Williams, Greg; McEwan, Blair --- "But for' what? Complex causation in product liability cases" [2020] PrecedentAULA 15; (2020) 157 Precedent 10

---

‘BUT FOR’ WHAT?

COMPLEX CAUSATION IN PRODUCT LIABILITY CASES

By Greg Williams and Blair McEwan

Causation is a necessary element of the reasoning process of tort and product liability law – a point on a continuum in which there is no bright-line division.^[1] We are taught as law students and reminded constantly as litigators that there is little utility in proving a breach of the law unless it is possible to show that it resulted in some loss to the client.

Claims in negligence causation, as codified in the (relatively) uniform state-based legislation,^[2] have two elements: factual causation and scope of liability (also referred to as legal causation). In Wallace v Kam,^[3] the High Court explained the two-stage approach under s5D of the Civil Liability Act 2002 (NSW) (CLA) as:

‘a question of historical fact as to how particular harm occurred; and a normative question as to whether legal responsibility for that particular harm occurring in that way should be attributed to a particular person ... In particular, the application of the first question, and the existence of the second, have been obscured by traditional expressions of causation for the purposes of the common law of negligence in the conclusory language of "directness", "reality", "effectiveness" and "proximity".’^[4]

THE ‘BUT FOR’ TEST

This ‘but for’ test, codified in s5D(1)(a) and comparable legislation,^[5] requires the court to be satisfied that a step taken (or not taken) by a defendant more likely than not caused the relevant harm. It has been described as ‘the threshold test for determining whether a particular act or omission qualifies as a cause of the damage sustained’.^[6] As the High Court demonstrated in March v Stramare (E & MH) Pty Ltd,^[7] a ‘but for’ test of factual causation has limitations. Specifically, the ‘but for’ test is difficult to apply in circumstances where there is more than one possible cause of the damage in issue (including multiple potential tortfeasors).

There is no equivalent to s5D in the Australian Consumer Law (ACL).^[8] However, identical problems arise regarding the application of the ‘but for’ test even without the codified construction of s5D of the CLA. In particular, courts frequently wrestle with the challenge posed by the traditional approach to causation because:

• the evidence suggests a statistical (epidemiological) rather than a mechanical link between the contravening conduct and the harm; or

• the harm is caused by successive exposure to risk factors, only some of which are the responsibility of defendants before the court.

Such analysis is often made more complex because it is interwoven with a consideration of whether a defect exists in the product that increases its propensity to cause harm. In such cases, issues that arise on the question of defect may overlap with those that arise on causation. Care and rigour are required to ensure that the two questions are not conflated.

Under Australian law, the orthodox view remains that demonstrating an increase in risk of injury is not sufficient to establish causation.^[9] As French CJ said in Amaca Pty Ltd v Booth, ‘[t]he risk of an occurrence and the cause of the occurrence are quite different things’.^[10] It is not enough to establish that the relevant conduct was merely capable of causing the damage suffered; the particular risk must have eventuated and caused the injury suffered.^[11]

However, causation may in some instances be established through accumulated inference, whether it be ‘the strength of association as measured by relative risk ratios’^[12] or gathered facts like ‘strands in a cable’,^[13] so that a court is prepared to conclude on the balance of probabilities that a risk caused an injury, notwithstanding that the evidence is unable to offer definitive proof.

CAUSATION UNDER THE STATUTORY PRODUCT LIABILITY REGIME

The leading case in Australia on the application of the causation test to the statutory product liability regime remains Merck Sharp & Dohme (Australia) Pty Ltd v Peterson (Peterson),^[14] a case which illustrates both of the challenges set out above. In Peterson, both the primary judge and the Full Court adopted a ‘strand’-based approach to causation but reached different conclusions as to the sufficiency of each strand in establishing a causative link, based on the strength of the evidence led.

The Full Federal Court found that consumption of the drug Vioxx being ‘in the mix’ of possible causes was not enough to show that it was a necessary condition for the plaintiff’s heart attack, and therefore causation was not established:^[15]

‘As Sir Frederick Jordan said in Carr v Baker [1936] NSWStRp 20; (1936) 36 SR (NSW) 301 at 306: “Conjecture may range from the barely possible to the quite possible”. We accept that it is quite possible the Vioxx was a necessary condition of Mr Peterson’s heart attack. But in the light of the causative potential of the other factors to which we have referred, we consider that a conclusion that Mr Peterson would not have had his heart attack but for the consumption of Vioxx based on the [evidence], which was itself subject to unresolved objection, is a matter of conjecture rather than reasonable inference on the balance of probabilities.’^[16]

The Court reaffirmed that the different words of the product liability regime in the Trade Practices Act 1974 (Cth) (now part of the ACL), such as ‘because’ and ‘by reason of’, do not give rise to a different test to that applied at common law.

This is in sharp contrast to the first instance decision where, on the basis of circumstantial evidence, an inference was drawn that the consumption of Vioxx made a material contribution to the occurrence of Mr Peterson’s injury.^[17]

CAUSATION IN OTHER CONTEXTS

There are also many other cases where plaintiffs have fallen short on issues of causation because the courts have adopted this orthodox approach. These cases show that in complex factual scenarios, the evidence can sometimes only take you so far. For example:

• In Forbes v Selleys Pty Ltd^[18] (Forbes v Selleys), the NSW Court of Appeal upheld the first instance decision that Mr Forbes did not prove that his exposure to a particular chemical component of the ‘Selleys Space Invader’ product caused or materially contributed to his seizure. The NSW Court of Appeal relevantly observed:

‘The appellant was critical of the trial judge for not taking a “robust and pragmatic” attitude to the scientific evidence. But this criticism is misplaced. His Honour was fully aware of his capacity to infer probable cause from the epidemiological possibility evidence. But he was not bound to do so. Robust and pragmatic decisions are not always made in the plaintiff’s favour.’^[19]

• In Adeel’s Palace Pty Ltd v Moubarak,^[20] a failure by the restaurant to have additional (reasonable) security measures in place was not found to be sufficient to avoid harm caused by a criminal activity, where an intoxicated patron entered the premises with a loaded gun: ‘Recognising that changing any of the circumstances in which the shootings occurred might have made a difference does not prove factual causation”.^[21]

• In Amaca Pty Ltd v Ellis (Amaca v Ellis),^[22] it was not proven that industrial asbestos exposure was a necessary condition for Mr Ellis’ lung cancer, in circumstances where the plaintiff was also a long-term smoker. As the Court said, ‘Knowing that asbestos can cause cancer does not entail that in this case it probably did’.^[23]

• In Roads and Traffic Authority v Royal^[24] (RTA v Royal), the High Court overturned the NSW Court of Appeal’s finding of contributory negligence by the RTA in relation to a vehicle collision. The High Court found that the Court of Appeal had failed to address whether there was a causative link between the collision and the negligent design of the intersection.

RISK AND CAUSATION

Plaintiffs have put forward various constructions to address the difficult nexus between risk and causation. Below, we explore four ways that courts have wrestled with the high threshold required by the ‘but for’ test and the desire to adopt a ‘common sense’ approach to causation.^[25]

Recognising where a ‘but for’ test is not appropriate

Despite remaining the accepted starting point for causation analysis, the court has recognised that there are instances where sole reliance on a counterfactual satisfaction of the ‘but for’ test is not appropriate.

This was explored by the NSW Supreme Court of Appeal in Allianz Australia Ltd v Sim; WorkCover Authority (NSW) v Sim etc^[26] (Sim) in the context of successive exposure to the same risk, rather than (as per Amaca v Ellis) concurrent exposure to different risks. Mr Sim developed asbestosis and lung cancer, as a result of exposure to asbestos dust and fibre.

That cumulative exposure to dust and fibre caused Mr Sim’s asbestosis was not disputed on appeal. However, the excessive exposure which had resulted in asbestosis had occurred over a prolonged period of time and in the employ of four different entities. As such, a ‘cumulative theory of causation’, had been advanced, and was the subject of significant expert evidence. An appeal was lodged in respect of the finding of culpability for Mr Sim’s lung cancer, which was found at first instance (and ultimately by the Court of Appeal) to have been caused by the asbestosis (which in turn was caused by exposure to asbestos).

The Court of Appeal in Sim explored approaches taken in the US and the UK in relation to the doctrine of material contribution, in particular Bonnington Castings Ltd v Wardlaw^[27] (Bonnington), but ultimately rejected the appellants’ argument, in this context, that the conduct of each contributing party must be ‘a sine qua non [an essential condition] with respect to the harm suffered’.^[28] In particular, the Court said:

‘The appellants identified no authority for the proposition that substantial successive and cumulative tortious conduct, independently engaged in by several defendants, did not render each liable for the consequential and individual harm, in circumstances where individually, the tortious conduct was neither necessary nor sufficient to cause the harm. The line of authorities ... demonstrates that evidence of causation satisfying those conditions is sufficient to establish liability according to general law principles.’^[29]

In other words, there will be some circumstances where causation can be established notwithstanding that the ‘but for’ test has not been satisfied. However, such an approach begs the question – what is the criteria that makes it permissible to avoid ‘but for’? Sim and Bonnington suggest that one such circumstance is where there are multiple instances of successive exposure to a risk factor. This is an approach which is applied in practice in many cases concerning career-long exposure to asbestos, where the defendant in question is only responsible for a portion of the exposure.

However, this is not the approach courts have adopted where it is possible to differentiate between the extent of the risk posed by successive periods of exposure. For example, in Bendix Mintex Pty Ltd v Barnes,^[30] the NSW Court of Appeal determined that causation had not been established in circumstances where the evidence showed that the exposure for which the defendant was responsible carried a much lower risk than that for which it was not.^[31]

Specific policy exception – s5D(2) of the CLA and equivalent legislation

The House of Lords’ decision in Fairchild v Glenhaven Funeral Services Ltd^[32] created an exception in the interest of public policy for certain personal injury (asbestos) claims, such that courts can relax the threshold for establishing causation to a material contribution.^[33] The exception was specifically limited to mesothelioma claims (and more recently has included asbestos-related lung cancer^[34]) where, due to the limitations of medical science, it is not possible to determine the cause of the claimant’s illness.

Australian courts have never gone this far and indeed have rejected such an approach (see Amaca v Ellis^[35]). However, Parliament has created a statutory exception under s5D(2) of the CLA and equivalent legislation. It is notable that such an exception was not available in either Peterson or Sim (because Peterson was a case under the then Trade Practices Act 1974 (Cth) and Sim a dust diseases case excluded from the operation of the CLA).

Perhaps surprisingly, it remains very difficult to find cases where s5D(2) or its equivalents have been applied.^[36] However, in the Victorian decision of Powney v Kerang^[37] the Court identified two circumstances where the evidentiary gap provision under s51(2) of the Wrongs Act 1958 (Vic) (being the equivalent provision to s5D(2) of the NSW CLA) might be applied, being:

1. exposures to a particular agent on multiple occasions, all of which contributed to a disease process but where factual causation could not be attributed to a specific exposure; and

2. where scientific evidence in relation to the level of exposure to a particular agent necessary to produce injury may still be developing.^[38]

Importantly, the Court said that, before s51(2) is engaged, it must be shown that negligence cannot be established as a necessary condition of the harm for the purpose of factual causation; in other words, that the primary test of causation cannot be established.^[39] But this does not mean that s51(2) is automatically available as a ‘fall back’ if factual causation cannot be made out. Rather, it is necessary to show that the case in question is an ‘appropriate’ (in the words of the Victorian statute) or ‘exceptional’ (in the words of the NSW statute) one to which the exception applies.

Arguing for a shift in onus

Plaintiffs sometimes try to argue that there are certain circumstances (such as breach by omission or failure to warn) which give rise to a presumption of causation, shifting the onus to the defendant. Plaintiffs seeking to make this argument have sought to rely on commentary by Kirby J in Chappel v Hart^[40] (endorsing Lord Wilberforce in McGhee v National Coal Board^[41]) and Gaudron J in Bennett v Minister of Community Welfare.^[42] In TC by his tutor Sabatino v New South Wales^[43] (Sabatino), Mason P provides a useful summary of this approach.

In actual fact, and evidenced by Mason P’s commentary in Sabatino, this is a position which has been consistently rejected by Australian courts:

‘In Bendix Mintex I concluded that the ultimate legal onus of proof rested with the plaintiff. It is not sufficient that a plaintiff prove that the defendant negligently exposed the plaintiff to a risk of injury: liability depends upon the plaintiff persuading the trier of fact that it was probable that the risk came home (see at 318) ...

I do not think that later High Court authority has concluded otherwise as regards reversal of onus of proof. If anything, there has been an endorsement of the traditional view as to the plaintiff bearing the ultimate legal onus, albeit in a context that has emphasised the propriety of a trier of fact taking a robust and pragmatic approach to causation.’^[44]

This approach was also explored and rejected in RTA v Royal^[45] and Forbes v Selleys.^[46]

Frame ‘defect’ in such a way that it establishes causation

The recent UK decision of Colin Gee and Ors v DePuy International Limited^[47] (Gee v DePuy), concerned an action against the manufacturer of the Pinnacle Acetabular Cup System, a metal-on-metal (MoM) total hip replacement product, brought under the UK’s implementation of the European Product Liability Directive 1985 (85/374/EEC).^[48]

The claimants argued that the safety of the Pinnacle product was not of a standard that persons are generally entitled to expect (the same test which applies in a safety defect claim brought under s138 of the ACL), and that this caused them personal injury for which DePuy was liable. The Court considered a preliminary issue: whether the defendant was liable to the claimant, subject to any development risk defence.^[49] Causation needed to be shown for the claimants to succeed.

Complex medical devices like artificial hips present significant challenges in questions of causation due to the high likelihood that a patient has pre-existing issues and other co-morbidities. Combined with numerous other risk factors such as the choice of product by the treating physician, surgical skill, post-operative care and compliance of the patient (to name but a few), it is exceedingly difficult to demonstrate on a ‘but for’ standard that a medical device was the cause ‘at large’ of the harm to an individual.

The claimants in Gee v DePuy alleged that the Pinnacle product had an ‘abnormal potential for damage’^[50] in the form of a risk of an adverse reaction to metal wear debris when compared to other (non-MoM) hip replacement products. They argued that if the abnormal potential for damage was found to constitute a defect, causation would be determined by asking what would have happened if the device implanted ‘did not have the identified abnormal potential for damage’.^[51] That is, the claimants framed the causation question in such a way that if a defect was proved, then causation would be proved.^[52]

The Court rejected this construction and instead proposed an orthodox approach to causation. That is, in the Court's view the correct question was whether, on the balance of probabilities, the claimant would have suffered the damage complained of (in this case, the early revision of their hip implant, noting that any artificial hip will likely require replacement during a patient’s lifetime) if the hip had not carried an increased risk of early failure.^[53]

Ultimately the Court found that the Pinnacle product did not carry a materially increased risk of early failure. As such, causation did not need to be established. However, most interesting for our purpose is the Court’s rejection of the claimants’ approach to equate ‘inherent potential for damage’^[54] with defect and therefore causation, which was seen to ‘[disregard] the burden on the injured party of proving the defect as well as the causal relationship between defect and damage ... One cannot divorce the defect from the concept of defectiveness in the manner suggested; they are two sides of the same coin.’^[55]

CONCLUSION

The ‘but for’ test has persevered as the starting point for questions of causation in Australian law under both statutory and common law regimes. Plaintiffs must therefore be cautious to ensure that all elements are satisfied where applying abductive reasoning to product liability claims. Causation must be established in fact and not assumed as ‘given’ where an injury exists.

Greg Williams (Partner) and Blair McEwan (Lawyer) are members of Clayton Utz’s Commercial Litigation (Product Liability) team. They advise clients from a range of industries including health and life sciences, financial services, automotive and consumer goods, including to defend class actions, interact with regulators and engage in other forms of dispute resolution. PHONE [phone] EMAIL [e-mail] / [e-mail].

---

^[1] Seltsam v McGuiness (2000) 49 NSWLR 262 (Seltsam), [84] per Spigelman CJ.

^[2] See Civil Liability Act 2002 (NSW), ss5D and 5E; Wrongs Act 1958 (Vic), ss51 and 52; Civil Liability Act 2003 (Qld), ss11 and 12; Civil Liability Act 1936 (SA), ss34 and 35; Civil Liability Act 2002 (WA), ss5C and 5D; Civil Liability Act 2002 (Tas), ss13 and 14; Civil Law (Wrongs) Act 2002 (ACT), ss45 and 46.

^[3] [2013] HCA 19; (2013) 250 CLR 375.

^[4] Ibid, [11].

^[5] Strong v Woolworths Ltd t/as Big W (2012) 285 ALR 420; 86 ALJR 267; [2012] HCA 5, [18].

^[6] March v E & M H Stramare Pty Ltd [1991] HCA 12; (1991) 171 CLR 506 (March), 530.

^[7] Ibid.

^[8] Competition and Consumer Act 2010 (Cth), sch2.

^[9] See eg, Amaca Pty Ltd (under NSW administered winding up) v Booth [2011] HCA 53; (2011) 246 CLR 36 (Amaca v Booth); Bendix Mintex Pty Ltd v Barnes (1997) 42 NSWLR 307; Seltsam v McGuiness (2000) 49 NSWLR 262; Strong v Woolworths Ltd t/as Big W [2012] HCA 5; (2012) 285 ALR 420; Wallace v Kam [2013] HCA 19; (2013) 250 CLR 375; Powney v Kerang and District Health [2014] VSCA 221; (2014) 43 VR 506; New South Wales v Mikhael [2012] NSWCA 338.

^[10] Ibid, Amaca v Booth, [41].

^[11] Seltsam, above note 1, [108]–[118].

^[12] Amaca) v Booth, above note 9, [43].

^[13] (2000) 49 NSWLR 262, [91].

^[14] [2011] FCAFC 128.

^[15] Ibid, [104].

^[16] Ibid, [124].

^[17] Ibid, [9]–[10].

^[18] [2004] NSWCA 149.

^[19] Ibid, [107].

^[20] Adeels Palace Pty Ltd v Moubarak (2009) 239 CLR 420 [2009] HCA 48.

^[21] Ibid, [50].

^[22] (2010) 240 CLR 111; [2010] HCA 5.

^[23] Ibid, [68].

^[24] (2008) 245 ALR 653; [2008] HCA 19.

^[25] See March, above note 6, 515–516 per Mason CJ.

^[26] [2012] NSWCA 68 (Sim).

^[27] [1956] UKHL 1; [1956] AC 613.

^[28] Sim, above note 26, [134].

^[29] Ibid, [145].

^[30] (1997) 42 NSWLR 307.

^[31] See also Wallace v Kam [2013] HCA 19; (2013) 250 CLR 375 and Seltsam v McGuiness (2000) 49 NSWLR 26.

^[32] [2002] UKHL 22; [2003] 1 AC 32.

^[33] Also see Barker v Corus UK [2006] UKHL 20.

^[34] Carl Heneghan (Son & Executor of James Leo Heneghan, Deceased) v Manchester Dry Docks Ltd & Ors [2016] EWCA Civ 86.

^[35] (2010) 240 CLR 111; 263 ALR 576; [2010] HCA 5.

^[36] See for example Nominal Defendant and Ors v Bacon [2014] NSWCA 275; (2014) 67 MVR 425; Smythe v Burgman [2015] NSWSC 150 and Bettergrow Pty Ltd v NSW Electricity Networks Operations Pty Ltd (No 2) [2018] NSWSC 514.

^[37] [2014] VSCA 221; (2014) 43 VR 506.

^[38] Ibid, [96].

^[39] Ibid, [85].

^[40] (1998) 195 CLR 232.

^[41] [1972] UKHL 7; [1973] 1 WLR 1.

^[42] [1992] HCA 27; (1992) 176 CLR 408.

^[43] [2001] NSWCA 380, [66]–[74].

^[44] Ibid, [62]–[63].

^[45] [2008] HCA 19; (2008) 245 ALR 653, [143].

^[46] [2004] NSWCA 149, [66] and [94]–[103].

^[47] [2018] EWHC 1208 (QB) (Gee v DePuy).

^[48] Directive implemented in England and Wales by Part 1 of the Consumer Protection Act 1987.

^[49] Gee v DePuy, above note 47, [15].

^[50] Ibid, [247]–[248].

^[51] Ibid, [184].

^[52] Ibid, [184]–[185].

^[53] Ibid, [185].

^[54] Ibid, [101].

^[55] Ibid, [98].