Negligence Causation Material Contribution To Harm 3

5
Material Contribution to Injury
Illustrative Cases: Bonnington, Bailey, O2, Rahman, Fitzgerald
Factual Basis
This analysis is applied where:

• there are multiple factors (not all of which need to be tortious);
• the case is not one of overdetermination or pre-emption;
• it has been established that the tortious factor/s have had an actual effect1 on the
claimant’s position at trial; and either
• the injury is divisible in principle, but it is not possible to attribute constituent
parts to particular factors on the facts of a given case (Bonnington); or
• the injury is indivisible (Bailey, O2, Rahman).
A ‘material contribution to injury’ analysis is appropriate where it is more likely
than not that at least one defendant’s breach has made a difference to the claim-
ant’s outcome, but it is not possible to isolate the physical effects of individual
breaches from one another. This impossibility precludes the application of basic
causal principles. For instance, in a factual scenario like the one in Performance
Cars Ltd v Abraham,2 there were two physically distinct instances of damage to the
claimant’s car, each of which could be linked discretely to a particular defendant.3
By contrast, those situations in which a court should resort to a ‘material contribu-
tion to injury’ analysis are those where the fact4 of a defendant’s contribution to
1
The question of what this amounts to is a contentious one and will be examined in detail below. In
summary, under the NBA, a factor must have been a necessary one in the claimant’s injury occurring as
and when it did in order to amount to a material contribution. This does not accord with the idea,
expressed in Sienkiewicz v Greif [2011] UKSC 10, [2011] 2 AC 229 at [90] 265 (per Lord Phillips) and in
Bonnington Castings v Wardlaw [1956] AC 613 (HL) at 621 (per Lord Reid) that such a factor need only
have made a contribution in excess of a de minimis level in order to be deemed a material contribution.
2
Performance Cars Ltd v Abraham [1956] 1 QB 33 (CA).
3
In this case, such an exercise was a straightforward one because the damage to the claimant’s car
occurred in two consecutive chronological stages, but this is not necessary for an injury to be divisible.
Were a car to have been hit by two other vehicles simultaneously, it might of course still be possible to
link separable parts of the total damage to each particular collision.
4
In the sense of the claimant having proven, on the balance of probabilities, that but for the defend-
ant’s breach, her natural course of events would have been unaffected.
Factual Basis 95
damage has been established, but there are other factors involved, and there is
simply no means of discretely assigning constituent parts of that damage to par-
ticular factors. In the case of indivisible injuries, this is because such damage is
binary in nature and so cannot be broken down into constituent parts. In the case
of injuries divisible in theory, the problem arises because it is not possible as a
matter of evidence to identify any clean correspondence between constituent parts
and multiple potential factors.
Injury Is Divisible in Principle but It Is Not Possible to Attribute

Constituent Parts to Particular Factors
The classic example of this category of case is Bonnington Castings v Wardlaw,5 in

which the claimant contracted pneumoconiosis during the course of his employ-
ment by the defendants. There were two factors identified as contributing to this
disease: first, those particles of silica dust in the workplace atmosphere which had
emanated from swing grinders and second, those particles of silica dust which had
come from pneumatic hammers. Whilst both types of workplace machinery were
the legal responsibility of the defendants,6 they had only breached their duty in
relation to the first, since there was no known or practicable means of reducing the
dust escaping from the latter. This meant that the particular question for the court
in this case was whether the dust resulting from the defendant’s breach could be
causally linked to the claimant’s injury. The essence of this problem was identified
by Lord Keith:
The disease is a disease of gradual incidence. Small though the contribution of pollution
may be for which the defenders are to blame, it was continuous over a long period. In
cumulo, it must have been substantial, though it might remain small in proportion. It
was the atmosphere inhaled by the pursuer that caused his illness and it is impossible, in
my opinion, to resolve the components of that atmosphere into particles caused by the
fault of the defenders and particles not caused by the fault of the defenders, as if they
were separate and independent factors in his illness. Prima facie the particles inhaled are
acting cumulatively, and I think the natural inference is that had it not been for the
cumulative effect the pursuer would not have developed pneumoconiosis when he did
and might not have developed it at all.7
The specific problem posed for the causal inquiry by Bonnington stems from the
fact that each potential causal factor (ie the ‘innocent’ dust and the ‘guilty’ dust)
5
Bonnington Castings v Wardlaw [1956] AC 613 (HL). Although, as Lord Rodger states in Fairchild
v Glenhaven Funeral Services [2002] UKHL 22, [2003] 1 AC 32 at [129] 100: ‘The idea of liability based
on wrongful conduct that had materially contributed to an injury was . . . established long before
Wardlaw. But Wardlaw became a convenient point of reference, especially in cases of industrial disease’.
6
Covered specifically by regulation 1 of the Grinding of Metals (Miscellaneous Industries)
Regulations 1925.
7
[1956] AC 613 (HL) at 626 (per Lord Keith).
96 Material Contribution to Injury
was operating on the claimant concurrently.8 As a direct result of such concur-

rence, the causal inquiry could not rely on the incremental nature of the disease’s
development to attribute causal valence to particular factors in the way that it
could if the factors had been operating consecutively.
Bonnington may represent a departure from the . . . orthodox approach in the context
of a particular evidentiary gap: namely, where it is known that the victim’s total condi-
tion is a divisible one but there is no acceptable evidentiary basis on which the disability
due to the separate insults to the body could be apportioned to the individual sources,
the claimant is allowed to recover for the total condition . . . the pursuer could prove an
orthodox causal connection between breach and a part of the divisible injury, he just
could not quantify it.9
An ‘acceptable evidentiary basis’ of the type lacking in Bonnington would exist

where the relevant potential causal factors affected the claimant during separate
consecutive periods, meaning that each could be linked to a particular stage of
development, and therefore divisible part, of the injury. Such is the factual basis of
cases in which orthodox apportionment can be carried out. Take, for example,
Thompson v Smiths Shiprepairers (North Shields) Ltd,10 in which Mustill J (as he
then was) reasoned as follows about the situation which would represent
Bonnington were the innocent and guilty dust factors to have operated consecu-
tively, as opposed to concurrently:
Next, one must consider how this approach can be applied to a case where either (a)
there are two successive employers, of whom only the second is at fault, or (b) there is a
single employer, who has been guilty of an actionable fault only from a date after the
employment began . . . Employer B has . . . ‘inherited’ a workman whose hearing is
already damaged by events with which that employer has had no connection, or at least
no connection which makes him liable in law. The fact that, so far as the worker is con-
cerned, the prior events unfortunately give him no cause of action against anyone should
not affect the principles on which he recovers from employer B. Justice looks to the
interests of both parties, not to those of the plaintiff alone.11
This distinction between the concurrent nature of the factors in Bonnington and
the consecutive operation of the factors in Thompson explains why, although both
injuries are divisible in principle, only the damage in the latter case was divisible
in practice. Therefore, Bonnington requires a material contribution to injury anal-
ysis, whereas Thompson does not. Looked at in this way, the material contribution
analysis appears to be relatively simple. There are, however, two complications
that have arisen in relation to it. The first is a question mark over what effect the
imposition of liability for material contribution to injury should have on a conse-
8
‘[C]oncurrent in effect, if not necessarily in time’ – Thompson v Smiths Shiprepairers (North
Shields) Ltd [1984] QB 405 (QB) at 442 (Mustill J).
9
J Stapleton, ‘Unnecessary Causes’ (2013) 129 LQR 39, 52–53. See also J Stapleton, ‘Lords a’Leaping
Evidentiary Gaps’ (2002) 10 Torts Law Journal 276, 283 onwards.
10
[1984] QB 405 (QB). See Ch 3, text to n 24.
11
[1984] QB 405 (QB) at 438.
Factual Basis 97
quent award of damages, and the second is whether a material contribution to

injury analysis is an application of, or an exception to, the But For test.
The first question no doubt arises because, as we see from Bonnington, injuries
which are divisible in principle will sometimes call for liability to be assessed on a
material contribution to injury basis where that divisibility is not possible in prac-
tice, but where there have been multiple potential causal factors. Facts such as
those arising in Bonnington therefore occupy something of a halfway house. It is
trite negligence law that, where possible, defendants should only be held liable for
that part of the claimant’s ultimate damage to which they can be causally linked,
as is clear from cases such as Thompson and Performance Cars. It is equally trite
that, where a defendant has been found to have caused or contributed to an indi-
visible injury, she will be held fully liable for it, even though there may well have
been other contributing causes:12
[I]t is . . . hard – and settled law – that a defendant is held liable in solidum even though
all that can be shown is that he made a material, say 5%, contribution to the claimant’s
indivisible injury. That is a form of rough justice which the law has not hitherto sought
to smooth, preferring instead, as a matter of policy, to place the risk of the insolvency of
a wrongdoer or his insurer on the other wrongdoers and their insurers.13
Since the injury in Bonnington is theoretically divisible, it seems not to fit into this
second category, but it can no more fit into the first because there is, as we have
seen, no practical basis on which any sensible division could be made.14 This was
not a question which troubled the Court in Bonnington, since the defendants’ case
was that they were not liable for the damage at all; they made no plea for any
apportionment to be made on the basis that their breach of duty was not the only
causal factor involved in triggering pneumoconiosis. The very fact, however, that
no apportionment was made, has led to questions being asked subsequently as to
whether that conclusion was the purely the result of the conduct of that particular
case, or whether the same outcome would have occurred for substantive reasons,
had the defendants requested that it be considered.15
The answer is that there should be no apportionment in cases which require, on
the basis outlined in this chapter, a material contribution to injury analysis. The
argument that apportionment of damages is not appropriate in cases in which a
defendant has materially contributed to an indivisible injury is both well estab-
lished and easy to justify.16 There is, however, no obvious or defensible reason why
a different approach should be taken where a practical segmentation is no more
feasible, despite the fact that the damage in question might be theoretically divisible
12
See, eg, Dingle v Associated Newspapers [1961] 2 QB 162 (CA) at 188 (per Devlin LJ), Hotson v East
Berkshire Health Authority [1987] AC 750 (HL) at 783 (per Lord Harwich) and J Smith, ‘Causation – the
Search for Principle’ [2009] Journal of Personal Injury Law 101, 103. See also Baldwin & Sons Pty Ltd v
Plane (1998) 17 NSWCCR 434 (NSWCA), Bendix Mintex Pty Ltd v Barnes (1997) 42 NSWLR 307
(NSWCA) and Gates v Howard Rotavator Pty Ltd (2000) 20 NSWCCR 7 (NSWCA).
13
Barker v Corus [2006] UKHL 20, [2006] 2 AC 572 at [90] 607–08 (per Lord Rodger).
14
Stapleton, ‘Lords a’ Leaping’, above n 9 at 283.
15
See, for example, Stapleton, ‘Unnecessary Causes’, above n 9 at 52.
16
See nn 12 and 13 above.
in nature.17 As illustrated above, had the factors in Bonnington operated consecu-

tively, the damage therein would have been both theoretically and practically
divisible. This is because the cumulative nature of pneumoconiosis (which is what
makes it divisible in theory) would have lent itself to being divided up chrono-
logically according to the extent of its development during the period of exposure
to each successive factor. A material contribution to injury approach would not
therefore have been necessary because such facts would have been amenable to
conventional causal analysis along the same lines as Thompson.18 The very fact that
such division was not possible is what makes it an appropriate case for analysis on
the basis of material contribution to injury. Once this practical impossibility exists,
there is no effective means of distinguishing between divisible and indivisible inju-
ries, since a court is no more able sensibly to divide up the one than the other. A
defendant who has been found to have materially contributed to such an injury,
therefore, should be held liable for 100 per cent of the claimant’s damages.19
The answer to the second question, of whether the material contribution to
injury analysis is an application of, or an exception to, the But For test, can be
made equally emphatically: it adheres to, and does not depart from, the basis of
But For causation. As will become clear, the first stage of the NBA, which is based
on aggregate But For causation, has still to be satisfied where a material contribu-
tion to injury analysis is applied, just as it does in other types of case.
1 – Is it more likely than not that a defendant’s breach of duty changed the
claimant’s normal course of events so that damage (including constituent parts
of larger damage) occurred which would not otherwise have done when it
did?
Under this analysis, a defendant’s breach has either to have part-caused an indivis-
ible injury, or caused part of a (theoretically) divisible injury. Unless, however, a
defendant has made a difference to the claimant’s course of events in this way,
there will be no liability.20
Injury Is Indivisible
An indivisible injury is one which cannot be broken down into separable constitu-
ent parts. It is obvious, given this characteristic, why a quantification of respective
17
But see Sienkiewicz v Greif [2011] UKSC 10, [2011] 2 AC 229 at [90] 265 (per Lord Phillips). And,
in order for a material contribution to injury analysis to be appropriate, the injury must by definition
be one which cannot practically be divided up amongst discrete causal factors.
18
[1984] QB 405 (QB) at 438.
19
According to Jane Stapleton, this is the approach adopted by the US courts in asbestosis claims –
see J Stapleton, ‘The Two Explosive Proof-of-Causation Doctrines Central to Asbestos Claims’ (2009)
74 Brooklyn Law Review 1011. This is also subject, as outlined below (see text to n 38) to a possible
reduction in certain heads of damage if the court decides that the injury might have occurred at some
point in the future owing to factors unrelated to the defendant’s breach of duty.
20
This argument will be addressed fully below, where the case of Bailey v Ministry of Defence [2008]
EWCA Civ 883, [2009] 1 WLR 1052 is discussed.
Factual Basis 99
contributions cannot be carried out amongst multiple factors.21 Some of the more
obvious examples of indivisible injury include a limb broken in one place at a time,
psychiatric injury22 and death. Psychiatric injury was the damage for which the
claimant sought recovery in a case which has come to be regarded as one of the
most difficult in this area: Rahman v Arearose Ltd.23 An NBA, however, clarifies
the main issues.
In Rahman, the claimant was working for the first defendants as the manager of
a fast food restaurant when he was attacked by two black youths. The injuries he
sustained during that attack necessitated his having a bone graft in his eye at the
second defendant’s hospital. The first defendants breached their duty by not pro-
viding sufficient protection for their employee, whilst the second defendant’s
breach lay in the negligent performance of the operation on his eye. Ultimately,
there were several dimensions to the claimant’s damage. First, he was blind in the
eye on which the operation had been performed. In addition, he developed post-
traumatic stress disorder, a severe depressive disorder of psychotic intensity, a
specific phobia of Afro-Caribbean people with paranoid elaboration, and an
enduring personality change; all of which left him unable to work, or even to func-
tion normally in society. It was this psychiatric aspect of his damage which was
forensically contentious. As a result of the Court of Appeal’s consideration of the
case,
the claimant got judgment against each defendant for part of his loss only. For his pain
and suffering/loss of amenity he obtained judgment for £7.5k against the employer and
£55k against the hospital, and for his economic loss of about £500k, one quarter against
the employer and three-quarters against the hospital . . .24
Laws LJ’s judgment, with respect, makes the causation issues in the case appear to
be far more complex than in fact they are. An application of the NBA leads to the
simple conclusion that both defendants should have been jointly and severally
liable for the whole of the damage following the negligent operation.25 The first
21
This is the correct way to analyse the well-known US case of Ybarra v Spangard 154 P 2d 687 (Cal
1945), in which the defendant suffered a traumatic arm injury whilst under sedation for an unrelated
surgical procedure. The alleged difficulty in his case was that he could not identify exactly how or by
which member of the team in theatre his injury was caused. Applying res ipsa loquitur, however, the
Supreme Court of California decided in favour of the claimant, and held, inter alia, that he did not have
to prove which party was responsible for the physical trauma because, in any event, each practitioner
had breached a duty to ensure his safety whilst in theatre. This is correct, and precisely the result that
would be reached on an NBA: since all members of the surgical team had breached a duty, there was at
least aggregate But For causation and, since his safety was compromised in theatre, every breach of that
duty was operative on him at the time his injury occurred. On this view, it is a straightforward case of
material contribution to an indivisible injury.
22
As a result of Rahman v Arearose Ltd [2001] QB 351 (CA), some doubt has been cast on this clas-
sification. As the analysis below will make clear, however, psychiatric injury is indeed indivisible.
23
[2001] QB 351 (CA).
24
T Weir, ‘The Maddening Effect of Consecutive Torts’ (2001) 60 CLJ 237, 238.
25
As conceded by counsel for the first defendants, and acknowledged at [2001] QB 351 (CA) at [36]
(per Laws LJ), a distinction must be made between the pre-operation and post-operation losses: since
the effect of the second defendant’s breach could clearly not be operative on the claimant until it
occurred, the question of multiple factors simply does not arise prior to the second breach occurring.
stage of the NBA is clearly satisfied, since it is more likely than not that at least one
defendant’s breach of duty changed the claimant’s normal course of events for the
worse. The second stage, asked of the two defendants in turn, would also be satis-
fied in each case, since the effects of both were still operative on the claimant when
he suffered the damage for which he claimed. Contrary to the view of Laws LJ, the
claimant’s psychiatric injury should properly have been classed as indivisible
harm, since his ongoing suffering could not be disaggregated and assigned to sep-
arate causes. Whilst Laws LJ gave detailed consideration to the question of whether
the psychiatric injury resulting from the two defendants’ breaches could be said to
be ‘the same damage’ for the purposes of the Civil Liability (Contribution) Act
1978, he ultimately held that it was not. His Lordship’s decision appeared to give
much weight to
an absurd report confected jointly by the experts for the three parties, who tentatively
divided up the victim’s present condition in terms of the two causes. They should not
have been asked to do this, and their answer should have been ignored, for there is no
scientific basis for any such attribution of causality: the claimant is not half-mad because
of what the first defendant did and half-mad because of what the second defendant did,
he is as mad as he is because of what both of them did. His mania is aetiologically indis-
cerptible . . . 26
The conclusion then reached was that
on the evidence the respective torts committed by the defendants were the causes of
distinct aspects of the claimant’s overall psychiatric condition, and it is positively estab-
lished that neither caused the whole of it . . . one cannot . . . draw a rough-and-ready
conclusion to the effect that this is really an indivisible injury and therefore ‘same dam-
age’ within section 1(1) of the 1978 Act.27
With respect, it is difficult to see how the claimant’s ultimate psychiatric damage
could be divided into ‘distinct aspects’, and Weir’s analysis is the more coherent.
Whilst Rahman did indeed suffer from more than one manifestation of mental
trauma, in that he had PTSD, depression and phobia, it would be neither authen-
tic nor feasible to regard these as having been the separate and discrete results of
individual breaches of duty. Not only is it far more likely that the two defendants’
actions worked synergistically to bring about the claimant’s ongoing injury, but
the tenor of more recent case law appears to support the view that such damage is
indivisible.28 For instance, Hale LJ in Hatton v Sutherland 29 refers to Rahman as a
On basic causation principles (see Ch 3), the first defendant is liable for all of the claimant’s damage up
until the time the negligent operation was performed. Applying the NBA confirms this since, under its
second stage of analysing the pre-operation damage (loss of earnings and removal expenses), the sec-
ond defendant’s breach would be found (obviously) to be not yet operative on the claimant. See also
Wright v Cambridge Medical Group [2011] EWCA Civ 669, [2013] QB 312 at [52] 328 and [129] 347.
26
Weir, ‘The Maddening Effect’, above n 24 at 238.
27
[2001] QB 351 (CA) at [23]–[24] 364–365.
28
Although some heads of damage, such as the first three years’ loss of earnings and reasonable
removal expenses, were attributed solely to the defendant employer since they pre-dated the effects of
the negligent medical treatment. This is standard practice and, as such, is unremarkable.
29
Hatton v Sutherland [2002] EWCA Civ 76, [2002] 2 All ER 1 at [37]–[40] 17–18.
Factual Basis 101
case involving indivisible injury,30 as does Smith LJ in Dickins v O2 Plc.31 In an

extra-judicial context, this latter view is reiterated in no uncertain terms, when
psychiatric damage is described as ‘par excellence an indivisible injury’.32 In one
sense, Laws LJ concedes this in an indirect way when he says of the contentious
report of the experts:
It is true that this agreed evidence does not purport to distribute causative responsibility
for the various aspects of the claimant’s psychopathology between the defendants with
any such degree of precision as would allow for an exact quantification by the trial court;
no doubt any attempt to do so would be highly artificial. But the lack of it cannot drive
the case into the regime of the 1978 Act to which, in principle, it does not belong . . . The
fact-finding court’s duty is to arrive at a just conclusion on the evidence as to the respec-
tive damage caused by each defendant, even if it can only do it on a broad-brush basis
which then has to be translated into percentages.33
In essence, what his Lordship prepares to do here is to apportion liability between

defendants on a basis apparently unconnected to the factual matrix of the case.
Such apportionment (one-quarter to the employers and three-quarters to the
health authority) was instead carried out according to an apparently impression-
istic account of the relative culpability of the defendants. Despite the strikingly
heterodox nature of this approach, it was validated by obiter remarks made by Hale
LJ in the Court of Appeal in Hatton34 and, although the House of Lords expressly
declined to offer a view on this point when they considered the same case,35 the
suggestion was implemented in Dickins v O2 Plc.36
The claimant in Dickins was suing her former employer for psychiatric injury
caused by excessive stress at work. The evidence suggested that she had been pro-
moted to a position beyond her natural capabilities and that this had, over time,
led to her suffering from mental health problems, variously characterised as anxi-
ety and depression. Once the Court of Appeal accepted that her work problems
stemmed from a breach of duty, it was faced with a potential problem of causation
in that, besides the situation at work, the claimant had a vulnerable personality,
and had suffered from mental health issues in the past. In addition to this, she suf-
fered from IBS, was at the material time experiencing difficulties in her relation-
ship with her partner (although the evidence was inconclusive as to whether these
phenomena were causes or effects of her stress), and had also, during the relevant
period, had to move out of her home for nine months as a result of flooding.
There existed, therefore, several non-breach factors which could potentially have
30
[2002] EWCA Civ 76, [2002] 2 All ER 1 at [37] 17 and again at [40] 18.
31
Dickins v O2 Plc [2008] EWCA Civ 1144, [2009] IRLR 58 at [45] 64. See also Weir, ‘The Maddening
Effect’, above n 24 at 239, where he says of Rahman that ‘the harm was not incremental, but the indivis-
ible result of a synergistic or catalytic concatenation of events’.
32
Smith, ‘Causation’, above n 12 at 103.
33
[2001] QB 351 (CA) at [23] 364.
34
[2002] EWCA Civ 76, [2002] 2 All ER 1.
35
[2004] UKHL 13, [2004] 2 All ER 385 at [63] 405 (per Lord Walker).
36
[2008] EWCA Civ 1144, [2009] IRLR 58.
contributed to the illness to which she eventually succumbed. In the Court of

Appeal, Smith LJ summarised the trial judge’s means of dealing with this point:
Following the guidance given in Hatton, the judge took those other matters into account
when apportioning the damages as to 50% being due to the tort and 50% due to the
non-tortious factors. Before the judge, both parties had accepted that it was right to
apportion the damages. The dispute between them was only as to how they should be
apportioned.37
As a result of the nature of the dispute between the parties, and therefore of the
grounds of appeal, the Court of Appeal in Dickins was unable conclusively to rec-
tify the trial judge’s mistake, which lay in his apportionment of liability in respect
of an indivisible injury to which the defendant made a material contribution. As
in Hatton and Rahman, this was apparently done on a ‘broad brush’ basis, accord-
ing to intuitive estimations of relative culpability, and was as inappropriate in this
case as it was in the judgments from which it took its lead. Fortunately, those who
heard the Dickins appeal seemed minded to do what they could to arrest the devel-
opment of this novel and ill-advised practice. For instance, although their com-
ments could only be obiter, there was commendable force in the remarks of two of
their Lordships on the apportionment point. Smith LJ stated:
I respectfully wish (obiter) to express my doubts as to the correctness of Hale LJ’s
approach to apportionment. My provisional view (given without the benefit of argu-
ment) is that, in a case which has to be decided on the basis that the tort has made a
material contribution but it is not scientifically possible to say how much that contribu-
tion is (apart from the assessment that it was more than de minimis) and where the
injury to which that has led is indivisible, it will be inappropriate simply to apportion
damages across the board. It may well be appropriate to bear in mind that the claimant
was psychiatrically vulnerable and might have suffered a breakdown at some point in the
future even without the tort. There may then be a reduction in some heads of damage
for future risks of non-tortious loss. But my provisional view is that there should not be
any rule that the judge should apportion the damages across the board merely because
one non-tortious cause has been in play.38
In concurring with that judgment, Sedley LJ added:

I am troubled by the shared assumption about the appropriateness of apportionment on
which the case has proceeded. While the law does not expect tortfeasors to pay for dam-
age that they have not caused, it regards them as having caused damage to which they
have materially contributed. Such damage may be limited in its arithmetical purchase
where one can quantify the possibility that it would have occurred sooner or later in any
event; but that is quite different from apportioning the damage itself between tortious
and non-tortious causes. The latter may become admissible where the aetiology of the
injury makes it truly divisible, but that is not the case.39
37
[2008] EWCA Civ 1144, [2009] IRLR 58 at [39] 63.
38
[2008] EWCA Civ 1144, [2009] IRLR 58 at [46] 64.
39
[2008] EWCA Civ 1144, [2009] IRLR 58 at [53] 65.
Factual Basis 103
Furthermore, Tony Weir points out that this is not ‘just a matter of aesthetics.
Consequences ensue. If, in the present case, either defendant had been insolvent,
the claimant would not have been fully indemnified’.40 As both Smith and Sedley
LJJ make clear,41 full indemnity in such a case as this may well take the form of
reduced damages, to account for the possibility of the same injury occurring in the
future as a result of non-tortious causes. This is the orthodox legal approach, and
is far superior to one which employs apportionment carried out on ‘the basis of
speculation or guesswork’:42
It is important conceptually to differentiate apportionment or divisibility from another
perfectly common process in the assessment of loss, which is to take account of the vicis-
situdes of life or contingencies as applicable to the individual claimant. If, for example,
a claimant suffered from a natural disease from which he was likely to die in five years,
the court would take that fact into account when limiting damages to a period of five
years . . . In truth, . . . [this] is not apportionment at all – what the court is doing is tak-
ing a snapshot of the claimant, at a point immediately prior to the accident, which
incorporates at that time all the particular negative or positive factors in the claimant’s
own past or future, as well as the future factors which might afflict persons generally.
The court, when taking into account the contingencies and vicissitudes of life, makes its
award in an attempt to reproduce the snapshot, the object being to restore the claimant,
warts and all, to the position he was in before the tort was committed.43
To understand exactly why this is so, it is necessary to consider the temporal

dimension of the causal inquiry. When we ask whether the claimant’s damage
would have occurred but for a defendant’s breach, this can only meaningfully be
understood as meaning ‘But for a defendant’s breach, would the claimant’s dam-
age have occurred when it did?’44 Unless it is understood in this way, a But For
inquiry is either impossible to answer, or legally meaningless, or both. Consider,
as an example, a claim in which death forms the gist of the damage. Clearly, here,
but for the defendant’s breach, the claimant would definitely have died. At some
stage.45 The point of legal relevance is of course whether the claimant would have
died when she did but for the defendant’s breach.46 Where death is concerned,
40
Weir, ‘The Maddening Effect’, above n 24 at 238.
41
And Smith LJ reiterates extra-judicially in Smith, ‘Causation’, above n 12 at 103.
42
The correct approach was taken in Fitzgerald v Lane [1989] AC 328 (HL), in which two defendants,
acting independently, were found to have been equally responsible for the ultimate indivisible injury
(partial tetraplegia), alongside a substantial contribution from the claimant himself. After reducing the
total damages by 50% for contributory negligence, the defendants were held jointly and severally liable
for the remaining 50%.
43
L Caun, ‘Multiple Causes of Injury’ [2003] Journal of Personal Injury Law 96, 107–08.
44
As already apparently recognised in academic literature, but rarely translated into express practi-
cal applications – see R Wright, ‘The NESS Account: Response to Criticisms’ in R Goldberg (ed),
Perspectives on Causation (Oxford, Hart Publishing, 2011) n 48 and J Stapleton, ‘Choosing What We
Mean by “Causation” in the Law’ (2003) 73 Missouri Law Review 433, 452–53.
45
See D Lewis, ‘Causation as Influence’ in J Collins, N Hall and LA Paul (eds), Causation and
Counterfactuals (Cambridge, MA, The MIT Press, 2004) 86. The significance of this point becomes even
more acute in cases involving epidemiological evidence; see below, text to n 50.
46
Although it is sometimes obvious that this is really what the But For inquiry is asking, this is by
no means universally the case (nor is it universally accepted as being appropriate).
therefore, a non-temporally specific But For inquiry will always be easy to answer,
but it will tell us nothing of value to the causal inquiry.47 Where, on the other hand,
the gist of the claim is damage of some other type, such as a broken leg, it will be
impossible to predict whether a claimant would ever have suffered such an injury
at any point in their life, were it not for the defendant’s breach. An inquiry such as
this one is therefore both impossible to conduct and lacks any legal purchase.
Facile examples these may be, but they illustrate clearly how important it is that
any But For inquiry is imbued with temporal specificity. Without this, as we have
seen, no such test will be able to distinguish between overdetermined and pre-
empted causal inquiries.48 It is difficult to discern a reason, therefore, for omitting
this crucial qualification from the express formulation of the test.49
In addition to this conceptual argument, there are practical justifications for
why the causal inquiry should have as its focus the stage at which a claimant
incurred the damage of which she ultimately complains. First, the specific ques-
tion of whether an individual would have incurred damage when she did is an
established part of epidemiological causal theory:
One definition of the cause of a specific disease occurrence is an antecedent event, con-
dition or characteristic that was necessary for the occurrence of the disease at the
moment it occurred, given that other conditions are fixed. In other words, a cause of a
disease occurrence is an event, condition or characteristic that preceded the disease
onset and that, had the event, condition or characteristic been different in a specified
way, the disease either would not have occurred at all or would not have occurred until
some later time.50
It is easy to identify several significant reasons why claimants should want to

remain undamaged for as long as possible. Clearly, most of us would want to die
later rather than sooner, and this applies not only to our emotional and physiolog-
ical perspectives, but also to the material effect that a longer life will often have on
our estate. Where illness is concerned, similar arguments apply, in that people
generally will want to have as much of their life as possible unaffected by pain,
suffering and infirmity. In the context of property damage, particularly where that
property is fungible and replaceable, the significance of timing is not necessarily so
striking from the subjective viewpoint of the claimant. Nevertheless, it remains
legally pertinent because, as with personal injury, determining the point at which
a defendant’s breach affected the claimant allows a court either to divide damage
up amongst several factors (where this is possible) or, in any event, to determine
47
See text to n 124 below.
48
Wright, ‘The NESS Account’, above n 44 at n 125.
49
Which is what Richard Wright has long argued for in relation to the US Restatement: American
Law Institute, 82nd Annual Meeting, Proceedings 2005 (Philadelphia, PA, American Law Institute, 2006)
81–84 (thus far in vain – see American Law Institute, Restatement of the Law, Third, Torts: Liability for
Physical and Emotional Harm (St Paul, MN, American Law Institute, 2010) § 26).
50
K Rothman, S Greenland, T Lash, Modern Epidemiology 3rd edn (Philadelphia, PA, Wolters
Kluwer, 2008) 6. The statistical implications of this for the legal test are explored further below.
Factual Basis 105
exactly what a claimant has lost as a result of her injury.51 The first stage of the NBA
explicitly includes this temporal condition by asking ‘but for at least one breach of
duty, would the claimant’s damage have occurred when it did?’
This timing issue is the principal difference between material contribution to
injury cases and pre-empted cause situations. As we have seen,52 the latter type of
situation is characterised by the non-duplicative effects of the factors concerned,
since pre-empted factors are those which never have an effect on the claimant. A
defining feature of material contribution to injury cases, on the other hand, is the
multiple effect of different factors on the claimant, and there is no requirement for
all of the factors concerned to be breaches of duty. So, even if a claimant is actually
affected by both breach factors and non-breach factors, and even if non-breach
factors would eventually have led to the same damage occurring at any point in the
future,53 a defendant will remain liable if her breach hastened the occurrence. The
quantification of what such a factor has caused a claimant to lose is a subsequent,
and non-causal, question of damages.54 In the well-known US case of Dillon v
Twin State Gas & Electric Co,55 for example, the defendant was held liable for neg-
ligently allowing a 14-year-old boy to be electrocuted by its electric cables. The fact
that, had the deceased not grabbed a cable, he would have fallen to his death or at
least to serious injury, made no difference to the causal question. It was relevant
only to the conceptually distinct issue of how much the defendant had to pay in
order to redress its wrong.56
Consequently, in Dickins,57 where the claimant had been affected by stress at
work, her own vulnerable personality, IBS, relationship issues and domestic flood-
ing, and where her depressive illness was ongoing, as it was at trial, the better
analysis of the case would have been that suggested in the obiter remarks of Smith
and Sedley LJJ.58 The defendant employer’s breach led to Ms Dickins’ breakdown
happening when it did, but the other factors affecting her concurrently made it
likely that she would have suffered the same damage at some point in the future
anyway, meaning that the defendant employer’s liability could be described
as hastening damage to which she was anyway vulnerable.59 It would, in such
51
If it were possible to determine that a breach of duty accelerated the occurrence of damage that
would have occurred at some time in the future in any event, the quantification of the claimant’s dam-
ages should reflect this – see M Green, ‘The Intersection of Factual Causation and Damages’ (2006) 55
DePaul Law Review 671, 677–80.
52
See Ch 4.
53
Even a moment later.
54
See Green, ‘The Intersection’, above n 51.
55
Dillon v Twin State Gas & Electric Co 163 A 111 (NH 1932).
56
See also R Stevens, Torts and Rights (Oxford, OUP, 2007) 134.
57
[2008] EWCA Civ 1144, [2009] IRLR 58.
58
See above, text to n 39.
59
[2008] EWCA Civ 1144, [2009] IRLR 58 at [46] 64. See also J King Jr, ‘Causation, Valuation and
Chance in Personal Injury Torts Involving Pre-Existing Conditions and Future Consequences’ (1981)
90 Yale Law Journal 1353.
circumstances, be open to a court to reduce the damages payable accordingly.60

This has been referred to as the ‘crumbling skull doctrine’:
The so-called ‘crumbling skull rule’ simply recognizes that the pre-existing condition was
inherent in the plaintiff’s ‘original position’. The defendant need not put the plaintiff in
a position better than his or her original position. The defendant is liable for the injuries
caused, even if they are extreme, but need not compensate the plaintiff for any debilitat-
ing effects of the pre-existing condition which the plaintiff would have experienced any-
way. The defendant is liable for the additional damage but not the pre-existing damage
. . . if there is a measurable risk that the pre-existing condition would have detrimentally
affected the plaintiff in the future, regardless of the defendant’s negligence, then this can
be taken into account in reducing the overall award . . . This is consistent with the general
rule that a plaintiff must be returned to the position he would have been in, with all of its
attendant risks and shortcomings, and not a better position.61
In the case from which this excerpt is taken, Athey v Leonati,62 the Supreme Court
of Canada dealt meticulously with the particularities of the material contribution
to injury means of analysis. In that case, the claimant, who had a history of back
problems, had been injured in two successive car accidents, each resulting from
the defendants’ breach of duty.63 Subsequently, whilst performing a routine
stretch, he suffered a disc herniation for which he required surgery. As a result, he
was forced to take lower-paid employment so that he could avoid heavy manual
work. The trial judge awarded the claimant 25 per cent of the global damages fig-
ure on the basis that the two accidents were not the sole cause of the herniation,
(because of the pre-existing back problems) but that they played a causative role,
estimated to be in the region of 25 per cent. The Court of Appeal agreed with this
assessment, but the Supreme Court rectified this mistake conclusively and con-
structively in awarding full damages against the defendants. It did so primarily on
the basis that the trial judge had concluded that ‘the plaintiff has proven, on a
balance of probabilities, that the injuries suffered in the two earlier accidents con-
tributed to some degree to the subsequent disc herniation’.64 Major J, in giving the
judgment of the Court, went on to say:
Had the trial judge concluded (which she did not) that there was some realistic chance
that the disc herniation would have occurred at some point in the future without the
accident, then a reduction of the overall damage award may have been considered. This
is because the plaintiff is to be returned to his ‘original position’, which might have
included a risk of spontaneous disc herniation in the future. However, in the absence of
such a finding, it remains ‘speculative’ and need not be taken into consideration.65
60
[2008] EWCA Civ 1144, [2009] IRLR 58 at [47] 64. See also A Dugdale and M Jones, Clerk &
Lindsell on Torts, 20th edn (London, Sweet & Maxwell, 2010) 2-161.
61
Athey v Leonati [1996] 3 SCR 458 (SCC) at [35].
62
Athey v Leonati [1996] 3 SCR 458 (SCC).
63
Although there were two defendants in fact, both were represented as one at the trial.
64
[1996] 3 SCR 458 (SCC) at [44].
65
[[1996] 3 SCR 458 (SCC) at [48]. See also Graham v Rourke (1990) 75 OR (2d) 622 (Ont CA),
Malec v JC Hutton Proprietary Ltd [1990] HCA 20, (1990) 169 CLR 638 and Schrump v Koot (1977) 18
OR (2d) 337 (Ont CA).
Factual Basis 107
Even more helpfully, the Supreme Court explicitly summarised the principles on
which it reached its decision:
If the injuries sustained in the motor vehicle accidents caused or contributed to the disc
herniation, then the defendants are fully liable for the damages flowing from the her-
niation. The plaintiff must prove causation by meeting the ‘but for’ or material contribu-
tion test.66 Future or hypothetical events can be factored into the calculation of damages
according to degrees of probability, but causation of the injury must be determined to
be proven or not proven. This has the following ramifications:
1. If the disc herniation would likely have occurred at the same time, without the
injuries sustained in the accident, then causation is not proven.
2. If it was necessary to have both the accidents and the pre-existing back condition
for the herniation to occur, then causation is proven, since the herniation would not
have occurred but for the accidents. Even if the accidents played a minor role, the
defendant would be fully liable because the accidents were still a necessary contribut-
ing cause.
3. If the accidents alone could have been a sufficient cause, and the pre-existing
back condition alone could have been a sufficient cause, then it is unclear which was
the cause-in-fact of the disc herniation. The trial judge must determine, on the bal-
ance of probabilities, whether the defendant’s negligence materially contributed to
the injury.67
This last paragraph is slightly ambiguous. If it means that, because the judge in
such a situation is unable to make such a determination on the balance of proba-
bilities there can be no liability, it is correct. Otherwise, it is difficult to follow.
Nonetheless, the Athey judgment as a whole is to be welcomed for its generally
concise (at only 53 short paragraphs), straightforward and accurate exposition of
how to analyse a situation in which there have been material contributions to an
injury.
At the other end of the spectrum, a judgment apt to cause much confusion is
that of the English Court of Appeal in Bailey v Ministry of Defence.68 The claimant
in this case suffered from severe brain damage, resulting from a cardiac arrest
brought on by her aspirating her own vomit. She had attended a hospital managed
by the defendants in order that she might undergo a procedure, known as an
ERCP, to examine and treat a suspected gall stone in her bile duct. It was in rela-
tion to the claimant’s post-operative care that the defendant breached its duty of
care, since there was a failure to resuscitate the claimant during the night following
the procedure, leading to her being very unwell by the following morning. At the
same time, and unrelated to the defendant’s breach of duty, the claimant also
developed pancreatitis, an illness which is known to occur in some patients follow-
ing an ERCP. More than a fortnight after her initial operation, the claimant
66
Emphasis added – these italics highlight an unfortunate flaw in the Court’s otherwise coherent
reasoning: as should now be apparent, the But For and material contribution to injury analyses are not
mutually exclusive. Rather, the latter is a specialised application of the former.
67
[1996] 3 SCR 458 (SCC) at [41].
68
Bailey v Ministry of Defence [2008] EWCA Civ 883, [2009] 1 WLR 1052.
aspirated her vomit, leading to her cardiac arrest and, ultimately, to hypoxic brain
damage. For the purposes of the appeal, the pertinent issue was simply that of
whether or not the defendant’s breach of duty in failing to give proper post-
operative care, was causative of the claimant’s brain damage.
There is no doubt, according to the criteria outlined above, that the facts of this
case require a material contribution to injury analysis. For a start, there were mul-
tiple factors in the form of the defendant’s negligent aftercare and the naturally
occurring pancreatitis. Furthermore, this was not a case of overdetermination or
pre-emption.69 The potential causal factors in Bailey could only be described as
interdependent because it is not possible to establish what effects negligent after-
care and pancreatitis would have if suffered separately. Their effects are best
described as synergistic, or at least potentially so, meaning that Bailey belongs
outside of the duplicative cause category. Since brain damage is indivisible in prin-
ciple, it was therefore necessary to establish whether the defendant’s breach of duty
had been a partial cause of the ultimate injury. Unfortunately, although the facts
of Bailey were correctly deemed to require a material contribution to injury analy-
sis, this is not what followed. Whilst, with respect, there is much to lament in the
judgment as a whole,70 the essence of the mistake is encapsulated in the following
statement by Waller LJ:
In a case where medical science cannot establish the probability that ‘but for’ an act of
negligence the injury would not have happened but can establish that the contribution
of the negligent cause was more than negligible, the ‘but for’ test is modified, and the
claimant will succeed.71
This proposition, which formed the basis of the Court of Appeal’s dismissal of the
defendant’s appeal against liability, is dangerously misleading. As Stapleton points
out,
until that flawed proposition is disapproved it threatens to have an explosive impact
in the field of medical negligence. This is because, for example, it may often be the case
that a breach by a medical provider increases the weakness of a patient, by some non-
negligible but un-assessable degree, before the patient suffers an indivisible injury that
would have been avoided had the patient been of adequate strength . . . Clearly, if such
claimants are entitled to succeed under that proposition it would expose medical pro-
viders to a radically expanded realm of liability.72
This explains why, in order to be deemed a legally relevant cause, a factor must
either be a necessary part-cause of an indivisible injury, or the necessary cause
69
Since it does not fit the criteria outlined in Ch 4.
70
Such as equating material contribution to injury with the exceptional material contribution to
risk analysis developed in Fairchild v Glenhaven Funeral Services [2002] UKHL 22, [2003] 1 AC 32 and
a description of that case as one in which the evidence established that one fibre caused the claimant’s
injury. For a more detailed criticism of these arguments, see Ch 6.
71
[2008] EWCA Civ 883, [2009] 1 WLR 1052 at [46] 1069.
72
Stapleton, ‘Unnecessary Causes’, above n 9 at 58. See also M Stauch, ‘“Material Contribution” as a
Response to Causal Uncertainty: Time For a Rethink’ (2009) 68 CLJ 27, 28–29.
Medical Negligence 109
of part of a (theoretically) divisible injury. The first stage of the NBA must be satis-
fied in order for there to be liability on the grounds of a material contribution to
injury.
of larger damage) occurred which would not otherwise have happened when it
did?
In his proposition, Waller LJ appears to conflate But For with de minimis non curat
lex, by seeing them as mutually exclusive, or at least as alternatives. Since both are
necessary, this may well be the source of his confusion. Where indivisible injury,
such as that suffered in Bailey, is concerned, a claimant must establish, on the bal-
ance of probabilities, that the defendant’s breach was a part-cause of that damage.
In other words, she must prove that, but for the defendant’s breach, her injury
would not have occurred when it did. It may well be the case that, even where a
claimant has established that a defendant’s breach was, in But For terms, a part-
cause of her injury, a court may decide that that contribution was so small as to
absolve the defendant from liability on the basis of the de minimis principle, but
the two principles are complementary and not alternative. The facts in Bailey
should have led to a conclusion of no liability because the experts therein were
unable to say that, on the balance of probabilities, but for the negligent care,
Ms Bailey would have avoided having the cardiac arrest (which led to the brain
damage) when she did.73 The medical evidence did, however, suggest that the
defendant’s breach had a more than negligible chance of being a cause of the claim-
ant’s ultimate damage and this is what, erroneously, Waller LJ regarded as being
sufficient to find liability on a material contribution to injury basis. As a result,
Bailey is a confused, confusing, and ultimately unhelpful decision.
Medical Negligence
The performance of a material contribution to injury analysis requires particular

care in situations involving medical negligence. The medical context provides, in
any event, a special kind of challenge for the causal inquiry because, almost by
definition, medical practitioners deal with individuals who are already injured or
damaged in some way. Often, it is the task of extricating the breach from the non-
breach factors which makes this area of the law so difficult. An added dimension
to this problem is the fact that human physiology is unpredictable, imperfectly
understood and often makes it very hard to determine where the effect of one fac-
tor ends and another begins.
73
For an Australian perspective, see Tubemakers of Australia Ltd v Fernandez (1976) 50 AJLR 720
(HCA) at 724.
Wright v Cambridge Medical Group provides an example of these difficulties.74

In this case, the claimant’s GP, who worked for the defendant partnership, was
consulted by the claimant’s mother by telephone, and subsequently failed to see
the claimant or to refer her to hospital as (the defendants conceded) he should
have done. The claimant, who was 11 months old, was actually suffering from a
bacterial super-infection contracted during an earlier visit to hospital for chicken
pox treatment. She was finally referred to hospital two days later. Once there,
however, she received inadequate medical care, which would doubtless have been
deemed to be in breach of duty had the hospital been joined as a defendant to the
action which, inexplicably, it was not. The causal question for the court, therefore,
was whether the GP’s breach of duty was causative of the claimant’s ultimate
injury (a permanently unstable hip, restricted movement range, leg length dis-
crepancy and restricted mobility). It had to consider whether the hospital’s inad-
equate treatment broke the chain of causation between the defendant’s breach and
the claimant’s injury, and whether it would be reasonable to hold, as the trial judge
had done, that even a timely referral would have made no difference to the claim-
ant’s position, since the hospital would have treated her inadequately, leading to
her injuries occurring in any event.
Ultimately, the Court of Appeal (Elias LJ dissenting) allowed the claimant’s
appeal and found the defendants liable in full for the claimant’s permanent dam-
age.75 The essential basis of this decision can be found in the judgment of Lord
Neuberger of Abbotsbury MR:
In the present case, I consider that the defendants’ negligence was a causative factor of
the claimant’s permanent injury. In other words, as in Rahman’s case [2001] QB 351,
para 34, I have concluded that the negligence of the defendants and the failings of the
hospital had a ‘synergistic interaction, in that each tends to make the other worse’, and
accordingly it seems appropriate to proceed on the basis that both were causative of the
damage suffered by the claimant.
I do not consider the hospital’s failure to treat the claimant properly once she was
admitted . . . was of such significance that it justifies a finding that the defendant’s neg-
ligence was not causative of the claimant’s injury – or indeed a finding that it broke the
chain of causation between the defendant’s negligence and the claimant’s injury. It was
not such an egregious event, in terms of the degree or unusualness of the negligence, or
the period of time for which it lasted, to defeat or destroy the causative link between the
defendant’s negligence and the claimant’s injury.76
Although there was no mention in any of the majority judgments77 of a material

contribution to injury analysis, the facts of the case meant that this would have
74
Wright v Cambridge Medical Group [2011] EWCA Civ 669, [2013] QB 312.
75
Although it also made the point that the pain and suffering endured by the claimant after the time
of the negligence but before her eventual admission to hospital did not form part of the ultimate per-
manent damage, and the defendants should not therefore be liable for it. See [2011] EWCA Civ 669,
[2013] QB 312 at [52]–[53] 328 and [92] 336–37.
76
[2011] EWCA Civ 669, [2013] QB 312 at [36]–[37] 325 (per Lord Neuberger).
77
Elias LJ refers to it implicitly, see [2011] EWCA Civ 669, [2013] QB 312 at [92] 336–37 and explic-
itly at [96] 337–38.
Medical Negligence 111
been appropriate: the claimant’s permanent damage was indivisible, there were
multiple potential factors, and this was not a case of duplicative causation because
of the lack of simultaneity and the interdependence of the effects of those factors
(meaning that there was no several sufficiency).78 It should have fallen, therefore,
to establish, as the final part of the jigsaw, whether the defendant’s breach of duty
had, on the balance of probabilities, been a part-cause of that damage. The Court
decided it had, but it was a conclusion not couched in material contribution to
injury terms. So, whilst the Court of Appeal’s analysis and one based on material
contribution to injury would have led to the same outcome, use of the latter
approach would probably have led to the case being more easily aligned with the
line of authority to which it belongs. In any event, on an NBA, the same answer is
easily reached.
of larger damage) occurred which would not otherwise have happened when it
did?
This question is easy to answer on the facts of Wright because it was clear on the
evidence that, in the absence of both breaches, the claimant’s damage would not
have occurred when it did.
2 – Was the effect of this defendant’s breach operative when the damage occurred?
This was less easy to answer than the first because of the interdependence of the
factors in this case. The defendant’s breach of duty and the consequent failure of
the hospital to diagnose and treat Wright accordingly were interdependent
because, first, it was not possible to say that, had the defendant made the referral
at the correct time, the hospital’s treatment would have been as bad as it was when
the referral was eventually made. There might, for instance, have been more com-
petent staff available on the earlier day. In fact, given that the claimant should have
been referred on a Wednesday, but was eventually referred on a Friday, and that
the Court found there were more consultants available during the week than there
were on weekends, the effects of the defendant’s failure to refer appear likely to
have continued, and operated in combination with those created by the hospital’s
poor treatment.79 Secondly, it was generally accepted on the evidence that the
claimant’s damage became permanent after six days of insufficient treatment. Had
the defendant made an earlier referral, therefore, the hospital would have had a
larger ‘window’ during which to reach a proper diagnosis and begin treatment
accordingly. Moreover, Lord Neuberger also made the point that
the judge’s conclusion did not take into account the agreed expert evidence . . . which,
in fairness to him, seems to have been overlooked in the argument before him (and,
indeed, the argument before us). In my judgment, the effect of that evidence is to
See Ch 4, under sub-heading ‘Several Sufficiency’.

78
See [2011] EWCA Civ 669, [2013] QB 312 at [68] 331 and [72] 332.
79
establish, at the lowest, that it is more likely than not that any permanent damage the
claimant would have suffered due to the inept treatment, which the judge found that she
would have received if she had been referred on 15 April [the Wednesday], would have
been significantly less than that which she did suffer. Indeed, I think that this evidence
established that there is a reasonable chance that she would have suffered relatively little
long-term damage if she had been referred in the late afternoon or early evening of
15 April.80
When the claimant’s damage became permanent, the risk created by the defend-
ant’s breach was still operative.81 It is clear, therefore, why Lord Neuberger consid-
ered the effects of the two factors to have been operating synergistically. As a factor
which partially contributed thereby to an indivisible injury, the breach of duty was
rightly held to be a legally relevant cause and the defendant liable for full damages.
Had the hospital been joined as a defendant, it would have been appropriate for
the Court to have held both parties jointly and severally liable, since both would
have materially contributed to the claimant’s injury.
It is worth re-emphasising a point already made in Chapter 4 on Duplicative
Causation. Like the case of Elayoubi 82 mentioned there,83 Wright is a case properly
categorised as one involving a material contribution to injury because the multiple
factors therein were not independent from one another. If, however, the GP’s fail-
ure to refer in Wright had in no way affected the treatment provided by the hospi-
tal, and if the delay had made no difference to the ultimate outcome, the factors
would have been independent, which would have made the case one of pre-empted
cause. Had this been the case, the defendants would not have been liable because
the second stage of the NBA would not have been satisfied in relation to it: the
effects of its breach would not ever have affected the claimant, since they would
have been pre-empted by the effects of the hospital’s sub-standard treatment. It is
easy to see, therefore, how significant interaction between factors (or lack of it) can
affect the outcome of the causal inquiry.
A material contribution to injury analysis is also appropriate to the harrowing
facts of Paroline v United States et al.84 Although a case principally about criminal
restitution in the US,85 the judgment makes significant reference to causation in
tort law. The claimant in that case had as a young girl been the victim of sexual
abuse, which had been filmed and distributed extensively online. Her hurt and
humiliation were therefore set to continue into the future, as more and more
individuals witnessed the material on the Internet. The defendant, whilst not the
individual who produced the material or first put it into circulation, was charged
with possessing child pornography, including images of the claimant, and the US
Supreme Court was asked to consider
80
[2011] EWCA Civ 669, [2013] QB 312 at [73] 332 (per Lord Neuberger).
81
[2011] EWCA Civ 669, [2013] QB 312 at [65]–[79] 330–34 (per Lord Neuberger).
82
Elayoubi v Zipser [2008] NSWCA 335.
83
See Ch 4, text to n 134.
84
Paroline v United States et al No 12-8561, April 23, 2014 (USA).
85
Under the Violence against Women Act 1994.
The ‘Doubling of the Risk’ Test 113
the theory of ‘aggregate causation,’ one formulation of which finds factual causation
satisfied where a wrongdoer’s conduct, though alone ‘insufficient . . . to cause the plain-
tiff’s harm,’ is, ‘when combined with conduct by other persons,’ ‘more than sufficient
to cause the harm.’ 1 Restatement (Third) of Torts: Liability for Physical and Emotional
Harm §27, Comment f. 86
Since the claimant’s mental distress, as well as her need for ongoing counselling
and lost income, amounted to indivisible damage, and since the act of each indi-
vidual viewing the material contributed to that damage in an interdependent way,
this was a case in which the defendant had materially contributed to the claimant’s
injury.87
The ‘Doubling of the Risk’ Test88
Cases classified as those in which a defendant has materially contributed to a

claimant’s injury are just one of the instances in which an application of the ‘dou-
bling of the risk’ (DTR) test has been mooted. This test, conceived as an epide-
miological device, performs a very specific function, and has unfortunately been
misapplied in a forensic context several times. The following excerpt from Novartis
Grimsby Ltd v Cookson89 exemplifies a factual phenomenon which is particularly
common in cases in which the claimant’s damage takes the form of a disease.
Besides the occupational exposure to aromatic amines in the dyestuffs industry, which
has been recognised as a cause of bladder cancer for many years, it is also known that
cigarette smoking can be a cause, as can the ingestion of certain drugs. More than one
potential cause was present in this case. As well as working for the Appellant for many
years, Mr Cookson had been a moderate cigarette smoker (10 to 20 a day) for about 20
years. He had given up the habit in about 1980. He had also taken potentially harmful
drugs for a time but it was not suggested by either side that those drugs had had any
significant effect on the causation of his bladder cancer. Cigarette smoke contains
amines and the amines from both sources act on the body in the same way. It was
accepted by both sides that the two forms of exposure would have had at least an
additive, if not multiplicative, effect. The argument between the parties was about the
relative potency of the effects of smoking and occupational exposure. In essence, the
argument was whether the occupational exposure was sufficient to have caused or mate-
rially contributed to the development of the cancer.90
86
No 12-8561, April 23, 2014 (USA) at B.
87
The defendant in that case was not held liable for all of the restitution claimed by the defendant.
Instead, the Court attempted to establish what was the defendant’s relative causal contribution – see No
12-8561, April 23, 2014 (USA) at B.
88
With thanks to Mark Ingham for checking the statistical accuracy of what follows.
89
Novartis Grimsby Ltd v Cookson [2007] EWCA Civ 1261, [2007] All ER (D) 465 (Nov).
90
[2007] EWCA Civ 1261, [2007] All ER (D) 465 (Nov) at [44] (per Smith LJ).
The essence of the issue from a forensic point of view is that there are not only
multiple potential causes, but that those causes might operate synergistically,
rather than independently. ‘Interaction is common, and exposures that simulta-
neously protect against the effect of other exposures when they cause a disease are
rare’.91
The significance of such ‘synergistic’ effects has unfortunately been exaggerated
in some recent decisions, and attempts to apply the DTR test to deal with the per-
ceived problem are misguided. The practice appears to have started in Smith LJ’s
judgment in Novartis. As far as the extract above goes, the issue was correctly
identified but it is not clear that the DTR test was in any way determinative of, or
necessary for, the result:
The evidence . . . was that occupational exposure accounted for 70% to 75% of the total.
Put in terms of risk, the occupational exposure had more than doubled the risk due to
smoking. In my view, if . . . the correct test for causation in a case such as this is the ‘but
for’ test and nothing less will do, that test is plainly satisfied on the facts as found. The
natural inference to draw from the finding of fact that the occupational exposure was
70% of the total is that, if it had not been for the occupational exposure, the Respondent
would not have developed bladder cancer. In terms of risk, if occupational exposure
more than doubles the risk due to smoking, it must, as a matter of logic, be probable that
the disease was caused by the former.92
Even without any consideration of whether a risk had been doubled, it seems likely
that the defendant in Novartis would have been held liable, on the basis that the
occupational exposure accounted for 70–75 per cent of the total risk involved.93
This is strong evidence on which a tribunal could base a belief probability94 that
the occupational exposure was more likely than not to have part-caused the cancer
and so formed a material contribution to the claimant’s injury. Given that the
DTR test was irrelevant to that result, therefore, it was unfortunate that it was
mentioned at all because the ‘test’ is simply not independently determinative of
But For causation; it merely provides one indication of a potential statistical asso-
ciation between a given factor and a given result.95 Sadly, however, the Novartis
decision has led to the DTR test being either applied or requested in other, equally
unsuitable, contexts since. It was, for instance, applied in Shortell v BICAL
Construction,96 and received judicial acceptance at some point (albeit in both cases
91
A Broadbent, ‘Epidemiological Evidence in Proof of Specific Causation’ (2011) 17 Legal Theory
237, 259.
92
[2007] EWCA Civ 1261, [2007] All ER (D) 465 (Nov) at [74] (per Smith LJ).
93
Although the Court of Appeal decision in AB v Ministry of Defence [2012] UKSC 9, [2013] 1 AC
78 at [153] 135 (per Smith LJ) suggests at [153] that the DTR test was in fact determinative of this case,
such an approach is, with respect, not the most helpful analysis of the evidence therein.
94
See Ch 2, under heading ‘The Balance of Probabilities’.
95
C McIvor, ‘The “Doubles the Risk” Test for Causation’ in S Pitel, J Neyers and E Chamberlain (eds),
Tort Law: Challenging Orthodoxy (Oxford, Hart Publishing, 2013).
96
Shortell v BICAL Construction, Liverpool District Registry (unreported) 16 May 2008. The defend-
ant’s employee had died from lung cancer, having been exposed at work to asbestos and having been a
smoker for a number of years of his life. On the basis that the exposure to asbestos more than doubled
the risk of contracting lung cancer, Mackay J found for the claimant.
by the same person) in both the Court of Appeal in Sienkiewicz v Greif 97 and AB v
Ministry of Defence.98
As the name suggests, the DTR test purports to equate a finding that a factor had
the effect of exposing a claimant to twice the risk of suffering damage, relative to a
claimant not exposed to that factor, with a finding of probable causation. It is
traceable to Mackay J’s judgment in XYZ v Schering.99 That case, however, posed a
highly specific question, expressly concerned with relative instances of injury. In
Sienkiewicz v Greif, a decision which suggests an appropriate limitation on the use
of the DTR test, Lord Philips recognises this fact:
XYZ v Schering Health Care Ltd 70 BMLR 88 is a lengthy and complex judgment devoted
exclusively to a preliminary issue on the effect of epidemiological evidence. The issue
was whether a second generation of oral contraceptives more than doubled the risk of
causing deep vein thrombosis (DVT) that was created by the first generation of oral
contraceptives. It was common ground that, if the claimants in this group litigation
could not establish this, their claims under the Consumer Protection Act 1987 were
doomed to failure. I do not believe that Smith LJ has correctly identified the relevance
of this issue. It was not whether the DVT suffered by the claimants had been caused by the
second generation of oral contraceptives which they had taken. It was whether the sec-
ond generation of contraceptives created a significantly greater risk than the first. The
experts appear to have been in agreement that the ‘doubles the risk’ test was the proper
one to apply in order to resolve this issue. Thus I do not believe that that decision affords
any direct assistance to the question of whether the ‘doubles the risk’ test is an appropri-
ate test for determining causation in a case of multiple potential causes. (Emphasis
added)100
In other words, the explicit question for the Court in XYZ, a product liability case,
was whether the product which had been prescribed for the claimants was more
than twice as likely as its earlier incarnation (the risks of which the claimants were
taken to have accepted) to cause its users harm. Here, asking whether the risk has
been doubled is clearly and directly appropriate because it answers this specific
question. As Lord Philips identified in Sienkiewicz, however, this is not the same as
asking whether it is more likely than not that one of several potential factors caused
a particular injury, and his Lordship was correct in ruling that the DTR ‘test’ is not,
therefore, the correct approach to take on those facts:101
For reasons that I have already explained, I see no scope for the application of the ‘dou-
bles the risk’ test in cases where two agents have operated cumulatively and simultane-
ously in causing the onset of a disease. In such a case the rule in Bonnington applies.
Where the disease is indivisible, such as lung cancer, a defendant who has tortiously
contributed to the cause of the disease will be liable in full. Where the disease is divisible,
97
Sienkiewicz v Greif [2009] EWCA Civ 1159, [2010] QB 370 at [23] 379 (per Smith LJ).
98
AB v Ministry of Defence [2012] UKSC 9, [2012] 1 AC 78 at [132] 130, [140] 132, [146] 134, [151]
135, and [153] 135.
99
XYZ v Schering [2002] EWHC 1420, 70 BMLR 88 (QB).
100
[2011] UKSC 10, [2011] 2 AC 229 at [74] 261–62.
101
See also J Stapleton, ‘Factual Causation, Mesothelioma and Statistical Validity’ (2012) 128 LQR
221, 223.
such as asbestosis, the tortfeasor will be liable in respect of the share of the disease for
which he is responsible.102
With respect, whilst his Lordship’s conclusion is undoubtedly correct on this

point, the reasons he gives are neither complete nor fully accurate. In order to
understand why, the function of this epidemiological tool must be fully under-
stood.103 There are two principal reasons why the DTR test is not, without more,
an appropriate device for calculating probability of causation. The first is that the
relative ‘risk’ on which it is based is not in any technical sense a risk at all, but
merely a measure of incidence over a defined population:
[I]t is crucial to realize that ‘risk’ in this context means nothing more than incidence
over a specified time interval. It does not denote the product of the probability that the
harm will occur and the gravity of that harm; in fact it does not even denote a probabil-
ity, strictly speaking . . . It is a purely statistical measure of the relative frequency with
which a disease occurs in exposed and unexposed populations.104
To treat it as substitutive for probability, therefore, is to confuse two distinct con-

cepts, with potentially adverse consequences for the causal inquiry.
The second reason is that the result reached by the relative risk calculation
involved gives a ‘net result’. That is, it tells us how many extra cases of a disease are
caused by the factor being tested. It does not, therefore, include those cases in
which the adverse outcome would have happened anyway, just not when it did. In
other words, it only identifies cases which would never otherwise have happened,
and does not account for those which the factor merely accelerated. This is sig-
nificant since, as we have seen above,105 suffering injury sooner rather than later is
undoubtedly damage for which the law should provide compensation.
To illustrate this point, Alex Broadbent has provided a simple and effective
example:
Consider, for example, a fictitious group of Himalayan porters. One might imagine that
carrying heavy loads up and down mountains increases their risk of back injury. Among
those porters suffering back injury will be some who for anatomical reasons would have
developed a back injury anyway. But having a weak back does not protect these porters
from the effects of carrying heavy loads. On the contrary, it is quite possible that carrying
a heavy load will be a cause of back injury among many or even all those porters whose
backs were such that they would have developed a back injury even in a less physically
demanding profession. [Equating relative risk incidence with probability of causation]
amounts to endorsing the astonishing view that a person who would develop a disease
without the exposure in question is thereby protected from the effects of that exposure:
that having a weak back will protect a Himalayan porter from the harm that carrying a
102
[2011] UKSC 10, [2011] 2 AC 229 at [90] 265.
103
For a highly useful account, specific to its use in negligence, see McIvor, ‘The “Doubles the Risk”
Test’, above n 95. There also exists the question of how far such epidemiological evidence is relevant to
the causal inquiry in negligence, and what its limits are. This will be addressed in Ch 7.
104
Broadbent, ‘Epidemiological evidence’, above n 91 at 240.
105
See also Smith v Leech Brain [1962] 2 QB 405 (QB) at 413 (per Lord Parker CJ).
heavy load would otherwise do to his back. It does not take any great conceptual sophis-
tication to see that this is an error.106
What this does tell us is that the DTR test and the measures on which it relies can
underestimate a factor’s effect on a population. What it cannot do, however, is
overestimate that effect:
The . . . problem is that the exposure dose at which the probability of causation exceeds
50% (the point at which exposure causation is more likely than not) may fall well below
the ‘doubling dose’ (the dose at which the incidence of the disease is doubled) . . . When
an effect of exposure is to accelerate the time at which disease occurs, the rate fraction107
. . . will tend to underestimate the probability of causation because it does not fully
account for the acceleration of disease occurrence. In particular, and contrary to com-
mon perceptions, a rate fraction of 50% [which equates to a finding that the ‘risk’ has
been doubled by the factor in question]108 does not correspond to a 50% probability of
causation.109
The conclusion, therefore, that the incidence of injury has been doubled by the
factor being tested amounts to a single indication that there might exist a statistical
association between that factor and the injury.110 It should also be clear, however,
that this is not the same thing as establishing on the balance of probabilities that
the factor caused the injury: first, such a conclusion does not amount to a ‘fact
probability’ of 50 per cent or more and second, for reasons examined elsewhere in
this book, fact probabilities are themselves not independently sufficient to estab-
lish causation in negligence.111 Rather, such statistics are just one component of the
range of evidence on which a court reaches its decision, and, eo ipso, tell us nothing
about what happened in this particular case. Wright is insistent, for example, that
non-particularised statistics are ‘mere ad hoc distributions not related to any
106
Broadbent, ‘Epidemiological Evidence’, above n 91 at 256.
107
That proportion of the totality of disease within a population which is attributable to the factor
being tested ([incidence rate among exposed population minus incidence rate among non-exposed
population] divided by incidence rate among exposed population).
108
For example, suppose that 40% of those exposed to brick dust get dermatitis and 20% of those
not so exposed suffer the same fate. On these figures, it is easy to see how the incidence or ‘risk’ is dou-
bled by exposure and how 50% of cases are attributable to exposure, which is what Greenland’s ‘rate
fraction’ refers to. For a more detailed account of the relationship between the two measures, see
A Broadbent, Philosophy of Epidemiology (Basingstoke, Palgrave Macmillan, 2013) ch 3.
109
S Greenland, ‘Relation of Probability of Causation to Relative Risk and Doubling Dose: A
Methodology Error That Has Become a Social Problem’ (1999) 89 American Journal of Public Health
1166, 1166 and 1168–69. See, for instance, Jones v Secretary of State for Energy and Climate Change
[2012] EWHC 2936 (QB), [2012] All ER (D) 271 (Oct).
110
Conditional language is employed here deliberately to account for all of the potential errors to
which such a conclusion is in any event vulnerable, such as poorly designed experiments and sampling
errors (both of which are independent dangers – see D Barnes, ‘Too Many Probabilities: Statistical
Evidence of Tort Causation’ (2001) 64 Law and Contemporary Problems 191). This is a topic on which
there has been extensive academic discussion in the US in particular: for both a contribution and a
brief review of some of the arguments, see MD Green, ‘The Future of Proportional Responsibility’ in
S Madden (ed), Exploring Tort Law (New York, NY, CUP, 2005).
111
See, for instance, discussion of belief probabilities in Ch 2, under heading ‘The Balance of
Probabilities’.
causal generalisation’.112 Indeed, the fallacy of equating these to conclusions about

individualised instances of causation is now well known, thanks to the famous
examples of the taxi cab problem113 and the Gatecrasher paradox.114
Broadbent, however, has different concerns:
There are two serious mistakes in judicial and academic literature on this topic. The first
is that RR > 2 is necessary for proof of specific causation.115 The second is that epide-
miological evidence is never sufficient for proof of causation: that no matter how strong
the evidence for a general causal link between a wrong and a harm, we are never thereby
warranted in deciding that a particular claimant’s harm was caused by the wrong. Both
errors have led to injustices . . . if epidemiological evidence is not capable of bearing on
individual cases then it would lead to the radical misuse of evidence that is clearly admis-
sible, such as that generated by medical tests. It follows that, where epidemiological
evidence to the effect that RR > 2 is the only evidence, it is capable of proving causation
as more likely than not . . .116 What epidemiological evidence cannot do, however, is
disprove a causal claim where RR < 2.117
In referring to the situation in which epidemiological evidence is the ‘only’ evi-

dence available, Broadbent draws attention to an area in which Wright’s hostility
to the independent influence of such evidence requires the most scrutiny.118 If all
a court has at its disposal is that with which epidemiology provides it, should it use
this as a basis for its decision? The following extract from one of Wright’s earliest
and most comprehensive works on causation suggests that there is less disagree-
ment on this point than might first appear:
112
R Wright, ‘Proving Causation: Probability v Belief ’ in Goldberg (ed) Perspectives on Causation,
above n 44 at 210.
113
In which an individual is knocked down by a taxi whose colour is not observed. The incident
occurs in a town where there are only two taxi firms: one which has three blue cabs, and another which
has one yellow cab. The example is supposed to show that it would be inappropriate to infer from those
facts alone that it was more probable than not that the accident involved a blue cab, despite the existence
of a 75% statistical probability of this. The example can be found in the dissenting judgment of
Brachtenbach J in Herskovits v Group Health Cooperative of Puget Sound 664 P 2d 474 (Wash 1983), a
decision of the Supreme Court of Washington. A more complex example, to which particularistic evi-
dence and reliability estimates have been added, can be found in A Tversky and D Kahneman, ‘Evidential
Impact of Base Rates’ in D Kahneman, P Slovic and A Tversky (eds), Judgement Under Uncertainty:
Heuristics and Biases (Cambridge, CUP, 1982) 153. See also Ch 7.
114
In which there is a rodeo with 1,000 attendees, 499 of whom have paid for admission. There are
no issued tickets, nor testimony as to whether A climbed over a fence to gain entry. There is, however,
a mathematical probability of .501 that A did not pay, which would mean, on a mathematical interpre-
tation of the forensic standard, that the organisers would be entitled to judgment against him. The
problem here is that they would also be entitled, on those grounds, to judgment against every attendee
on the same basis. LJ Cohen, The Probable and the Provable (Oxford, OUP, 1977) 75.
115
This refers to the distinction sometimes made between the question of whether an agent is capa-
ble of causing an injury of the type in question (general causation) and the question of whether a
particular claimant’s injury was caused by a particular defendant’s breach (specific causation). See
Seltsam Pty Ltd v McGuiness (2000) 49 NSWLR 262 (NSWCA) at [22] (per Spigelman CJ).
116
Since courts have to make a decision one way or another, any relevant evidence is capable of help-
ing to reach a belief probability.
117
Broadbent, Philosophy of Epidemiology, above n 108 at 206. Chapter 7 will elaborate further on
the general applicability of epidemiological evidence to the forensic exercise.
118
See MD Green, ‘The Future of Proportional Liability: The Lessons of Toxic Substances Causation’
in Madden (ed), Exploring Tort Law, above n 110 at 352.
A judgment on what actually happened on a particular occasion is a judgment on which

causal generalization and its underlying causal law was fully instantiated on the particu-
lar occasion. Particularistic evidence connects a possibly applicable causal generaliza-
tion to the particular occasion by instantiating the abstract elements in the causal
generalization, thereby converting the abstract generalization into an instantiated
generalization. Without such particularistic evidence, there is no basis for applying the
causal generalization to the particular occasion.119
In other words, on Wright’s analysis, evidence of a claimant’s exposure to a given

agent is particularised evidence sufficient to link epidemiological information
about a condition to that claimant.120 Where this is all that is available,
then the only possibly applicable causal generalisation with at least some particularistic
instantiation in the particular situation is the toxic-agent causal generalisation, which
fact could support the formation of a belief that it was the causal process actually at work
in the particular situation.121
Wright’s point is not, therefore, that epidemiology cannot be used or useful.

Rather, his entire conceptualisation of causation demands that there be a distinc-
tion between ‘mere ad hoc distributions’ and particular instances of causation.122
Logically, this must be correct. One means of illustrating this is to show how a RR
of < 2 is equally compatible with a 0 per cent and 100 per cent probability of cau-
sation. Greenland does this using the following illustration:
As an extreme example, suppose the damage done by exposure was that of accelerating
the development of disease in all individuals destined to contract disease. Then, when
considering the lifetime experience of the exposed cohort, all of the exposed occurrences
of disease would be accelerated cases . . . In other words . . . the excess fraction would be
0 [and the RR would be 1], incorrectly suggesting that there was no exposure effect, and
yet the probability of causation would be 100%.123
In conclusion, it would seem that the epidemiological device of estimating the

effect of a given factor can be forensically useful where that effect is shown at least
to have doubled the rates of incidence as against background factors. It is equally
clear, however, that this method should in no way constitute a routine or stand-
alone test which is treated as conclusive of the causal inquiry.124 The decision of the
High Court of Australia in Amaca Pty Ltd v Ellis125 is an object lesson in how to deal
119
R Wright, ‘Causation, Responsibility, Risk, Probability, Naked Statistics, and Proof: Pruning the
Bramble Bush by Clarifying the Concepts’ (1988) 73 Iowa Law Review 1001, 1051.
120
Although see Sienkiewicz v Greif [2011] UKSC 10, [2011] 2 All ER 857 at [158] 906–07 (per Lord
Rodger) and at [170] 910 (per Baroness Hale).
121
Wright, ‘Proving Causation’, above n 112 at n 67.
122
See also Sienkiewicz v Greif [2011] UKSC 10, [2011] 2 All ER 857 at [96] 888.
123
Greenland, ‘Relation of Probability’, above n 109 at 1168. Greenland further points out that,
whilst these facts may seem farfetched, it is precisely what will be seen if the outcome in question is
death rates in a population followed for its entire lifetime, such as the atomic bomb survivors in Japan.
124
See Seltsam Pty Ltd v McGuiness (2000) 49 NSWLR 262 (NSWCA) at [78]–[89] and [102] (per
Spigelman CJ). Although this was a material contribution to risk case, the point remains relevant what-
ever the causal question.
125
Amaca Pty Ltd v Ellis [2010] HCA 5, (2010) 240 CLR 111.
appropriately with such epidemiological evidence.126 The deceased in that case had
died from lung cancer, having been a smoker for over 25 years and also having
been exposed by the defendant employers to asbestos in breach of their duty of
care. Appropriately, the High Court made no mention of the DTR test, but instead
considered the epidemiological evidence presented to it as an evidential compo-
nent in its inference from the facts127 as to what, on the balance of probabilities,
caused the deceased’s cancer:
If the relative risks and probabilities derived from epidemiological studies were to be
treated as revealing what was a probable explanation of what caused Mr Cotton’s cancer,
those analyses support two conclusions. First, it is more probable than not that smoking
was a cause of (in the sense that it was a necessary condition for) Mr Cotton’s cancer.
Second, the risks and probabilities associated with asbestos, whether alone or in con-
junction with smoking, are low and not sufficient to found the inference which the
plaintiff sought to have made: that it is more probable than not that exposure to respi-
rable asbestos fibres was a cause of Mr Cotton’s cancer.128
Since the findings of liability in both Novartis129 and Shortell 130 were based on evi-
dence that the respective breaches of duty more than doubled the risk of the harm
occurring, the formation of a belief probability that those breaches more likely
than not caused the injuries seems unobjectionable.131 Nevertheless, the presenta-
tion of these results as being predicated on satisfaction of the DTR test is regret-
table.
The relative risks permits such an inference only if it is credible, which means that the
belief probability is greater than fifty percent. The preponderance of the evidence stan-
dard is not met without a belief probability of greater than fifty percent and a risk ratio
greater than 2.0.132
For the reasons just outlined, therefore, courts should refrain from asking experts
to present their evidence in the form of an answer to the DTR test, but should
instead ensure that, where possible, such evidence takes account of what epidemi-
ologists refer to as the etiologic fraction. In essence, this covers both the excess
126
See also D Hamer, ‘Mind the “Evidential Gap”: Causation and Proof in Amaca Pty Ltd v Ellis’
(2009) 31 Sydney Law Review 465 (criticising the contrasting approach of the Court of Appeal).
127
See Clements v Clements [2012] SCC 32, [2012] 2 SCR 181 at [38] (per McLachlin CJ).
128
[2010] HCA 5, (2010) 240 CLR 111 at [64] 134. These facts would not, therefore, satisfy the first
stage of the NBA: see above, text to n 20. See also Evans v Queanbeyan City Council (2011) 9 DDCR 541,
[2011] NSWCA 230 which follows Ellis in all aspects material to the current discussion. In Evans, the
Court of Appeal made a misleading and unfortunate reference to the Fairchild principle, in saying that
it could not be applied by them, but would be a matter for the High Court. The facts of Evans would
not, in any event, be suitable for a correct application of that principle – see Ch 6 generally, and Ch 7,
n 31.
129
[2007] EWCA Civ 1261, [2007] All ER (D) 465 (Nov).
130
Liverpool District Registry, (unreported) 16 May 2008.
131
See also J Stapleton, ‘Factual Causation, Mesothelioma and Statistical Validity’, above n 101 at 223
and 227 for an explanation of why the DTR test is only applicable where conclusions are sought as to
which of several mutually exclusive or ‘competing alternative’ mechanisms caused the injury in ques-
tion.
132
Barnes, ‘Too Many Probabilities’, above n 110 at 207.
fraction already referred to (those cases which would never have happened but for
the exposure) as well as that fraction of cases which would still have happened but
for the exposure, but not until later (those accelerated by the exposure). This
paints a comprehensive picture of all of those cases affected by the exposure, and
does not underestimate it in the way that DTR does.133
To explain the problem in algebraic terms, suppose that AT exposed persons contracted
the disease during the time period in question and that, of these individuals, A0 are
unaffected, A1 were accelerated by exposure, and A2 represented all-or-none occurrences
of disease [those that would not have occurred at all but for the exposure]. By definition,
exposure . . . harmed persons whose disease was either accelerated or all or none. Hence,
the fraction of exposed persons with the disease who were harmed by the exposure is
(A1 + A2)/A T . This quantity is the etiologic fraction. Furthermore, if we randomly select
an exposed person with disease from the total AT, the chance that exposure harmed that
person (i.e., the chance that the person had an accelerated or all-or-none occurrence) is
also (A1 + A2)/A T. The latter quantity is thus also the probability of causation.134
In Merrell Dow Pharmaceuticals v Havner,135 the Supreme Court of Texas, in a

judgment referred to with apparent approval by Lord Philips in Sienkiewicz,136
made the methodologic error to which Greenland refers:
Assume that a condition naturally occurs in six out of 1,000 people even when they are
not exposed to a certain drug. If studies of people who did take the drug show that nine
out of 1,000 contracted the disease, it is still more likely than not that causes other than
the drug were responsible for any given occurrence of the disease since it occurs in six
out of 1,000 individuals anyway. Six of the nine incidences would be statistically attrib-
utable to causes other than the drug, and therefore, it is not more probable that the drug
caused any one incidence of disease. This would only amount to evidence that the drug
could have caused the disease. However, if more than twelve out of 1,000 who take the
drug contract the disease, then it may be statistically more likely than not that a given
individual’s disease was caused by the drug.137
This statement assumes, of course, that anyone who was ever likely to have suf-
fered from the condition in question would have been immune to the effects of the
drug (and thereby fails to recognise those cases where the drug would accelerate
the development of the disease in those who would have contracted it at some
point). Whilst this biological pattern can probably not be ruled out as ever occur-
ring, it should certainly not be treated as if it were a normal occurrence. The
excerpt above, therefore, demonstrates a forensic underestimation of the effects of
the drug in question. This error is unfortunate, not only for its intrinsic logical
flaw, but also because it detracts from a decision which is otherwise constructive
in its treatment of epidemiological evidence (and which, anyway, did not rely on
133
It also emphasises the importance of the temporal dimension to the But For question, as dis-
cussed above.
134
Greenland, ‘Relation of Probability’, above n 109 at 1167–68.
135
Merrell Dow Pharmaceuticals v Havner 953 SW 2d 706 (Tex 1997).
136
[2011] UKSC 10, [2011] 2 AC 229 at [85] 264.
137
953 SW 2d 706 (Tex 1997) at 717.
this mistaken interpretation of the evidence for its ultimate result of no liability).
Elsewhere, for instance, Owen J states:
We recognize, as does the federal Reference Manual on Scientific Evidence, that a disease
or condition either is or is not caused by exposure to a suspected agent and that fre-
quency data, such as the incidence of adverse effects in the general population when
exposed, cannot indicate the actual cause of a given individual’s disease or condition
. . . But the law must balance the need to compensate those who have been injured by
the wrongful actions of another with the concept deeply imbedded in our jurisprudence
that a defendant cannot be found liable for an injury unless the preponderance of the
evidence supports cause in fact. The use of scientifically reliable epidemiological studies
and the requirement of more than a doubling of the risk strikes a balance between the
needs of our legal system and the limits of science . . .We do not hold, however, that a
relative risk of more than 2.0 is a litmus test or that a single epidemiological test is legally
sufficient evidence of causation. Other factors must be considered. As already noted,
epidemiological studies only show an association. There may in fact be no causal rela-
tionship even if the relative risk is high. . . . Likewise, even if a particular study reports a
low relative risk, there may in fact be a causal relationship. The strong consensus among
epidemiologists is that conclusions about causation should not be drawn, if at all, until
a number of criteria have been considered.138
This excerpt forms only part of the judgment’s considerable evaluation of the use
of epidemiological evidence, and the various ways in which it can be interpreted
and used. Unlike the English cases which rely on epidemiological data, this deci-
sion cites much academic commentary on the issue, and highlights points of disa-
greement and uncertainty.139 This explains, at least in part, the measured and
cautionary approach taken by the court to the use of such information, and it is
unfortunate that the English courts have instead chosen to implement the special-
ised DTR test without any comparable reference to expert evaluation.
138
953 SW 2d 706 (Tex 1997) at 718. Although the Reference Manual referred to also makes the
interpretative mistake highlighted here, as Alex Broadbent points out in Broadbent, ‘Epidemiological
Evidence’, above n 91 at 254. See, however, Restatement (Third) of Torts: Liability for Physical Harm § 28
comment c.
139
953 SW 2d 706 (Tex 1997) at 715–21 in particular.

Negligence Causation Material Contribution To Harm 3

Uploaded by

Copyright:

Available Formats

Negligence Causation Material Contribution To Harm 3

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Negligence Causation Material Contribution To Harm 3

Uploaded by

Copyright:

Available Formats

5

Material Contribution to Injury

Illustrative Cases: Bonnington, Bailey, O2, Rahman, Fitzgerald

This analysis is applied where:

Injury Is Divisible in Principle but It Is Not Possible to Attribute

The classic example of this category of case is Bonnington Castings v Wardlaw,5 in

was operating on the claimant concurrently.8 As a direct result of such concur-

An ‘acceptable evidentiary basis’ of the type lacking in Bonnington would exist

quent award of damages, and the second is whether a material contribution to

in nature.17 As illustrated above, had the factors in Bonnington operated consecu-

case involving indivisible injury,30 as does Smith LJ in Dickins v O2 Plc.31 In an

In essence, what his Lordship prepares to do here is to apportion liability between

contributed to the illness to which she eventually succumbed. In the Court of

In concurring with that judgment, Sedley LJ added:

To understand exactly why this is so, it is necessary to consider the temporal

It is easy to identify several significant reasons why claimants should want to

circumstances, be open to a court to reduce the damages payable accordingly.60

The performance of a material contribution to injury analysis requires particular

Wright v Cambridge Medical Group provides an example of these difficulties.74

Although there was no mention in any of the majority judgments77 of a material

See Ch 4, under sub-heading ‘Several Sufficiency’.

The ‘Doubling of the Risk’ Test88

Cases classified as those in which a defendant has materially contributed to a

With respect, whilst his Lordship’s conclusion is undoubtedly correct on this

To treat it as substitutive for probability, therefore, is to confuse two distinct con-

causal generalisation’.112 Indeed, the fallacy of equating these to conclusions about

In referring to the situation in which epidemiological evidence is the ‘only’ evi-

A judgment on what actually happened on a particular occasion is a judgment on which

In other words, on Wright’s analysis, evidence of a claimant’s exposure to a given

Wright’s point is not, therefore, that epidemiology cannot be used or useful.

In conclusion, it would seem that the epidemiological device of estimating the

In Merrell Dow Pharmaceuticals v Havner,135 the Supreme Court of Texas, in a

You might also like