What happened
Reema Tabet was six years old in January 1991 when she contracted chickenpox from her brother. She had a pre-existing history of headaches and vomiting dating back to 18 December 1990. On 28 December she was seen by Dr Mansour, who treated her for a suspected streptococcal infection and discharged her. She was re-admitted between 29 and 31 December with continuing symptoms but no papilloedema or neurological signs; chickenpox rash appeared and she was again discharged. On 11 January 1991 her condition had deteriorated and she was referred to the appellant, Dr Maurice Gett, a locum paediatrician. Dr Gett provisionally diagnosed post-chickenpox meningoencephalitis, arranged immediate admission, and directed a lumbar puncture (which was unsuccessful due to distress). The child remained in hospital. On 13 January at approximately 11 am she suffered a neurological episode described by her father as "staring and unresponsive". Nursing notes recorded unequal pupils (right non-reactive), drowsiness, irritability and headache. Dr Gett was informed, examined her, and directed an urgent lumbar puncture which was performed at 11.30 am. On 14 January she suffered a major seizure. An urgent CT scan revealed a medulloblastoma (a posterior fossa tumour that had been growing for over two years). A ventricular drain was inserted, followed by partial surgical resection on 16 January, chemotherapy and radiotherapy. The respondent suffered irreversible brain damage attributable in part to the tumour itself, in part to the events of 14 January (seizure and raised intracranial pressure), and in part to the treatment.
Proceedings were brought against both Dr Mansour and Dr Gett. Studdert J at first instance ([2007] NSWSC 36) found Dr Mansour not negligent. He found Dr Gett breached his duty on 13 January by failing to order a CT scan once the neurological incident occurred in the context of an unexplained headache history predating the chickenpox prodrome. However, his Honour was not satisfied on the balance of probabilities that the breach caused the full extent of the damage referable to 14 January (assessed at 25% of the total disability). Instead, applying Rufo v Hosking [2004] NSWCA 391, he awarded damages for loss of a 40% chance of a better medical outcome, producing a judgment of $610,000. Dr Gett appealed on breach, causation and the applicability of Rufo. The respondent cross-appealed, contending that causation should have been found on the balance of probabilities for the whole 25% and that breach occurred earlier on 11 January. The Court of Appeal (Allsop P, Beazley and Basten JJA jointly) allowed the appeal, entered judgment for Dr Gett, dismissed the cross-appeal and expressly declined to follow Rufo and Gavalas v Singh [2001] VSCA 23.
The factual controversy at trial turned heavily on expert evidence. The appellant's experts (including Mr Johnston, Dr Allen and Professor Watson) considered the provisional diagnosis reasonable and the lumbar puncture appropriate even after the 13 January incident, given the absence of papilloedema and the consistency of symptoms with viral encephalitis. The respondent's experts (Dr Knight, Dr Hopkins, Mr Wallace, Mr Klug and Dr Maixner) emphasised the unexplained headache history from 18 December, the neurological signs on 13 January, and the need for urgent CT imaging before lumbar puncture to exclude raised intracranial pressure. The trial judge preferred the latter group after a detailed review of the records and oral evidence. On appeal the Court conducted its own review under Supreme Court Act 1970 (NSW), s 75A, giving weight to the trial judge's advantages in assessing expert persuasiveness but ultimately concluding there was no error on breach. Causation, however, failed on the balance of probabilities because the evidence (particularly from Mr Johnston) showed that even with a CT scan on 13 January the likely initial treatment would have been steroids rather than immediate drainage, and any benefit in preventing the 14 January deterioration was speculative. The Court also held that the 40% chance assessment had in any event misapplied the loss of chance analysis by overweighting the possibility of drainage.
Why the court decided this way
The Court began by reiterating the appellate task on a rehearing. Citing Fox v Percy [2003] HCA 22 and CSR Ltd v Della Maddalena [2006] HCA 1, it emphasised that the appellate court must conduct a "real review" of the evidence and inferences but must make "proper allowance" for the trial judge's advantages in seeing and hearing witnesses, assimilating a large body of expert evidence over an extended trial, and making evaluative judgments that cannot be fully articulated in reasons. In a borderline case such as this, where the trial judge's conclusion on breach was not implausible and was supported by a substantial body of expert opinion, the Court would not intervene unless satisfied the conclusion was wrong. After a meticulous chronological review of the clinical records, the appellant's evidence (including concessions that he had not fully explored the headache history), and the competing expert views, the Court concluded there was no error in finding breach on 13 January. The provisional diagnosis was reasonable on 11 January given the acute presentation and chickenpox history, but by 13 January the combination of the pre-chickenpox headache history (which the appellant either knew or ought to have known from the records and parents) and the new focal neurological signs required urgent imaging before lumbar puncture. The absence of papilloedema was important but not decisive; the history itself was a "red light" that the diagnosis of viral encephalitis could not fully explain.
On causation the Court was not satisfied that the breach had been shown to have materially contributed to the 25% of disability attributable to the 14 January events. Even if a CT scan had been performed on 13 January and the tumour discovered, the likely neurosurgical approach at that hospital (Mr Johnston's evidence) was initial steroids rather than immediate drainage. Mr Johnston's evidence was that any benefit in preventing the seizure was "speculative" and "entirely possible" that the same deterioration would have occurred. The Court rejected the respondent's attempt to shift the burden of proof under principles derived from Betts v Whittingslowe and Chappel v Hart, citing Flounders v Millar, Roads and Traffic Authority v Royal and Seltsam v McGuinness for the proposition that material increase in risk is not equated with material contribution; the plaintiff must still prove on the balance of probabilities that the risk came home. The references in the trial judgment to earlier intervention being "better" were too general to support a finding of divisible damage caused by the delay on 14 January.
The decisive legal step was the reconsideration of Rufo. The Court first set out the constitutional and doctrinal framework for precedent in an integrated national system. Intermediate courts are not bound by their own decisions but should depart only on a principled basis. The test is whether the earlier decision is "plainly wrong" (strong conviction of error, not mere preference for another view) and whether there are "compelling reasons" (both jurisprudential coherence and pragmatic considerations of certainty and predictability). Rufo was plainly wrong because it permitted recovery for loss of a chance of a better medical outcome without requiring the plaintiff to prove on the balance of probabilities that the breach caused actual harm. This altered the conventional causation rule, was inconsistent with the statutory definitions of "harm" and causation in the Civil Liability Act 2002 (NSW) (and equivalent legislation elsewhere), and lacked any clear limiting principle. Sellars and Malec were distinguished: they concerned the valuation of lost commercial opportunities or the assessment of contingencies once causation of some loss is established on the balance of probabilities. The Court noted that the UK authorities (Hotson, Gregg v Scott, Barker v Corus) and Canadian authority (Laferrière v Lawson) pointed against the Rufo approach. Policy considerations—coherence of tort law, the proper role of the High Court in reforming causation principles, and the absence of any insurance or societal adaptation to Rufo—provided compelling reasons to depart. Because causation was not established on orthodox principles, the appeal had to be allowed and judgment entered for Dr Gett. The cross-appeal on earlier breach and on loss of marriage benefits failed for the reasons given at first instance.
Before and after state of the law
Before Gett v Tabet the law on loss of a chance in Australian medical negligence was unsettled. Sellars v Adelaide Petroleum NL [1994] HCA 4 had recognised loss of a commercial opportunity as compensable damage whose value could be assessed according to possibilities rather than the balance of probabilities. Malec v JC Hutton Pty Ltd [1990] HCA 20 had established that once causation of some loss is proven, future hypothetical events and contingencies are assessed proportionally. Dicta in Chappel v Hart [1998] HCA 55 (Kirby J) and Naxakis v Western General Hospital [1999] HCA 22 (Callinan J) suggested that loss of a chance might be available in medical cases, but Gaudron J in Naxakis expressly rejected the approach where physical injury had eventuated. In this Court, Gavalas v Singh (Victorian Court of Appeal, but followed in NSW) and Rufo v Hosking appeared to accept that a less-than-50% chance of a better medical outcome could sound in damages, although in Rufo the point was conceded and the Court did not analyse the tension with orthodox causation. Lower courts and commentators treated the law as at least open to a loss of chance analysis in medical negligence.
After Gett v Tabet the position is clear in New South Wales and, by the reasoning on national integrated precedent, across Australia unless the High Court intervenes. A plaintiff in a medical negligence case must prove on the balance of probabilities that the breach of duty caused or materially contributed to actual harm (physical or mental injury or death, or economic loss flowing therefrom). Mere creation of a risk or loss of a chance of a better outcome is not compensable harm. Rufo and Gavalas are not to be followed. The Civil Liability Acts in all Australian jurisdictions (e.g. Civil Liability Act 2002 (NSW), ss 5, 5D-5E) are now to be read as confirming that position; their definitions of "harm" and the express requirement that the plaintiff bears the onus on causation on the balance of probabilities are inconsistent with a loss of chance doctrine. Causation remains an "all or nothing" inquiry once the threshold of material contribution is crossed, subject only to ordinary vicissitudes in assessment of damages (Malec). Appellate courts must continue to apply the "plainly wrong/compelling reasons" test when reconsidering their own or co-ordinate authority, with particular emphasis on coherence with statute and the national common law. The decision reinforces the boundary between commercial lost opportunity cases and personal injury, preserving the latter within orthodox tort principles.
Key passages with plain-English translation
[3]: "An appeal by way of rehearing under the Supreme Court Act 1970 (NSW), s 75A requires the Court of Appeal to correct errors of law and to assess the evidence, inferences and evaluative judgments made to determine the case. This requirement is subject to limitations on the Court's ability to comprehend fully the evidence and process of the trial."
Plain English: On appeal the judges must look at the whole case afresh, but they weren't in the courtroom watching the experts give evidence for days, so they must give real weight to the trial judge's view unless it is clearly wrong.
[22]: "To succeed here, the appellant must satisfy this Court that the trial judge came to the wrong conclusion. Where no error can be identified and the conclusion of his Honour is not itself implausible, this Court should properly give weight to the view formed by the trial judge, because to do so is to acknowledge the fact that in various respects, his Honour enjoyed a beneficial position in resolving conflicting evidence, drawing inferences and making the ultimate evaluative judgment."
Plain English: The doctor can only win the appeal if we are convinced the trial judge got it wrong. If the decision looks reasonable and we can't spot a clear mistake, we should respect the trial judge's advantage in seeing the witnesses.
[273]: "Intermediate appellate courts are not legally bound by their own earlier decisions, but should only depart from such authority or the authority of courts of co-ordinate jurisdiction within the national system if they are of the view that the decision is 'plainly wrong' and, such an error having been identified, there are 'compelling reasons' to depart from the earlier decisions."
Plain English: NSW appeal courts can overrule their own past decisions, but only if they are convinced the old decision was clearly mistaken and there are strong legal and practical reasons to change it.
[321]: "Damages should only be awarded if the Court is satisfied on the balance of probabilities that the breach of duty by the tortfeasor materially contributed to the harm. If the Court is so satisfied, then the tortfeasor will be held liable for the whole of the harm so caused."
Plain English: You only get damages if it is more likely than not that the doctor's mistake actually caused real injury. If that test is met, the doctor pays for all the injury, not just a percentage.
[389]: "For the above reasons and as summarised below, we consider the approach adopted in Rufo and Gavalas to have involved a departure from conventional principles. There were no compelling reasons to adopt that approach and to do so was, in all the circumstances in our respectful view, plainly wrong... Eighthly, there is no evidence that insurance companies or members of the public have adapted their personal or commercial relations in reliance upon these authorities."
Plain English: Rufo was a wrong turn in the law. It broke long-established rules about what counts as injury and how you prove a doctor caused it. There are no good reasons to keep it, and no one has organised their affairs around it, so we are overruling it.
What fact patterns trigger this precedent
Gett v Tabet will be triggered in any medical negligence claim (or indeed any personal injury negligence claim) where the plaintiff seeks to recover damages by characterising the injury as the "loss of a chance" of a better outcome rather than proving on the balance of probabilities that the breach caused actual physical or mental harm. Typical triggers include:
- Delayed diagnosis of tumour, infection or other progressive condition where earlier intervention might have improved the outcome but the evidence does not prove it would probably have done so.
- Failure to order imaging (CT, MRI) or other investigations in the face of ambiguous symptoms, followed by deterioration, where the plaintiff relies on expert evidence that the chance of avoiding the deterioration was, say, 30-49%.
- Cases involving multiple possible causes (tumour plus infection, as here) where the plaintiff cannot disentangle the effect of the breach from the underlying disease process.
- Any attempt to rely on Rufo or Gavalas directly; courts must now treat those decisions as overruled on the loss of chance point.
The precedent does not apply to pure commercial or contractual lost opportunity cases (Sellars territory) or to failure-to-warn cases where the plaintiff can prove on the balance of probabilities that a warning would have led to a different course that would have avoided the injury (Chappel v Hart). It also has no direct application to statutory compensation schemes or where divisible injury can be proven on orthodox causation principles. Fact patterns involving very young children (as here) or where marriage or dependency benefits are claimed will continue to be assessed under the vicissitudes approach rather than as a separate head of loss.
How later courts have treated it
Although the judgment is from 2009, subsequent Australian authority has consistently treated Gett v Tabet as settling the law against recovery for loss of a chance of a better medical outcome. In State of New South Wales v Burton (No 2) [2008] NSWCA 319 (decided shortly after but referring to the then-recent Gett reasons), the Court applied the principle that increased risk is not material contribution. In Dobler v Halverson [2007] NSWCA 335 (which predates Gett but was decided on orthodox principles) the Court had already signalled caution. Post-Gett decisions in other states have followed the "plainly wrong" analysis: see, for example, the Victorian Court of Appeal in RJE v Secretary to the Department of Justice [2008] VSCA 265 (cited with approval in Gett itself on the precedent test) and Western Australian decisions applying the same restraint. Trial judges now routinely strike out or refuse to entertain loss of chance claims in medical cases, directing plaintiffs to prove causation on the balance of probabilities.
The High Court has not yet been required to consider Gett directly, but the reasoning is consistent with the Court's repeated insistence in cases such as Roads and Traffic Authority v Royal [2008] HCA 19 and Amaca Pty Ltd v Booth [2011] HCA 53 that material increase in risk is not equivalent to material contribution. Academic and textbook writers (Luntz, "Assessment of Damages for Personal Injury and Death", 4th ed; Sappideen, "Torts: Commentary and Materials") treat Gett as the current authoritative statement. No subsequent intermediate court has sought to distinguish or limit Gett; rather, it is cited for the proposition that any expansion of liability for increased risk is a matter for the High Court or legislature. The decision has been influential in reinforcing the uniformity of the common law of Australia on causation under the various Civil Liability statutes.
Still-open questions
Several important questions remain after Gett v Tabet. First, the precise boundary between "commercial opportunity" cases (still governed by Sellars) and personal injury cases is not fully mapped. Could a lost chance to pursue a lucrative career that was itself dependent on physical health be characterised as a commercial loss? Second, the interaction with the Civil Liability Acts' provisions on proportionate liability and the statutory definition of "harm" may still generate argument in mixed economic/physical injury claims. Third, the Court left open whether a very high probability (say 49%) of a better outcome might still be treated as too speculative; the judgment suggests any chance short of balance of probabilities is insufficient, but the exact threshold for "real and substantial" (as opposed to speculative) remains for future cases. Fourth, the position in respect of failure-to-warn cases where the plaintiff cannot prove she would have declined the procedure but can prove a lost chance of a better outcome is untouched. Fifth, the quantum of damages in cases where causation is proven only for a divisible part of the injury (as the 25% was characterised here) continues to require careful factual analysis; Gett does not prescribe a methodology beyond orthodox principles. Finally, whether the High Court will one day accept a limited loss of chance doctrine in medical cases—perhaps confined to cases of indivisible injury with clear statistical evidence—remains an open policy question. Until that occurs, Gett represents the law: actual harm, proven causation on the balance of probabilities.