AustLII Home | Databases | WorldLII | Search | Feedback

Precedent (Australian Lawyers Alliance)

You are here:  AustLII >> Databases >> Precedent (Australian Lawyers Alliance) >> 2020 >> [2020] PrecedentAULA 18

Database Search | Name Search | Recent Articles | Noteup | LawCite | Author Info | Download | Help

Graham, Duncan --- "Causation problems in medical negligence cases - Part 2" [2020] PrecedentAULA 18; (2020) 157 Precedent 26


PART 2: EPIDEMIOLOGICAL CAUSATION PROBLEMS

By Dr Duncan Graham SC

In medical negligence cases, there are two situations in which epidemiology is important. The first is in proving that an injury was caused by a treatment provided (for example, did a wrongly prescribed drug cause a heart attack or were unrelated factors responsible?). The second, and more problematic, is in proving as part of the counterfactual analysis that the treatment that should have been provided would have avoided the injury (for example, if breast cancer had been diagnosed earlier, the prognosis would have been better).

There is often confusion about what epidemiology is and whether it is of any use in determining causation. This is surprising considering it is one of the most important fields of scientific research to determine the safety and efficacy of pharmaceuticals and medical devices.

It is thus useful to know what it is and what its limitations may be.

WHAT IS EPIDEMIOLOGY?

In Seltsam Pty Ltd v McGuinness,[1] Spigelman CJ (with whom Davies AJA agreed) described epidemiology as follows:

‘Epidemiology is the study of the distribution and determinants of disease in human populations. It is based on the assumption that a disease is not distributed randomly in a group of individuals. Accordingly, subgroups may be identified which are at increased risk of contracting particular diseases.

Epidemiological evidence identifies associations between specific forms of exposure and the risk of disease in groups of individuals. Epidemiologists do make judgments about whether a statistical association represents a cause-effect relationship. However, those judgments focus on what is sometimes called in the epidemiological literature "general causation": Whether or not the particular factor is capable of causing the disease. Epidemiologists are not concerned with "specific causation": Did the particular factor cause the disease in an individual case?’[2]

In Amaca Pty Ltd v Booth,[3] Gummow, Hayne and Crennan JJ referred to the following passage from the UK decision, Sienkiewicz v Grief (UK) Ltd; Knowsley Metropolitan Borough Council v Willmore,[4] to describe how epidemiological evidence is used to answer issues of causation:

‘Epidemiology is the study of the occurrence and distribution of events (such as disease) over human populations. It seeks to determine whether statistical associations between these events and supposed determinants can be demonstrated. Whether those associations if proved demonstrate an underlying biological causal relationship is a further and different question from the question of statistical association on which the epidemiology is initially engaged.

Epidemiology may be used in an attempt to establish different matters in relation to a disease. It may help to establish what agents are capable of causing a disease, for instance that both cigarette smoke and asbestos dust are capable of causing lung cancer, it may help to establish which agent, or which source of an agent, was the cause, or it may help to establish whether or not one agent combined with another in causing the disease.’[5]

It is only through understanding what epidemiology is that we are able to understand what it can and cannot prove. In practice, this means appreciating the difference between general and individual causation. Too often this distinction is not appreciated by plaintiff lawyers.

ASBESTOS CASES

The leading authorities on the use of epidemiology are asbestos cases. A good starting point is Seltsam Pty Ltd v McGuinness, where the court was required to determine whether the plaintiff’s kidney cancer was caused by asbestos exposure. The trial judge found that it was but the NSW Court of Appeal disagreed. The Court of Appeal decision was based on the quality of the epidemiological studies relied on in evidence.

Chief Justice Spigelman described the use of epidemiological evidence in a number of oft-cited paragraphs:

‘Epidemiology is, as I have noted above, concerned with the study of disease in human populations. It is not, of itself, directed to the circumstances of an individual case. For the purpose of determining whether exposure to a particular substance is the legal cause of a particular disease, epidemiology only provides evidence of possibility.

Evidence of possibility, including expert evidence of possibility expressed in opinion form and evidence of possibility from epidemiological research or other statistical indicators, is admissible and must be weighed in the balance with other factors, when determining whether or not, on the balance of probabilities, an inference of causation in a specific case could or should be drawn. Where, however, the whole of the evidence does not rise above the level of possibility, either alone or cumulatively, such an inference is not open to be drawn.’[6]

‘In my opinion, evidence of possibility, including epidemiological studies, should be regarded as circumstantial evidence which may, alone or in combination with other evidence, establish causation in a specific case.’[7]

‘Causation, like any other fact can be established by a process of inference which combines primary facts like "strands in a cable" rather than "links in a chain", to use Wigmore's simile ...’[8]

This approach is consistent with viewing causation as involving a general and an individual question. Epidemiology is relevant to the first question, but is rarely on its own capable of answering the second. Chief Justice Spigelman in essence was saying that epidemiological evidence establishes that a particular factor can cause harm (general causation) but not that it in fact did so in the particular case (individual causation). The evidence establishes there is a risk of harm, but something more is required to prove that the risk ‘came home’ in the individual case.[9]

His Honour then analysed the quality of the epidemiology evidence. Many studies showed no increased risk of kidney cancer associated with asbestos exposure. In the studies that did show an increase, the increased risk was only low to moderate. The extent of the increased risk was too small to justify an inference of causation, either alone or in combination with other factors such as biological plausibility, laboratory experiments or professional opinion. Chief Justice Spigelman concluded that the epidemiological evidence only proved on the general causation question a possibility of a causal association.[10]

In Amaca Pty Ltd v Ellis,[11] the High Court considered the use of epidemiological evidence in assessing whether the deceased’s lung cancer was caused by his asbestos exposure and/or his cigarette smoking. The Court stated:

‘To draw an inference about causation from what was established by the epidemiological studies, it would be necessary to decide whether the particular case under consideration should be treated as conforming to the pattern described by the epidemiological studies. Absent evidence which suggests that the individual may stand apart from the ordinary, there may be sufficient reason to assume conformity, but ... the first step that must be taken, if an inference is to be drawn from epidemiological studies, is to relate the results of studies of populations to the particular case at hand. That step is not inevitable.’[12] [footnote omitted.]

In Amaca Pty Ltd v Booth, the High Court found that the plaintiff’s exposure to asbestos from brake linings caused his mesothelioma. Chief Justice French considered how causation can be inferred from epidemiological evidence:

‘In summary, a finding that a defendant’s conduct has increased the risk of injury to the plaintiff must rest upon more than a mere statistical correlation between that kind of conduct and that kind of injury. It requires the existence of a causal connection between the conduct and the injury, albeit other causative factors may be in play. As demonstrated by medical evidence in this case ... A causal connection may be inferred by somebody expert in the relevant field considering the nature and incidents of the correlation. The Bradford Hill criteria provide a guide to the kind of considerations that lead to an inference of causal connection. As noted above, they may include reference to relative risk ratio as an indicator of the strength of the association. Where the existence of a causal connection is accepted it can support an inference, in the particular case, when injury has eventuated, that the defendant’s conduct was a cause of the injury ... Where such an inference is drawn, the probability that it is correct is not to be determined only by reference to epidemiologically based ex ante probabilities.’[13] [footnote omitted.]

DRAWING A CAUSAL INFERENCE FROM EPIDEMIOLOGY

In the representative proceedings regarding the drug Vioxx, Merck Sharp and Dohme (Australia) Pty Ltd v Peterson,[14] the Full Court overturned a finding of causation on the basis that the applicant, Mr Peterson, stood apart from the ordinary case as set out in the epidemiological studies. His myocardial infarction was readily explained by personal risk factors that were independent of the possible effects of Vioxx, such that the strength of the epidemiological evidence as a strand in the cable of circumstantial proof was insufficient to establish causation. Chief Justice Keane, Bennett and Gordon JJ said:

‘It is desirable to begin by referring to the legal principles governing the scope of the legitimate use of epidemiological evidence in relation to proof of causation. We begin with the proposition that proof of what may be expected to happen in the usual case is of no value unless it is proved that the particular applicant is indeed "the usual case".’[15]

In order for epidemiological evidence to support an inference of causation in an individual case, it is therefore necessary that the particular case conform to the pattern described by the epidemiological studies. In some circumstances, absent evidence which suggests that the individual may stand apart from the ordinary, there may be sufficient reason to assume conformity and to draw an inference of causation.[16] But it is rarely so straightforward.

Two recent cases expose these problems: Pierce v Metro North Hospital and Health Service[17] and King v Western Sydney Local Health Network.[18] Pierce is a case where epidemiology was relevant to proving whether the actual treatment caused the injury (a negligent act). King is a case where epidemiology was relevant to the counterfactual analysis – determining whether the hypothetical treatment that should have occurred would have avoided the injury (a negligent omission).

Negligent omission

In King, the plaintiff was seriously injured in utero (fetal varicella syndrome (FVS)) after her mother developed chickenpox during pregnancy. The defendant was found to be negligent for failing to recommend and give varicella-zoster immunoglobulin (VZIG) to the plaintiff’s mother. The question when considering the counterfactual analysis was whether administration of VZIG would have prevented the mother’s chickenpox and consequently prevented FVS. This question entailed an analysis of epidemiological studies about the efficacy of VZIG in preventing chickenpox.

The largest study indicated that 54 per cent of a maternal population avoided the consequences of chickenpox if they were administered VZIG. The accepted dose of VZIG in Australia was 600 international units (IU). The study tested the efficacy of a dose of 1750–2000IU. Accordingly, but for the defendant’s breach, the mother would have received a dose of 600IU, one third of the dose proven to be effective. The trial judge reduced the percentage chance of avoiding chickenpox that was reported in the study to something less than 50 per cent given the disparity between the doses. In other words, the mother did not conform to the study population (she did not receive the same dose). He also reduced the percentage because 10 per cent of pregnant women would not develop chickenpox regardless of whether VZIG had been administered.

At first blush, it seemed illogical to think that Australian authorities would recommend an ineffective dose of VZIG or that doctors would order VZIG if they thought that it was of no benefit. Nonetheless, the majority in the Court of Appeal (Hoeben and Ward JJA) accepted the trial judge’s reasoning as correct. Justice Ward scrutinised the epidemiological evidence in detail and concluded that the studies, particularly the large study central to the case, only addressed the issue of whether giving VZIG might have prevented injury to the fetus rather than whether it would probably have done so.[19] There was roughly a 40 to 50 per cent chance that VZIG would have prevented the negative consequences of chickenpox in the mother. This was insufficient to prove that the administration of VZIG would have prevented maternal chickenpox. In any event, the chance of a fetus developing FVS once the mother contracted chickenpox was very small (less than 1 per cent). The statistical correlation between prevention of chickenpox in the mother and prevention of FVS in the baby meant that regardless of whether VZIG was given, there was only a very small possibility that the fetus would contract FVS.

Negligent act

In Pierce v Metro North Hospital and Health Service, the question for the court was whether complex partial status epilepticus (CPSE) caused brain damage and worsening epilepsy. The plaintiff had been admitted to Royal Brisbane Hospital for investigation of the site of origin of her disabling epilepsy. A fit was induced, and electroencephalogram traces were obtained for interpretation (telemetry), but the hospital failed to then terminate the fit by administering appropriate drugs. The plaintiff suffered a prolonged period of CPSE of almost two and three-quarterhours. Following the episode, the plaintiff’s epilepsy worsened and she suffered cognitive impairment.

The defendant disputed any causal connection and asserted that the deterioration was due to the natural progression of the underlying condition. In large part this argument rested on an absence of epidemiological studies demonstrating a connection between CPSE and brain damage or aggravation of epilepsy. There was a contest between the neurologists called by the parties. Prof Dunne for the defendant asserted that the scientific literature did not support the hypothesis that CPSE of less than three days duration causes discrete brain or neuronal damage. He conceded, however, that for there to be a change ‘in frequency and character ... there must be a change in some way in the electrical activity of the relevant neurons within the brain’.[20] He agreed that there was no scientific reason why the electrical activity of CPSE ‘cannot cause neuronal damage’.[21] He also accepted that there were shortcomings with the epidemiology. The trial judge found that the epidemiology studies did not provide a sure guide as to whether the plaintiff’s CPSE could produce the type of change, or electrical disturbance, in the brain that could have itself lead to a definite progression, acceleration or aggravation of epilepsy.[22]

The trial judge accepted that he could draw a causal inference, based on all the evidence including the epidemiological evidence, that it was possible CPSE could cause neuronal damage. This was despite the fact that no study had yet been performed to prove definitively that CPSE of two and three-quarterhours duration caused damage and worsened the underlying disease.

The defendant appealed (Metro North Hospital and Health Service v Pierce[23]). Justice White, speaking for the Court, accepted the following submission of counsel for Ms Pierce:

‘The real difference between the experts on the issue of causation related to matters of high-level scientific proof. In the absence of scientific data to prove a causal link categorically, Prof Dunne was not prepared to accept one. Dr Fong was of the view that scientific possibility and a temporal link were a more likely explanation than natural progression, coincidentally at a time of serious insult to the Respondent’s brain. Having found there was a change linked in time to the telemetry event, it was entirely proper for the Primary Judge to then infer that the more likely explanation was neuronal damage caused by the CPSE, which was the hypothesis preferred by Dr Fong.’[24]

This was orthodox reasoning based on a long line of cases concerning causal inferences, including McGuiness, EMI (Australia) Ltd v Bes,[25] Fernandez v Tubemakers of Australia Ltd,[26] and Adelaide Stevedoring Co Ltd v Forst.[27] Justice White cited the following passage from Dixon J’s dissenting judgment in Adelaide Stevedoring Co Ltd:

‘First, I think that upon a question of fact of a medical or scientific description, a court can only say that the burden of proof has not been discharged where, upon the evidence, it appears that the present state of knowledge does not admit of an affirmative answer, and that competent and trustworthy expert opinion regards an affirmative answer as lacking justification, either as a probable inference or as an accepted hypothesis.’[28]

Having regard to Prof Dunne’s concessions in cross-examination, White JA said that Prof Dunne regarded it as an accepted hypothesis that the telemetry event of 5 January 2010 materially contributed to the increased frequency of the subsequent seizures suffered by Ms Pierce.[29] Dr Fong regarded that as a probable inference. In the circumstances, it was reasonable for the trial judge to have drawn the same causal inference.

Pierce is not an unusual case; frequently, there may be no directly applicable epidemiological study available. For example, corticosteroids may be routinely prescribed for intracerebral inflammation caused by a parasite even though there are no reliable studies establishing the efficacy of such treatment. Does this mean that early diagnosis of the inflammation and the prescription of steroids would have been to no avail? The absence of a specific study proving the point does not mean the treatment is ineffective. There may be many reasons for the absence of a definitive study – it may be unethical to test the theory, or the numbers required to see a real difference may be huge and impossible to assemble. In such instances, recourse to first principles and other studies or experiments coupled with weak epidemiological studies may establish causation.

In order to assess epidemiological evidence and its applicability to a case, it may be necessary to engage an epidemiologist or biostatistician. It is also useful to engage in a degree of self-education on the topic. As a starting point, it is important to understand the quality of the studies relied on by experts. Studies may be observational (cross-sectional surveys, cohort studies or case-controlled studies) or experimental (randomised control trials (RCTs) or quasi-experiments). Generally, RCTs and meta-analyses are considered the highest or best level of evidence.

Familiarity with the Bradford Hill criteria[30] is also advisable. These are an appropriate and well-accepted set of criteria that can be applied to epidemiological evidence to assess the causal hypothesis. The criteria provide a guide for the conditions necessary to provide adequate evidence of a causal relationship between an act or omission and a pathology.

As King demonstrates, paying attention to the methodology used in a study is crucial. Do not be surprised if a study’s conclusions are not supported by its statistical findings.

Dr Duncan Graham SC is a barrister at Maurice Byers Chambers, Sydney. PHONE (02) 8233 0306 EMAIL d.graham@mauricebyers.com.


[1] (2000) 49 NSWLR 262 (Seltsam).

[2] Ibid, [59]–[60].

[3] [2011] HCA 53; (2011) 246 CLR 36 (Booth).

[4] [2011] UKSC 10, [80]–[81] per Lord Phillips.

[5] Booth, above note 3, [86].

[6] Seltsam, above note 1, [78]–[79].

[7] Ibid, [89].

[8] Ibid, [91].

[9] Bendix Mintex Pty Ltd v Barnes (1997) 42 NSWLR 307, 318.

[10] Ibid, [174]–[175], [183].

[11] [2010] HCA 5; (2010) 240 CLR 111 (Ellis).

[12] Ibid, [62]–[63].

[13] Booth, above note 3, [57].

[14] [2011] FCAFC 128.

[15] Ibid, [106].

[16] Ellis, above note 11, [62].

[17] [2016] NSWSC 1559 (Pierce).

[18] [2013] NSWCA 162.

[19] Ibid, [168].

[20] Pierce, above note 17, [149].

[21] Ibid.

[22] Ibid, [150].

[23] [2018] NSWCA 11 (Pierce appeal).

[24] Ibid, [152].

[25] [1970] 2 NSWR 238, 242.

[26] [1975] 2 NSWLR 190, 197

[27] (1940) 64 CLR 538.

[28] Pierce appeal, above note 23, [142].

[29] Ibid, [143].

[30] BA Hill, ‘The environment and disease: association or causation?’, Proceedings of the Royal Society of Medicine, Vol. 58(5), 1965, 295–300.


AustLII: Copyright Policy | Disclaimers | Privacy Policy | Feedback
URL: http://www.austlii.edu.au/au/journals/PrecedentAULA/2020/18.html