing recovery for cost of medical surveillance, but not for risk
itself). Early detection of diseases such as cancer can reduce the likelihood that contraction will
be fatal. Ibid. at 458-59.
[L]iability awaits only an injured victim. Two such victims appear: A, who has con-
tracted cancer as a result of taking the drug, seeks recovery for all losses suffered at
the time of trial plus the present value of estimated future losses (future medical
expenses, work losses, pain and suffering); B, who is at risk from having taken the
drug, seeks compensation for the present, actuarial value of the possible future losses.
Practical problems aside, is there any moral difference in these claims? Put somewhat
differently, what moral argument would [the drug company] have against B that it does
not have against A?
Robinson, “Probabilistic Causation”, supra, note 4 at 790.
1990]
RISK EXPOSURE AS INJURY
Moreover, awarding the cost of premiums may soothe fears that defendants will
be swamped with speculative and spurious claims.219
Such an award maintains the integrity of the risk as injury classification.
“If insurance is available at an actuarially fair rate, the premiums charged will
equal the potential injury loss discounted by its probability …. ,22 The plaintiffs
are allowed to insure themselves fully against the eventuality of contracting dis-
ease, so there is no windfall or shortfall;”l and the cost of that insurance should
exactly match the total sum which would be properly awarded to all those who
do eventually suffer injury, so there should be perfect deterrence.
The preceding approach is formulated to alleviate discomfort associated
with awards to plaintiffs who may never suffer physical injury. Risk-based lia-
bility in its pure form is untroubled by such considerations, because risk expo-
sure is a protected interest. If the theory advanced in this article is accepted, lia-
bility for risk of future losses is an inevitable consequence.222 In fact, there is
support for the proposition that Canadian courts have already moved in this
direction.’
VII. Loss of a Chance
In the loss of a chance or indeterminate harm cases, the plaintiff is suing
for the destruction of an opportunity to better, or to avoid a worsening of, her
overall condition.’ Medical negligence is the most common such circum-
stance. For example, by negligently failing to diagnose or properly treat a pro-
gressive disease the doctor deprives the patient of a chance to halt the advance
of that disease.’
To recover under an all or nothing standard, the plaintiff must show that the
disease probably would have been halted or slowed with proper care. The
Herskovits court recognized that adoption of this approach would constitute “a
blanket release from liability for doctors and hospitals any time there was less
than a fifty percent chance of survival, regardless of how flagrant the negli-
gence.””
219Andrues, supra, note 4 at 2104-05; Pegno, supra, note 38 at 460.
2 2-Rosenberg, supra, note 4 at 886.
221At the risk of stating the obvious, there would be no payout if the plaintiff never develops
the disease.
222For a formula for calculating awards for future losses, see infra, notes 252-53.
223See Cooper, supra, note 4 at 90-96.
4 See generally King, supra, note 4. See also, Brennwald, supra, note 4; Cooper, ibid. For an
opposing viewpoint, see Coote, supra, note 4.
225See Brennwald, ibid. at 749-51. See also Herskovits, supra, note 3; Hotson, supra, note 3.
226Ibid. at 477.
McGILL LAW JOURNAL
[Vol. 35
Joseph King further illustrates the necessity of moving to risk-based liabil-
ity with the following hypothetical:’ Suppose there is a jar which, it is estab-
lished, probably contains some pennies. It is estimated that the jar contains forty
coins, so the jar’s value would be forty cents. If the jar were negligently lost,
its owner would recover forty cents under the more likely than not standard.
Now suppose a plaintiff is deprived of a chance to pick one coin from a
jar holding one hundred coins, forty of which are worth one dollar, and sixty of
which are worth nothing. Anyone picking one coin at random will have a forty
percent chance of finding a dollar. Thus the estimated value of such a chance
to take a coin is forty cents, identical to the value of the first jar. The traditional
causation standard would deny recovery to the second plaintiff because it is
probable that a valueless coin would be drawn from the jar; such an analysis,
however, ignores that forty people in a group of one hundred would have found
a dollar.’m Where a chance is lost, one cannot know whether an actual injury
was suffered, so each plaintiff should be compensated for the risk of loss created
by the defendant.
Recovery for loss of a chance has been accepted by a number of courts.
The classic example is Chaplin v. Hicks, 9 where damages were awarded for the
loss of a chance to compete for a prize in a beauty contest.30 More recently, in
Herskovits, the Supreme Court of Washington was faced with circumstances
where the defendant doctor negligently failed to diagnose lung cancer in the
decedent, reducing the latter’s chance for survival from thirty-nine percent to
twenty-five percent.” The Herskovits court rejected the suggestion that the
plaintiff must fail without a showing that Herskovits had a fifty-one percent
hypothetical.
227Supra, note 4 at 1376-77. See also, Brennwald, supra, note 4 at 773-74. I have simplified the
2 2 Brennwald, ibid.
-29Supra, note 3.
tmThe injury addressed was the loss of the opportunity to win a prize, rather than the loss of
the prize itself. Chaplin, ibid. at 791, 795 & 798. See also Cooper, supra, note 4 at 197-99.
The award in Chaplin does not seem to have been probabilistically connected to the plaintiff’s
actual likelihood of winning a prize but the commitment to compensation for lost chances is unde-
niable; see Brennwald, supra, note 4 at 769 n.195.
For other cases where loss of an economically valuable chance was redressed, see Coote, supra,
note 4 at 767-68; Cooper, ibid. at 199-201.
23lSupra, note 3 at 475. Thus, it was likely the decedent would not have survived even with
proper treatment.
Epidemiological evidence is used to calculate risk contribution in medical negligence cases, For
example, the progression of cancer is classified according to the three developmental stages of the
disease. Cancer statistics are based on survival rates for each stage, representing how long victims
have “survived after a given date, whether it be the date of diagnosis or of the beginning of treat-
ment.” Estimates are phrased, for example, in terms of what percentage of Stage One victims of
a particular cancer will survive for five years (the standard testing period). Brennwald, supra, note
4 at 749 n.23-24.
1990]
RISK EXPOSURE AS INJURY
chance of survival with proper treatment,232 concluding that “the loss of
a…chance [should be recognized] as an actionable injury.” 3
In Hotson v. East Berkshire Area Health Authority,’ the Court of Appeal
permitted a plaintiff whose hip was injured and negligently treated, resulting in
permanent deformity, to recover for the loss of a twenty-five percent chance that
the permanent injury could have been averted with proper treatment.”3 The
House of Lords reversed that decision, but on the specific point that there was
no chance of avoiding the permanent injury. 6 Their Lordships declined to
decide whether an action could lie for loss of a chance, 7 but Lord Mackay con-
sidered Herskovits at length and treated it with some sympathy.”8
A. Janiak: The Foundation for Risk-based Liability
The Supreme Court of Canada was confronted in a unique context with an
argument that it should recognize the value of a chance. In Janiak v. Ippolito,”9
the plaintiff was negligently injured by the defendant in a car accident. It was
estimated that surgery to correct his back injuries carried a seventy to seventy-
five percent chance of success, but the plaintiff refused to undergo the opera-
tion. In doing so, he was unreasonably failing to mitigate his losses. The court
was asked to decide whether the plaintiff could still claim recovery for the pos-
sibility that the remedial measures would have failed.
Madame Justice Wilson was not prepared to treat the seventy to seventy-
five percent chance of success as a legal certainty, and compensated the plaintiff
for the remaining risk of continuing loss.41 In doing so, she was influenced by
a statement of Lord Diplock in Mallet v. McMonagle:
In determining what did happen in the past a court decides on a balance of prob-
abilities. Anything that is more probable than not it treats as certain. But in
assessing damages which depend upon its view as to what will happen in the
future or would have happened in the future had something not happened in the
past, the court must make an estimate as to what are the chances that a particular
thing will or would have happened and reflect those chances, whether they are
more or less than even, in the amount of damages it awards.242
232Herskovits, ibid. at 479 & 486-87.
2331bid. at 487. The majority explicitly adopts King’s approach at 486-87. See also Waffen v.
United States, 799 F.2d 911 (4th Cir. 1986).
234Supra, note 3.
2351bid. at 761, 764 & 768-69.
236Ibid. at 782, 789-90 & 782.
237Lord Bridge did note that he saw “formidable difficulties” in sustaining an analogy between
the circumstances in Chaplin, supra, note 3 and Hotson, ibid. at 782.
2381bid. at 786-89.
239Supra, note 12.
240Ibid. at 15-52.
241Ibid. at 169-72.
24211970] A.C. 166 (H.L.) [hereinafter McMonagle] at 176.
REVUE DE DROIT DE McGILL
[Vol. 35
Put simply, Lord Diplock is explaining that courts must discount awards
for future loss by the chance that the loss will not occur. This principle cannot
be limited to mitigation of damages, but speaks to future contingencies of any
nature. Madame Justice Wilson’s decision, and her endorsement of Lord
Diplock’s statement, must be seen as indicating support on the part of the
Supreme Court for a probabilistic approach to future loss. Pardy makes the point
even more strongly, arguing that as a result of Janiak, “proportional recovery for
a risk of future loss…is now the law in Canada. ”2 3
Such a development carries with it consequences which affect the whole
risk exposure thesis. Janiak speaks to future loss, but what if the event (or
injury) in question has already occurred?2′ It would be anomalous for courts to
compensate for risk of future loss, as in Janiak, but not for negligently imposed
risk which possibly created a present loss. This may be more easily understood
by examining the toxic tort (indeterminate plaintiff) cases, where recovery for
as yet unmanifested injuries (the paradigm of risk of future loss) is much more
controversial than recovery where the injuries have actually been suffered by
the plaintiff.US It is logical, then, that the McMonaglelJaniak principle be
extended to permit a proportional approach to damages assessment for not only
future but also existing injuries and losses.
One could argue that Wilson J. explicitly precludes such a conclusion with
the following statement in Janiak:
[Tihe balance of probabilities test is confined to determining what did in fact hap-
pen in the past. In assessing damages the Court determines not only what will hap-
pen but what would have happened by estimating the chance of the relevant event
occurring….
But while it appears that she is upholding the all or nothing standard for past
events, she does recognize that “[t]he potential for over or under compensation
is…a pervasive difficulty with the present ‘once and for all’ method of awarding
2 3Pardy, supra, note 4 at 291. See also Schrump v. Koot (1977), 18 O.R. (2d) 337, 82 D.L.R.
(3d) 553 (C.A.); Hearndon v. Rondeau (1984), 54 B.C.L.R. 145,29 C.C.L.T. 149 (C.A.). See also
Cooper-Stephenson and Saunders, supra, note 92, c. 3.
The Quebec Court of Appeal recently permitted recovery for loss of a chance of obtaining proper
care as a result of the physician’s failure to advise of biopsy results. See Laferrire v. Lawson
(1988), 49 C.C.L.T. 309 (leave to appeal to the Supreme Court of Canada granted June 8, 1989).
mSuch circumstances would include not only the medical negligence scenario, but also the
entire mass exposure pattern. It is worth repeating that the “indeterminate defendant”, “indetermi-
nate plaintiff’, and “loss of a chance” sub-patterns are merely classes of circumstances falling
within the one broad pattern of risk exposure cases, and the analysis in this section is applicable
to all of the sub-patterns. In other words, establishment of a theoretical foundation for any one of
the sub-patterns validates the other two sub-patterns and the entire risk exposure pattern.
5See supra, notes 205-07.
24
246Supra, note 12 at 170-71.
19901
RISK EXPOSURE AS INJURY
tort damages.” 7 Moreover, the court was not asked to review that standard in
a general sense. Most importantly, one must be remember the all or nothing
standard is to be used under the risk exposure thesis and specifically, to what
it is to be applied. Whether the plaintiff was actually exposed to a tortious risk
(and thus made the victim of a legal injury) must still be established on the bal-
ance of probabilities. Quantification of the risk is reserved for damages assess-
ment, where probabilistic analysis logically extends to future and existing
losses. Thus, it is submitted that Janiak represents the foundation for the future
development of risk-based liability in Canada.
B. Valuing the Loss of a Chance
In the indeterminate plaintiff context, Steve Gold argues that risk exposure
should be estimated subjectively by the trier of fact.” The simplicity of this
method, however, is outweighed by the desire to measure as accurately as pos-
sible the extent of the risk created by the defendant, particularly where medical
evidence of survival rates is available.”49 A better approach is to “measure a
compensable chance as the percentage probability by which the defendant’s tor-
tious conduct diminished the likelihood of achieving some more favorable out-
come.” This “simple probability” method of valuation awards damages pro-
portionally to the chance lost by the plaintiff.”1 If a defendant negligently denies
a plaintiff a thirty percent chance of healing his injured arm, the court would
assess the value of a healthy arm, and discount that value by seventy percent.
The simple probability method is ideal for determining awards where all
chance of achieving a beneficial result has already been lost, such as in a
Chaplin or Hotson situation. Where the case is one of prospective future loss,
King has developed a still more accurate method of valuation, the “weighted
mean” method. 2 He distinguishes its operation from that of simple probability
in the following example. The defendant’s negligence creates a thirty percent
chance that a healthy plaintiff will become blind in the future. Most likely, if
blindness does result, it will be at age fifty. The value of the plaintiff’s eyesight
at age fifty is $100,000. Under simple probability the court would award
$30,000, which is thirty percent of $100,000.
The weighted mean method takes into account the possibility that blind-
ness might develop earlier or later than at age fifty. Suppose it is established that
the plaintiff’s chances of becoming blind are twenty-five percent at age fifty,
2471bid. at 172.
248Supra, notes 202-03. This was the approach used in Chaplin, supra, note 3.
249Brennwald, supra, note 4 at 783.
25King, supra, note 4 at 1382.
25Brennwald, supra, note 4 at 783.
252King, supra, note 4 at 1384.
McGILL LAW JOURNAL
[Vol. 35
four percent at forty, and one percent at thirty. There is still a thirty percent
chance that the plaintiff will be blind by fifty. The value of the plaintiff’s vision
at age fifty is $100,000; at forty, $200,000; and at thirty, $300,000. Under the
weighted mean method, the total loss would be arrived at by “aggregating [each
of] the possible outcomes discounted to reflect their degree of likelihood. ” “i
The intricacy of the preceding method is justifiable. At this point in the
analysis the paramount goal is to reduce the number and size of errors in val-
uation. More complex testimony will be required than for the subjective
approach; but more important than the potential complexity is the danger that
the court will be forced to arrive at an award without meaningful guidance.’
VIII. Judicial or Legislative Reform?
Some argue that solutions such as the risk as injury thesis involve radical
departures from traditional notions of individual justice, and that where “such
a break from the whole tradition of our culture” is at stake, it is the place of leg-
islatures only, and not the courts, to intervene by introducing reform.”s The
answer to this is now almost a legal cliche, for (as must have been argued by
practically every freshman tort law student at one time or another) if this argu-
ment had been given weight by Lord Atkin in deciding Donoghue v. Stevenson,
the face of tort law would look rather different today.
This is not to say that legislative reform would not be welcome.
Rosenberg, a strong advocate of proportional liability, has undertaken a thor-
ough examination of how public law mechanisms (such as class actions and
insurance fund judgments) could be utilised in the causal indeterminacy con-
text.” But while he has concerns that judicial (as opposed to legislative) reform
might impair the productivity and efficiency of the justice system,u7 Rosenberg
also recognises that:
In effect, legislatures have declined the invitation to formulate a comprehensive
administrative solution –
in part because, at least for the foreseeable future, such
a solution seems politically infeasible. Thus the problem of the tortious use of
2531bid. To calculate, add $25,000 (twenty-five percent of $100,000), $8,000 (four percent of
$200,000), and $3,000 (one percent of $300,000) to reach a total of $36,000. Here the award is
higher than under simple probability, but that is merely a product of the probabilities and dates
used.
251bid. at 1385. Moreover, “[t]he calculations involved.., are no more complex than those under-
taken in a sophisticated claim for lost future wages, where future bonuses, benefits, and promotions
must be factored into a final damages award.” Brennwald, supra, note 4 at 784.
255Fleming, supra, note 4 at 669.
256Supra, note 4 at 905-24.
257The common assumption is that a rule of proportional recovery would invite a flood of lit-
igation. This may well be true, but Rosenberg argues that exactly the opposite might occur. He sug-
gests that lawyers will profit little from risk exposure claims, and that proportional recovery will
actually reduce the volume of mass exposure litigation by making claims less marketable.
Rosenberg, supra, note 4 at 887-905.
1990]
RISK EXPOSURE AS INJURY
toxic agents is unavoidably on the tort system’s doorstep; the personal and social
dimensions of the problem are simply too momentous to allow the courts to abdi-
cate authority altogetherP 8
In addition to political infeasibility, legislatures may be faced with another
obstacle. While the three sub-patterns operate under the same broad theoretical
umbrella, the actual circumstances which they deal with tend to be extremely
varied –
from toxins in the environment to negligent diagnosis of an injury
depriving the victim of a chance of recovery.” It must be questioned whether
a legislative scheme can be designed which can at once both give effect to the
risk exposure principle, and adequately and efficiently answer all the specific
practical demands of the sub-patterns.
Courts are much more experienced than legislatures in the creation and
application of broad principles. Thus it appears that they have a vital responsi-
bility to attempt to resolve the injustices imposed upon victims of causal
indeterminacy.
Conclusion
This article has had three primary objectives. First, it has established the
unity of the three sub-patterns. This having been accomplished, it follows (along
the lines of basic principles) that any developments in the jurisprudence regard-
ing any one or more of the sub-patterns ought to be capable of immediate exten-
sion to the others. Second, this article has demonstrated that the risk exposure
thesis is the only tort solution to causal indeterminacy which can operate within
traditional causation requirements and still provide accurate compensation and
deterrence (while never imposing liability upon a non-negligent defendant).
And third, it has provided a Canadian perspective to the causal indeterminacy
problem and shown, through Janiak, that there is a foundation in Canadian
jurisprudence for recovery based upon risk exposure.
By now it should be clear that it is necessary to classify risk as an injury
to maintain the integrity of of the causation inquiry. Courts have explored alter-
native methods of compensating plaintiffs,” but too often the solution involves
relaxing proof requirements. The unsatisfactory nature of this approach is best
illustrated by McGhee,” where Lord Wilberforce shifts the onus to the defend-
2581bid. at 926. Even Fleming realises that legislative intervention is not forthcoming, describing
it as “unlikely”: supra, note 4 at 663.
290r any number of lost chance circumstances. For the broad range of examples, see Cooper,
supra, note 4.
260See Brennwald, supra, note 4 at 758-59; T.M. Dworkin, “Fear of Disease and Delayed
Manifestation Injuries: A Solution or a Pandora’s Box?” (1984) 53 Fordham L. Rev. 527.
261Supra, note 14.
REVUE DE DROIT DE McGILL
[Vol. 35
ant to disprove causation notwithstanding that the defendant was in no better
position to carry the onus than the plaintiff.2
Risk-based liability recognizes modem reality and dispenses with the fic-
tion of searching for certainty where none exists. The balance of probabilities
test is still applied, but to an issue which is quite capable of an either/or deter-
mination. Rather than asking the plaintiff to point out a particular defendant as
the cause of his or her injury, the law should in appropriate cases require only
that each defendant be identified as creating a risk to which the plaintiff was
exposed. Risk as injury is the only theory flexible enough to permit broad
application.
There has been very little litigation in Canada of the patterns discussed in
this paper, but this does not mean that risk-based liability will operate in a vac-
uum. Traditional causation principles have prevented disputes from even reach-
ing the courts, so the problem is less apparent than it might be.263 If courts move
to resolve the dilemma facing risk exposure plaintiffs, many more claims will
come to light. In order for this to occur, it is time for the law to fall in line with
other disciplines in recognizing the reality of how causal relationships are
established.
262See supra, notes 182-91.
263Rosenberg, supra, note 4 at 893.
