Paul v Cooke

2. 175 CLR 479 Rosenberg v Percival [2001] HCA 18; (2001) 205 CLR 434 Ruddock v Taylor [2003] NSWCA 262; (2003) 58 NSWLR 269 Sibraa v Brown [2012] NSWCA 328 Sienkiewicz v Greif (UK) Ltd [2011] UKSC 10; [2011] 2 AC 229 State Transit Authority of New South Wales v Fritzi Chemler [2007] NSWCA 249; (2007) 5 DDCR 286 Stephens v Giovenco; Dick v Giovenco [2011] NSWCA 53 Strong v Woolworths Ltd [2012] HCA 5; (2012) 246 CLR 182 Suttor v Gundowda Pty Ltd (1950) 81 CLR 418 Tabet v Gett [2010] HCA 12; (2010) 240 CLR 537 Victims Compensation Fund Corporation v Brown [2003] HCA 54; (2003) 77 ALJR 1797 Wallace v Kam [2013] HCA 19; (2013) 87 ALJR 648 Wyong Shire Council v Vairy [2004] NSWCA 247; (2004) Aust Torts Reports 81-754 Zanner v Zanner [2010] NSWCA 343; (2010) 79 NSWLR 702 Zotti v Australian Associated Motor Insurers Ltd [2009] NSWCA 323 Texts Cited: James Allsop, "Causation in Commercial Law" in Degeling, Edelman and Goudkamp (eds), Torts in Commercial Law (2011) Lawbook Co, 269 Nicola Bodor, "A consideration of 'scope of liability' within the Restatements" (2012) 20 The Tort Law Review 163 Lord Hoffmann, "Causation" (2005) 121 Law Quarterly Review 592 David Ipp, "Problems and Progress in Remoteness of Damage" in Paul Finn (ed), Essays on Damages (1992) Law Book Company, 14 Barbara McDonald, "The impact of the Civil Liability legislation on fundamental policies and principles of the common law of negligence" (2006) 14 Torts Law Journal 268 Jane Stapleton, "Legal Cause: Cause-in-Fact and the Scope of Liability for Consequences" 54 Vanderbilt Law Review 941 (2001) Jane Stapleton, "Unpacking 'Causation'" in Cane and Gardner (eds), Relating to Responsibility: Essays for Tony Honore on his Eightieth Birthday (2001) Hart Publishing, 145 Jane Stapleton, "Choosing what we mean by 'Causation' in the Law" 73 Missouri Law Review 433 (2008) Jane Stapleton, "Unnecessary Causes" (2013) 129 Law Quarterly Review 39 Category: Principal judgment Parties: Christine Paul (Appellant) Kenneth Cooke (Respondent) Representation: Counsel: D Graham SC / R O'Keefe (Appellant) J Kirk SC / V Thomas (Respondent) Solicitors: Maurice Blackburn (Appellant) Ashurst (Respondent) File Number(s): 2012/261627 Decision under appeal Citation: Paul v Cooke [2012] NSWSC 840 Date of Decision: 2012-07-25 00:00:00 Before: Brereton J File Number(s): 2008/289586

HEADNOTE [This headnote is not to be read as part of the judgment] In 2003, Ms Paul underwent a scan to determine whether she had an intracranial aneurysm. Her radiologist Dr Cooke negligently failed to diagnose the aneurysm. In 2006, following a further scan, the aneurysm was diagnosed. On the advice of her treating practitioners and informed of the risks involved, Ms Paul underwent an operation to remove it. During the course of that operation, and without any lack of skill or care on the part of the surgeons, the aneurysm ruptured, causing her to have a stroke and suffer serious injuries. If Dr Cooke had diagnosed the aneurysm in 2003, Ms Paul would have undergone surgery then. The procedure Ms Paul underwent in 2006 (endovascular surgery) was different to the procedure she would have undergone in 2003 (open neurosurgery). Based on statistical evidence that the overall risk of stroke following rupture during either procedure was less than 1%, it was highly likely that Ms Paul would have suffered no harm had a procedure been performed in 2003; that is, "but for" Dr Cooke's failure to diagnose the aneurysm in 2003, Ms Paul would have had the aneurysm safely removed in 2003 and therefore would not have had the surgery and suffered the harm in 2006. The delayed diagnosis did not of itself increase the risks associated with surgery, in that the aneurysm did not change in size, shape or propensity to rupture during those three years. Ms Paul claimed damages from Dr Cooke for his negligence in failing to diagnose the aneurysm in 2003. The primary judge found that Dr Cooke's negligence had not caused Ms Paul's injuries, because, applying s 5D(1)(b) of the Civil Liability Act 2002 (NSW), it was not appropriate for the scope of Dr Cooke's liability to extend to those injuries. The primary judge also gave reasons as to why he would reject Dr Cooke's contention that he was not liable for Ms Paul's injuries pursuant to s 5I, because those injuries were suffered as a result of the materialisation of an inherent risk. Ms Paul appealed to this Court in relation to the primary judge's determination that causation was not made out; Dr Cooke supported the primary judge's reasoning on causation and also contended that his Honour had erred in rejecting Dr Cooke's argument based on s 5I. Held, dismissing the appeal:

1. The harm suffered by Ms Paul resulted from the materialisation of a risk occurring that could not be avoided by the exercise of reasonable care and skill. The "reasonable care and skill" referred to in s 5I(2) is not necessarily limited to that of the defendant; whether it will be in any particular case depends upon the relationship between the act of negligence and exposure inherent risk, and there was so such relationship here: Basten JA at [7]-[9], Leeming JA at [70]-[78], Ward JA agreeing with both. Wallace v Kam [2013] HCA 19; (2013) 87 ALJR 648, applied. Mahony v J Kruschich (Demolitions) Pty Ltd (1985) 156 CLR 522, considered.

Judgment 1BASTEN JA: In 2006 the appellant, Christine Paul, was diagnosed with an aneurysm in her right anterior cerebral artery. The aneurysm comprised a berry-shaped bulge in the wall of the artery. To avoid the risks of a spontaneous rupture (a similar event having killed her sister) the appellant underwent an operation to remove the aneurysm. The operation (known as "coiling") involved depositing a number of tiny coils into the aneurysm, via a micro-catheter inserted through an artery in the groin. In the course of the operation the aneurysm ruptured, without any lack of skill or care on the part of the operating practitioners. The appellant suffered serious and permanent disabilities which, if recoverable by way of damages, were agreed in the figure of $1 million. 2The respondent to the proceedings, and the defendant at trial, was Dr Kenneth Cooke, a radiologist who had failed to diagnose the aneurysm when an angiogram had been undertaken in 2003. It was agreed that he failed to exercise reasonable skill and care in not identifying the aneurysm which in 2003 was the same size and shape as in 2006. The appellant brought proceedings against Dr Cooke, claiming damages for his negligence. 3As the case was ultimately presented at trial, the sole basis upon which the respondent was said to have been legally responsible for the adverse outcome of the operation was a statistical assessment of the likelihood of the risk of rupture during the operative procedure. The risk was in the range of 1%-2%. Therefore, the appellant contended, had the operation been undertaken in 2003, or at any other time, on the probabilities, she would not have suffered the rupture. Accordingly, the respondent's negligence "caused" the damage which in fact eventuated and she was entitled to recover from him. 4The trial judge, Brereton J, rejected that reasoning: Paul v Cooke [2012] NSWSC 840. The appellant contended in this Court that the conclusion of the trial judge was erroneous. Leeming JA has concluded that it was not and that the appeal should be dismissed with costs. I agree with that conclusion and with his analysis of the evidence and the operation of Part 1A of the Civil Liability Act 2002 (NSW). However, the conclusion can be reached on a more confined basis. 5The factual premises on which the appeal must be decided were as follows: (a) the appellant had suffered an aneurysm before the angiogram undertaken in 2003; (b) there was no physical change in the aneurysm over the period from the angiogram in 2003 until the operation in 2006; (c) unsurprisingly, there was no deterioration in the standards or success of available medical procedures over that period, and (d) although there may have been a change in medical practice as to the preferred form of treatment, the risk of rupture associated with the "coiling" procedure adopted in 2006 remained the same, namely 1%-2%. 6If a negligent delay in diagnosis led to an increased risk of an adverse outcome, which in fact occurred, it might be said that the patient had lost the chance of a better medical outcome. However, damages are not recoverable for the loss of such a chance unless the better outcome is the probable result: Tabet v Gett [2010] HCA 12; 240 CLR 537; Gett v Tabet [2009] NSWCA 76; 254 ALR 504. It should follow that where the conduct of a medical practitioner, albeit negligent, did nothing at all to increase the risk of particular harm materialising, that practitioner cannot be liable for the adverse outcome which in fact eventuated. 7It is now necessary to articulate that conclusion by reference to the provisions of the Civil Liability Act, Pt 1A. The short answer is, as explained by Leeming JA, that the harm suffered by the appellant resulted from the materialisation of a risk that could not be avoided by the exercise of "reasonable care and skill". Section 5I provides that a person is not liable in negligence for such harm. 8The trial judge held that the "reasonable care and skill" referred to in s 5I(2) is that of the defendant, and not that of "some subsequent intervener whose intervention is occasioned by the defendant's negligence": at [118]. Accordingly, the trial judge concluded that the respondent could not rely upon s 5I: at [119]. 9This construction may well be correct in some circumstances. For example, if a radiologist mistakenly identified a condition from which the patient did not suffer, as a result of which the patient underwent an operation with an inherent risk of a catastrophic outcome and the outcome eventuated, through no lack of care on the part of the surgeon, the radiologist might be liable. However, that is not this case and the outcome depends, not on reading some implied restriction into the provision, but on the relationship between the act of negligence and exposure to the inherent risk. In this case there was none. (It is not necessary in the present case to consider how s 5I operates in circumstances where the negligence increases an inherent risk.) 10The second possible characterisation of the conclusion is that causation was not established for the purposes of s 5D of the Civil Liability Act. 11Section 5D(1) is designed to ensure that courts considering other than routine issues of causation distinguish between "factual causation" (s 5D(1)(a)) and policy considerations as to the appropriate "scope" of the negligent person's liability: s 5D(1)(b). However, the determination of "factual causation" is not a value-free activity undertaken without reference to context. The variables in the present case relevant to the harm suffered, were the physical condition of the aneurysm and the effect of the insertion of two micro-catheters into the vicinity of the aneurysm. The negligence of the respondent did not affect (on the facts found) either element. His negligence was no more a cause of the harm than was the death of the appellant's sister, which no doubt affected her choice to undergo medical treatment, rather than do nothing. However, the respondent did not challenge the trial judge's finding of "factual causation" and it is not appropriate to decide the case on this basis. 12In the alternative, the absence of any relationship between the negligent act and the harm suffered made it inappropriate to impose liability on the respondent for the harm which was aptly described as subsequent to, but not consequent upon, his negligent failure to diagnose. Assuming a rigid dichotomy between "factual causation" and "scope of liability", the appellant contended that factors establishing "factual causation" were to be disregarded when considering policy issues identifying the appropriate scope of the respondent's liability. Even if the test of "factual causation" is to be treated as a value-free exercise, it is not possible to pretend that policy considerations are not highly fact-specific and dependent upon the circumstances of the case. The submission that factual considerations establishing 'but-for' causation become irrelevant in considering the proper scope of liability was misconceived. 13Under either s 5I or s 5D, and under either approach to s 5D(1), the respondent was not liable. The risk which eventuated was relevantly unconnected with his breach of duty. The appellant identified no social, moral or economic principle or policy which favoured an imposition of liability on the respondent in these circumstances. 14WARD JA: I agree with Leeming JA's reasons and with the additional observations of Basten JA. 15I should add that in Cox v Fellows [2013] NSWCA 206 the question as to what was, for the purposes of s 5I, to be unable with the exercise of reasonable care and skill to be avoided was the "risk" or the "occurrence" was not a question on which the argument focussed. Instead the argument there was as to what the inherent risk was said to be. 16I agree with Leeming JA that there is not a dichotomy in s 5I between the "occurrence" being unavoidable and the "risk" being unavoidable. What must be identified is the particular risk that cannot be avoided with the exercise of reasonable care and skill. The section refers to the risk of "something occurring". It is in that context that one focusses on the occurrence, to see whether what has occurred is the materialisation of such a risk. 17Here, the risk of intra-operative rupture was always present and could not with reasonable care and skill be avoided once the decision was made to have an operation of whichever kind the appellant elected to have. It was a risk that could only have been avoided had the decision to undergo surgery not been made. 18I agree with Basten and Leeming JJA that the appeal should be dismissed with costs. 19LEEMING JA: The respondent's negligent failure to diagnose an aneurysm meant that the appellant underwent treatment in 2006, rather than in 2003. In 2006, the small chance of harm materialising from the procedure eventuated, although there was no negligence by the surgeons involved and the appellant had been fully warned of the risks. It was highly likely that had a procedure been performed in 2003, the appellant would have suffered no harm. However, the delayed diagnosis did not of itself increase the risks associated with treatment. In those circumstances, was the respondent liable for the harm suffered by the appellant? 20The primary judge found that the respondent's negligence did not cause the appellant's harm as required by s 5D of the Civil Liability Act 2002 (NSW) (Act), and accordingly dismissed the proceedings. I agree with his reasons and conclusion on legal causation. However, I respectfully disagree with the primary judge's reasoning, which was obiter, on whether the appellant's harm resulted from an "inherent risk" within the meaning of s 5I of the Act. In my view, s 5I is an additional, and indeed preferable, basis for confirming the primary judge's orders, in accordance with the respondent's notice of contention.

Factual background 21An intracranial aneurysm, often referred to as a "berry aneurysm", occurs when an artery within the skull contains a balloon-shaped bulge which distorts the flow of blood. An aneurysm may spontaneously rupture. The risk of rupture is far from fully understood, but correlates with the aneurysm's size and location. In 2003, the appellant Ms Paul had a 7-8mm anterior communicating artery aneurysm. It was common ground that, untreated, there was a 1-2% annual risk of spontaneous rupture, in which event there was a 60% chance of death, and a significant risk of permanent neurological deficit if she survived. Ms Paul's twin sister had died in 1989, aged 37, without warning, as the result of a ruptured berry aneurysm. 22On 15 July 2003, Ms Paul consulted her general practitioner, complaining of headaches, seeing flashing lights, nausea and feeling faint. He referred her for an angiogram, which was performed on 21 July 2003, and reviewed and reported on by the respondent Dr Cooke on the same day. Dr Cooke did not diagnose the aneurysm. Dr Cooke admitted that he failed to exercise reasonable care and skill when he reported that the examination of the cerebral CT angiogram of 21 July 2003 was normal. Ms Paul's berry aneurysm thus was left undiagnosed for nearly three years. 23In 2003 and 2006, there were three courses available for treating a diagnosed berry aneurysm. All were risky. The medical evidence, including an article in the Lancet dated 12 July 2003 on which all the experts relied, was that "the management of unruptured intracranial aneurysms is controversial". That article, which assessed the histories of 4,060 patients in European and North American hospitals, did not point to any clear conclusion, save that whichever course was chosen, there was a significant chance of death or serious physical harm. 24One course was described as "conservative management" - in effect, monitoring the aneurysm with a view to intervention if it changed in shape or size or if symptoms (such as a "warning headache") occurred. In the Lancet study there were 390 patients diagnosed with aneurysms sized between 7-12mm who undertook conservative management. Within that group, there was a 14.5% five-year cumulative rupture risk, which very commonly had disastrous consequences. 25A second course was the medical procedure known as "clipping". This involved an open neurosurgical operation, under general anaesthetic, by which a band would be placed around the neck of the aneurysm, ultimately eliminating it. There was a risk that the aneurysm would rupture during the procedure of some 5-10%. If that occurred, there was a relatively small risk (5%) of stroke. But there were many other risks. The article in the Lancet concluded that the general mortality and morbidity rates associated with clipping were 13.7% after 30 days, and 12.6% after one year. That is to say, of 1,591 patients who underwent the clipping procedure, 218 had died or were suffering moderate to severe neurological disability 30 days after the surgery, while after a year, 200 of the 1,591 had died or were suffering moderate to severe neurological disability (happily, the condition of some patients who suffered neurological disability in the month after surgery improved in the following 11 months). 26The third course was a procedure known as "coiling". Coiling was only 80% likely to eliminate the aneurysm permanently, but it was much less invasive. Although it was performed under a general anaesthetic, it involved inserting two micro catheters from the groin into the cranial artery, through one of which a balloon was inflated, and through the other was inserted a number of coils which would fill the aneurysm. The Lancet article examined 409 patients who had undertaken this endovascular procedure. Of those patients, 38 (9.3%) died or suffered moderate to severe neurological disability within 30 days, rising to 40 (9.8%) after one year. 27Some caution is warranted, in my view, before relying upon the percentages obtained from the Lancet article, not least because they are aggregated. A person confronted with the invidious choice faced by Ms Paul would want to know things like: is the Australian population comparable to the European and North American populations analysed in the Lancet? How many were women? How many were women of Ms Paul's age? And, perhaps especially, how many had had an aneurysm present for three years without any apparent ill effects? 28Ultimately, what matters for present purposes is the expert evidence relating to the position of Ms Paul in 2003, and 2006, contained in the joint report tendered at trial, that:

Procedural background 34Following a five day trial, the primary judge rejected Ms Paul's argument, for reasons which will be addressed in more detail when dealing with the appeal and the notice of contention. In brief, his Honour found that there was s 5D(1)(a) factual causation, but found that the s 5D(1)(b) scope of liability requirement was not made out, so that Ms Paul failed on the issue of causation. His Honour also rejected Dr Cooke's defence that the harm was a materialisation of a s 5I inherent risk. 35Three other minor matters may be mentioned briefly. First, the claim was advanced on appeal that the doctor who would have carried out the hypothetical clipping in 2003, but for Dr Cooke's negligence, would have been more experienced than the surgeon who performed the coiling in 2006. This was not squarely raised at trial, and the attempt to do so on appeal was resisted on a Suttor v Gundowda Pty Ltd (1950) 81 CLR 418 basis. In my opinion, this is a case where evidence could have been given which might have prevented the point from succeeding, and so cannot be raised on appeal. But in any event, where as here there was no evidence whatsoever of the benefits of experience as between clipping and coiling, and there was no allegation of negligence in 2006, the claim even if it were permitted to be made would not undermine the agreed risk profile of the two procedures. 36Secondly, Dr Cooke said in his notice of contention that the primary judge was wrong to proceed on the basis that there had been an admission of a duty of care which extended to taking reasonable steps to prevent intra-procedural harm, as opposed to a duty to exercise reasonable care and skill in the reporting of the angiogram (as the duty was formulated in the proceedings). It is not necessary to form a view as to whether anything could turn on that difference in the circumstances of this appeal, because the premise is wrong. As the appellant pointed out, Dr Cooke formally admitted (in a Scott Schedule filed 15 June 2011) that the only issue was causation; there was no error in the primary judge proceeding on that basis. 37Thirdly, an attempt was made late in the trial on behalf of Ms Paul to claim that she had suffered a "warning headache" in 2006, from which it was to be inferred that there was an increased susceptibility to rupture in 2006 as opposed to 2003. That was rejected at trial and was not challenged on appeal. Hence it is appropriate to proceed on a basis which will, perhaps, be relatively uncommon, namely, that the negligent failure to diagnose did not worsen Ms Paul's condition.

Part 1A of the Civil Liability Act 38In order to explain why I agree with the primary judge's rejection of the appellant's argument, it is convenient to commence with some more general observations about the language in Part 1A of the Act, before addressing the two provisions which are critical to this appeal, ss 5D and 5I.

(a) "Negligence" as defined by Part 1A of the Act 39First, Part 1A is titled "Negligence", and s 5A(1) provides for its mandatory application (subject to specified exclusions in sub-s (2)): "This Part applies to any claim for damages for harm resulting from negligence, regardless of whether the claim is brought in tort, in contract, under statute or otherwise." 40Negligence is defined exhaustively to mean "failure to exercise reasonable care and skill": s 5. Although definitions may be displaced by a contrary context, there is no occasion for that to occur here. In particular, s 5A confirms that "negligence" is not a reference to the tort, but to a category of conduct, which may be an element of a cause of action in tort, in contract, under statute or otherwise. One example giving content to "or otherwise" in s 5A(1) may be seen in admiralty: Astley v Austrust Ltd [1999] HCA 6; (1999) 197 CLR 1 at [83]; Balnaves v Smith [2010] QSC 39; [2011] 2 Qd R 17 at [122]ff. I put to one side whether and if so how the Part might apply in equity, where causation is addressed differently: O'Halloran v R T Thomas & Family Pty Ltd (1998) 45 NSWLR 262 at 271-278. It is essential to bear in mind that where the statute is directed to "duty of care" (such as in ss 5H and 5M), it is confined in its operation to causes of action in respect of which duty of care is an element. When the Act is directed specifically to tort, it says so (see ss 3C and 5Q). But where the Act provides that a person "is not negligent" (s 5B), or "is not liable in negligence" (ss 5I and 5L), or "does not incur a liability in negligence" (s 5O), then it is directed to any claim for damages for harm resulting from a failure to exercise reasonable care and skill. 41Thus the effect of s 5A is that Part 1A of the Act applies uniformly to a class of claims for damage, irrespective of how the cause of action has been formulated, so long as the damage results from a failure to exercise reasonable care and skill. The effect of sections like ss 5I, 5L and 5O is to provide a complete answer to any claim falling within Part 1A. In short, Part 1A elides the traditional categorisation of causes of action and instead imposes uniform rules and principles wherever there is a claim for harm resulting from a failure to exercise reasonable care and skill. That policy of unification was anticipated by Justice Ipp, writing extrajudicially of the divergent tests of remoteness, in "Problems and Progress in Remoteness of Damage" in Paul Finn (ed), Essays on Damages (1992) Law Book Company Ltd, 14 at 29-41. It would seem to follow that no longer is there a different test for, say, remoteness of damage caused by a failure to exercise reasonable care and skill, irrespective of whether the cause of action is in tort or contract (cf Astley v Austrust Ltd at [47] and [76]-[80]), and the same may be true for statutory causes of action (there may also be questions of construction and further analysis will be required in the case of federal statutes). That in turn demonstrates that care must be taken in evaluating statements in the extrinsic materials that the Act was not intended to change the law. In large measure, such statements at best reflect an opinion that it was not intended to change the substantive law of the tort of negligence. There is no doubt that the Act changed the law insofar as principles primarily developed in the area of negligence have been imposed upon other causes of action which involve a failure to exercise reasonable care and skill. Further, although little turns on it for present purposes, to the extent that it is directed to admissibility (s 5D(3)(b)), or onus (s 5E), or the ways in which courts are to determine issues (notably, s 5D(2) and s 5D(4)), plainly enough the Act effects a substantive change, aspects of which have been described by Professor Barbara McDonald: "The impact of the Civil Liability legislation on fundamental policies and principles of the common law of negligence" (2006) 14 Torts Law Journal 268. In relation to s 5D, twice now it has been said that whether or not s 5D will produce a different result from common law has not been decided: Adeels Palace Pty Ltd v Moubarak [2009] HCA 48; (2009) 239 CLR 420 at [44]; Strong v Woolworths Ltd [2012] HCA 5; (2012) 246 CLR 182 at [19] and [28].

(b) Warnings are treated differently 42Secondly, claims involving breach of a duty to use reasonable care and skill to give a warning are given special treatment by the Act. The provisions may be summarised as follows. There is no duty to warn of an "obvious risk", absent a request from the plaintiff or a statutory obligation or the defendant being a professional and the risk being of death or personal injury: s 5H. There is no duty to a person who engages in a "recreational activity" if there had first been a "risk warning": s 5M. There is no liability at all for an "inherent risk", save that liability is not excluded in connection with a duty to warn of such a risk: s 5I. And although a professional will be held to have satisfied a duty if he or she acted in a manner which was widely accepted by peer professional opinion by reason of s 5O, that does not apply to liability arising in connection with the failure to give a warning, advice or other information in respect of the risk of death or personal injury: s 5P. 43Before the Act came into force, cases turning upon negligent failure to warn were treated differently from other claims based on a failure to exercise reasonable care and skill. Such cases squarely invoke what the High Court has described as a duty to warn based on the "underlying common law right of the patient to choose" and a policy to "protect the patient from the occurrence of physical injury the risk of which is unacceptable to the patient": Wallace v Kam [2013] HCA 19; (2013) 87 ALJR 648 at [36]. The element of choice which is inherent in the common law right and the underlying policy is quite different from many other cases where harm is caused by a failure to take reasonable care. That reflects what the High Court described in Rogers v Whitaker (1992) 175 CLR 479 at 489 as a "fundamental difference between, on the one hand, diagnosis and treatment and, on the other hand, the provision of advice or information to a patient", a difference which in part explains the continuing controversy over the outcome and reasoning in Chappel v Hart [1998] HCA 55; (1998) 195 CLR 232 and Chester v Afshar [2004] UKHL 41; [2005] 1 AC 134. It will be seen below that the different treatment by the Act of negligent failures to warn tells against some of the appellant's submissions.

(c) Three different statutory expressions for causal connection 44Thirdly, three different expressions are used within Part 1A to express relationships of causal connection. The expressions are "caused", "resulted from" and "arising from". The difference in language is of some significance, for "the habit of a legal draftsman is to eschew synonyms", as Lord Diplock said in Prestcold (Central) Ltd v Minister of Labour [1969] 1 WLR 89 at 97. Although as Gibbs J said this can often be "of no great weight" (King v Jones (1972) 128 CLR 221 at 266), in my opinion it is worthy of note, and bears upon the question whether s 5I is a defence and how the proceeding is to be resolved. 45The language of "caused" and "causation" is used in ss 5D and 5E. In both cases, what is specified is a relationship between particular harm and a failure to take reasonable care and skill. The particular harm is to be read as that harm which comprises the claimant's claim to damages which engages s 5A(1). Section 5E confirms that, consistently with all areas of tortious, contractual and statutory liability to which the Part applies, there is an "issue" of causation. That issue is the determination in accordance with s 5D that the failure to take reasonable care and skill caused the particular harm. That is the traditional language of the common law describing an element of liability (which has been much criticised: see for example Professor Jane Stapleton, "Choosing what we mean by 'Causation' in the Law" 73 Missouri Law Review 433 at 441 (2008) ("this diversity of usage resulted in notorious confusion")). 46Elsewhere in Part 1A the language of harm "resulting from" or "as a result of" something is used. It may be seen in s 5I(1) (harm suffered as a result of the materialisation of an inherent risk), ss 5J(1) and 5M(7) (harm resulting from a recreational activity) and s 5L (harm suffered as a result of the materialisation of an obvious risk). It is also seen in s 5N(1) and (3) and (6) dealing with the qualified scope for contracting out of contractual liability that falls within the Part. 47The language of "arising from" or "arising in connection with" is used in s 5O(1) and s 5P. The language in those provisions is connecting something quite different from that connected by "caused by" and "resulting from". Whereas the latter address the causal connection between two factual matters (the failure to exercise reasonable care and skill and the harm), "arising from" and "arising in connection with" are used to link a liability with some conduct by the defendant. The connection between a legal concept (liability) and conduct by a defendant, is quite different from the connection between a factual concept (harm) and conduct by the defendant; hence the different language. 48Having noticed that third category reflected in ss 5O and 5P, it may be put to one side. What matters for present purposes is the distinction between the "issue of causation" mentioned in s 5E for which s 5D prescribes rules, and the excluding provisions in ss 5I, 5J and 5L. 49Part 1A applies only to claims for damages for harm resulting from failure to exercise reasonable care and skill. Within that class, the issue of causation will be decided in favour of the plaintiff in the circumstances stated by s 5D and in accordance with the onus in s 5E when particular harm is found to have been caused by the failure to exercise reasonable care and skill. However, there are classes of harm resulting from the failure to exercise reasonable care and skill which are stated not to give rise to liability. 50That language of "resulting from" or "as a result of" conveys a broader notion of causality than is seen in s 5D. I am conscious that the looseness of language in ordinary speech can make the words "equivalent and interchangeable" (cf Fitzgerald v Penn (1954) 91 CLR 268 at 274) and that in some contexts the words have been held to be confined to proximate cause (as to which see Zotti v Australian Associated Motor Insurers Ltd [2009] NSWCA 323 at [42]-[44]), but textual differences in statutory language matter; the working assumption is that differences are meaningful. The language of "resulting from" or "as a result of" is necessarily broader because it mirrors the language of s 5A which defines the application of the Part. For a claim only falls within the scope of Part 1A if it is a claim for damages for harm resulting from a failure to exercise reasonable care and skill. But not every such claim is one in which the claimant will succeed in establishing what s 5D requires, namely, that negligence has caused the particular harm complained of. 51It follows that once it is established that the particular harm is as a result of the materialisation of an inherent risk, or an obvious risk of a dangerous recreational activity engaged in by the plaintiff, then there is no liability for that harm under Part 1A. That is to say, the issue of causation, determined in accordance with s 5D, does not arise once it is determined a claim falls within s 5I or 5L. Putting the matter differently, ss 5I and 5L do not merely deny s 5D causation. Instead, they provide a complete answer to any claim falling within Part 1A.

Inherent risk 52In my opinion, the appropriate starting point in this appeal is s 5I, which is in these terms: "5I No liability for materialisation of inherent risk (1) A person is not liable in negligence for harm suffered by another person as a result of the materialisation of an inherent risk. (2) An inherent risk is a risk of something occurring that cannot be avoided by the exercise of reasonable care and skill. (3) This section does not operate to exclude liability in connection with a duty to warn of a risk." 53If a case can conveniently be decided under s 5I, it should be. The language of s 5I reflects the elements of liability which the plaintiff needs to establish. That is why it is framed in terms of the broader causal language of "as a result of", reflecting the language of s 5A(1) rather than of s 5D(1), and why its opening words are "A person is not liable in negligence". That is reinforced by s 5I(3), which carves out from the operation of the section "to exclude liability" a class of liability connected with a duty to warn. Section 5I does not deny s 5D causation; rather it answers the implicit question posed by the "claim" contemplated by s 5A(1) negatively: the defendant is not liable for that claim for damages for harm resulting from negligence. 54The reasons for my view that s 5I should be applied if it is available are as follows. First, once s 5I is engaged, there is no liability for a failure to exercise reasonable care and skill. The entire inquiry under Part 1A comes to an end. 55Secondly, s 5I is unyieldingly prescriptive. In contrast, the confining element in s 5D(1) of "scope of liability" is burdened with evaluative and contestable elements: the "appropriate" extension of scope of liability, and "whether or not and why responsibility for the harm should be imposed". Whenever a trial or an appeal is determined by reference to s 5D, it is necessary to address those contestable issues, which reflect what McHugh J called the "common experience of the relevant community": Dovuro Pty Ltd v Wilkins [2003] HCA 51; (2003) 215 CLR 317 at [34], applied by Campbell JA in Lym International Pty Ltd v Marcolongo [2011] NSWCA 303 at [252]-[253]. Predictably, when the appeal was heard, submissions on s 5D occupied far more time and many more pages than submissions on s 5I. Chief Justice Allsop has noted that s 5D might tend to increase the cost and complexity of litigation ("Causation in Commercial Law" in Degeling, Edelman and Goudkamp (eds), Torts in Commercial Law (2011) Lawbook Co, 269 at 296); the same concerns which underlie the overriding purpose in s 56 of the Civil Procedure Act 2005 (NSW) support relying on s 5I where it is applicable. 56Thirdly, it can never be "appropriate" pursuant to s 5D(1)(b) to extend scope of liability to circumstances excluded by s 5I (or any other of the express exclusions in Part 1A). That is so because, although little guidance is given as to what are the "relevant things" in s 5D(4) which inform what is appropriate, one such thing must surely be a provision in the self-same Part of the Act which negates liability in the first place. 57I note that the approach which I favour resembles the "straightforward path" followed in Action Paintball Games Pty Ltd (in liq) v Barker [2013] NSWCA 128 at [25] (Basten JA, Hoeben and Ward JJA agreeing).

The reasons at first instance on s 5I 58The dispositive reasoning of the primary judge (at [118]-[122]) referred to the position at common law that a tortfeasor whose negligence exposes a victim to an inherent risk that materialises will be liable, such as that occasioned by non-negligent medical treatment considered in Mahony v J Kruschich (Demolitions) Pty Ltd (1985) 156 CLR 522. On the basis that not lightly should a legislative intention be imputed to alter that position, his Honour considered that the "reasonable care and skill" in s 5I(2) was that of the defendant, and not that of some subsequent intervener. It followed that s 5I(1) did not apply where the plaintiff would not have been exposed to the risk if the defendant had used reasonable care and skill. Secondly, his Honour approached s 5I(2) on the basis that it was "the occurrence, not the risk of it" that must be unavoidable by exercise of reasonable care and skill". He said that this accorded with the focus of s 5I(1) being not the risk, but its materialisation. His Honour hence posed the question as being whether in this case (in 2006) the risk of intra-operative rupture could have been avoided by the use of reasonable care and skill. Although there was no finding (nor was it even alleged) that the surgeons in 2006 had been negligent, he concluded that the occurrence of rupture during the procedure could and would have been avoided by the exercise of reasonable care and skill on the part of Dr Cooke. 59I respectfully disagree that there is a dichotomy in s 5I between the occurrence being unavoidable and the risk being unavoidable, and I do not consider that s 5I is confined to cases where the reasonable care and skill in s 5I(2) is that of the defendant.

(a) Section 5I is about unavoidable risks of things occurring, not about unavoidable occurrences 60I start with the definition of inherent risk: "a risk of something occurring that cannot be avoided by the exercise of reasonable care and skill." That definition is circular and ambiguous. 61The definition is circular because although "inherent" is defined, "risk" is not. As Professor Stapleton has observed, "'Risk' is a term that tends to be used ambiguously ... The term is best avoided unless the sense in which it is used is made explicit": "Unpacking 'Causation'" in Cane and Gardner (eds), Relating to Responsibility: Essays of Tony Honoré on his Eightieth Birthday (2001) Hart Publishing, 145 at 170. In my view it is used here in the sense of chance or possibility. That was the sense used by Tobias JA in Wyong Shire Council v Vairy [2004] NSWCA 247; (2004) Aust Torts Reports 81-754 (although that case was not decided under the Act) at [164] when he said that that "An inherent danger is a danger (or risk) attaching to a condition or activity that cannot be removed by the exercise of due care". 62The definition is ambiguous, because it is unclear, as a matter of grammar, what the relative pronoun "that" in the defining relative clause refers to - "risk" or "something occurring"? Although it is normal for the antecedent to cling to the relative pronoun, either construction is available here. Such syntactical ambiguities recur in ordinary English, and are mostly resolved by semantics (which is to say, meaning). Contrast "A tree laden with fruit that hasn't been picked" and "A tree laden with fruit that hasn't been pruned"; the meaning of each sentence enables the identification of the noun which is the antecedent of "that". 63But the definition in s 5I(2) is different; it is like the sentence "A tree laden with fruit that hasn't been sprayed". One cannot tell from the meaning what the antecedent of "that" is - whenever the tree is sprayed, so too is the fruit, and vice versa. The same applies in s 5I(2). To say that the occurrence of something cannot be avoided by the exercise of reasonable care and skill is to say nothing more than that there is a risk of it. The reason that it is impossible to resolve the syntactical ambiguity is that it does not make sense to do so. In truth, it makes little sense to speak in the abstract of a risk; the immediate question is, a risk of what? 64Further, the operation of s 5I(1) and (2) is quite different. The definition in s 5I(2) is necessarily forward-looking, to something which may or may not occur in the future. Writing of s 5B, Campbell JA said that "As a matter of ordinary language a 'risk of harm' relates to harm that has not yet happened": Sibraa v Brown [2012] NSWCA 328 at [41]. The forward-looking, or prospective, nature of the inquiry is why there is a risk (viz a chance or a possibility or a danger) of it occurring, which risk cannot be avoided by the exercise of reasonable care and skill. In contrast, s 5I(1) is necessarily backwards-looking. For the premise of s 5I(1) is that there has been a materialisation of an inherent risk, which has resulted in harm. 65It follows that merely labelling the focus of s 5I(1) upon the materialisation of the risk fails to do justice to the way in which the definition and the operative provision operate. The better approach is to recognise that the compound term "risk of something occurring" is the natural semantic unit in the definition, that the definition in s 5I(2) looks forward into the uncertain future, and that although the premise of the substantive provision in s 5I(1) is that harm has resulted from the materialisation of a risk, to identify whether that risk was an "inherent risk" requires an analysis of the position before it materialised. 66It is as well to elaborate what that approach involves. A risk of something occurring in the future, if it cannot be avoided by the exercise of reasonable care and skill, is an "inherent risk". That is the sense in which the language was used in earlier cases, notably in Rogers v Whitaker at 483 where the joint judgment recorded that: "The principal issue in this case relates to the scope and content of the appellant's duty of care: did the appellant's failure to advise and warn the respondent of the risks inherent in the operation constitute a breach of this duty?" However, whenever a claim is made in response to which a defendant asserts that s 5I is the answer, then it is being said that a risk has materialised as a result of which harm has been suffered. To that extent, the focus then falls on the occurrence, as opposed to the risk. But even so, in order to determine whether s 5I is engaged, which is to say, whether the harm was the result of the materialisation of a risk that is an inherent risk, one must return to the definition in s 5I(2) and examine the position in a forward-looking way. 67In my opinion, by focussing on the occurrence, and posing as the ultimate question in [122] whether in this case the intra-operative rupture could have been avoided by the exercise of reasonable care and skill, the primary judge posed the wrong question. Of course, had Dr Cooke exercised reasonable care and skill, Ms Paul would not have been operated upon in 2006, and so the particular occurrence of intra-operative rupture followed by stroke in 2006 would not have taken place. But in my respectful opinion, the question posed by the statute is whether the "risk of something occurring", ie the "risk of intra-operative rupture followed by stroke", could have been avoided by the exercise of reasonable care and skill. That had to be determined from Ms Paul's position before she underwent surgery. There was always a small but unavoidable risk of intra-operative rupture followed by stroke. In those circumstances, s 5I applies. 68I am conscious that not merely did the primary judge conclude that s 5I focussed on the occurrence, as opposed to the risk of the occurrence, but also that his approach was endorsed in Cox v Fellows [2013] NSWCA 206 at [189]. The point was not considered in any detail, and may not even have been argued. In contrast, Tobias JA (with whom Mason P agreed) said (as quoted above) in Wyong Shire Council v Vairy at [164] that "An inherent danger is a danger (or risk) attaching to a condition or activity that cannot be removed by the exercise of due care", and it is plain that his Honour's formulation is the same as that preferred by me (for "that cannot be removed by the exercise of due care" cannot refer to the "condition or activity" and must relate to the "danger or risk"). Moreover, Heydon and Callinan JJ reproduced that passage and said that they did "not disagree with the descriptions of Tobias JA of inherent and obvious dangers" in the course of dismissing an appeal in Mulligan v Coffs Harbour City Council [2005] HCA 63; (2005) 223 CLR 486 at [73]. There is no reason to suppose that reasoning (of the position at common law) is inapplicable to construing s 5I. 69In contrast with those earlier decisions, this Court has enjoyed the benefit of argument on the point because (seemingly for the first time) it has arisen in a context where it matters and indeed was dispositive of Dr Cooke's reliance on s 5I. It seems to me that the better course is to determine the issue in accordance with principle (cf Lemery Holdings Pty Ltd v Reliance Financial Services Pty Ltd [2008] NSWSC 1344; (2008) 74 NSWLR 550 at [45]).

(b) Inherent risk not confined to those which cannot be avoided by reasonable care and skill by the defendant 70Secondly, his Honour confined "inherent risk" to something that cannot be avoided by the exercise of reasonable care and skill by the defendant. I respectfully disagree. 71Textually his Honour's construction faces some obstacles. First, it involves reading additional words into the Act in order to confine words of prima facie general meaning; this amounts to a judicial gloss narrowing the legal meaning of the text. In Wallace v Kam at [14] the High Court cautioned against judicial glosses on s 5D; in my view s 5I is in no different position. 72Secondly, there are examples where the identity of the person performing the conduct in a "no liability" section is stated expressly. Thus s 5L refers to dangerous recreational activities engaged in by the plaintiff. Section 5M(6) requires a risk warning to be given by the defendant or the occupier of the place where the recreational activity is engaged in. Section 5M(8) restricts the operation of risk warnings where there has been a contradiction of the warning by representations made by or on behalf of the defendant. Section 5M(9) precludes the defendant from relying on a risk warning if the plaintiff was required to engage in the activity by the defendant. (Hence an operator of a recreational activity could rely on its risk warning even if a social club visited and told its members that the activity was not dangerous and compelled its members to engage in the activity.) In short, other sections appear to be drafted on the basis that where their operation is confined to conduct of a particular person, the section says so. Section 5I is not one of those sections. I acknowledge that this is in the nature of an expressio unius consideration, to be deployed cautiously. 73Thirdly, and to my mind most influentially, the reasons of the court below reflect a concern that if the "reasonable care and skill" in s 5I were not confined to that of the defendant, then there would be a sharp and unheralded change to the common law. In my respectful view, that concern was unwarranted. 74In the class of case exemplified by Mahony v J Kruschich (Demolitions) Pty Ltd, the defendant creates the initial injury and also therefore the unavoidable risk of complications in its treatment. In those circumstances, the defendant's liability may continue even if the treatment is itself undertaken negligently. "[T]he first tortfeasor may be liable in negligence for a subsequent injury and its consequences although the act or omission of another tortfeasor is the more immediate cause of that injury .... The original injury can be regarded as carrying some risk that medical treatment might be negligently given": Mahony at 528-529. 75The present case is quite different. Ms Paul's condition pre-dated Dr Cooke's negligent diagnosis, and the relevant risk only arose after the aneurysm had been diagnosed. (Of course, Dr Cooke's negligence did leave in place the risk of a spontaneous rupture after 2003 when no treatment was undertaken, but plainly enough that was a risk which could have been avoided, so there is no question of s 5I subtracting from Dr Cooke's liability if that risk had eventuated.) Dr Cooke did not create the relevant risk - the occasion for Ms Paul to run (if she so chose) the risk of intra-operative rupture. That risk could not be avoided by reasonable care on the part of Dr Cooke; that risk was always present once Ms Paul developed an aneurysm. It could never materialise until it was diagnosed and Ms Paul chose to undergo surgery. 76Put simply, whether or not Dr Cooke exercised reasonable care, Ms Paul always faced the risk of intra-operative rupture if she chose to undergo surgery. Either way, the whole of the harm suffered by Ms Paul was as a result of the materialisation of the risk in 2006 she chose to run on the operating table. It is no answer to say that she would have, but for Dr Cooke's negligence, chosen to run that risk (or for that matter a slightly different risk) in 2003. Either way, Ms Paul faced an unavoidable risk once she chose to undergo a procedure following the diagnosis of her aneurysm. That is exactly the circumstance to which s 5I applies. Her harm resulted from the materialisation of a risk which could not be avoided by the exercise of reasonable care and skill. 77It follows that the view which I favour does not detract from the well-established principle that a defendant is liable not merely for injuries caused by his or her negligence, but also for complications (whether negligent or non-negligent) from any resulting medical treatment. Mr Kirk SC gave the example of a driver whose negligence causes a pedestrian to be hospitalised needing a general anaesthetic. The unavoidable anaesthetic risk in that case is not an "inherent risk" within the meaning of s 5I, because the occasion for administering an anaesthetic was created by the defendant's negligence. It would be inaccurate and incomplete to describe the pedestrian's harm as resulting from the materialisation of the unavoidable anaesthetic risk. The example given by Basten JA at [9] of a patient being harmed by the materialisation of an unavoidable risk from treatment which the patient only underwent by reason of a negligent diagnosis is similar: s 5I would not apply because the negligence created the occasion for the materialisation of the risk. Only where, as here, the risk pre-dates and is neither created nor increased by the defendant's negligence would the section apply. I agree with Basten JA that it is not necessary to consider how s 5I operates where negligence increases an inherent risk. 78Mr Kirk said that acceptance of this submission would, or might, amount to taking a step beyond the statements in the extrinsic materials to the effect that s 5I did not change the law. I am far from sure that that is the case; neither party pointed to any decision closely analogous to the facts of the present case (whose salient characteristic is that the delay caused by the negligence did not create or increase the risk) which would be decided differently if the construction I favour were applied. But in any event, statements in the course of the enactment of the Act do not detract from the statutory text, especially where it is clear that not all of the subtleties in this Act, which is replete with subtlety, were fully appreciated at the time. Such statements tend to be made at too high a level of abstraction, which ordinarily precludes them from being "capable of assisting in the ascertainment of the meaning of the provisions": Harrison v Melhem [2008] NSWCA 67; (2008) 72 NSWLR 380 at [12].

Is s 5I a "defence"? 79Submissions were made as to whether s 5I was a "defence", although strictly this did not arise on the notice of appeal or notice of contention. Nothing turns on this question for the purposes of this appeal - Dr Cooke squarely pleaded and relied on the section, and there was ultimately no dispute as to the facts. True it is that the appellant had advanced a case at trial that Dr Cooke had failed to establish whether intra-operative risk in 2006 could have been reduced, in support of a submission that "cannot be avoided" meant "cannot be avoided or minimised" but this was not pressed on appeal. 80It follows that in its application to this appeal the question whether s 5I is a "defence" is an arid one on which nothing turns. Moreover, as was indicated in CTM v The Queen [2008] HCA 25; (2008) 236 CLR 440 at [6]-[8], the language of "defence" is itself loosely used: it may denote a defence in the strict sense, with an evidentiary and legal onus resting with the defendant, or it may instead denote a ground of exculpation, which once raised does not shift the legal onus of displacing it from the plaintiff. It seems clear from its broad language that s 5I is a complete answer to liability which is governed by Part 1A. It resembles what in other contexts has been called a "safe harbour": cf Bond v Barry [2008] FCAFC 115; (2008) 173 FCR 106 (s 65A of the Trade Practices Act 1974 (Cth) which disapplies a range of other provisions in that Act where it is satisfied and which lends itself to early determination). Other examples may be seen in tax (MTAA Superannuation Fund (RG Casey) Building Property Pty Ltd v Commissioner of Taxation [2012] FCAFC 89; (2012) 203 FCR 415) and copyright (Roadshow Films Pty Ltd v iiNet Ltd [2012] HCA 16; (2012) 86 ALJR 494 at [25]-[26]). If a defendant wishes to take advantage of the protection afforded by that section, it is for the defendant to allege it and, I am inclined to think, to prove it. That is to say, I do not consider that it is for the plaintiff in every case where s 5I is raised to negate its applicability (contrary to the ordinary position in criminal legislation where the legal onus ordinarily remains with the prosecutor). That said, it unquestionably remains for the plaintiff to establish s 5D legal causation, and the logical consequence of so doing may (and often will be) to negate s 5I.

Section 5D causation 81In deference to the reasons of the primary judge whose decision was based on s 5D and the arguments which were at the forefront of the appeal, and, in accordance with Kuru v New South Wales [2008] HCA 26; (2008) 236 CLR 1 at [12], I now turn to that section lest the foregoing be wrong. 82Section 5D(1) and (2) is relevantly in these terms: "5D General principles (1) A determination that negligence caused particular harm comprises the following elements: (a) that the negligence was a necessary condition of the occurrence of the harm (factual causation), and (b) that it is appropriate for the scope of the negligent person's liability to extend to the harm so caused (scope of liability). (2) In determining in an exceptional case, in accordance with established principles, whether negligence that cannot be established as a necessary condition of the occurrence of harm should be accepted as establishing factual causation, the court is to consider (amongst other relevant things) whether or not and why responsibility for the harm should be imposed on the negligent party. ... (4) For the purpose of determining the scope of liability, the court is to consider (amongst other relevant things) whether or not and why responsibility for the harm should be imposed on the negligent party." 83Section 5D "is the applicable statutory provision that must be applied" in any case to which Part 1A applies: Adeels Palace at [44]. Section 5D(1) divides legal causation into two elements, "factual causation" and "scope of liability": Adeels Palace at [42]; Wallace v Kam at [12]. 84It might be thought at first blush that s 5D(1) had the effect that legal causation required there to be a favourable determination of each of the two "elements" identified in paragraphs (a) and (b). However it is plain that factual causation is not a necessary condition for legal causation. That is the effect of s 5D(2), and acknowledges those cases (described as "exceptional") where legal causation is established even though the plaintiff cannot prove that the defendant's negligence was a necessary condition of the occurrence of the harm: cf Fairchild v Glenhaven Funeral Services Ltd [2002] UKHL 22; [2003] 1 AC 32 and the issues discussed by Basten and Ward JJA in King v Western Sydney Local Health Network [2013] NSWCA 162 at [34] and [214]-[222], as well as Sienkiewicz v Greif (UK) Ltd [2011] UKSC 10; [2011] 2 AC 229; see also Professor Stapleton's article "Unnecessary Causes" (2013) 129 Law Quarterly Review 39. 85Thus, when s 5D(1) and (2) are read together, it is clear that legal causation may be established in cases when there is factual causation, and in (exceptional) cases where factual causation is not established. 86It is also clear that a plaintiff will not obtain a favourable verdict merely by establishing breach and loss which would not have been suffered but for the breach. That is to say, factual causation is not a sufficient condition for legal causation. That was plainly the position at common law (hence for example notions of remoteness and novus actus interveniens) and the Act, especially s 5D(1)(b) and 5D(4), proceeds on that premise, which is confirmed by Wallace v Kam at [21]. Scope of liability is the means by which what Professor Stapleton described as the "voraciousness" of negligence may be controlled by a court: "Legal Cause: Cause-in-Fact and the Scope of Liability for Consequences" 54 Vanderbilt Law Review 941 at 956 (2001), which is especially relevant when the trier of fact is a jury, as in the United States of America. 87It may also be noted that factual causation falls first to be determined, for scope of liability in s 5D(1)(b) refers to the harm "so caused", and it is difficult to see how an assessment of what is "appropriate" could take place without first having determined factual causation. 88The primary judge found that but for Dr Cooke's negligence, Ms Paul would have chosen promptly to undergo surgery. She would then have accepted the recommendation of her doctors (which would have been clipping) and, on the basis of the agreed risk statistics, it would have been much more likely than not that there would have been no intra-procedural rupture, and no stroke or any other injury as were sustained by her in 2006. In short, factual causation was established and was not challenged on appeal. 89The primary judge then turned to the "scope of liability" inquiry under s 5D(1)(b) and 5D(4), and approached the question by reference to principles developed by the common law, in accordance with what had been said in Ruddock v Taylor [2003] NSWCA 262; (2003) 58 NSWLR 269 at [89] (Ipp JA), noting the reservation on the part of Allsop P in Zanner v Zanner [2010] NSWCA 343; (2010) 79 NSWLR 702 at [11] ("What its statutory content is and the extent of any continuity with developing common law concepts awaits judicial elucidation"). His Honour then addressed seven topics (in a slightly different order from what follows), rejecting the various matters raised by Ms Paul as "appropriate" and "relevant" for the purposes of s 5D(1)(b) and (4).

(a) The relevant rule of responsibility 90Recognising that the analysis of causation may be decisively influenced by, in Lord Hoffmann's phrase, the "rule by which responsibility is being attributed" (Environment Agency v Empress Car Co (Abertillery) Ltd [1999] 2 AC 22 at 29, and see State Transit Authority of New South Wales v Fritzi Chemler [2007] NSWCA 249; (2007) 5 DDCR 286 at [37] and Stephens v Giovenco; Dick v Giovenco [2011] NSWCA 53 at [16]), his Honour turned to the role of diagnosis. Correct diagnosis is always or at least usually a necessary precondition to effective treatment and relief; hence harm that can be avoided or alleviated by prompt diagnosis is harm of the kind which the rule of responsibility was intended to protect. In contrast, he said (at [65]): "However, diagnosis is a precursor to treatment. Such treatment will usually carry some risks. Harm from the very treatment that prompt and proper diagnosis was intended to enable is not harm of the kind against which the relevant rule of responsibility was intended to protect a patient; to the contrary, it arises from performance of the duty to diagnose, rather than from breach of it." 91His Honour rejected the submission that the duty to diagnose was analogous to the duty to warn of material risks. He said (at [66]-[67]): "In failure to warn cases, breach of the duty results in the patient undergoing a procedure and incurring its associated unacceptable risks; in a failure to diagnose case, breach of the duty results in a condition not being discovered and treated, so that the condition persists and potentially deteriorates, continuing to cause harm that could otherwise be avoided or becoming more difficult to treat in the future. Whereas the rationale of the duty to warn (as explained by Allsop P in Wallace v Kam [2012] NSWCA 82]) is to protect the patient from harm from material inherent risks that are unacceptable, the rationale of the duty in connection with diagnosis (as explained above) is to protect the patient from harm from illness or injury that can be avoided or alleviated by treatment. Unlike the duty to warn, its purpose is not truly to enable a patient to make an informed choice about treatment or to submit to what otherwise would be an assault; rather, it is to enable the appropriate treatments to be identified. ... Harm from a spontaneous rupture which such treatment following prompt diagnosis would have avoided is harm of the kind from which the relevant duty of care is intended to protect the patient. However, the very treatment that diagnosis would enable - whether clipping or coiling - itself carries risks, including of intra-procedural rupture. Harm occasioned by the materialisation of those risks is not harm of the kind from which the relevant rule of responsibility is intended to protect the patient; to the contrary, it is harm to the risk of which the patient is exposed by treatment consequent on performance of the duty." 92His Honour concluded (at [68]) that the harm suffered by Ms Paul was not harm of the kind from which the relevant rule of responsibility was intended to protect her. 93I agree with the entirety of his Honour's reasoning, which in my respectful opinion cuts to the heart of the matter. Ms Paul submitted that the "duty to diagnose cannot be, and should not be, partitioned from the duty to inform" because the patient's right to know what he or she is suffering from, and to decide for himself or herself to undergo treatment, depended upon both diagnosis and warning. I do not agree. The duty to warn is treated differently by the Act, just as it was treated differently by the common law. In my view, neither Chappel v Hart nor Chester v Afshar assists the appellant. Those cases are failure to warn cases. The common law attaches considerable value to the individual's autonomous, and informed, choices, and vindicates the negligent denial of a right to informed choice by attaching liability. That is the right to an informed and therefore rational choice to which the High Court referred in Wallace v Kam at [8]: "The policy underlying the imposition of that component of the duty is to equip the patient with information relevant to the choice that is the patient's to make. The duty to inform the patient of inherent material risks is imposed to enable the patient to choose whether or not to run those inherent risks and thereby 'to avoid the occurrence of the particular physical injury the risk of which [the] patient is not prepared to accept'." 94But the present case is the opposite of the case where a poorly informed patient subjects himself or herself to a medical procedure involving a risk which he or she would have been unwilling otherwise to run. The negligent failure to diagnose means there is no medical procedure. The case therefore turns on whether there has materialised a risk other than from the medical procedure by reason of the delay, and the evidence in the present case was that Ms Paul's condition did not worsen between 2003 and 2006. 95The difference can be illustrated by considering different scenarios involving a negligent diagnosis. If following a negligent diagnosis, a patient is wrongly identified with a disease or condition (a "false positive"), and suffers harm before it is correctly diagnosed, then there is a ready basis for a cause of action. 96Conversely, suppose a negligent diagnosis wrongly identifies a disease, but the error is corrected before any harm is suffered (such as where the diagnosis is misplaced and not delivered until after the patient receives the correction). There could be no claim in those circumstances. So too in the present case: the negligent diagnosis failed to identify the berry aneurism, but in circumstances where the condition remained stable and unavoidable risks remained in place, there is no liability. 97The appellant further submitted that the law would serve no useful purpose if it did not reinforce the important obligation of a diagnostic radiologist to make accurate and timely diagnoses, and that Dr Cooke's negligence did not merely postpone the time for Ms Paul's surgery, for had he not been negligent, the risk of stroke would have been eliminated. Again, I disagree; the submission proves too much. A plaintiff who establishes negligence, harm and factual causation will not necessarily succeed. That was the position at common law, and it is unambiguously the position under Part 1A. It follows that arguments directed to the confining element of s 5D(1)(b) scope of liability need to be based, if they are to be persuasive, on more than the fact of breach or the fact of factual causation. 98The High Court has repeatedly acknowledged the importance attributed by the common law to the patient's right to choose, so that there is informed decision-making before undergoing medical procedures which have unavoidable risks. The central difficulty Ms Paul faces is that she was fully warned in 2006, so that common law policy is not available to support her claim based on Dr Cooke's 2003 negligence.

(b) The failure to warn cases 99His Honour addressed Chappel v Hart and Chester v Afshar and concluded that their rationale was that the risk which materialised was one to which the patient was exposed because of the negligent failure to warn (at [99]-[100]): "In such circumstances, the law regards the materialisation of the relevant risk that should have been the subject of a warning and which the patient so warned would not then have accepted, as causally connected to the failure to warn. This reflects that the duty to warn serves to protect the patient from harm from material inherent risks that are unacceptable, by enabling the patient to make an informed choice as to whether to undergo the procedure at all, or on the occasion and at the hands of the surgeon in question. These considerations are not applicable to the duty to use reasonable care in and about diagnosis, which serves a different purpose - to protect the patient from harm from an ongoing or deteriorating illness or injury, by enabling its timely treatment. The purpose of diagnosis is to enable appropriate treatment, not to enable the patient to give fully informed consent to it; the information diagnosis provides is of the conditions from which the patient suffers that may require treatment and from which the plaintiff may suffer harm if not treated, not of inherent risks associated with the proposed treatment in order fully to inform a patient's decision whether or not to proceed." 100I respectfully agree. The appellant said, correctly, that the failure to warn cases highlight the fundamental right of patients to make informed, free choices about what treatment to undergo and when to undergo it. But that does not subtract from the key difference that diagnosis is not so much a part of informing a patient of his or her choices before an activity is undertaken, but an anterior step, and that the occasion for a warning in a case like the present one only arises upon the (non-negligent) performance of the diagnosis.

(c) Foreseeability 101The primary judge dealt at [69]-[74] with the submission that surgical treatment was the foreseeable consequence of diagnosis, and so damage occasioned by it ought fall within the scope of liability. He considered that that missed the point: for treatment was a foreseeable consequence of performance of the duty, and was not a foreseeable consequence of failing to do so. Similarly, his Honour addressed the considerations that Dr Cooke would be liable if Ms Paul had suffered a stroke while driving a car, or if there had been no diagnosis for many years, such that there was an increase in the propensity to intra-procedural rupture. Those were reasonably foreseeable consequences of the breach of duty, but they were not the harm Ms Paul suffered. 102I respectfully agree with the entirety of the reasons of the primary judge on this issue. The appellant submitted that the primary judge erred by defining foreseeability too narrowly. The appellant said that the relevant and foreseeable risk was not intra-procedural rupture, or intra-procedural rupture occasioning stroke, but stroke simpliciter (whether in hospital or spontaneous). But I cannot agree that those risks should be conflated. Intra-procedural rupture is a risk which occurs at the time and place chosen by the patient, presumably following his or her fully informed and autonomous choice, because of an invasive procedure undertaken under general anaesthetic. Spontaneous rupture may be a risk chosen by a patient (whose aneurysm is diagnosed but who has chosen conservative management) or it may be risk run unknowingly, as it was by Ms Paul for at least three years prior to 2006 (and for all that is known perhaps many years longer). The causal mechanism leading to spontaneous rupture is poorly understood. It seems entirely artificial to my mind to conflate such different things, even though the ultimate harm (stroke) is the same. And the appellant provides no basis for concluding that stroke is the appropriate level of abstraction. Why not more generalised harm (such as the mortality and morbidity assessed in the Lancet article)? That is supported not only by commonsense, but also by the discussion of principle by Lord Caplan in Moyes v Lothian Health Board 1990 SLT 444 at 447 which the High Court endorsed in Wallace v Kam (at [33]-[34]): "The ordinary person who has to consider whether or not to have an operation is not interested in the exact pathological genesis of the various complications which can occur but rather in the nature and extent of the risk. The patient would want to know what chance there was of the operation going wrong and if it did what would happen."

(d) Causative irrelevance of delay in diagnosis 103His Honour addressed at [75]-[78] the argument that, perhaps unusually, this was a case where the greater the delay did not lead to a greater harm (contrast the delayed diagnosis of cancers). Ms Paul's argument at its essence turned on probabilities. If there had been a non-negligent diagnosis, she would have undergone surgery in 2003, and probably avoided rupture and stroke. However, his Honour observed that on the facts before him, where the risks remained the same, the argument would have applied equally in 2004 and 2005 and 2007 and 2008 - and indeed at any other time. His Honour concluded that the delay in diagnosis had no meaningful causal relationship to the harm suffered by Ms Paul. 104That reasoning is, in my view, plainly correct. The appellant construed the primary judge's reasons to amount to a finding that the failure to diagnose was causally irrelevant, which was an improper revisiting of factual causation. True it is that the primary judge said that there was no "meaningful or commonsense way" in which the negligence related to the catastrophe, and that the failure "was not related - except in the barest 'but for' sense" to the harm. But far from improperly revisiting the factual causation question, his Honour was instead undertaking precisely the narrowing inquiry required by s 5D(1)(b). This aspect of Ms Paul's argument once again in effect amounted to utilising the favourable finding of factual causation to pre-empt the assessment of scope of liability, while it is plain that the point of s 5D(1)(b) is to impose a further separate necessary condition before there can be a finding of causation.

(e) The scope of the risk 105The primary judge relied on what McHugh J had said in March v Stramare (E & M H) Pty Ltd (1991) 171 CLR 506 at 535 which is now reflected in s 5D(1)(b). Applying what had there been said as to what was fairly to be regarded as within the risk created by the negligence, Ms Paul said that Dr Cooke's negligence created the risk of spontaneous rupture. That is true in the loose sense that his negligence created the risk that the aneurysm, which otherwise would have been clipped or coiled in 2003, might while it was left untreated rupture spontaneously in 2004 or 2005. However, as the primary judge observed, on no view did Dr Cooke's negligence create the risk of intra-operative rupture. More recently, in Wallace v Kam at [24], the High Court has confirmed the "limiting principle" that "the scope of liability normally does not extend beyond liability for the occurrence of such harm the risk of which it was the duty of the negligent party to exercise reasonable care and skill to avoid". It was no part of Dr Cooke's duty to avoid the risk of intra-operative rupture. In my view this is a clear case for the application of that limiting principle rather than its displacement through the prism of s 5D(1)(b) "appropriateness". 106The appellant said that there was a doubling of the risk of rupture and stroke between 2003 and 2006 which was material; for if a 1:14,000 risk in Rogers v Whitaker was material, a doubling of risk must be material to the attribution of responsibility in this case. Now it is true that if the focus is on intra-operative rupture and stroke, the risk doubled by reason of the change from clipping in 2003 to coiling in 2006 (it will be recalled that there was a much greater risk of stroke following a rupture if the patient chose coiling). But the short point is, as the primary judge said, that Dr Cooke's negligence did not create any intra-operative risk, and the appellant is being understandably but artificially selective in focussing upon but one of the heightened risks from coiling, rather than the unfortunate truth that coiling and clipping, whether in 2003 or 2006, carried with them a chance of 9.8% or 12.6% of mortality or serious morbidity (assuming the Lancet study referred to above was applicable).

(f) No relevant increase in risk from different procedure 107There was, on the agreed statistics, a greater risk of stroke associated with the 2006 procedure than with the 2003 procedure. That is a consequence of the finding that Ms Paul in 2003 would have chosen clipping, with its different risk profile, one element of which is a reduced risk of stroke. The primary judge gave two responses. The first was to observe that either way the risk of intra-operative stroke was less than 1%, although his Honour (with respect correctly) noted that that could not be determinative of scope of liability. The second was to observe that the breach of duty meant that Ms Paul had an additional option available to her, with different advantages and disadvantages. He said (at [86]): "Where advances in medicine during the period while her aneurysm remained undiagnosed resulted in Mrs Paul having a wider range of treatment options available, and she chose for sound reasons and on appropriate advice an option that was in many ways more attractive than the one she would have chosen in 2003, the fact that a particular risk of the chosen option that was less likely in the alternative option materialised and caused harm is not fairly or logically to be seen as attributable in any but the barest 'but for' sense to the delayed diagnosis." I agree that those considerations do not engage s 5D(1)(b). 108The appellant submitted that the primary judge ought to have held that Ms Paul "was forced to accept the risks of the recommended treatment" in 2006 "because of the failure to diagnose the aneurysm in 2003 and the loss of any choice to have it obliterated then". On analysis, and stripped of the emotive (and inaccurate) language of compulsion, that submission is nothing more than a reiteration of the proposition that factual causation was established. 109The appellant submitted that the "strength of the factual causal connection was irrelevant to the application of section 5D(1)(b)". This overlapped with a submission that there was an erroneous "blurring" by the primary judge of the questions to be decided under s 5D(1)(a) and (b), and that once factual causation had been found: "there was no place for reconsidering that question under section 5D(1)(b) or in considering the same factors again. They had played themselves out." 110That simply cannot be so, in my opinion. It would be most surprising if scope of liability, which is central to establishing the limits of negligence, was confined to purely normative considerations. True it is that there are statements that the determination of factual causation is entirely factual and the determination of scope of liability is entirely normative, notably in Wallace v Kam in the High Court at [14]. What precisely flows from those two elements being delineated by the Act, something emphasised by the High Court, remains to be fully worked out. There is to my mind no basis in the statute or any decision on it to support the notion that the strength of the causal connection (or any other factual matter relevant to factual causation) is irrelevant to the determination of what is "appropriate". Where as here the causal connection is weak, that is far from an irrelevant consideration in determining whether it is appropriate for a defendant to be liable for the harm so caused. To give the simplest counterexample, the actions of the driver who ignores the red light and of the speeding ambulance driver racing to the hospital who follows the green light are both necessary conditions of their collision, yet the scope of liability of each is very different. The different considerations informing the scope of liability of the ambulance driver, for example, include facts (such as speed and whether the siren was operating) as well as matters grounded in policy. As Basten JA said in King v Western Sydney Local Health Network at [34]: "questions of factual causation and scope of liability, as separately identified in s 5D, do not readily fall into separate and independent watertight compartments." 111That echoes concerns raised by Lord Hoffmann over many years, notably in his lecture "Causation" published in (2005) 121 Law Quarterly Review 592; see also Rosenberg v Percival [2001] HCA 18; (2001) 205 CLR 434 at [56]-[57] (Gummow J). To my mind they are well founded. Only if human activities readily fell into well-defined pigeonholes would it be possible to erect a taxonomy of purely legal rules determining when it was appropriate to restrict a defendant's scope of liability. That would be contrary to the common law tradition and the "ordinary processes of legal reasoning" which it employs: cf Lumbers v W Cook Builders Pty Ltd (in liq) [2008] HCA 27; (2008) 232 CLR 635 at [85]. The appellant's submission amounts to an implied exclusion from the s 5D(1)(b) analysis of any factual matter which plays a part in factual causation. There is in my opinion no warrant in the text of the section (which uses open-ended language of "appropriate" and "relevant") or its purpose to support such an implication. It is to be recalled that in its New South Wales incarnation, unlike in the United States, factual causation and scope of liability will be determined by judges, who will give reasons, which will (hopefully) create a coherent body of law, a point made by Nicola Bodor, "A consideration of 'scope of liability' within the Restatements" (2012) 20 The Tort Law Review 163 at 168.

(g) Policy arguments 112Ms Paul complained that her various policy arguments were given short shrift by the primary judge. I do not think that complaint is made out. I mean no criticism, for it is probably a natural consequence of the open-ended analysis required by s 5D(1)(b), but there was a deal of overlap in the submissions that were advanced (just as there was a deal of repetition in their refutation by the primary judge, and for that matter in these reasons on appeal - it is difficult without repetition to respond to submissions with recurring themes). 113The flavour of these arguments may be gained from the following summary (which is not exhaustive). It was said that community expectations were that a wrongdoer, especially a specialist, should be liable. That amounts to an assertion - without a skerrick of evidence - that there should be no limiting principle such as s 5D(1)(b) expressly contemplates. It was said, again without evidence, that there would be a greater cost to the community if Dr Cooke were not held liable, and an increased incidence of complications from undiagnosed aneurysms. Those assertions are also, to my mind, questionable (cf Tabet v Gett [2010] HCA 12; (2010) 240 CLR 537 at [59] and [102]), and do not advance the legal argument. It was said that it would be "surprising" if scope of liability did not extend to damage which could have been identified and treated earlier, but that submission does not grapple with the key difference between harm from spontaneous rupture and harm from an informed choice to undergo surgery following diagnosis. It was said that there was a specific inquiry from Ms Paul, who would have had the aneurysm obliterated earlier, but that does not amount in my view to more than a reiteration of factual causation. It was said that Dr Cooke was relatively blameworthy, and insured, and performed his duty for commercial profit, and that the consequences for Ms Paul were serious and life-changing. All that is true, but to my mind is misdirected. 114It may be seen that the primary "policy" argument, reflected in many of the points summarised in the previous paragraph, was that negligent medical practitioners should be liable for harm which is causally connected with their breach. That argument is not correct. It ignores the subtractive effect of s 5D(1)(b) and of the previous doctrines developed by the law to constrain the limits of liability for negligence which inform its operation. The primary judge was, with respect, correct to say (at [104]) that "that amounts to no more than an invitation to disregard the requirement to establish causation, in order to reinforce the importance of the duty of care." In this respect, the position is no different from Victims Compensation Fund Corporation v Brown [2003] HCA 54; (2003) 77 ALJR 1797, where Heydon J said for the Court at [33]: "The legislation has endeavoured to define [the circumstances in which compensation is payable] in precise language which does not permit universal recovery; and hence '[t]he Court is not required to give the most expansive possible interpretation of such circumstances'." 115The nuanced language of the Act calls for a more careful analysis than that for which the appellant contended. If anything is clear from the statutory language, it is that the Legislature has confirmed the position at common law that breach of duty coupled with factual causation does not entitle a plaintiff who has been harmed to recover. As Windeyer J said in Kavanagh v The Commonwealth (1960) 103 CLR 547 at 584: "Law may at times seem to flounder somewhat when it approaches questions of causation, professing to avoid philosophic consideration of what is essentially a problem of philosophy. Law must, for its purposes, extract one or more circumstances out of the whole complex of antecedent conditions of an event as its cause." Naturally, s 5D changes the framework in which that process takes place, although it refrains from providing clear guidance. That said, Windeyer J's statement acknowledging the conceptual messiness of determining the limits of legal liability through the means of "scope of liability" retains force. 116Given that the law demands more than breach of duty and factual causation to require a defendant to compensate for harm, this is a clear case, in my view, to conclude that it is not appropriate for the scope of Dr Cooke's liability to extend to Ms Paul's harm. If Dr Cooke had exercised reasonable care and skill, and diagnosed the berry aneurysm in 2003, then Ms Paul would still have faced the risk of a procedure she willingly chose to undergo. Conversely, the only thing that would have been avoided had Dr Cooke exercised reasonable care and skill was the three year period from 2003 to 2006 which, it is now known beyond any doubt, did not result in any harm to Ms Paul. 117There will be many cases where s 5D(1)(b) considerations are finely balanced. This is not one of them. Although no one could review the facts of this case without feeling sympathy for Ms Paul, the harm she has suffered is not within the scope of Dr Cooke's liability. Accepting, as the Act commands, that there must be some occasions where breach of duty and factual causation are not sufficient, the absence of any increased risk and the fact that Ms Paul's harm resulted from her informed exercise of choice to undergo a risky procedure combine to make this a clear case. I propose that the appeal be dismissed, with costs.

Amendments 02 June 2014 - Typographical errors corrected Amended paragraphs: 8, 41, 50 and 114 02 June 2014 - Citation for Strong v Woolworths Ltd [2012] HCA 5; (2012) 246 CLR 182 corrected Amended paragraphs: Coversheet 02 June 2014 - "what" inserted after the words "in accordance with" in the first sentence Amended paragraphs: 89 02 June 2014 - The words "of America" have been deleted in the last sentence of the paragraph Amended paragraphs: 111 02 June 2014 - In the quotation from [86] of the primary judgment, the words "Mrs Paul" have been substituted for the words "Mrs Cooke". Amended paragraphs: 107 02 June 2014 - Quotation marks in fourth sentence moved so that they appear before the word "not" rather than before the word "did" Amended paragraphs: 68 02 June 2014 - Title of publication amended from "Essays in honour for Tony Honore on his Eightieth Birthday" to "Essays in honour of Tony Honore on his Eightieth Birthday" Amended paragraphs: 61 DISCLAIMER - Every effort has been made to comply with suppression orders or statutory provisions prohibiting publication that may apply to this judgment or decision. The onus remains on any person using material in the judgment or decision to ensure that the intended use of that material does not breach any such order or provision. Further enquiries may be directed to the Registry of the Court or Tribunal in which it was generated. Decision last updated: 02 June 2014