Paul v Royal Wolverhampton NHS Trust

"direct perception of some of the events which go to make up the accident as an entire event, and this includes ... the immediate aftermath ..."

Lastly, as regards the means by which injury is caused to the claimant, Lord Wilberforce noted that communication of the news by a third party had never been regarded as sufficient and commented that "this is surely right". He said, at p 423A:

"The shock must come through sight or hearing of the event or of its immediate aftermath. Whether some equivalent of sight or hearing, eg through simultaneous television, would suffice may have to be considered."

Alcock

These limits on recovery were considered in depth by the House of Lords in Alcock . The claims in Alcock arose out of the disaster in which 95 people were crushed to death and over 400 more sustained injuries as a result of severe overcrowding in the Hillsborough football stadium. The Chief Constable of South Yorkshire, who was responsible for the policing of the match, admitted liability in negligence for the deaths and physical injuries. Of the ten claimants whose appeals were heard by the House of Lords, all had relatives who were killed in the disaster. Two of the claimants had been present at the ground; the others either saw the disaster unfold on live television or (in two cases) watched recorded television later. At least three of the claimants went to the mortuary to identify the body of their relative.

The House of Lords unanimously held that none of the claimants was entitled to succeed. Four of the five members of the appellate committee separately expressed their reasons for the decision. However, it is the speech of Lord Oliver of Aylmerton to which reference has most often been made in later cases and which was the main focus of the submissions made about Alcock on these appeals.

Lord Oliver's speech is the genesis of a distinction drawn in the case law between "primary" and "secondary" victims. He divided the cases into two broad categories, namely, "those cases in which the injured plaintiff was involved, either mediately or immediately, as a participant, and those in which the plaintiff was no more than the passive and unwilling witness of injury caused to others": [1992] 1 AC 310 , 407. He described claimants in the latter category as "secondary victims" (p 411A) and used the term "primary victim" to refer to the person whose injury is witnessed by the claimant in such a case (p 410A). Lord Keith of Kinkel drew a similar distinction when he described the injuries suffered by the claimants in Alcock as "a secondary sort of injury brought about by the infliction of physical injury, or the risk of physical injury, upon another person" (p 396G).

The focus in Alcock was on the requirements which must be satisfied for a claim by a secondary victim to succeed. Lord Oliver began by identifying the common features of all the reported cases in which such claims had previously succeeded (p 411F-H). These were:

"... first, that in each case there was a marital or parental relationship between the plaintiff and the primary victim; secondly, that the injury for which damages were claimed arose from the sudden and unexpected shock to the plaintiff's nervous system; thirdly, that the plaintiff in each case was either personally present at the scene of the accident or was in the more or less immediate vicinity and witnessed the aftermath shortly afterwards; and, fourthly, that the injury suffered arose from witnessing the death of, extreme danger to, or injury and discomfort suffered by the primary victim. Lastly, in each case there was not only an element of physical proximity to the event but a close temporal connection between the event and the plaintiff's perception of it combined with a close relationship of affection between the plaintiff and the primary victim."

Lord Oliver considered that it must be "from these elements that the essential requirement of proximity is to be deduced" (p 411H).

Lord Oliver did not suggest that the elements which he identified should be regarded as fixing rigid lines of demarcation between claims which can or cannot succeed. In particular, he saw no logic or policy reason for requiring a marital or parental relationship between the claimant and the primary victim and for precluding recovery in other cases where a sufficiently close relationship in terms of love and affection is present. But equally he considered that "further pragmatic extensions of the accepted concepts of what constitutes proximity must be approached with the greatest caution" (p 417F).

Of the two claimants who were present at the ground, one lost two brothers in the disaster and the other lost his brother-in-law. Their claims failed because a sufficiently close tie of love and affection with the deceased could not be presumed from their family relationship and had not been proved by evidence. Claimants who had seen their son's body in the mortuary after the disaster for the purpose of identification were held not to come within the scope of the "aftermath" of the disaster. Claims based on seeing the disaster on television failed because the televised images of the unfolding tragedy (which did not depict the suffering of recognisable individuals) could not be treated as equivalent to being present at the stadium.

Frost

The third of this trilogy of cases, Frost , also arose out of the Hillsborough disaster. The issue was whether police officers who were present at the stadium and who suffered psychiatric illness could recover compensation. It was common ground that they could not do so if their claims had to satisfy the requirements for claims by secondary victims established by the decision in Alcock .

The members of the House of Lords were substantially agreed about what those requirements were. Lord Steyn stated them, at p 496D-E, as being:

"(i) that [the claimant] had a close tie of love and affection with the person killed, injured or imperilled; (ii) that he was close to the incident in time and space; (iii) that he directly perceived the incident rather than, for example, hearing about it from a third person."

Lord Hoffmann summarised the requirements, at p 502G-H, in very similar terms as follows:

"(1) The plaintiff must have close ties of love and affection with the victim. Such ties may be presumed in some cases (e.g. spouses, parent and child) but must otherwise be established by evidence. (2) The plaintiff must have been present at the accident or its immediate aftermath. (3) The psychiatric injury must have been caused by direct perception of the accident or its immediate aftermath and not upon hearing about it from someone else."

Lord Goff of Chieveley gave a similar summary of the requirements (p 472E). Lord Griffiths adopted Lord Hoffmann's summary (p 462G-H), and Lord Browne-Wilkinson agreed with the speeches of both Lord Steyn and Lord Hoffmann (p 462B-C).

None of the claimants in Frost (who were present at the stadium in their capacity as police officers) satisfied these requirements as none of them had a close tie of love and affection with any of those killed or physically injured. They argued, however, that the Alcock requirements did not apply to them, either because the Chief Constable was in breach of a duty of care owed to them as their employer or because they were not bystanders or spectators but rescuers involved as participants in the disaster and, as such, fell within Lord Oliver's description of primary victims. A majority of the House of Lords rejected both arguments for reasons given by Lord Steyn and Lord Hoffmann, with which Lord Browne-Wilkinson agreed. They held, in particular, that the category of primary victims is limited to persons exposed (or who perceived themselves to be exposed) to physical danger and for that reason did not include the claimants.

Lord Steyn referred to policy considerations. He identified what he saw as four distinctive features of claims for psychiatric harm which in combination may account for treating them differently from claims for physical injury (pp 493F-494E). In summary these were: (1) the complexity of drawing the line between acute grief and psychiatric harm; (2) concern that greater availability of compensation and consequent litigation would act as an unconscious disincentive to recovery; (3) concern that relaxing the restrictions on recovery would greatly increase the class of persons who can recover damages in tort; and (4) concern to avoid imposing a burden of liability on defendants which is disproportionate to their fault.

With the exception of the last point, which was also made by Lord Wilberforce in McLoughlin , we would not ourselves give significant weight to these considerations. We in any case do not accept the premise that, aside from the requirement to show a medically recognised psychiatric illness, there are different rules for the recovery of compensation for psychiatric harm and physical injury. Lord Steyn referred to the rule that "bystanders at tragic events, even if they suffer foreseeable psychiatric harm, are not entitled to recover damages" (p 493A). As we have indicated, we do not consider that there is such a special rule restricting compensation for psychiatric harm. Rather, the inability of bystanders to recover damages even where they suffer foreseeable harm (of any kind) is a consequence of the general rule that the law does not grant remedies for the effects - whether psychological, physical or financial - of the death or injury of another person.

A point of general importance, which was critical to the decision in Frost , is the need in defining the limits on the recovery of damages by secondary victims to avoid distinctions which would offend most people's sense of justice. In McLoughlin that concern had persuaded the House of Lords to extend the class of eligible claimants to close relatives who did not see or hear an accident but came upon its aftermath. In Frost such reasoning operated in the opposite direction. The majority was unwilling to uphold the claims of the police officers when the claims of bereaved relatives had been rejected in Alcock : see [1999] 2 AC 455 , 499H (Lord Steyn) and 510E-F (Lord Hoffmann). Lord Hoffmann said that such differential treatment would be "unacceptable to the ordinary person" because:

"[such a person] would think it unfair between one class of claimants and another, at best not treating like cases alike and, at worst, favouring the less deserving against the more deserving. He would think it wrong that policemen, even as part of a general class of persons who rendered assistance, should have the right to compensation for psychiatric injury ... while the bereaved relatives are sent away with nothing."

Matters not in issue on these appeals

In case it was necessary to decide whether to depart from any of these decisions of the House of Lords, in particular Alcock , these appeals have been heard by a panel of seven Justices. But in the event none of the parties has urged us to depart from any earlier decision reached at this appellate level. The claimants have based their submissions squarely on the existing case law. They argue that allowing recovery in the present cases either follows from or is at any rate consistent with what has previously been decided by the House of Lords.

A feature of the law as it has evolved is the distinction between "primary" and "secondary" victims: see para 40 above. Precisely how the distinction is or should be drawn, if at all, potentially raises difficult questions: see eg the Law Commission Report on Liability for Psychiatric Illness (1998) (Law Com No 249) [1998] EWLC 249 , paras 5.52, 5.54; Harvey Teff, "Liability for negligently inflicted psychiatric harm: justifications and boundaries" (1998) 57 CLJ 91 ; Chris Hilson, "Liability for psychiatric injury: primary and secondary victims revisited" (2002) 18 PN 167 . In W v Essex County Council [2001] 2 AC 592 , 601, the House of Lords expressed the view that the concept is still to be developed in different factual situations. But it is not an issue on these appeals. It is common ground that the claimants are to be classified as "secondary victims". The essential point is that the harm for which they are claiming compensation in each case is harm brought about indirectly by injury caused to another person.

Relevance of McLoughlin, Alcock and Frost to non-accident cases

McLoughlin , Alcock and Frost , like the earlier authorities discussed in those cases, all involved injuries to the primary victim sustained as a result of an accident. In McLoughlin the relevant event was a road accident, which is perhaps the paradigm. In Alcock and Frost the event was on a scale so large that it is more naturally described as a "disaster", but nothing turns on that linguistic difference. In each case the event was not "accidental" in the sense that no one was to blame for it since it was caused by the defendant's negligence; but it was an "accident" in the sense already mentioned that it was an unexpected and unintended event which caused injury (or a risk of injury) by violent external means to one or more primary victims.

That is not the usual situation in medical negligence cases such as those under appeal. In these cases, the event (or its aftermath) witnessed by the secondary victim is generally not an accident; it is the suffering or death of their relative from illness. As a shorthand and without intending it to be a term of art, we will refer to such an event as a "medical crisis". The question raised by these appeals is whether witnessing a negligently caused medical crisis (or its aftermath) can in principle found a claim for damages by a secondary victim or whether such a claim can lie only where the triggering event is an accident in the sense we have described.

Counsel for the claimants have argued that this question is answered in their favour by the decision in Alcock . They highlight passages in the speeches which refer to liability arising where the claimant witnessed "the event of injury" to the primary victim (p 410H), or "the injury" suffered by the primary victim (p 411G), or simply "the event" (p 411G). They submit that this language is wide enough to encompass cases where the event is a medical crisis. We are unimpressed by this argument. For every passage in which expressions such as "the injury" or "the event" are used, another can be found referring to "the accident". For example, in the passage quoted from Lord Oliver's speech at para 41 above the third element identified is that the claimant "was either personally present at the scene of the accident or was in the more or less immediate vicinity and witnessed the aftermath shortly afterwards". The same ambiguity can be seen in Frost in the passages quoted at para 45 above, where in summarising the Alcock requirements Lord Hoffmann referred to the "accident" whereas Lord Steyn used the potentially broader term "incident". The plain fact is that the question whether damages can in principle be recovered in a case where there is no accident did not arise in Alcock (or Frost ) and was not considered by the House of Lords. In these circumstances it is fallacious to fasten selectively on particular forms of words used in various passages of the speeches and then deploy those quotations out of context in support of an argument which was not in the contemplation of the law lords and to which their reasoning was not addressed.

A similar point can be made about a passage in Lord Wilberforce's speech in McLoughlin on which the claimants also seek to rely. Lord Wilberforce said, at p 421A-B:

"We must then consider the policy arguments. In doing so we must bear in mind that cases of 'nervous shock,' and the possibility of claiming damages for it, are not necessarily confined to those arising out of accidents on public roads. To state, therefore, a rule that recoverable damages must be confined to persons on or near the highway is to state not a principle in itself, but only an example of a more general rule that recoverable damages must be confined to those within sight and sound of an event caused by negligence or, at least, to those in close, or very close, proximity to such a situation."

We do not read this passage as asserting that any kind of event caused by negligence can give rise to a claim for damages. The point made by Lord Wilberforce was only that claims in which damages are recoverable "are not necessarily confined to those arising out of accidents on public roads". That does not preclude a requirement that there must be an accident which need not be a road accident but might, for example, be a rail accident or an industrial accident (a type of event specifically mentioned by Lord Wilberforce at p 421D). All the cases discussed by Lord Wilberforce were cases that involved accidents and, once again, the question whether or when damages are recoverable in cases arising out of any other type of event was not in issue, was not the subject of any argument and was not addressed by Lord Wilberforce or any of the other law lords.

We would equally reject the opposite argument made on behalf of the defendants that the use of the word "accident" and the focus on accidents in these cases must mean that only an accident can be a qualifying event capable of giving rise to a claim for damages by a secondary victim. The defendants have submitted, for example, that, of the elements identified by Lord Oliver in Alcock in the passage quoted at para 41 above, the third (presence at the scene of the accident or its aftermath) would be otiose and subsumed by the fourth (direct perception of the death of or injury to the primary victim) if no accident was required and a medical crisis could be a qualifying event. We have already noted, however, that Lord Oliver's list was a list of features found in the previous cases and was not intended to set the law in stone. Lord Oliver made it clear that he was "not dissenting from the case-by-case approach advocated by Lord Bridge" in McLoughlin (p 418C) but was seeking to steer a middle way between the two "extreme positions" of drawing rigid lines based on policy considerations and limiting recovery by reference to reasonable foreseeability of damage alone (pp 413H-415C).

We do not agree with Lord Hoffmann's lament in Frost , at p 511B, that "in this area of the law, the search for principle was called off in Alcock ". Alcock , like the speech of Lord Wilberforce in McLoughlin , emphasised the need to set limits on the recovery of damages by secondary victims based on the concept of proximity and recognised that these limits are influenced by practical and policy considerations rather than purely analogical development of the law. But the common law strives for coherence and whenever a question arises as to whether or how what was decided in an earlier case should be applied in a different factual situation, it is always necessary to exercise judgment about whether the factual differences should be regarded as legally significant. Such a question arises here as to whether or how what was decided in McLoughlin , Alcock and Frost should be applied in cases where the event witnessed by the claimant is not an accident.

In the period of some 30 years since Alcock was decided by the House of Lords a number of claims made by secondary victims in medical negligence cases have come before the courts. But although there has been some discussion of whether damages can in principle be recovered in such cases in the absence of an accident caused by the defendant's negligence, the question has never been examined in any depth or authoritatively decided.

Taylor v Somerset

The question was raised in a case decided shortly after Alcock . In Taylor v Somerset Health Authority [1993] PIQR P262 the claimant's husband suffered a heart attack at work and died after being taken to the defendant's hospital. The claimant went to the hospital within the hour and was told of her husband's death. She later went to the mortuary and saw his body. The defendant admitted that Mr Taylor's death was caused by its clinical negligence in failing, many months earlier, to diagnose or treat his serious heart disease. It was also accepted that Mrs Taylor had developed a recognised psychiatric illness and that there was a causal link between what she witnessed and her illness.

This was a case, therefore, where the relevant event was a medical crisis. Auld J held that the claim failed for two reasons. The first was that "the test required some external, traumatic, event in the nature of an accident or violent happening" (p 267), and there had not been such an event. Rather, Mr Taylor's death was the culmination of the natural process of heart disease. The second reason was that, even if her husband's death could be regarded as a qualifying event, the doctor's communication of it to the claimant at the hospital and her subsequent sight of her husband's body in the mortuary did not come within the "aftermath" extension recognised in McLoughlin as an exception to the general rule that the claimant must have perceived the event as it happened. Although Mrs Taylor's visit to the mortuary had occurred within an hour of Mr Taylor's death, the judge found that its purpose was principally to settle the claimant's disbelief in the fact of his death and that his body "bore no marks or signs to her of the sort that would have conjured up for her the circumstances of his fatal attack" (p 268). Mrs Taylor was therefore in a different position from the claimant in McLoughlin who came upon her injured and distressed husband and children in very much the same condition as they were at the scene of the accident.

Sion

The first of the reasons given by Auld J for rejecting the claim in Taylor v Somerset - that "some external, traumatic, event in the nature of an accident" was required - was doubted by Peter Gibson LJ in Sion v Hampstead Health Authority [1994] 5 Med LR 170 (" Sion "). The claimant in Sion was the father of a young man who was fatally injured in a motor-cycle accident. The defendant to the claim was not the person responsible for that accident, however, but the hospital which treated the victim for his injuries. It was alleged - and assumed for the purpose of an application to strike out the claim - that the hospital had negligently failed to diagnose substantial and continuing bleeding from the young man's left kidney and that this had resulted in his death. The claimant, Mr Sion, sat by his son's bedside for 14 days watching his gradual deterioration until he fell into a coma and died.

One issue was whether, for the claim to succeed, the claimant needed to prove that he had experienced a sudden shock. The Court of Appeal held that there was such a requirement and that, as the facts alleged did not disclose such a shock, the claim was doomed to fail. Peter Gibson LJ, however, also commented (obiter) on a submission based on Auld J's judgment in Taylor v Somerset that Mr Sion's claim could not succeed because the alleged breach of duty on which the claim was based did not involve a sudden and violent incident. He said, at p 176:

"I see no reason in logic why a breach of duty causing an incident involving no violence or suddenness, such as where the wrong medicine is negligently given to a hospital patient, could not lead to a claim for damages for nervous shock, for example where the negligence has fatal results and a visiting close relative, wholly unprepared for what has occurred, finds the body and thereby sustains a sudden and unexpected shock to the nervous system."

Walters

A case strongly relied on by the present claimants and appellants is North Glamorgan NHS Trust v Walters [2002] EWCA Civ 1792 , [2003] PIQR P16 (" Walters "). Mrs Walters' infant son died following two days of illness which she witnessed. The claimant was sleeping in the same room as her son in the defendant's hospital when he suffered a major epileptic seizure leading to coma and irreparable brain damage. Appropriate treatment was delayed by a misdiagnosis. The baby was eventually transferred to a London hospital; but nothing could be done to save him and, approximately 36 hours after the seizure, he died in his mother's arms once the life support machine was switched off.

The defendant admitted that the baby's death was a result of its negligence. It was further agreed that the claimant had developed a recognised psychiatric illness. The trial judge held that the claimant was a secondary victim, and this finding was not challenged on appeal. The judge identified the essential issue as being whether what happened to cause the claimant's illness constituted "a sudden appreciation by sight or sound of a horrifying event" as opposed to "an accumulation over a period of time of more gradual assaults on the nervous system". The judge held that it did, as the entire 36-hour period could be regarded as one horrifying event for this purpose and the claimant's appreciation of the event was sudden within that temporal context. This decision was affirmed by the Court of Appeal.

In explaining his reasons for this conclusion, Ward LJ, with whom Clarke LJ and Sir Anthony Evans agreed, used phrases which have been relied on in later cases. He said, at para 35:

"In my judgment on the facts of this case there was an inexorable progression from the moment when the fit occurred as a result of the failure of the hospital properly to diagnose and then to treat the baby, the fit causing the brain damage which shortly thereafter made termination of this child's life inevitable and the dreadful climax when the child died in her arms. It is a seamless tale with an obvious beginning and an equally obvious end . It was played out over a period of 36 hours, which for her both at the time and as subsequently recollected was undoubtedly one drawn-out experience." (emphasis added)

Having concluded by this reasoning that the claimant's experience over a period of 36 hours was to be regarded as "one entire event", the judgment then sought to explain how the claimant's appreciation of this drawn-out event could be characterised as "sudden". The explanation given was that an "entire event" can be made up of "one or more discrete events" (para 34) and that each of the three events which the claimant experienced were sudden and unexpected and "had their impact there and then" (para 42).

Shorter

In Shorter v Surrey and Sussex Healthcare NHS Trust [2015] EWHC 614 (QB) , (2015) 144 BMLR 136 (" Shorter ") the claimant's sister, Mrs Sharma, died in hospital from a subarachnoid haemorrhage caused by an aneurysm in a cerebral artery. She had negligently been released from hospital earlier because her brain scans were not properly assessed. The claimant's pleaded case alleged that, starting with the news of the serious deterioration of her sister's condition, she had experienced a "seamless single horrendous event" which ended with her sister's death a day later and caused her to suffer from a psychiatric illness (para 8). The judge rejected that characterisation of the facts, finding that there had been no "seamless single horrifying event" similar to that experienced by the claimant in Walters but rather a series of events over a period of time. Only some of the individual events involved the claimant actually witnessing her sister's condition and none of those could be regarded as an event which would be recognised as "horrifying" by a person of ordinary susceptibility, nor as sudden or unexpected (para 218). The claim was therefore dismissed.

Ronayne

In Liverpool Women's Hospital NHS Foundation Trust v Ronayne [2015] EWCA Civ 588 , [2015] PIQR P20 (" Ronayne ") the claimant's wife underwent a hysterectomy and was readmitted to hospital a few days later with complications including septicaemia and peritonitis caused by the defendant's negligence. Shortly before she underwent emergency surgery, Mr Ronayne saw his wife connected to various machines including drips and monitors. After surgery he saw her unconscious, connected to a ventilator and with antibiotics being administered intravenously. Her arms, legs and face were very swollen. She remained in intensive care in hospital for some nine weeks but eventually made a complete recovery. Mr Ronayne's claim for damages for psychiatric illness caused by his experience succeeded at trial, but that decision was overturned on appeal.

Tomlinson LJ who gave the reasons of the Court of Appeal recorded, at para 8, that the appeal had concentrated on two interrelated points: (a) whether the events concerned were "in the necessary sense 'horrifying'"; and (b) whether the sudden appreciation of the events caused the claimant's psychiatric illness. He held, at paras 38-41, that there was no sudden appreciation of an event because the judge had been wrong to characterise what happened as one event. There was no "seamless tale" or "inexorable progression" as there had been in Walters . Rather, there was a series of events giving rise to an accumulation of gradual assaults on the claimant's mind as he came to realise that his wife's life was in danger. At each stage, Mr Ronayne was conditioned for what he was about to perceive. There was nothing sudden or unexpected about being ushered in to see his wife and finding her connected to medical equipment before or after the operation. Nor were the events horrifying by objective standards, as they were not exceptional in nature.

Must the claimant experience a sudden shock?

Sion , Walters , Shorter and Ronayne were not cases in which there had been an external, traumatic, event in the nature of an accident caused by the defendant's negligence. In none of these cases, however, did the court decide the question whether in principle the rules developed in accident cases ought to be applied. In the latter three cases this question was not even raised or mentioned: it was simply assumed that the same rules applied. Instead, the judgments in all these cases focused on whether it could be said that the claimant had suffered psychiatric illness because of a "sudden shock" or a "sudden appreciation of a horrifying event".

Those phrases reflect language used in Alcock . Lord Oliver, at p 411F, identified as one of the features of all the reported cases that the injury "arose from the sudden and unexpected shock to the plaintiff's nervous system". Lord Ackner said, at p 401F, that:

"'Shock', in the context of this cause of action, involves the sudden appreciation by sight or sound of a horrifying event, which violently agitates the mind. It has yet to include psychiatric illness caused by the accumulation over a period of time of more gradual assaults on the nervous system."

Those observations were no doubt true as descriptions of how historically the causation of psychiatric illness in accident cases was understood. The very term "nervous shock", which was still commonly used in court proceedings when Alcock was decided, embodies such an understanding. The crude mechanical model which attributes psychiatric illness in such cases to an "assault on the nervous system" has, however, long since been discredited: see eg the criticism made in the Law Commission Report, para 5.29.

The remarks of Lord Ackner and Lord Oliver quoted above were not necessary to the decision of the House of Lords in Alcock and we do not consider that those dicta establish an additional restriction on the recovery of damages by secondary victims nor that the law is tied to an outdated theory of the aetiology of psychiatric illness. The requirements established by the decision in Alcock were, in our view, accurately and authoritatively summarised in Frost (see para 45 above). They do not include a requirement that the claimant's psychiatric injury must have been caused by a "sudden shock to the nervous system". None of the law lords in Frost endorsed such a requirement and Lord Goff expressly stated, at p 489E-F, that "the nature of PTSD illustrates very clearly the need to abandon the requirement of nervous shock in these cases, and to concentrate on the requirement that the plaintiff should have suffered from a recognised psychiatric illness".

With regard to causation, it is sufficient for a claimant who was present at the scene of the accident (or its immediate aftermath) in which a loved one was killed, injured or imperilled to show that there is a causal connection between witnessing that event and the illness suffered. It is not necessary (even were it possible) to demonstrate the neurological or psychological mechanism by which the illness was induced.

Must there be a "horrifying event"?

Lord Ackner's description of "shock" in Alcock (quoted at para 72 above) has also given rise to a notion that, to succeed in a claim as a secondary victim, it is necessary to prove that the event perceived was "horrifying". In Shorter and Ronayne Lord Ackner's use of this word was elevated to the status of a formal requirement and treated as requiring the court to decide whether the relevant event was "horrifying by objective standards". It is of course necessary for a claimant to show that it was reasonably foreseeable that the defendant's negligence might cause her injury. If, for example, a claimant with a history of psychiatric illness develops such an illness after witnessing a minor accident in which his wife sustains some cuts and bruises, his claim might fail that test. But we can see no justification for super-imposing an additional, separate requirement that the event witnessed by the claimant was "horrifying".

Although said to involve an "objective standard", such a test is in truth unavoidably subjective. There is no available Richter scale of horror. The test pushes judges to compare the facts of the case before them with the facts of other cases and engage in an exercise of deciding whether or not the facts are similarly "horrifying". As Tomlinson LJ put it in Ronayne , at para 15, the judge's task is to "allocate to [the] case its appropriate place on the spectrum between circumstances which attract compensation and those which do not".

A judge adopting this approach in the present cases would therefore have to ask the question: is Mrs Purchase's experience of coming upon her daughter's still warm but dead body at home and listening to her dying breaths in a voicemail message as horrifying as the experience of Mrs Walters waking up to find her baby having an epileptic fit and choking up blood? Or if the claimant seeing her sister lying in intensive care on the verge of death could not be regarded as a "horrifying event", as the judge found in Shorter , how does that experience rank on the spectrum of horror with that of Saffron and Mya Paul seeing their father fall to the ground after suffering a fatal heart attack in the street? Such questions are invidious and not susceptible to any proper answer.

Again, this test is not one of the requirements established by Alcock and confirmed in Frost . In engrafting onto those requirements additional requirements of needing to prove that the claimant's injury was caused by the mechanism of a "sudden shock to the nervous system" and was a sufficiently "horrifying event", the law has in our opinion taken an unfortunate wrong turn which these appeals enable us to correct.

What counts as one event?

Another unsatisfactory development has been the emergence of a legal test of whether what the claimant witnessed should be regarded as one event or several separate events. This was viewed as the central issue in Walters , where the Court of Appeal managed to conclude that the whole period of 36 hours from the time when Mr Walters' baby suffered an epileptic seizure to the time of his death was to be regarded as one event, although at the same time - supposedly without contradiction - this period could also be regarded as comprising a number of separate events some of which were sudden and unexpected. In the absence of any coherent test of what counts as one event, parties and judges in later cases have adopted phrases used by Ward LJ in Walters , at para 35, when he described what happened in that case from the moment when the epileptic fit occurred to when the baby died as "an inexorable progression" and "a seamless tale with an obvious beginning and an equally obvious end". In Shorter (paras 8 and 218) and Ronayne (paras 35 and 36) these phrases were deployed as if they amounted to a legal test.

The phrases are vague and are not of much help. The phrase "inexorable progression" suggests that something had happened which made death inevitable (at least absent an early medical intervention). The reference to a "seamless tale" seems to reflect Mrs Walters' subjective experience "both at the time and as subsequently recollected" (see the full passage quoted at para 66 above). Implicit in both phrases is a suggestion that what happened should be regarded as one event because it followed a dramatic arc which makes for a compelling story. Tomlinson LJ spelt this out in Ronayne , at para 35, when he said of the facts of Walters :

"The working out of the tragedy, with the raising of hopes, the journey up the motorway to London following in the wake of the ambulance, and the dashing of hopes and then their final destruction was almost Sophoclean in its seamlessness."

We find it hard to see why the defendant's legal liability should turn on the court's impression of whether or not the facts of the case fit the dramatic pattern of a Greek tragedy.

The medical negligence cases treating the traumatic scenes witnessed over several days as themselves comprising an event founding liability appear also to have led to an extension of the aftermath of true accidents to a period far beyond what was contemplated by Lord Wilberforce in McLoughlin . It has created the separate problem which is not simply identifying when the aftermath finishes but also when the triggering event stops and its "aftermath" starts. Two cases illustrate the development, both of them road accident rather than medical negligence cases. In Taylorson v Shieldness Produce Ltd [1994] PIQR P329, the parents of a 14-year old boy who suffered serious head injuries in a road accident went to the hospital where he was being treated. The parents saw their son being rushed on a trolley into surgery and then sat by his bedside until the life support machine was switched off almost three days after the accident. The defendant was the employer of the driver whose negligence caused the accident. The Court of Appeal upheld the dismissal of the parents' claims for damages for psychiatric illness on the grounds that they did not witness the accident or its immediate aftermath and also that their illnesses could not be attributed to one shocking event but "grew out of a whole sequence of events extending over an appreciable period of time" (P335).

By contrast in Galli-Atkinson v Seghal [2003] EWCA Civ 697 , [2003] Lloyd's Rep Med 285 (" Galli-Atkinson "), decided a few months after Walters, the claimant's daughter was a pedestrian killed by the defendant's negligent driving. About an hour later the claimant learnt that there had been an accident and went to the scene. The police did not allow her through the highway cordon so she did not see the crash site but she was told that her daughter was dead. Subsequently she went to the mortuary and saw her daughter's body. Her claim against the defendant driver for damages for psychiatric injury was rejected by the judge on the ground that the claimant had not witnessed the road accident or its immediate aftermath. However, the claim succeeded on appeal. Relying on Walters , Latham LJ who gave the lead judgment held, at para 26, that the "immediate aftermath" extended from the moment of the accident until the moment the claimant left the mortuary. The trial judge had erred because he "artificially separated out the mortuary visit from what was an uninterrupted sequence of events." In Galli-Atkinson , therefore, the theory derived from Walters that "an uninterrupted sequence of events" is capable of being characterised as one entire event was relied on to extend the concept of the "aftermath" to include the claimant's visit to see her daughter's body in the mortuary more than two hours after the accident in which her daughter was killed. The Court of Appeal sought to distinguish Alcock , where mortuary visits by relatives within hours of the Hillsborough disaster were held by the House of Lords not to fall within the scope of the aftermath, on the basis that the visit in Galli-Atkinson was made "not merely to identify the body" but "to complete the story" so far as the claimant was concerned. We cannot regard the claimant's perceived motivation for seeing her daughter's body as a satisfactory criterion for determining the defendant's liability.

Although not a medical negligence case, the decision of the Court of Appeal in Taylor v A Novo (UK) Ltd [2013] EWCA Civ 194 , [2014] QB 150 (" Novo ") has featured prominently in the reasoning of the courts below and in the arguments on these appeals. The case therefore requires close scrutiny. The defendants rely on the decision, while the claimants invite us to distinguish or overrule it.

The facts and reasoning in Novo

The facts were that the claimant's mother, Mrs Taylor, sustained injuries in an accident at work when a stack of racking boards fell on top of her. The defendant, her employer, admitted liability in negligence for the accident. Mrs Taylor was apparently recovering well when three weeks later she unexpectedly collapsed and died at home. Her death was caused by a pulmonary embolism resulting from a deep vein thrombosis which was itself due to injuries sustained in the accident. The claimant did not witness the accident, but she witnessed her mother's death and as a result developed post-traumatic stress disorder. The defendant accepted that the claimant was a secondary victim who met all but one of the requirements to succeed as such. The sole defence was that she was not present at the scene of the accident or its immediate aftermath. The claim succeeded at trial. The judge held that the requirement of physical proximity was satisfied because Mrs Taylor's collapse and death was a qualifying event at which her daughter was present.

This decision was reversed by the Court of Appeal. Lord Dyson MR, with whom Moore-Bick and Kitchin LJJ agreed, said that the use of the word "event" had a tendency to distract. There had been one accident, the falling of the stack of racking boards, with two consequences: the initial injuries to Mrs Taylor and her death three weeks later. To allow the claimant to recover as a secondary victim when she had not been in physical proximity to her mother at the time of the accident would be to go too far (para 29). This was for two reasons. The first was that this would mean that the claimant could have recovered damages even if her mother's death had occurred months or possibly years after the accident. By contrast, if Mrs Taylor had died at the time of the accident and the claimant had suffered psychiatric illness as a result of coming on the scene shortly after what constitutes the "immediate aftermath", damages could not have been recovered. The idea that the claimant could recover in the first of these situations but not in the second "would strike the ordinary reasonable person as unreasonable and indeed incomprehensible" (para 30).

The second reason was closely connected with the first, namely that to allow recovery would extend the scope of liability to secondary victims considerably further than in previous cases and policy reasons articulated by the House of Lords in Frost militated against any such extension.

As regards earlier authorities, Lord Dyson MR agreed with the decision of Auld J in Taylor v Somerset that the kind of case in which a claimant can recover damages as a secondary victim is one involving an accident which (i) more or less immediately causes injury or death to a primary victim and (ii) is witnessed by the claimant. In such a case the relevant event is the accident. It is not a later consequence of the accident. Lord Dyson discounted the observations of Peter Gibson LJ in Sion as obiter dicta which were therefore not binding. And he distinguished Walters on the ground that the only question in that case was whether there was a single event and the question whether the death, when held to be a separate event from the sustaining of the injuries, was a relevant event for the purposes of a claim by a secondary victim had not been addressed.

The reasoning of the courts below

In his clearly reasoned judgment in Paul , Chamberlain J analysed the ratio of Novo as being that, in a case where the defendant's negligence results in an "event" giving rise to injury in a primary victim, a secondary victim can claim for psychiatric injury "only where it is caused by witnessing that event rather than any subsequent, discrete event which is the consequence of it, however sudden or shocking that subsequent event may be": [2020] EWHC 1415 (QB) , [2020] PIQR P398 , para 73 (emphasis in original). In Paul , there was only one pleaded event, namely, Mr Paul's collapse and death from a heart attack, at which his daughters had been present. Unlike in Novo , there had been no previous "event" since the negligent failure to diagnose Mr Paul's heart condition could not be described as an event akin to the racking boards falling onto Mrs Taylor. Chamberlain J concluded that Novo could be distinguished on this basis and was not a bar to recovery.

The Court of Appeal did not agree with this analysis. In their view Novo decided that a secondary victim cannot claim for psychiatric injury caused by witnessing a "horrific event" involving injury to the primary victim resulting from the defendant's negligence if the horrific event is a separate event removed in time from the defendant's negligence: [2022] EWCA Civ 12 , [2023] QB 149 , paras 12 and 96 (Sir Geoffrey Vos MR) and para 104 (Underhill LJ). In each of the present cases the horrific event witnessed by the claimant (the death of the primary victim or its immediate aftermath) occurred an appreciable time after the omissions which constituted the defendant's negligence. The Court of Appeal therefore concluded that, as it is bound by its own previous decisions, it was bound by the decision in Novo to hold that the present claims cannot succeed.

What Novo decided

This court is not so bound, but we agree with the Court of Appeal that the present claims cannot succeed unless we conclude that Novo was wrongly decided. We disagree, however, both with their interpretation and with that of Chamberlain J of what was decided in Novo . In our view, the analyses of both courts below share the common flaw that they treat Novo simply as a case in which there were two events, separated in time, in which injuries caused by the defendant's negligence occurred or became manifest, and view the fact that there was an accident as if it were an incidental feature of the facts, not material to the decision. This ignores the insistence in the judgment of Lord Dyson MR that what mattered was not the number of "events" but the fact that there had been an accident. It also ignores his express endorsement of Auld J's reasoning in Taylor v Somerset which identified as necessary conditions for the recovery of damages: (i) an external, traumatic, event which immediately causes injury or death to a primary victim; and (ii) direct perception of the event (or its immediate aftermath) by the claimant. The reason why the claim in Novo failed was that, although there was an external, traumatic, event (ie "an accident") which immediately caused injury to Mrs Taylor, the claimant did not witness that event and the event which she did witness and which caused her psychiatric illness was not an accident. The proximity (or lack of it) of the claimant to an accident was therefore critical to the court's reasoning.

In disagreement with the Court of Appeal, we do not read the judgment in Novo as suggesting that the length of time between the defendant's negligent act or omission and the event witnessed by the claimant and which caused her psychiatric injury was a relevant factor. Lord Dyson MR did not doubt that, if the claimant had been in physical proximity to her mother at the time of the accident and had suffered psychiatric illness as a result of seeing the accident and the injuries sustained by her mother, she would have qualified as a secondary victim on established principles: see [2014] QB 150 , paras 29 and 32. But, as Chamberlain J pointed out, Lord Dyson said nothing to suggest that the position would have been any different if the accident in which the stack of boards fell over onto Mrs Taylor had been caused by negligent stacking weeks or months before the accident occurred.

We agree with the Court of Appeal that Novo is authority for the proposition that no claim can be brought in respect of psychiatric injury caused by a separate event removed in time from the accident. But we do not agree with the suggestion in paras 12 and 96 of the judgment that Novo decided anything about distance in time between the event which caused psychiatric injury and the original negligence.

More important, however, is that neither of the two alternative analyses canvassed by the courts below provides an acceptable justification for the outcome in that case, as they themselves recognised in their analysis of Novo .

Must the event be close in time to the negligent act or omission?

Although they considered themselves bound by Novo to apply such a test, the Court of Appeal could see no good reason why the gap in time (short or long) between the negligence and the horrific event caused by it should affect the defendant's liability. Nor can we. Sir Geoffrey Vos MR postulated a case of a negligent architect who designs a door in a load-bearing wall without specifying an RSJ, causing masonry to fall on a primary victim's head years later (paras 79-80). These facts are similar to those of the actual case of Clay v AJ Crump & Sons Ltd [1964] 1 QB 533, where an architect who was responsible for the safety of a building site negligently left a wall standing when a building was demolished. The architect was held liable to compensate a person working on the site who was injured when over two months later the wall collapsed. In agreement with the Court of Appeal, we see no reason why, in a case of this kind, the gap in time between the negligence and the accident should prevent a claim by a secondary victim when it does not prevent a claim by a primary victim. If, for example, a mother who was present and saw masonry fall on her child's head suffered psychiatric injury, her ability to make a claim cannot rationally depend on the length of time between the negligence and the accident.

Typically in accident cases, the accident and the defendant's negligent act or omission which caused the accident occur at much the same time. That is almost inevitably so in cases such as McLoughlin involving road accidents. We agree, however, with Chamberlain J that there is nothing in any of the House of Lords authorities to suggest that the right to recover damages for personal injury caused by witnessing a person's death or injury in an accident is affected by the length of time between the negligent act or omission and the accident. The requirements established by the decision in Alcock include closeness in space and time to, and direct perception of, the accident (or "the event caused by the defendant's breach of duty to the primary victim", per Lord Oliver at p 416E). They do not include any requirement of closeness in space and time to the defendant's breach of duty. There is no suggestion in Alcock and Frost that the timing of the negligent acts or omissions was a relevant consideration in those cases. In Frost Lord Goff mentioned that the immediate cause of the Hillsborough disaster was the decision of a senior police officer to open an outer gate to the stadium without cutting off the crowd's access to two pens in which crushing then occurred: [1999] 2 AC 455 , 465H-466C. But nothing was said by any of the law lords (or the lower courts) to suggest that the claims of either relatives or police officers would be affected if the operative negligence lay in decisions on crowd control or police deployment taken in the days before the match.

Although in the present cases the defendants' stance on this point appears to have fluctuated (compare para 63 of Chamberlain J's judgment with para 7 of the judgment of the Court of Appeal), in his oral submissions on their behalf in this court Mr Simeon Maskrey KC made it clear that the defendants do not contend that there is any requirement of closeness in time between the defendant's negligence and the accident which caused the claimant psychiatric injury. In our opinion, that concession was rightly made.

Must the event be the first manifestation of damage to the primary victim?

As we have indicated, Chamberlain J explained the decision in Novo on the basis that there had already been an "event" in which Mrs Taylor suffered injury three weeks before she collapsed and died. In his view, the fact that the earlier event was an accident was not legally significant. The position would have been the same if the earlier event had been "internal to the primary victim" (para 75). What mattered was that in Novo the occasion when Mrs Taylor collapsed and died was not the first occasion when damage to her "became manifest" (paras 79-80). By contrast, on the facts alleged in Paul , where Mr Paul's collapse in the presence of his daughters was the "first manifestation" of damage which would have been avoided by proper diagnosis and treatment, there is no earlier event which bars recovery.

In this court counsel for the claimants in Paul have argued that this approach is in principle correct. They submit that there can only be one qualifying event capable of giving rise to a claim for damages by a secondary victim, and this event must involve the first manifestation of the damage which it was the defendant's duty to prevent.

This argument is not supported by the claimants in Polmear and Purchase , who make common cause with the defendants in this respect. It is easy to see why these claimants do not support a "first manifestation of damage" test. It is apparent that both Esmee Polmear and Evelyn Purchase had shown significant symptoms of illness (which on the claimants' case would have been avoided by proper diagnosis and treatment) before the medical crisis occurred on which the claims are founded. Therefore, if the test for which the claimants in Paul contend represents the law, the claims in Polmear and Purchase might well fail.

We do not think it justifiable to differentiate on this ground between the claims made in these three cases. We have already explained why the first manifestation of damage test, although compatible with the result reached in Novo , is inconsistent with the reasoning in that case. There is no precedent for applying such a test in any authority cited to us. Nor can we see good reason to introduce it. We agree with the Court of Appeal that the test "would create unprincipled and complex factual disputes" (Sir Geoffrey Vos MR at para 82) and would be "both unprincipled and unworkable" (Underhill LJ at para 105).

The proposed test would in the first place create a new layer of factual complexity in proceedings by inviting investigation of whether any and, if so, what symptoms were manifested by the primary victim at any time during what might be a long period between misdiagnosis and the event witnessed by the claimant. As counsel for the claimant in Purchase pointed out, it is also unclear what "manifest damage" means for this purpose. Suppose that in the case of Paul Mr Paul had experienced symptoms of angina on an occasion before he collapsed and died. Would this disqualify the claim and, if so, why should it? Would it matter whether anyone else was present when these symptoms were experienced? If so, would it make a difference whether the person present was the claimant or another relative or a stranger? Would it make a difference whether the symptoms were serious or minor and, if so, how serious would they need to be to count as "the first manifestation of damage"?

In their written case counsel for the claimants in Paul sought to address the last of these questions. They submitted that, even if in the intervening period between Mr Paul's admission to hospital in November 2012 when the alleged breaches of duty occurred and his heart attack in January 2014 there had been minor, ongoing symptoms of his underlying coronary artery disease, "these were the ordinary effects of the untreated illness". By contrast, the cardiac arrest in January 2014 "represented the first infliction of the damage which should have been avoided". It is unclear, however, why ordinary effects of the untreated illness are not damage which should have been avoided if the illness had been treated or why the cardiac arrest should not be regarded as such an ordinary effect.

There is no rational answer to any of the questions we have posed because there is no principle which justifies any version of the proposed test. We agree with the Court of Appeal that it is illogical to make the liability of a defendant for injury caused to a secondary victim depend on whether the event witnessed by the claimant was or was not the "first manifestation of damage" to the primary victim.

Should damages be recoverable in the absence of an accident?

Having rejected the two justifications for the result reached in Novo canvassed by the courts below, we must consider whether Novo was correctly decided. In our opinion it was, for the reason given by Lord Dyson MR, namely, that the claim could not succeed because the claimant was not present at the scene of the accident or its immediate aftermath and the event which she witnessed was not an accident.

We think it relevant to note first that the occurrence or manifestation of injury is not part of what defines an accident. An accident is an external event which causes, or has the potential to cause, injury: it is not the injury, if there is one, caused by that event. In the many cases which have involved accidents, the right to claim damages has depended on whether the claimant was present at and directly perceived the accident (or its immediate aftermath). Witnessing injury caused by the accident has not been treated as either necessary or sufficient. It is not sufficient because Alcock and other cases in that line of authority have held that, where the claimant was not present at the scene of the accident (or its immediate aftermath) but saw the injured victim or the body of the victim afterwards, damages cannot be recovered. Nor is witnessing injury caused by the accident necessary because a claim may succeed where the claimant fears for the safety of another person but no injury is in fact suffered by that person. Lord Oliver made this point when he said in Alcock , at p 412A:

"There may, indeed, be no primary 'victim' in fact. It is, for instance, readily conceivable that a parent may suffer injury, whether physical or psychiatric, as a result of witnessing a negligent act which places his or her child in extreme jeopardy but from which, in the event, the child escapes unharmed."

Since witnessing injury sustained by another person is neither a necessary nor sufficient condition for a claim as a secondary victim in an accident case, no ready or obvious analogy can be drawn from such cases to cases where the claimant witnesses injury that has not been caused by any external accident.

Why witnessing an accident is legally significant

To pursue the comparison further, we will identify three ways in which the occurrence of an accident is integral both to the reasons for recognising the category of claims by secondary victims arising from an accident and in defining the limits of this category.

First, an accident is, by definition, a discrete event in the ordinary sense of that word, meaning something which happens at a particular time, at a particular place, in a particular way. Whether someone was present at the scene and whether they directly perceived an accident are in most cases questions which admit of a clear and straightforward answer. These criteria for determining whether a person is eligible to claim compensation as a secondary victim therefore have the great merit of providing legal certainty. The clarity and certainty of these tests have been compromised to some extent by the decision in McLoughlin that the claimant in that case had witnessed the "immediate aftermath" of the accident even though she saw the injured members of her family in a different place from the accident site more than two hours after the accident had occurred. Nevertheless, a reasonably clear line can be drawn if heed is paid to the observations of Lord Wilberforce that allowing the claim in McLoughlin (a) was "upon the margin of what the process of logical progression would allow" (p 419G) and (b) depended critically on the evidence that, when the claimant came upon the members of her family, "they were in the same condition [as they had been at the roadside], covered with oil and mud, and distraught with pain" (p 419F).

Second, witnessing an accident involving a close family member is itself likely to be a disturbing and upsetting event even if that person in fact escapes unharmed and all the more so if that person is physically injured or killed. It is easy to appreciate the psychological trauma caused to a mother who sees her child run down by a car or to a husband who comes upon the badly injured body of his wife immediately after an accident. Most people would, we think, accept that, if a line is going to be drawn between cases where illness consequent on bereavement is compensable and cases where it is not, distinguishing between claimants who suffered the ordeal of actually witnessing the accident in which a close relative was killed and those who did not is an intelligible place to draw it.

A third significant feature of accident cases is that it is often difficult or arbitrary in such cases to distinguish between primary and secondary victims. Once the courts accepted that compensation can be recovered for psychiatric injury sustained without any physical impact, no distinction could reasonably be drawn between injury caused by fear for the claimant's own safety and by fear for the safety of a close family member. It was this reasoning which led the Court of Appeal in Hambrook to reject the view expressed (obiter) by Kennedy J in Dulieu [1901] 2 KB 669 , 675, that the right to recover compensation is limited to cases where psychiatric injury "arises from a reasonable fear of immediate personal injury to oneself". In Hambrook [1925] 1 KB 141, 157, Atkin LJ said of this suggested limitation:

"It would result in a state of the law in which a mother, shocked by fright for herself, would recover, while a mother shocked by her child being killed before her eyes, could not, and in which a mother traversing the highway with a child in her arms could recover if shocked by fright for herself, while if she could be cross-examined into an admission that the fright was really for her child, she could not. In my opinion such distinctions would be discreditable to any system of jurisprudence in which they formed part."

We agree. We would add that in a case where, for example, both a mother and her child are put in physical peril, it would not only be unjust but practically impossible to distinguish between the mother's emotions of fear for herself and fear for her child.

The contrast with non-accident cases

None of these three significant features of the accident cases is applicable where the claimant suffers illness from witnessing physical injury or illness in another person but does not witness any accident.

First, as we have seen, in many such cases there is no discrete event comparable to an accident. This is not always true. If a person suddenly and unexpectedly collapses and dies after suffering a cardiac arrest or some other medical crisis as happened in Paul and Novo , such an event can be identified with the same degree of certainty as an accident can. But the length of time for which symptoms of injury or disease last before a person recovers or dies is entirely variable. It may be minutes, hours, days or weeks. In Walters , for example, the period was 36 hours; in Shorter it was around 24 hours (measuring from first manifestation of injury rather than the inception of the underlying cause); in Sion it was 14 days. This gives rise to uncertainty about what qualifies as an "event" capable of founding a claim. We have discussed above the intractable difficulties involved in trying to answer that question in a way that is both reasonably certain and not entirely arbitrary. There are no comparable difficulties in determining whether a person has been killed, injured or put in peril in an accident.

Second, in cases where the claimant has not witnessed an accident but has witnessed the injury or illness of a close family member, the extent to which this experience is traumatic is also entirely variable. The facts of the present cases are at one extreme, as was Walters . But again there is a whole range of possible scenarios. The symptoms of injury or illness witnessed may be more or less severe. Unless the door is to be opened to claims based on direct perception of any symptoms however mild, some criterion is needed to distinguish between claims which are, or are not, admissible. We have discussed above three such criteria which have been applied or proposed: the "sudden shock", "horrifying event" and "first manifestation of damage" tests. For the reasons indicated, none of these, in our view, is an acceptable test. We have not been shown any cases in which comparable difficulties have arisen where the claim is based on presence at, and direct perception of, an accident.

Third, in cases where the claimant was not present at the scene of any accident, no question can arise of the claimant suffering psychiatric harm through fear for her own safety or bodily integrity. Any such harm which the claimant suffers can only be of a secondary nature caused by witnessing the injury, illness or death of another person. Allowing the claimant to recover compensation cannot therefore be justified by the practical impossibility and injustice of otherwise having to distinguish between injury caused by fear for the claimant's own safety and injury caused by fear for the safety of a close family member.

For these reasons, we do not consider that an analogy can reasonably be drawn between the situation with which McLoughlin , Alcock and Frost were concerned where illness is caused by witnessing an accident (or its immediate aftermath) involving a victim with whom the claimant has a close tie of love and affection and situations where the claimant does not witness an accident but suffers illness as a result of witnessing such a person suffering a medical crisis.

We also agree with the Court of Appeal in Novo that to extend the scope of allowable claims by secondary victims to situations where the claimant witnesses the death or illness of a relative from disease would give rise to unacceptable and unfair differences in treatment between different categories of claimant. It would be impossible to explain to an ordinary reasonable person why, for example, damages can be recovered by a daughter who sees her parent die from a heart attack or pulmonary embolism which should have been avoided, but compensation is denied to, say, a mother who did not witness a road accident in which her child was fatally injured or its "immediate aftermath" but identifies the mutilated body afterwards in the mortuary or is present at the hospital when her child dies many days later from injuries sustained in the accident. Such unjust differences in outcome could, of course, be avoided by removing the second and third requirements for claims by secondary victims established by the House of Lords in Alcock and confirmed in Frost . But it would not, in our view, be right to contemplate such a radical departure from settled law.

There is a further point which we consider important. The distressing experiences of the claimants in McLoughlin , Sion , Walters , Shorter , Ronayne and many other cases all occurred because the claimants attended the hospital where their child or other close relative was being treated. Quite rightly, no one criticised Mrs Walters for wanting to sleep in her baby's room on the night when he had an epileptic fit or for staying with him (except while he was being driven in an ambulance to a London hospital) for the next 36 hours or for holding her baby in her arms at his death. Nor do the cases discuss whether the hospital could or should have discouraged or even prevented Mr Sion from sitting by his son's bedside for 14 days or Mrs Shorter from sitting with her sister in her dying moments. Such decisions are very sensitive and difficult for both the relatives and the clinicians attending the primary victim. Everyone is seeking to do what is best both for the patient who may be comforted in their final moments by the presence of those dearest to them and for the relatives who may strongly want to be there. It is undesirable for decisions about end-of-life care to be complicated by the risk that, if it is said that the death ought to have been prevented, the hospital will be exposed to potential legal liability to family members as a result of them seeing and remaining with the patient.

Other jurisdictions

We have also considered whether any useful comparison can be made with the approach taken in other common law jurisdictions. However, there are significant differences between English, Australian, Canadian and New Zealand law, not to mention the laws of different states of the United States, concerning the recovery of damages for psychiatric harm suffered in connection with the death, injury or imperilment of another person caused by the defendant's negligence. These differences are such as to make it difficult and perhaps dangerous to draw any direct analogy. Furthermore, we have not had any Commonwealth authority cited to us which has addressed the recoverability of such damages in cases of medical negligence. In these circumstances we have not been able to derive assistance from comparison with the law applicable in other common law jurisdictions in considering the central question that we have to decide.

Earlier authorities revisited

Of the previous medical negligence cases to which we have referred, the only one in which a claim to recover damages as a secondary victim succeeded is Walters . In Novo , at para 35, the Court of Appeal distinguished Walters on the ground that what happened in that case was regarded as a single event and the court was not concerned with a situation, as in Novo , where the sustaining of the injuries and the death of the primary victim were separate events. Counsel for the claimants on these appeals and the courts below have attached significance to the fact that in Novo the Court of Appeal distinguished Walters rather than saying that it was wrongly decided. Given that the Court of Appeal cannot overrule its own previous decisions, however, this is unsurprising.

It is clear that on the facts of Walters the brain damage and death of Mrs Walters' baby were not caused by an accident. It follows from our conclusion that a claimant cannot recover damages for personal injury as a secondary victim unless the claimant witnessed an accident (or its immediate aftermath) that, had this defence been raised in Walters , the claim should have failed. As the defence was not raised and in consequence was not considered either by the trial judge or by the Court of Appeal, Walters cannot be regarded as an authority which weighs against our conclusion. Ward LJ cited the dicta of Peter Gibson LJ in Sion (which we have quoted at para 63 above) suggesting that there is no logical reason why a breach of duty causing an incident involving no violence could not lead to a successful claim for damages. But he did not refer to Taylor v Somerset or the requirement held by Auld J in that case to be necessary of an "external, traumatic event in the nature of an accident". Instead of raising that question, the argument in Walters was preoccupied with whether Mrs Walters' experience was to be regarded as "a sudden appreciation by sight or sound of a horrifying event". We have explained why we do not consider that an appropriate test.

We would therefore hold that Walters was wrongly decided on its facts and should not be followed.

It also follows that Sion , Shorter and Ronayne , although correctly decided, were decided on a wrong basis and that the claims in those cases should have been dismissed for the simple reason that the claimant did not witness an accident (or its aftermath) caused by the defendant's negligence. The confusion created in Galli-Atkinson by transposing the analysis in Walters to the aftermath of accident cases disappears once the aftermath concept is returned to the confines set in McLoughlin . The Galli-Atkinson approach should also not be followed.

The question was raised in argument of whether the rules governing claims by secondary victims arising from accidents could ever apply in a medical setting. The question does not arise in the present cases, as none of them involves an accident in the relevant sense. Various hypothetical examples were, however, posed in argument such as a scenario where a doctor injects a patient with a wrong dose or a wrong drug, inducing an acute adverse reaction which is witnessed by a close relative. In our view, the issues raised by such examples are best left to be addressed in a case where they actually arise on the facts.

As foreshadowed at the beginning of this judgment, we turn to test these conclusions by examining the general principles which govern the existence and scope of the duties of care owed by medical practitioners.

The need to establish an independent duty

In Alcock [1992] 1 AC 310 , 411A-B, Lord Oliver said that, although it is convenient to describe a claimant whose injury arises from witnessing the event of injury to another person as a "secondary" victim:

"that description must not be permitted to obscure the absolute essentiality of establishing a duty owed by the defendant directly to him - a duty which depends not only upon the reasonable foreseeability of damage of the type which has in fact occurred to the particular plaintiff but also upon the proximity or directness of the relationship between the plaintiff and the defendant."

The "absolute essentiality" in the present cases of establishing such a duty owed by the defendants directly to the claimants is not in dispute.

At the start of the oral argument on these appeals, Mr Robert Weir KC (whose submissions on this point were adopted by counsel for the other claimants) submitted that it is nevertheless neither necessary nor relevant to analyse the present claims by reference to the general principles which determine when a duty of care is owed by a doctor (or other professional person). He submitted that, in any case where the claimant is a secondary victim, the question whether the necessary proximity between the claimant and the defendant exists is governed by the rules established by the Alcock line of authority. He cited a passage from the speech of Lord Rodger of Earlsferry in D v East Berkshire Community Health NHS Trust [2005] UKHL 23 , [2005] 2 AC 373 , para 107, describing claims by secondary victims who suffer "nervous shock" as "a particular chapter of the law" and a "distinct line of authority". Lord Rodger suggested that "medical mishaps" fall within this category, giving Walters as an example.

Those observations of Lord Rodger were made in passing in a case which was not concerned with a "medical mishap" and where the correctness of the decision in Walters was not questioned, as it has been in these appeals. The key question which we have to decide is whether the rules that determine when the necessary proximity exists to give rise to a duty of care owed to a secondary victim in an accident case (or analogous rules) apply in cases of medical negligence where there is no accident. That question was not considered, let alone answered, in Alcock or other cases in that line of authority. Nor for that matter was it considered in Walters where (as noted above) the issue was not raised. In these circumstances to assert, as counsel for the claimants have, that the question whether the defendants owed a duty directly to the claimants in the present cases is governed by the rules established by the Alcock line of authority begs the central question raised on these appeals by assuming an answer to the very point in dispute.

It is also inconsistent with the point that Lord Oliver was making in the passage quoted at para 125 above. He was at pains to emphasise that, to justify a remedy, it is not sufficient to establish (1) a breach of a duty of care owed by the defendant to a primary victim and (2) an appropriate relationship between the primary victim and the claimant. Not only is this insufficient, but it is also unnecessary. Lord Oliver made it clear that there may in fact be no primary victim (see para 105 above). He said nothing to suggest that, for the purpose of establishing a duty of care owed by the defendant directly to the claimant, the general principles of the law of negligence that determine when the relationship between the parties is such as to give rise to a duty of care can be ignored or bypassed. Where the context is a medical crisis and the defendant is a medical practitioner, we think it essential to consider whether a duty of care is owed by reference to the general principles applicable to this type of case.

Proximity

As Lord Oliver emphasised, reasonable foreseeability of harm, although necessary, is not by itself enough to give rise to a duty of care. There must also exist the necessary "proximity" in the relationship between the parties to make it just to impose such a duty. The classic description of this requirement is that of Lord Atkin in Donoghue v Stevenson [1932] AC 562 , 580:

"You must take reasonable care to avoid acts or omissions which you can reasonably foresee would be likely to injure your neighbour. Who, then, in law is my neighbour? The answer seems to be - persons who are so closely and directly affected by my act that I ought reasonably to have them in contemplation as being so affected when I am directing my mind to the acts or omissions which are called in question."

The question is one of interpersonal justice. As Lord Nicholls of Birkenhead explained in Stovin v Wise [1996] AC 923 , 932:

"Proximity is convenient shorthand for a relationship between two parties which makes it fair and reasonable [that] one should owe the other a duty of care."

The need to show not only reasonable foreseeability of harm but proximity sufficient to give rise to a duty of care applies whatever the nature of the harm suffered by the claimant - whether it be financial loss, damage to property or personal injury: see eg Marc Rich & Co AG v Bishop Rock Marine Co Ltd [1996] AC 211 , 235-236. Donoghue v Stevenson itself, for example, was a case of alleged personal injury caused by consuming a defective product. The House of Lords did not decide that the manufacturer of a product owes a duty of care to a person who suffers personal injury from consuming or using the product whenever such injury is reasonably foreseeable. Rather, they held that the necessary proximity in the relationship between the parties exists, not in every such case, but where the product is sold in a form which shows that it is intended to reach and be used by the ultimate consumer without alteration or the reasonable possibility of intermediate examination: see Lord Atkin at [1932] AC 562 , 599; and Grant v Australian Knitting Mills Ltd [1936] AC 85 .

The relevant relationship between the manufacturer of a product and the ultimate consumer arises only at the point of use. Similarly, two road users one of whom injures the other by careless driving may have been complete strangers before the accident in which the injury is inflicted. Sometimes, however, proximity is established by a pre-existing relationship between the parties. A concept used to explain how such a relationship may give rise to a duty of care is that of assumption of responsibility. The core of this idea is that a person (A) who provides a service to another person (B) who reasonably relies on A's expertise in performing the service assumes a responsibility to B to perform the service with reasonable care and skill.

It is this principle which underlies the relationship of proximity between an architect or building contractor and their employer, between a lawyer or accountant and their client, and between a doctor or other medical practitioner and their patient. By providing a service, whether under a contract for reward between the doctor and the patient or - as in the case of a patient entitled to treatment under the National Health Service - where the doctor is paid by the state, the service-provider assumes a responsibility towards the person to whom the service is provided, which gives rise to a duty of care.

The scope of the duty will vary with the circumstances and will depend, critically, on the purpose for which the service is provided. The recent decision of this court in Meadows v Khan [2021] UKSC 21 , [2022] AC 852 illustrates the importance of this consideration in determining the scope of the duty of care owed by a doctor to a patient. The Supreme Court held that, where the purpose for which a doctor was consulted concerned a particular risk in a pregnancy (of giving birth to a child with haemophilia), the doctor was not liable for the consequence of an unrelated risk (that the child would suffer from autism). Consideration of the purpose for which services are provided is equally important in determining whether or when a duty of care is owed by a doctor to someone other than their patient.

Duties owed by doctors to non-patients

There are circumstances in which the duty of care owed by a medical practitioner may extend beyond the health of their patient to include other people. For example, some commentators suggest that a doctor who negligently sent home a patient with a highly infectious disease would owe a duty of care to members of the patient's household who contract the disease as a result: see Clerk & Lindsell on Torts , 24th ed (2023), para 9-10; Michael Jones, Medical Negligence , 6th ed (2021), paras 2-159 - 2-160. In the case cited for this proposition, however, the claim failed because on the facts the defendant local authority was found not to be vicariously liable for the conduct of the doctor: see Evans v Liverpool Corporation [1906] 1 KB 160 . It is likewise arguable that a doctor who negligently fails to diagnose or treat a sexually transmitted disease may owe a duty of care to the sexual partner of a patient who, in consequence, contracts the disease. Such a claim succeeded in the Australian case of BT v Oei [1999] NSWSC 1082 where the sexual partner of a patient contracted HIV. The duty of care to the patient's partner recognised in that case, however, was limited to a duty (owed also to the patient) to warn him of his HIV status and advise him that he had a statutory responsibility to warn prospective sexual partners of his condition. The court found that, had the patient been given such advice, he would have informed the claimant who would then have taken steps to avoid contracting HIV.

We express no view on the difficult questions raised by such cases, save to observe that, in relation to infectious disease, doctors are considered to have a responsibility to protect public health which is wider than their duty to protect the health of their patient and is reflected, for example, in statutory obligations which already existed when Evans v Liverpool Corporation was decided.

Family members who witness a patient's medical crisis

Here the question is whether a doctor who owes a duty of care to a patient also owes a duty to members of the patient's close family to take care to protect them against the risk of illness from the experience of witnessing the medical crisis of their relative arising from the doctor's negligence.

It cannot be said that a doctor who treats a patient thereby enters into a doctor-patient relationship with any member of the patient's family and thereby assumes responsibility for their health. As regards other factors relevant to whether the necessary relationship of proximity exists, the extent of the control which a doctor may be seen as having over the risk of injury to members of the patient's family and the directness of the causal link between the doctor's negligence and the materialisation of that risk will depend upon the particular facts of the case. In a case such as Paul , it may be happenstance whether, if the patient's untreated coronary artery disease leads at some unpredictable future time to a cardiac arrest, this happens to occur in the presence of close family members or elsewhere, such as at the person's workplace. By contrast, on the facts of Walters , the harm suffered by the claimant was a far more direct and obviously foreseeable consequence of the defendant's negligence. The same might be said about the facts of Purchase .

Common to all cases of this kind, however, is a fundamental question about the nature of the doctor's role and the purposes for which medical care is provided to a patient. We are not able to accept that the responsibilities of a medical practitioner, and the purposes for which care is provided, extend to protecting members of the patient's close family from exposure to the traumatic experience of witnessing the death or manifestation of disease or injury in their relative. To impose such a responsibility on hospitals and doctors would go beyond what, in the current state of our society, is reasonably regarded as the nature and scope of their role.

There is no doubt that witnessing the death from disease of a close family member can have a powerful psychological impact additional to the grief and deep distress caused by the fact of the death. Whether that impact is damaging or may even help the grieving process must depend on many factors, including the vulnerability and circumstances of the individual who witnesses the event and the place, time and other circumstances in which the death occurs. The experience of seeing a person die or discovering their dead body is rarer today than it once was. Most deaths in the United Kingdom now occur in hospitals or other institutions such as care homes. But although social attitudes and expectations may be changing, we would not accept that our society has yet reached a point where the experience of witnessing the death of a close family member from disease is something from which a person can reasonably expect to be shielded by the medical profession. That is so whether the death is slow or sudden, occurs in a hospital, at home or somewhere else, and whether it be peaceful or painful for the dying person. We do not mean in any way to minimise the psychological effects which such an experience may have on the person's parent, child or partner when we express our view that, in the perception of the ordinary reasonable person, such an experience is not an insult to health from which we expect doctors to take care to protect us but a vicissitude of life which is part of the human condition.

We return to the point with which we began this judgment, that the general policy of the law is opposed to granting remedies to third parties for the effects of injuries to other people. What therefore principally requires justification is not the narrowness of the category of cases in which a claimant who suffers personal injury which is secondary to the death or injury of another person can recover damages but the fact that it exists at all. Lord Oliver made this point in Alcock , when he said, at p 410H:

"What is more difficult to account for [than the general rule] is why, when the law in general declines to extend the area of compensation to those whose injury arises only from the circumstances of their relationship to the primary victim, an exception has arisen in those cases in which the event of injury to the primary victim has been actually witnessed by the plaintiff and the injury claimed is established as stemming from that fact."

Lord Oliver regarded the existence of this exception as "now too well established to be called in question" and so do we.

Unless the exception defined by the Alcock line of authority is to become the general rule, however, a line must be drawn somewhere to keep the liability of negligent actors for such secondary harm within reasonable bounds. Wherever the line is drawn, some people who suffer what may be serious illness in connection with the death or injury of another person will be left uncompensated. The mother who learns in a telephone call that her child has been killed in a road accident may suffer an illness no less severe than a mother who was present at the scene and saw the accident. But there is a rough and ready logic in limiting recovery by secondary victims to individuals who were present at the scene, witnessed the accident and have a close tie of love and affection with the primary victim. These limitations are justified, not by any theory that illness induced by direct perception is more inherently worthy of compensation than illness induced by other means; but rather by the need to restrict the class of eligible claimants to those who are most closely and directly connected to the accident which the defendant has negligently caused and to apply restrictions which are reasonably straightforward, certain and comprehensible to the ordinary person.

We have not been asked on these appeals to alter or abrogate the limits on the recovery of damages by secondary victims in accident cases established by the decision of the House of Lords in Alcock . Instead, this court is asked to recognise as analogous a category of cases in which illness is sustained by a secondary victim as a result of witnessing a death or manifestation of injury which is not caused by an external, traumatic event in the nature of an accident but is the result of a pre-existing injury or disease. For the reasons given, we do not consider that such cases are analogous. That conclusion is reinforced by our opinion that the persons whom doctors ought reasonably to have in contemplation when directing their minds to the care of a patient do not include members of the patient's close family who might be psychologically affected by witnessing the effects of a disease which the doctor ought to have diagnosed and treated. Hence there does not exist the proximity in the relationship between the parties necessary to give rise to a duty of care.

Result

No one could read or hear about the events which Saffron and Mya Paul, Lynette and Mark Polmear and Tara Purchase experienced without being moved by the terrible distress caused to them by the sudden deaths of, respectively, Parminder Singh Paul, Esmee Polmear and Evelyn Purchase and the shocking circumstances in which those deaths occurred. The thought that these tragic events could have been avoided if the hospital or doctor had exercised due care must, as in every case of wrongful death, add further to the agony and perhaps anger that they feel. The law cannot, however, impose duties and liabilities on the basis of sympathy, however strongly felt. For the reasons we have sought to explain, the claims for compensation made in these cases do not satisfy the legal requirements for the recovery of damages by secondary victims who suffer injury as a result of the death of another person. The appeals must therefore be dismissed.

lord Burrows ( dissenting ):

The common law has struggled to deal satisfactorily with negligently caused "pure" psychiatric illness (that is, psychiatric illness that is not consequential on physical injury to the claimant). In the context of secondary victims - who can be regarded, generally speaking, as those who suffer psychiatric illness as a result of another's death or injury or imperilment - the tort of negligence draws distinctions that are difficult to defend. It is arguable that the only truly principled solution, which would avoid any arbitrary line drawing, would be to impose a duty of care where, in general terms, it was reasonably foreseeable that psychiatric illness would be caused to the secondary victim (as a person of reasonable fortitude). But policy concerns, in particular the fear of opening the floodgates of litigation, mean that adoption of that principled solution would constitute a radical and giant leap forward for the common law. For that reason, at the present stage of development, that solution is not a realistic option open to the courts.

The Law Commission of England and Wales looked at this area over 25 years ago. In its report, Liability for Psychiatric Illness Law Com No 249 (1998), [1998] EWLC 249 (and I should declare that I was the Law Commissioner in charge of that project), the Law Commission recommended legislative reform. In essence, the recommendation was that, in addition to reasonable foreseeability of psychiatric illness (to a person of reasonable fortitude), the requirement would be retained that the secondary victim must have a close tie of love and affection to the primary victim (thereby, it was argued, avoiding opening the floodgates of litigation). But the other main restrictive common law requirements (closeness in time and space to the event, perception through one's own unaided senses, and the need for the event to be shocking) would be abandoned under the proposed legislation.

That recommendation for legislation was not accepted by the Government who, significantly for these appeals, preferred to leave the courts to continue to develop the law. It said, in Ministry of Justice, The Law of Damages, Response to Consultation , CP(R) 9/07, 1 July 2009, at p 51:

"The arguments in this complex and sensitive area are finely balanced. On balance the Government continues to take the view that it is preferable for the courts to have the flexibility to continue to develop the law rather than attempt to impose a statutory solution."

The Government has therefore thrown back to the courts the challenge of developing the law in this difficult area. There is no realistic prospect of legislation.

The three conjoined cases, with which we are here concerned, are the first cases coming before the highest court (whether House of Lords or Supreme Court) since the Government made clear that it was leaving to the courts the challenge of developing the law in this area and that a legislative solution would not be forthcoming. With a couple of exceptions that have little bearing on the matters before us ( Page v Smith [1996] AC 155 and W v Essex County Council [2001] 2 AC 592 ), this is also the first time that the highest court has had a chance to consider the law on negligently caused psychiatric illness to secondary victims since the Hillsborough disaster litigation which spawned the leading cases of Alcock v Chief Constable of South Yorkshire Police [1992] 1 AC 310 (" Alcock ") and Frost v Chief Constable of South Yorkshire Police [1999] 2 AC 455 (" Frost ").

In my view, in the light of the Government's explicit approach, it would be inappropriate for this court to continue to take the view that "thus far and no further", propounded by Lord Steyn in Frost, at p 500, some 25 years ago and at a time before the Government had spoken. It is clear that Lord Steyn made his statement against the backdrop of possible legislation and, although (in contrast to Lord Griffiths, Lord Goff and Lord Hoffmann in Frost ) he did not mention the work of the Law Commission, the decision in Frost was handed down at a time when the Law Commission's recommendations, although published, had yet to be considered by Government. Lord Steyn said this at p 500:

"In my view the only sensible general strategy for the courts is to say thus far and no further. The only prudent course is to treat the pragmatic categories as reflected in authoritative decisions such as the Alcock case ... as settled for the time being but by and large to leave any expansion or development in this corner of the law to Parliament. In reality there are no refined analytical tools which will enable the courts to draw lines by way of compromise solution in a way which is coherent and morally defensible. It must be left to Parliament to undertake the task of radical law reform."

The three conjoined cases, with which we are concerned, have common central features. In each there was medical negligence in failing to diagnose and treat the primary victim's life-threatening condition. This led, some time after the breach of duty, to the unexpected death of the primary victim. The secondary victim was a close relative who either witnessed the death or came upon the primary victim immediately after her death and it is not in dispute that the death in each case was shocking or horrific.

There were conflicting decisions in the lower courts in the three cases. Unlike the other two cases, in Paul v The Royal Wolverhampton NHS Trust, there was a first appeal to the High Court and, in an impressive judgment, Chamberlain J found for the claimants on the basis that the relevant event was the death and that it was not in dispute that, if the death were the relevant event, the limiting factors controlling liability were here all satisfied. But in all three cases, with some reluctance, the Court of Appeal held itself bound by the decision of the Court of Appeal in Taylor v A Novo (UK) Ltd [2014] QB 150 (" Novo ") to find for the defendants and to deny liability.

The appeals arise from applications to strike out the claims so that the facts, as pleaded by the claimants, must be assumed to be true for the purposes of these appeals.

(1) Paul v The Royal Wolverhampton NHS Trust

The relevant claimants (who are the appellants) are daughters of the deceased. Aged 12 and 9 at the time, they were with their father when he collapsed and died on 26 January 2014. Fourteen months before his death, on 9 November 2012, the deceased was admitted to the defendant's hospital complaining of chest and jaw pain. He was given treatment by a cardiology registrar for acute coronary syndrome. He was reviewed by a cardiology registrar on 11 November 2012 and it was recommended that he undergo echocardiogra phy as an inpatient to investigate the possibility of significant coronary artery disease. The deceased was then seen by a consultant cardiologist on 12 November 2012 and discharged as echocardiography was not available.

The alleged negligence was that, during that period in hospital, there was a failure to recognise that the deceased's presentation was typical for coronary artery disease and coronary angiography was therefore not arranged. Had coronary angiography been performed, it was alleged that it would have revealed significant coronary artery disease and the deceased would have been treated with coronary revascularisation. In this way, Mr Paul would have avoided suffering the heart attack in January 2014.

On 26 January 2014 the deceased was out shopping with the claimants. He mentioned that he felt ill. One of his daughters was walking slightly behind her father and the other slightly in front. The daughter in front (the first daughter) turned and saw her father leaning against a wall momentarily. She saw his eyes roll back. Both girls saw him fall backwards and hit his head on the floor. The claimants were alone with their father in the street. The first daughter tried to ring her mother and then an ambulance but in her distress was unsuccessful. There was no one immediately around and she began shouting for help until eventually a lady came and called an ambulance. The other daughter managed to make contact with her mother but was too distressed to be understood. The first daughter took the phone and told her mother what had happened. Both girls saw a man holding their father's head and there was blood on his hands.

Their mother arrived at the scene and the children were taken into a nearby church. The claimants remember hearing their mother outside screaming their father's name and going back outside. The claimants saw the ambulance crew put a foil blanket over their father. They were doing chest compressions. There was a crowd of people including the police. The claimants were taken to an aunt and uncle's house. The ambulance arrived at 15.57 and left the scene at 16.28, arriving at hospital at 16.43 but further resuscitation was felt to be futile and the deceased was declared dead at 16.51.

The claimants' pleaded case is that their experiences on 26 January 2014 were horrific and shocking and were the cause of the psychiatric illnesses for which they make their claims.

(2) Polmear v Royal Cornwall Hospital NHS Trust

The claimants, who are the appellants, are the mother and father of Esmee Polmear. Esmee, then aged 6, was seen by her GP on 19 August 2014 with a history of episodes during which she could not breathe, appeared pale and turned blue after a few minutes. Esmee's mother was reassured by the GP. On 10 September 2014, Esmee and the claimants returned to the GP because of worsening symptoms. Esmee was referred to a paediatrician at the hospital and was seen by a paediatric registrar on 1 December 2014 in the presence of both claimants. As a result, from 21 to 22 January 2015, Esmee underwent ambulatory echocardiogram monitoring, during which Esmee did not experience any episode of shortness of breath. At that point, the episodes were not occurring daily. The reviewing consultant paediatrician concluded that Esmee's symptoms were likely to be related to exertion and were physiological ''with nothing to suggest an underlying abnormality of her cardiac rhythm". He confirmed those conclusions in correspondence with the claimants. Esmee was seen again by her GP on 21 April 2015 accompanied by her father. Esmee was re-referred to the paediatrician at the hospital but the referral did not take place due to her death on 1 July 2015, the cause of which was pulmonary veno-occlusive disease.

On 1 July 2015, Esmee was due to attend a school trip to the local beach but did not feel well. It was agreed that her father would meet Esmee at the beach to take her back to school if required. When he later went to the beach, Esmee was not present. The father found Esmee with a teacher and another pupil. Esmee looked tired, pale and was breathless. Esmee wanted to sit down but was encouraged to try to walk. At one point, she stopped and vomited.

Her father resumed the walk to the school but Esmee seemed frightened at the thought of walking and had to stop frequently, causing him to carry Esmee to the school. She was white and clammy with some blueness around her lips. At the door of the school, Esmee said that she felt faint. Her father reassured and comforted her. He walked away but received a call asking him to return. He ran to the school and saw Esmee lying on the floor and a member of staff providing first aid.

Her father took over and attempted to give Esmee mouth-to-mouth resuscitation. She was not breathing. Her mother ran to the school and saw Esmee lying on the floor with members of staff attempting resuscitation which she could see was not working. Paramedics arrived and attempted resuscitation, which was witnessed by both claimants. The claimants went with Esmee in an ambulance to hospital. Attempts to revive Esmee continued at the hospital but they were unsuccessful. Esmee was 7 years old when she died.

As a result of witnessing the collapse, unsuccessful attempts to resuscitate and the death of Esmee, her mother developed post-traumatic stress disorder and major depression. Her father developed post-traumatic stress disorder, and major depression with addictive behaviour.

The defendant admits that Esmee's condition should have been diagnosed by mid-January 2015. It is the claimants' case that, with proper diagnosis and management, Esmee would not have collapsed or died on 1 July 2015 and would not have required resuscitation.

(3) Purchase v Dr Ahmed

The claimant, who is the appellant, is the mother of Evelyn Purchase. Evelyn, aged 20, died on 7 April 2013. The cause of her death was extensive bilateral pneumonia with pulmonary abscesses.

On 28 January 2013, Evelyn visited her GP with acute sinusitis. In February, Evelyn continued to feel unwell. She lost her appetite, resulting in weight loss. On 28 March 2013, Evelyn visited her GP and was prescribed medication for oral thrush and for a skin infection. She subsequently developed a cough and mouth ulcers. She lost her appetite and stopped eating.

By 4 April 2013, Evelyn was weak and generally unwell. The claimant took Evelyn to the out-of-hours clinic where she was examined by the defendant at around 21.58. Evelyn had difficulty walking into the clinic as a result of weakness, dizziness and problems in breathing which was rapid, shallow and noisy. The diagnosis made was respiratory tract infection with pleuritic pain, oral thrush and depression. Antibiotics and an antidepressant were prescribed. Evelyn was advised to contact her own GP if the problems did not resolve.

Evelyn's condition remained the same except that, by 6 April 2013, she was additionally complaining of heart palpitations. That evening the claimant attended a pre-planned event in London with her younger daughter. She discussed staying at home but Evelyn insisted she kept to her plans. Evelyn's father (the claimant's ex-husband) remained at home with her.

The claimant returned home at 4.50am on 7 April 2013. She found Evelyn lying motionless on the claimant's bed with the house telephone in her hand, staring at the ceiling not moving. Her skin was slightly warm, she looked alive but was not moving or blinking. The claimant felt stunned, panicked and began screaming. She was joined by her younger daughter and her ex-husband. All were screaming. The claimant attempted to call 999 but the phone would not work. The younger daughter called 999 and the family were advised to move Evelyn to the floor and to carry out cardiopulmonary resuscitation.

As the claimant did so, she noticed Evelyn had urinated. In attempting mouth-to- mouth resuscitation, the claimant opened Evelyn's mouth but this caused blood and bodily fluids to spill out of the mouth and nose. The claimant tipped Evelyn's body to one side and more fluid spilled out. Increasingly aware that her efforts would be in vain, the claimant attempted resuscitation until the arrival of paramedics. The claimant watched the paramedics' attempts at resuscitation. Within minutes, the claimant was told that her daughter had died.

The claimant realised that she had a missed call from Evelyn on her mobile phone and a voice message. It was the sound of Evelyn's dying breaths which continued for four minutes and 37 seconds. This caused the claimant to run out of the house and stand screaming in the street. The call was timed at 4.40am, and concluded approximately five minutes before the claimant saw Evelyn.

The claimant developed post-traumatic stress disorder, severe chronic anxiety and depression. It is the claimant's case that Evelyn had severe pneumonia when seen by the defendant on 4 April 2013. It is alleged that there was a negligent failure properly to assess and treat Evelyn's symptoms and that, but for the alleged breaches, the events that the claimant witnessed, and which it is alleged caused her psychiatric illness, would have been avoided.

(1) The decisions at first instance

The first of the three cases to come before the courts was Paul v The Royal Wolverhampton NHS Trust ("Paul") . Master David Cook [2019] EWHC 2893 (QB) decided that the claim should be struck out because, applying Auld J's decision in Taylor v Somerset Health Authority [1993] PIQR P262, as approved by the Court of Appeal in Novo , the relevant event could not be the death. The time lag of 14 ½ months between the negligence and the death meant that the necessary proximity in time and space was not satisfied. In contrast, in Polmear v Royal Cornwall Hospital NHS Trust (" Polmear" ), the same judge, [2021] EWHC 196 (QB) , refused to strike out the claim. This was because, subsequent to his decision in Paul , there had been a successful appeal in Paul to Chamberlain J and, in Polmear , Master David Cook considered himself bound by that decision. He could not say that the claim was bound to fail. Earlier in Purchase v Dr Ahmed ("Purchase") , and prior to Chamberlain J's decision, District Judge Lumb, sitting in the County Court at Birmingham (in a judgment dated 6 May 2020) held that, "not without some considerable regret" (para 32), he was bound by the authorities, in particular Novo , to strike out the mother's claim. The death was not the relevant event.

(2) The appeal to Chamberlain J in Paul

Chamberlain J [2020] EWHC 1415 (QB) , [2020] PIQR P19 , decided that the claim in Paul should not have been struck out by Master David Cook and therefore allowed the appeal. Having examined all the relevant cases, his essential reasoning was as follows:

(i) He explained that the key question was whether Mr Paul's death, 14 ½ months after the allegedly negligent treatment, could be the relevant "event". If it could be, then it was not in dispute that each of the "control mechanisms" was satisfied on the facts pleaded.

(ii) It was not an objection to the death being the relevant event that there was a significant time lag between the negligence (ie the breach of the duty of care) and the death. There was nothing in the authorities to suggest that a claim for psychiatric illness from witnessing a person's death or injury caused by, for example, the collapse of negligently erected scaffolding or electrocution as a result of negligent wiring would be affected by the date of the negligence.

(iii) It was not an objection to the death being the relevant event that the liability depended on an omission rather than an act. To draw such a distinction in this context would be unprincipled and, in any event, the claim in the medical negligence case of North Glamorgan NHS Trust v Walters [2002] EWCA Civ 1792 , [2003] PIQR P16 (" Walters ") succeeded even though that was a case of negligent omission.

(iv) It was not an objection to the death being the relevant event that the secondary victim must be present at the scene of the tort because here the scene of the tort, ie where the cause of action accrued, was where Mr Paul collapsed and died and the claimants were present at that scene.

(v) It was open to interpretation whether the reasoning of Auld J in Taylor v Somerset HA , requiring an external traumatic event, was referring to an event external to the primary victim or an event external to the secondary victim. But in so far as Auld J was referring to an event external to the primary victim, this would be inconsistent with the Court of Appeal's decision in Walters (as well as Swift J's reasoning in the medical negligence case of Shorter v Surrey and Sussex Healthcare NHS Trust [2015] EWHC 614 (QB) ) because the seizure in Walters was internal to the primary victim; and, while Lord Dyson MR in the Court of Appeal in Novo distinguished Walters , he did not say that it was wrongly decided. Moreover, Lord Dyson was careful to say that an accident case was a paradigm case and that allowed for non-paradigm cases where there was no accident, as in Walters .

(vi) Novo could be distinguished, and did not preclude liability, because in this case there was only one event, the death, which had been witnessed by the secondary victims. In contrast, in Novo there was an earlier event, the collapse of the racking boards onto the primary victim, which had not been witnessed by the secondary victim. If (contrary to Chamberlain J's view) it were necessary to identify a stopping point after which the consequences of negligence can no longer qualify as an event, the most obvious candidate would be the point when damage to the primary victim becomes manifest. Here, therefore, Novo would be no bar to recovery if it were shown that Mr Paul's death was the first occasion on which the damage caused by the negligent failure to diagnose and treat his heart disease became manifest.

(3) The Court of Appeal's decision on the conjoined appeals

The defendants in Paul and Polmear, and the claimant in Purchase, appealed to the Court of Appeal and the appeals were heard together. Sir Geoffrey Vos MR, Underhill LJ and Nicola Davies LJ [2022] EWCA Civ 12 , [2023] QB 149 , allowed the appeals of the defendants in Paul and Polmear and refused the appeal of the claimant in Purchase . That is, it was decided by the Court of Appeal that there was no liability for the negligently caused psychiatric illness in any of the three cases. The leading judgment was given by Sir Geoffrey Vos. Underhill LJ gave a short but penetrating judgment agreeing with Sir Geoffrey Vos. Nicola Davies LJ agreed with both judgments.

The reasoning of the Court of Appeal can be summarised as follows:

(i) The Alcock elements (or "control mechanisms" although that was a phrase which Sir Geoffrey Vos preferred to avoid) are concerned to establish the necessary legal proximity between the defendant and the secondary victim. In line with Lord Oliver's speech in Alcock, Sir Geoffrey Vos articulated five elements (although the last tends to replicate two of the earlier elements): a marital or parental relationship; sudden shock; presence at the scene or its immediate aftermath; witnessing the death or injury of, or extreme danger to, the primary victim; physical and temporal proximity to the event. The five elements apply to clinical negligence cases as well as to accident cases.

(ii) The important question raised by these cases is what constitutes the relevant horrific event.

(iii) Although advocated, respectively, by leading counsel for the claimant in Paul and leading counsel for the defendants, the relevant horrific event in these cases was neither the first manifestation of injury to the primary victim nor the accrual of the cause of action for negligence. As Underhill LJ put it at para 105, those two approaches are "unprincipled and unworkable".

(iv) Furthermore, applying Novo , the relevant horrific event could not be the death in these cases because the death was a separate event removed in time from the negligence (although Walters showed that the negligence and the horrific event could be part of a continuum). Novo was binding authority and, contrary to Chamberlain J's reasoning, could not be distinguished. In Underhill LJ's words at para 104:

"the fair reading of [the relevant paras in Novo ] seems to be that the ultimately decisive feature was simply that there had been an interval of time between the breach of duty, whether or not it occasioned any injury at the time, and the shocking event. In Novo itself the interval was three weeks, but the principle must be the same whatever the interval, provided it is not part of the same sequence of events as in cases of the Walters kind."

Similarly, Sir Geoffrey Vos said that the death could not be the relevant event because it was removed in time from the negligence or the accident or the first horrific event.

(v) Had the matter been free from authority, the Court of Appeal indicated that it would have taken the contrary view. Sir Geoffrey Vos said at para 12:

" Novo is binding authority for the proposition that no claim can be brought in respect of psychiatric injury caused by a separate horrific event removed in time from the original negligence, accident or a first horrific event. I accept that, although there is no logical reason for these rules, they are the way Auld J in Somerset and the Court of Appeal in Novo built upon the five elements and adapted them to the clinical negligence context. If I were starting with a clean sheet, I can quite see why secondary victims in these cases ought to be seen to be sufficiently proximate to the defendants to be allowed to recover damages for their psychiatric injury. Since, however, this court is bound by Novo, it is for the Supreme Court to decide whether to depart from the law as stated by Lord Dyson MR in that case." (emphasis added)

Similarly, Underhill LJ said the following at paras 102 - 103:

"Lord Oliver's references in Alcock to the need for 'physical and temporal propinquity' are not directed to the relationship between the breach of duty and the shocking event but rather to the need for the claimant to be close in space and time to the shocking event. It follows that if the point were free from authority I would be minded to hold that on the pleaded facts the claimants in all three cases should be entitled to recover... It would not involve going beyond the elements established in Alcock: rather, it would represent their application in a different factual situation."

Underhill LJ concluded at para 106 that, like Sir Geoffrey Vos, "My strong provisional view... is that the issues raised ... merit consideration by the Supreme Court."

The central submission of Henry Pitchers KC and David Tyack KC, leading counsel for the appellants in the Polmear and Purchase cases, is that one should treat the relevant event in these cases as the death of the primary victim; and that once one does treat the relevant event as the death of the primary victim, all the established proximity or control factors are here satisfied and do not need to be departed from in order for the claimants to succeed. That is, the secondary victim had a close tie of love and affection to the primary victim, the secondary victim was close in time and space to the death, the death was experienced by the secondary victim through his or her own unaided senses, and the death was shocking and horrific. In so far as one would be developing the law to allow recovery in these cases, the development would be an incremental one and within the accepted ambit of the judicial role. The law would be moving forward, if at all, only in the limited sense of recognising the death as the relevant event.

Robert Weir KC , leading counsel for the appellants in Paul , preferred to put forward the different central submission that the relevant event was the first manifestation of the injury to the primary victim. On the facts of Paul , the first manifestation of the injury was the death of the primary victim but the implication of the submission is that the first manifestation of the injury could be at an earlier time in other cases (and, presumably, that might have been so in the Polmear and Purchase cases).

To simplify matters, I should say at the outset that I prefer the central submission put forward by Henry Pitchers KC and David Tyack KC for Polmear and Purchase to that put forward by Robert Weir KC for Paul . In these cases, like Chamberlain J, I regard the central issue as whether or not the relevant event was the death of the primary victim. It would overcomplicate matters and lead to needless and, in practice, very difficult enquiries to treat the relevant event as the first manifestation of the injury. Indeed, I would regard the practical difficulties in pinpointing the first manifestation of the injury to be such that, like Underhill LJ, I would reject the central submission of Mr Weir as unworkable.

The approach advocated by the respondents, which emerged particularly clearly in Simeon Maskrey KC's oral submissions, was essentially that (putting to one side analogous imperilment cases) liability for secondary victims is confined to where the primary victim is injured or killed in an accident . The requirement for there to have been an accident, which Mr Maskrey argued was, in effect, synonymous with there being an event external to the primary victim, is crucial. It follows from this submission that, subject to rare exceptions, there can be no liability to secondary victims in the context of medical negligence because medical negligence rarely involves an accident. Mr Maskrey therefore submitted that the need for there to be an external event to the primary victim, relied on by Auld J in the medical negligence case of Taylor v Somerset Health Authority [1993] PIQR P262 to deny recovery, is of central importance. He also placed particular reliance on the Court of Appeal's decision in Novo because, although that was not a medical negligence case, the reasoning rejected treating the mother's death as the relevant event and confirmed and built on Auld J's restriction. The Court of Appeal in Novo decided that, in an accident case, the relevant event was the accident alone and that one could not treat the later death, which was witnessed by the secondary victim, as the event for these purposes.

The general law on the recovery of damages for psychiatric illness by secondary victims is to be gleaned from the leading cases in the House of Lords of McLoughlin v O'Brian [1983] 1 AC 410 , Alcock and Frost . There is no need on these appeals to consider the details of those leading cases because the important elements of the general law laid down in those cases are not in dispute. They can be stated, in summary form, as follows:

(i) The underlying question is whether the defendant owed a duty of care to the claimant (the secondary victim) not to cause that person a recognised (or recognisable) psychiatric illness consequent on the death, injury or imperilment of the primary victim.

(ii) The claimant (the secondary victim) must suffer a recognised psychiatric illness as distinct from mental distress (which includes upset, grief and anxiety).

(iii) It must have been reasonably foreseeable to the defendant that the claimant, as a person of reasonable fortitude, might suffer a psychiatric illness as a result of the defendant's negligent conduct which has led to the death, injury or imperilment of the primary victim.

(iv) There are four additional proximity factors, or controls, that the claimant must establish. Assuming for present purposes that the relevant event (ie the event that has resulted in the secondary victim's psychiatric illness) is an accident (which is the paradigm situation although whether the relevant event must be an accident is the central question in dispute in these appeals), these proximity factors, or controls, can be expressed as follows. First, that the claimant had a close tie of love and affection with the primary victim. Secondly, that the claimant was close to the accident in time and space or came across its immediate aftermath. Thirdly, that the claimant directly perceived the accident through his or her own unaided senses (rather than, for example, hearing about it from a third party). Fourthly, that the psychiatric illness was caused by a shock to the system: that is, the accident must have been shocking and horrific. As regards that last element, in Alcock , Lord Ackner said, at p 401: "'Shock', in the context of this cause of action, involves the sudden appreciation by sight or sound of a horrifying event, which violently agitates the mind. It has yet to include psychiatric illness caused by the accumulation over a period of time of more gradual assaults on the nervous system." And, in the words of Lord Oliver in Alcock at p 411, there was a requirement that "the injury for which damages were claimed arose from the sudden and unexpected shock to the plaintiff's nervous system".

There have been several medical negligence cases in this jurisdiction dealing with psychiatric illness to secondary victims. Three were focused on in the submissions made by the parties. I shall look at those three cases in some detail before briefly mentioning three other relevant medical negligence cases.

The first, which featured prominently in the submissions for the respondents, is Taylor v Somerset Health Authority . Here the primary victim died from a heart attack, suffered at work, which the defendant health authority admitted was caused by the negligent failure, over the previous months, to diagnose and treat the deceased's serious heart disease. The deceased's wife went to the hospital within an hour, and was told of his death by a doctor about 20 minutes after her arrival. She was shocked and distressed and could not believe the news. Shortly afterwards she went to the hospital mortuary and identified her husband's body, partly because she had been asked to do so, but partly because she could not believe that what she had been told was true.

Liability in the tort of negligence for the psychiatric illness suffered by the deceased's wife as a consequence of the death was denied by Auld J for two reasons. The first, which is the most important for our purposes, is that there was no "external, traumatic, event caused by the defendant's breach of duty which immediately causes some person injury or death" (P267). In this respect, Auld J earlier said that he agreed with counsel for the defendant's submission that what one needed was some external event in the nature of an accident whereas here the death was the culmination of the natural process of heart disease. It is tolerably clear from the context that the external event being referred to by Auld J was an event external to the primary victim .

The second reason given by Auld J for denying liability was that the claimant had not experienced the death through her own unaided senses. She was not present when her husband died and had first been told about the death by a doctor.

In Sion v Hampstead Health Authority [1994] 5 Med LR 170 (" Sion"), the claimant was the father of his 23-year-old son who had been injured in a motor-cycle accident. His son died in hospital 14 days later. The father had stayed at his son's bedside during that period and had watched him deteriorate in health, fall into a coma and die. He alleged medical negligence and that, as a result of that negligence, his son had died and he had suffered a psychiatric illness. His claim was struck out by Brooke J and that decision was upheld by the Court of Appeal. Staughton LJ and Peter Gibson LJ gave judgments, both of which were agreed with by Waite LJ. The claimant was held to have no realistic prospect of success because there was no sudden shocking event caused by the assumed medical negligence. Rather there was a process continuing for some time from first arrival at the hospital to the appreciation of medical negligence after the inquest; and, in particular, when the son's death occurred it was expected.

It is especially significant to the issues that we have to decide that Peter Gibson LJ, in obiter dicta, rejected a submission that one needed a shocking incident additional to the shocking injuries or death of the primary victim. Having said that Auld J in Taylor v Somerset HA had accepted a similar argument (and he was here referring to Auld J's first line of reasoning set out in para 182 above), he said at p 176:

"I am not persuaded by this argument. ... I see no reason in logic why a breach of duty causing an incident involving no violence or suddenness, such as where the wrong medicine is negligently given to a hospital patient, could not lead to a claim for damages for nervous shock, for example where the negligence has fatal results and a visiting close relative, wholly unprepared for what has occurred, finds the body and thereby sustains a sudden and unexpected shock to the nervous system."

The third medical negligence case is North Glamorgan NHS Trust v Walters [2002] EWCA Civ 1792 , [2003] PIQR P16 (" Walters "). The claimant's baby was admitted to hospital on 17 June 1996 with signs of jaundice. The hospital negligently failed to diagnose that he was suffering from acute hepatitis and needed a liver transplant. If he had received one, he would probably have lived. Instead, he was given other treatment during the week and allowed home at weekends. On the weekend of 26 July 1996, the claimant brought him back to hospital. On 30 July 1996, the claimant, who was sleeping in the same room as the baby, was awoken at about 03.00 to see and hear him having a fit. She was then told, incorrectly, that he had not suffered any serious damage as a result of the fit. He was later transferred to another hospital, where the claimant learned that he had in fact suffered severe brain damage, required a life support machine and that a liver transplant was inappropriate. The following day the claimant was told that the baby's brain damage was so severe that he would have no quality of life and would be unable to recognise his parents. The claimant and her husband were asked whether they thought it best to continue life support. Following on their decision, life support was terminated and the baby died in the claimant's arms approximately 36 hours after the seizure. The claimant sought damages for the pathological grief reaction that she suffered. In upholding the decision of Thomas J, the Court of Appeal held that the defendant health authority was liable for the claimant's psychiatric illness. The leading judgment was given by Ward LJ, with whom Clarke LJ and Sir Anthony Evans agreed.

The central legal issue in the case was whether the claimant's illness had been caused by shock and, in line with the formulation of Lord Ackner in Alcock , that required asking whether the illness arose "from the sudden appreciation by sight or sound of a horrifying event or its immediate aftermath" (Ward LJ at para 20). Ward LJ explained that the horrifying event need not be confined to "a frozen moment in time" (para 23) but could extend to a series of events making up the entire event. That was satisfied on the facts. As Ward LJ expressed it at para 35:

"In my judgment the law as presently formulated does permit a realistic view being taken from case to case of what constitutes the necessary 'event'. ... [O]n the facts of this case there was an inexorable progression from the moment when the fit occurred as a result of the failure of the hospital properly to diagnose and then to treat the baby, the fit causing the brain damage which shortly thereafter made termination of this child's life inevitable and the dreadful climax when the child died in her arms. It is a seamless tale with an obvious beginning and an equally obvious end. It was played out over a period of 36 hours, which for her both at the time and as subsequently recollected was undoubtedly one drawn-out experience."

Ward LJ considered that the application of the established law to medical negligence cases did not involve any new application of principle even if the facts were new; and that the decision here being reached did not involve taking an incremental step which advanced the frontiers of liability. Clarke LJ agreed with this, but added in his short concurring judgment that, even if it did involve such a step, he would take it.

There are three other medical negligence cases to which we were referred that I should briefly mention. Tredget v Bexley Health Authority [1994] 5 Med LR 178 was a decision of His Honour Judge White in the Central London County Court. The claimants, a mother and father, were awarded damages for psychiatric illness caused by the admitted negligence of the defendants during the labour and birth of their son, who was born with severe brain damage and died two days later. The labour and childbirth, in which significant problems were encountered which should have been avoided by the mother having a Caesarean section, were traumatic for both the mother and the father, who was present throughout. The judge rejected the submission that there was no sudden shocking event and held that, although lasting over 48 hours from the onset of labour to the death, this effectively was one event. It can be seen that this was somewhat similar to the subsequent approach of the Court of Appeal in Walters .

In contrast, in Shorter v Surrey and Sussex Healthcare NHS Trust [2015] EWHC 614 (QB) and Liverpool Women's Hospital NHS Foundation Trust v Ronayne [2015] EWCA Civ 588 , [2015] PIQR P20 , the claims for psychiatric illness consequent on seeing the illness and death in hospital of the claimant's sister in the first case and the serious illness in hospital of the claimant's wife in the second case, both failed because there was held to be no sudden shocking event. It was also reasoned in Ronayne that the event was not horrifying. In both cases, Walters was distinguished.

In terms of what we have to decide, three particular points of significance emerge from this survey of relevant medical negligence cases.

(i) The claims for psychiatric illness by the secondary victims succeeded in two cases, most importantly in Walters but also in Tredget v Bexley Health Authority . The claims failed in the other four cases mainly because there was held to be no sudden shocking event.

(ii) There is no support in the other five of the six cases for the idea, in Auld J's first line of reasoning in Taylor v Somerset HA, that there must be an event external to the primary victim (ie an accident). It would appear that in none of the six cases did the medical negligence comprise an event external to the primary victim and yet, with the exception of Taylor v Somerset HA, nothing in the courts' reasoning turned on that. So, for example, in Walters the Court of Appeal did not mention Taylor v Somerset HA and yet the decision is plainly inconsistent with Auld J's first line of reasoning because, on the facts, there was no event external to the primary victim. Furthermore, in Sion , Peter Gibson LJ in obiter dicta (see para 185 above) explicitly rejected an argument based on that first line of reasoning (although it may be said that his focus was more on the shocking, rather than the internal, aspect of the event).

(iii) In none of the six cases did anything turn on there being an omission. These were all omission cases in the sense that the medical negligence was a failure to benefit the primary victim. Yet there was no mention of this presenting any sort of problem in establishing liability for the secondary victims.

As explained in para 174 (iv) and (v) above, the Court of Appeal in the three cases with which we are concerned held somewhat reluctantly that, contrary to Chamberlain J's reasoning in Paul, Novo could not be distinguished so that it was bound by Novo to decide against the defendants. The Novo case therefore sits at the very heart of these appeals. What did it decide?

The primary victim was injured in an accident at work when a fellow employee caused a stack of racking boards to tip over on top of her. As a result she sustained injuries to her head and left foot. The defendant employer admitted its negligence for those injuries. The primary victim was apparently making a good recovery when, three weeks later, she suddenly and unexpectedly collapsed and died at home. Her sudden collapse and death were due to deep vein thrombosis and consequent pulmonary emboli, which themselves were due to the injuries that she had sustained in the accident. Her daughter, the secondary victim and the relevant claimant, did not witness the accident but she did witness her mother's death. It was not in dispute that, as a result of witnessing her mother's death, she suffered post-traumatic stress disorder. The only issue was whether the daughter was entitled to damages from the defendant employer for her psychiatric illness.

The trial judge held that she was so entitled because the relevant event that caused the psychiatric illness was the sudden death of her mother and all the relevant requirements for liability were satisfied in respect of that event, in particular because the daughter was present at her mother's death. It did not matter that she was not present at the scene of her mother's accident at work or its immediate aftermath.

The Court of Appeal (Lord Dyson MR with whom Moore-Bick and Kitchin LJJ agreed) overturned that decision and held that the daughter was not entitled to damages for her psychiatric illness. This was because the relevant event, in order to establish the necessary proximity of relationship between the defendant and the secondary victim, was the mother's accident and not the mother's death. Auld J's judgment in Taylor v Somerset Health Authority , including the need for an external event to the primary victim, was approved.

Lord Dyson gave two more specific inter-related reasons for reaching the conclusion that it would extend recovery too far to treat the relevant event as the death and not the accident. First, if one treated the event as the death, this could allow the recovery of damages even if the mother's death had occurred years after the accident. Yet if the daughter had come across the scene of the accident shortly after the "immediate aftermath" she could not recover. That distinction "would strike the ordinary reasonable person as unreasonable and indeed incomprehensible" (para 30).

Secondly, in line with Lord Steyn's approach in Frost of "thus far and no further" there should be no substantial extension of the common law. Lord Dyson said, at para 31:

"In the Frost case the House of Lords recognised that this area of the law is to some extent arbitrary and unsatisfactory. That is why Lord Steyn said 'thus far and no further' in the Frost case and Lord Hoffmann and Lord Browne-Wilkinson agreed with him. It is true that the issue in the Frost case was very different from that with which we are concerned in the present case. But that does not detract from the force of the general point that their Lordships were making. In my view, the effect of the judge's approach is potentially to extend the scope of liability to secondary victims considerably further than has been done hitherto. The courts have been astute for the policy reasons articulated by Lord Steyn to confine the right of action of secondary victims by means of strict control mechanisms. In my view, these same policy reasons militate against any further substantial extension. That should only be done by Parliament."

(1) The death is the relevant event in these three cases

In my view, the correct approach is to reject the focus in the respondents' submissions on accidents or events external to the primary victim and instead to focus in these three cases on the death of the primary victim as the relevant event. This was the approach advocated by counsel for Polmear and Purchase and was the view taken in the decision of Chamberlain J on the first appeal in Paul . It also appears to be the view favoured by the Court of Appeal albeit that it considered itself bound by Novo to reach the contrary decision.

There is good reason to treat the relevant event in these cases as the death of the primary victim not least because it was witnessing the death or its immediate aftermath that caused the psychiatric illness to the secondary victims. Moreover, it was reasonably foreseeable (treating the secondary victims as having reasonable fortitude) that they would suffer psychiatric illness as a consequence of the death. It then follows that there should be liability in these three cases because it is not in dispute that, once one treats the event as the death, all the established proximity or control factors are satisfied. That is: first, the secondary victim had a close tie of love and affection to the primary victim; secondly, the secondary victim was present at the primary victim's death or came across the immediate aftermath; thirdly, the psychiatric illness was brought about through the secondary victim's own unaided senses; and fourthly, on the facts set out in paras 154-155, 158-160, 167-169 above, the death was plainly shocking and horrific.

It is important to add that I am not suggesting that, in medical negligence cases, the only relevant event has to be a death. On the contrary, in a case where the primary victim becomes seriously ill as a result of medical negligence but does not die, an analogous approach should be applied. That is, the secondary victim who has suffered a recognised psychiatric illness caused by witnessing that serious illness should be able to recover damages in the tort of negligence provided foreseeability and the proximity/control factors are satisfied. But as we are concerned in these three cases with psychiatric illness consequent on the death of the primary victim, rather than the serious illness of the primary victim, I will put that variation to one side.

(2) Incremental development

To accept that the death is the relevant event in these cases may be said to fall within the existing parameters of the law because, as has just been shown, the existing legal requirements are all satisfied once one treats the relevant event as the death. Put another way, one might argue that the approach I am adopting does not represent any development of the law but is merely the correct application of the existing law to new facts.

However, even if the better view is that treating the death as the relevant event is a development of the common law, it is in my view a justified incremental step that falls well within the traditional judicial role in correcting errors and keeping the common law up-to-date. Indeed these appeals present a rare opportunity for this court to move the law forward, by traditional incremental development, to a more satisfactory position that is closer to what is arguably the only truly principled solution (see para 144 above).

Lord Hoffmann in Frost said at p 511:

"It seems to me that in this area of the law, the search for principle was called off in Alcock v Chief Constable of South Yorkshire Police [1992] 1 AC 310 . No one can pretend that the existing law, which your Lordships have to accept, is founded upon principle. I agree with Jane Stapleton's remark that 'once the law has taken a wrong turning or otherwise fallen into an unsatisfactory internal state in relation to a particular cause of action, incrementalism cannot provide the answer:' see The Frontiers of Liability , vol 2, p 87. Consequently your Lordships are now engaged, not in the bold development of principle, but in a practical attempt, under adverse conditions, to preserve the general perception of the law as [a] system of rules which is fair between one citizen and another."

With respect, that was a counsel of despair. There is no need to give up on the incremental development of the common law in this area. Moreover, the incremental development of the law is one that has been explicitly entrusted by the Government to the courts in this area: see para 146 above. Lord Steyn's approach in Frost of "thus far and no further" is no longer justified if it ever was (see para 148 above).

(3) Insisting on an accident (an event external to the primary victim) would needlessly deny recovery in almost all medical negligence cases

In the context of medical negligence, there will rarely be an accident, that is, an event external to the primary victim. Most situations will be ones where the primary victim is suffering from a naturally occurring illness or disease that has not been caused by any medical intervention. Even if there has been what one might class as an accident - for example, the injection of an incorrect drug that has made the primary victim's condition worse or leaving surgical equipment inside the primary victim's body after an operation - that accident or its immediate consequences will rarely be readily identifiable and observable.

Following from this, one approach, and that put forward by the respondents (see para 178 above), is to say that, subject to the rare exception where one might argue that there is an accident, there can be no liability for negligently caused psychiatric illness to a secondary victim consequent on medical negligence. On the respondents' submission, the law on secondary victims suffering psychiatric illness is confined to where there has been an accident (ie an event external to the primary victim). Heavy reliance is placed on Auld J's first line of reasoning in Taylor v Somerset HA .

Although by insisting on an accident one might be said to be applying the same legal principles to medical negligence as to other areas of negligence causing psychiatric illness to secondary victims (for example, negligent driving), adoption of that approach, as advocated by the respondents, would in practice result in a blocking off of medical negligence as an area where, subject to rare exceptions, there can be no liability for psychiatric illness suffered by secondary victims.

To block off this area of negligence is not an attractive approach where the Government has explicitly entrusted flexible development of the law to the courts and where what is arguably the only truly principled approach (see para 144 above) would lead in the opposite direction by allowing claims where there has been medical negligence.

Moreover, to adopt such an approach would contradict the authorities on medical negligence causing psychiatric illness to secondary victims. As we have seen (see para 186 above), there was no accident in Walters and yet recovery was allowed; and, with the exception of Auld J in Taylor v Somerset HA, none of the reasoning in the other relevant medical negligence cases (whether allowing or rejecting the claims) turned on whether there was an accident or not.

Not least because it is not supported by the other medical negligence cases, but also because it is applying an unnecessary restriction, that part of Auld J's reasoning requiring an event external to the primary victim (ie an accident) should be rejected. But the decision remains correct and should not be overruled because of the alternative ground relied on by Auld J: ie the claimant had not perceived the death through her own unaided senses.

There is a further linked point. What one means by an accident may be regarded as drawing a somewhat arbitrary line. Mr Maskrey defined an accident as an event external to the primary victim. However, one might alternatively look at the matter from the perspective of the secondary victim and define an accident as an event external to the secondary victim. Defined in that way, the death of the primary victim is an accident because it is an event external to the secondary victim. Therefore, an objection that can be raised against the insistence on an accident (defined as an event external to the primary victim) is by asking, what is the justification for adopting that definition of an accident and not another?; or, put another way, why do some accidents count and others do not?

(4) That one would be imposing a liability for omissions in these cases is not a valid objection

Following a request from the court prior to the hearing, counsel for both parties made brief oral submissions in relation to what one may describe as the conceptual basis of the liability owed to secondary victims for psychiatric illness caused by medical negligence. One particular possible problem is that in these three cases, in common with other medical negligence cases, the duty of care of the doctor to the patient covers omissions as well as acts. The inclusion of a liability for omissions - or, as one might more accurately put it, for a failure to benefit the claimant - needs specific justification because the general common law rule is that there is no duty of care owed in respect of a failure to benefit a claimant. See, generally, N v Poole Borough Council [2019] UKSC 25 ; [2020] AC 780 . In a standard medical negligence context, where the claim for personal injury is brought by the patient, there is no difficulty in including liability for a failure to benefit because the doctor has assumed responsibility to the patient; and the existence of an assumption of responsibility is a well-recognised exception to the general rule of no negligence liability for omissions. However, on the face of it, there may be difficulty in saying that the doctor has also assumed responsibility to the secondary victim. Does that therefore mean that recovery for negligently caused psychiatric illness to a secondary victim must be ruled out in almost all medical negligence cases, irrespective of the proximity or control factors set out in para 199 above being satisfied, because there is usually no assumption of responsibility by a doctor to a secondary victim?

Although in my view the answer to that question is clearly "no" (and it is fair to say that none of the counsel, even Mr Maskrey for the respondents, appeared to regard this issue as being of central importance to what the court has to decide) the precise reason for this is not straightforward, as I shall now seek to explain.

It is not in dispute that the essential conceptual basis of the tort of negligence for psychiatric illness caused to secondary victims rests on whether there was an independent duty of care owed to the secondary victim by the defendant. It is insufficient to show that a duty of care was owed to the primary victim and that the claim of the secondary victim is simply derived from that. This has been accepted explicitly or implicitly in all the leading cases. One of the clearest statements is that of Lord Oliver in Alcock at p 411:

"Although it is convenient to describe the plaintiff in such a case as a 'secondary' victim, that description must not be permitted to obscure the absolute essentiality of establishing a duty owed by the defendant directly to him — a duty which depends not only upon the reasonable foreseeability of damage of the type which has in fact occurred to the particular plaintiff but also upon the proximity or directness of the relationship between the plaintiff and the defendant."

This requirement for an independent duty of care may be contrasted with, for example, the Fatal Accidents Act 1976 where the rights of dependants to recover damages for financial loss or bereavement are derivative from the rights that the deceased had against the defendant prior to the death. That is, in the context of the tort of negligence, the rights of the dependants do not depend on a duty of care being owed to them by the defendant. Their rights derive from the breach of the duty of care owed to the deceased. Hence under section 1(1) of the 1976 Act, an action can only succeed if the wrongful act, neglect or default which caused the death "is such as would (if death had not ensued) have entitled the person injured to maintain an action and recover damages in respect thereof..."

That in the context of liability for psychiatric illness an independent duty of care to the secondary victim is required explains why the starting requirement, as set out in the summary in para 179 above, is the foreseeability of a psychiatric illness to the secondary victim - not the primary victim - as a person of reasonable fortitude.

The independence of the duty of care owed to the secondary victim also explains why defences that may negate the duty of care owed to the primary victim need not negate the duty of care owed to the secondary victim. For example, it would appear that the valid exclusion of liability to the primary victim need not rule out a claim by the secondary victim. It also explains why there appears to be a separate limitation period for claims by secondary victims so that such claims may not be out of time even if the claim by the primary victim is out of time.

Nevertheless, although this has rarely been analysed, it should be recognised that the claim of the secondary victim is not wholly independent of the duty of care owed to the primary victim and that there are some derivative features. In particular, at least in general terms, the standard of care is unaffected by the fact that there may be a claim by a secondary victim. For example, in a road accident case, such as McLoughlin v O'Brian , the standard of care owed to the primary victim was determinative of whether there was a breach of the duty of care owed to the secondary victim (the existence of that duty of care having been established independently). Similarly, in Alcock and Frost it would make no sense to treat independently the standard of care owed to the primary and secondary victims. Rather the standard of care owed by the Chief Constable to the primary victims was determinative of whether there was a breach of the duty of care owed by the Chief Constable to the secondary victims (the existence of that duty of care having been established independently).

Applying this to the medical negligence context, the claim by the secondary victim has both independent and derivative features. The requirement for an independent duty of care owed by the doctor to the secondary victim necessitates that the law summarised in para 179 above is applied. That requires the relevant foreseeability and the necessary proximity or control factors to be satisfied. But there are also derivative elements because, assuming a duty of care to the secondary victim has been so established, the determination of whether there has been a breach of that duty of care essentially depends on whether there has been the breach of a duty of care in relation to the primary victim.

In my view, an analogous approach should be applied in respect of the failure to benefit question albeit that that is a duty of care, rather than a standard of care, issue. Perhaps the simplest way of expressing this is that, by way of exception to the normal rule that there can be no liability for omissions, the secondary victim can derivatively rely on the assumption of responsibility exception that is being applied in relation to the primary victim. Once it has been established that, first, there was a duty of care owed to the patient, who is the primary victim, in respect of a failure to benefit the patient and, secondly, that an independent duty of care was owed to the secondary victim, applying the relevant foreseeability and proximity or control factors, there can be no objection that the liability to the secondary victim is being imposed for an omission/failure to benefit. In relation to the secondary victim, overcoming the failure to benefit objection is derivative.

This also explains why, as set out in para 191 (iii) above, in none of the relevant medical negligence cases has any objection been raised to the liability being one for an omission (ie a failure to benefit the primary victim). For example, it was held in Walters that there should be liability even though the relevant negligence comprised an omission (ie a failure to benefit the primary victim).

It may also be thought relevant that Chamberlain J in Paul referred to White v Lidl UK GmbH [2005] EWHC 871 (QB) as showing that, even in accident cases outside the medical negligence context, there is no problem about imposing a liability to secondary victims for omissions. There a crash barrier, which had not been properly maintained, came though the primary victim's car windscreen. She subsequently committed suicide and the secondary victim, her husband, found her hanging body and suffered psychiatric illness. Although the claim was struck out because that illness was not reasonably foreseeable and because the suicide was too far removed from the accident, none of the reasoning turned on the omission/act distinction. Indeed Hallett J made clear, at para 38, that a different result would have been reached (ie the claim would have succeeded) had the secondary victim witnessed the accident and suffered psychiatric illness as a consequence.

For completeness, I should add that there would be a problematic omission issue if, in the reverse of the position in the three cases with which we are concerned, there was no liability to the primary victim because, vis a vis that primary victim, one was dealing with an omission in relation to which no duty of care to the primary victim would be imposed. It would then be odd, as a matter of policy, if the secondary victim had a claim and the primary victim did not. The Law Commission gave an example of this in its report, Liability for Psychiatric Illness Law Com No 249 (1998), [1998] EWLC 249 at para 6.37:

"Where, for example, the defendant was a mere passer-by who failed to warn the immediate victim of some impending danger, the defendant would not normally owe a duty of care to the immediate victim because there is no general duty to act for the benefit of another. On the same reasoning (that there is no general duty to act for the benefit of another) the defendant should not normally owe a duty of care to a loved one who suffered psychiatric illness consequent on the immediate victim's injury."

(5) Allowing the claims would not favour a secondary victim who suffers psychiatric illness over a secondary victim who suffers a physical injury

It is clear that, in general terms, a "third party" is unable to recover for harm caused to a primary victim. So it is that, at common law, and putting to one side the Fatal Accidents Act 1976, a third party cannot recover for economic loss or mental distress consequent on the death or injury of the primary victim. This was stressed by, for example, Lord Oliver in Alcock at pp 408-411; and by Lord Rodger in D v East Berkshire Community Health NHS Trust [2005] UKHL 23 , [2005] 2 AC 373 , paras 100-107.

If one sees that as the general rule then the law of negligence on liability for psychiatric illness suffered by secondary victims is an exception to it (although the need for a separate duty of care to be owed to the secondary victim may be said to restore the general rule). But plainly it would be odd if the secondary victim who suffers a psychiatric illness is in a legally better position than the secondary victim who suffers a physical injury. On the contrary, one would have thought that one of the aims of the tort of negligence, given the modern medical understanding of psychiatric illness, would be to move the law on psychiatric illness towards assimilation with a less restrictive law on negligently caused physical injury.

However, there is no such oddity because allowing recovery in these three cases would not favour a secondary victim who suffers psychiatric illness over a secondary victim who suffers a physical injury. The difficulty in comparing the two situations is that it is extremely rare for a "secondary victim" to suffer physical injury as opposed to a psychiatric illness (leaving aside where, as in Hambrook v Stokes Brothers [1925] 1 KB 141, the secondary victim suffers a psychiatric illness which then goes on to cause the secondary victim physical injury). It follows that the law on physical injury to secondary victims has been little explored in the case law (although for exceptions, in the context of rescuers, see, eg, Videan v British Transport Commission [1963] 2 QB 650 and, in relation to infectious diseases, see Evans v Liverpool Corporation [1906] 1 KB 160 ). The rarity of physical injury to secondary victims might make it more difficult on the facts of a case to establish that the physical injury was reasonably foreseeable. But say, for example, one of the secondary victims in these cases suffered a heart attack (rather than a psychiatric illness) from seeing the death or injury of the primary victim and that was found to be reasonably foreseeable. In principle, the secondary victim would have a claim in the tort of negligence even if the law required the application of analogous proximity factors to those applying to the claim for psychiatric illness. It is therefore misleading to imagine that by allowing recovery in these cases, one would somehow be treating the person who suffers psychiatric illness more favourably than the person who suffers physical injury.

(6) That there is a significant time lag between the negligence and the death of the primary victim is not a valid objection to treating the death as the relevant event

The significant time lag between the breach of the duty of care and the death is not an objection because, even in an accident case, it is irrelevant (and is not supported by anything said by the highest court) whether or not there has been a long time lag between the breach of the duty of care and the relevant consequence for the primary victim (whether that be death, injury or imperilment).

Say, for example, a door has been negligently designed without an appropriate load-bearing joist or scaffolding has been negligently erected or the wiring of a building has been negligently carried out or a crash barrier in a car park has been negligently inspected; and that, many months later, masonry above the door falls onto X or the scaffolding collapses onto X or the faulty wiring electrocutes X or the crash barrier falls onto X . If C, the secondary victim, sees the masonry fall on X or the scaffolding collapse on X or the electrocution of X or the barrier falling on X, and suffers psychiatric illness as a consequence, it is not a bar to C's recovery that there was a long time lag between the breach of duty of care and X's injury or death.

In Paul , Chamberlain J made this point forcibly in the following passage at para 63:

"Although McLoughlin and Alcock were both cases where the negligence was close in time to the 'event', there is nothing in any of the House of Lords authorities to suggest that this must invariably be so. Lord Oliver said in Alcock at p 416 that the 'temporal propinquity' required was between the psychiatric injury and ' the event caused by the defendant's breach of duty to the primary victim' (emphasis added), not the breach of duty itself. As [counsel for the claimant] submitted, there is nothing in any of the House of Lords authorities considering the control mechanisms to suggest that a claim for psychiatric injury suffered as a result of witnessing a person's death or injury caused by (for example) the collapse of negligently erected scaffolding, or electrocution as a result of negligent wiring, would be affected by the date of the negligence. Taylor v A Novo does not suggest that it would. In that case, Lord Dyson MR made clear at para 29 that the secondary victim would have been able to recover if she had witnessed the accident with the racking boards. There is nothing to suggest that the position would have been any different if their collapse had been caused by being negligently stacked months or years beforehand."

The Court of Appeal below agreed that in principle it should not matter that there was a time gap, short or long, between the breach of duty of care and the death of or injury to the primary victim. Sir Geoffrey Vos MR said at para 80:

"Looking at the matter without regard to the authorities, it is hard to see why the gap in time (short or long) between the negligence (whether misdiagnosis or door design) and the horrific event caused by it should affect the defendant's liability to a close relative witnessing the primary victim's death or injury that it caused."

Lord Dyson MR's judgment in the Court of Appeal in Novo is open to interpretation on this time lag point. Chamberlain J at para 29 (set out at para 229 above) interpreted Lord Dyson as accepting that there could be a time lag between the breach of duty and the death or injury. On the other hand, the Court of Appeal thought that Lord Dyson appeared to have based his decision primarily on the time lag between the breach of duty and the death of the primary victim. As Sir Geoffrey Vos said, referring to Novo, a secondary victim cannot recover for psychiatric illness where the horrific event (eg the horrific death) is "a separate event removed in time from the negligence" (para 96). Underhill LJ made the same point at para 104 which has been set out in para 174 (iv) above.

Whatever the correct interpretation of Lord Dyson's judgment in Novo, the above hypothetical examples show that, in an accident case, the time lag between the breach of duty and the death or injury of the primary victim is not, and should not be, a bar to recovery. As this is not an objection in accident cases, it logically follows that the significant time lag between the negligence and the death also cannot be an objection in the non-accident cases with which we are concerned.

(7) That there is a significant time lag between the accrual of the primary victim's cause of action and the death of the primary victim is not a valid objection to treating the death as the relevant event

A separate possible objection that is similar to, but distinct from, that just considered is that there may be a significant time lag between the accrual of the primary victim's cause of action and the death of the primary victim (and hence the suffering of the psychiatric illness by the secondary victim). But again this is a flawed objection.

One reason that this is flawed is because the primary victim may not have any cause of action against the negligent defendant. This will most obviously be so where the situation is one of imperilment, as in the well-known runaway lorry case of Hambrook v Stokes Brothers [1925] 1 KB 141. In that case, there was no injury to the primary victim. Rather it was the secondary victim's assumption that her child must have been injured by the runaway lorry that caused her psychiatric illness.

Even where the primary victim has a cause of action, in an accident case, the injury caused may be latent. Let us assume, for example, that there is an accident in demolishing a building so that asbestos is negligently disturbed or there is a leak from a nuclear power station. The accident may not cause an observable injury or illness (eg mesothelioma or cancer) to the primary victim for many years. The tort of negligence recognises that a cause of action may accrue even though the claimant does not know and could not reasonably know that he or she has a cause of action because the injury is latent (see, eg, Cartledge v E Jopling & Sons Ltd [1963] AC 758). If a secondary victim suffers a psychiatric illness consequent on seeing the primary victim's horrific death or illness from mesothelioma or cancer - and even let us assume that the secondary victim is present when it is accepted that the serious illness first manifests itself - it cannot be relevant, as a matter of principle, that, subject to the claim of the secondary victim being time-barred by a limitation period, the primary victim's cause of action as a result of the accident accrued many years previously. As that is the position in relation to an accident case, it logically follows that the significant time lag between the accrual of the primary victim's cause of action and the death of the primary victim cannot be an objection in the non-accident cases with which we are concerned.

(8) Novo should be overruled

The facts and decision in Novo have been set out at paras 192-197 above. With great respect to a master of the common law, Lord Dyson MR's reasoning in Novo is flawed for at least the following reasons.

First, he applied the "thus far and no further" approach. However, that approach was put forward by Lord Steyn in Frost at a time when the courts saw the development of this area of the law as better achieved by legislation. As I have explained in paras 146 - 148 above, the explicit response of Government to the Law Commission's report has passed the baton back to the courts to develop this area of the law.

Secondly, Lord Dyson considered that the ordinary reasonable person would find it unreasonable and incomprehensible that a person who suffered psychiatric illness by coming too late to an accident causing injury or death could not recover, while a person who witnessed the later death from the accident could recover. However, there is a clear distinction between the two situations because, if one treats the accident and the death as separate events, the person who can recover is the same person in both situations ie the person with a close tie of love and affection, provided that that person witnesses (and is therefore closely proximate to) either the accident or the death. One is therefore treating the two situations in the same way and there is nothing unacceptable or irrational about so doing. In any event, applying the same hypothetical logic as Lord Dyson, I would suggest that the ordinary reasonable person would find it unacceptable and incomprehensible that, varying the facts of Novo, a daughter who witnessed the initial injury to the primary victim and suffered a psychiatric illness could recover whereas a daughter who witnessed the horrific death of the primary victim a few weeks later and suffered a psychiatric illness could not recover.

Thirdly, there was inconsistency in Lord Dyson's approach. This is because he accepted Auld J's reasoning in Taylor v Somerset HA on the need for an accident/event external to the primary victim and yet suggested that the Court of Appeal's decision in Walters could be distinguished. But if one applies Auld J's approach, Walters should have been regarded as wrongly decided.

Fourthly, it is be noted that, although Lord Dyson at para 30 in Novo appeared to be concerned, as one of the principal reasons for his decision, with the potentially long time lag between the negligence or accident in that case - the falling of the stack of racking boards - and the death of the mother, he did not consider other situations, such as the hypothetical examples given in paras 228 and 235 above, where the death or injury is also not more or less instantaneous with the negligence or accident.

In my view, for these reasons, the Court of Appeal's decision in Novo was incorrectly decided. Although on the face of it, one could distinguish it from these three cases, because it was not a medical negligence case, that would be a superficial and unprincipled distinction. If one is treating the death as the focus in medical negligence cases, so that the secondary victim who witnesses the death, can recover, it would be inconsistent to deny recovery where the secondary victim witnesses the death where there has been an earlier accident.

One might alternatively try to distinguish the cases with which we are concerned from Novo on the basis that, in that case, there were two shocking events, the initial accident and injury and the subsequent unexpected death, whereas in our cases there was only one shocking event, namely the death. This featured in Sir Geoffrey Vos MR's analysis of Novo and was the basis on which Chamberlain J sought to distinguish Novo . But while one can understand why Chamberlain J was trying valiantly to distinguish Novo, because he was otherwise bound to apply it, it seems faintly absurd to say that the three cases with which we are concerned are stronger cases for allowing recovery because the claim is in respect of only one shocking event namely the death, whereas in Novo the claim failed because there was more than one shocking event. Why should one allow recovery where there has been just one shocking event, while denying recovery where the relevant psychiatric illness is consequent on a second shocking event? The correct position is that it should not matter how many shocking events there have been because that is irrelevant to the principled outcome.

I therefore agree with the Court of Appeal below that Novo cannot be distinguished. As it cannot be distinguished and as there are good reasons for regarding it as incorrectly decided, it should be overruled.

(9) Conclusion

For all these reasons, it is my view that the relevant event in these three cases is the death of the primary victim. Once that is accepted, the claimants are entitled to succeed because foreseeability and the control or proximity factors are all satisfied.

(1) Academic writing and comparative law

In preparing this judgment, I have been assisted by academic writings, in particular, Peter Handford, Tort Liability for Mental Harm, 3 rd ed, (2017) especially chapter 22; Stelios Tofaris, "Limping into the future: negligence liability for mental injury to secondary victims" [2022] CLJ 452; and Imogen Goold and Catherine Kelly, "Time to start de Novo: the Paul, Purchase and Polmear litigation and the temporal gap problem in secondary victim claims for psychiatric injury" (2023) 39 Professional Negligence 24.

I have also looked briefly at the relevant law in Australia, New Zealand, Canada, the United States and Singapore. However, significant differences have emerged between the different jurisdictions (for example, in Canada - see, eg, Saadati v Moorhead [2017] 1 SCR 543 - it would appear that the approach in the leading House of Lords cases has been largely rejected; and there are legislative provisions in respect of tort liability in Australia set out in various Civil Liability Acts). In my view, it is therefore difficult and potentially misleading in this area to seek to draw lessons from the legal position in other common law jurisdictions.

(2) The need for the event to be shocking and horrific

Mr Maskrey submitted that an unfortunate consequence of treating the relevant event as the death would be that the court would face the invidious task of having to differentiate shocking horrific deaths from non-shocking and non-horrific deaths. Applying the established control mechanisms, only the former would trigger liability. But the courts already have to differentiate between treating some events as shocking and horrific and others as not being so. In my view, what this purported objection in truth raises is the question whether the requirement for the event to be shocking and horrific is a justified controlling factor. It can be strongly argued that it is not (although it may help in establishing that the psychiatric illness was reasonably foreseeable in a person of reasonable fortitude and also in establishing causation). This requirement may be said to derive from the early terminology of "nervous shock" and the focus on post-traumatic stress disorder. In relation to that type of psychiatric illness, a shocking event will often be necessary, factually, in establishing that the illness has been suffered. However, for other types of psychiatric illness the establishing of the illness is not dependent on there being a shocking event. All that matters is that a recognised psychiatric illness has been suffered.

The Law Commission considered the arguments for and against retaining this requirement and came down firmly in favour of recommending its legislative removal: see Liability for Psychiatric Illness , Law Com No 249 (1998), [1998] EWLC 249 paras 5.28 - 5.33. It is a separate question whether the common law should be developed by removing this restriction. These appeals have not been concerned with the pros and cons of such a development. We have had no submissions on this issue. Nevertheless, I agree with the thrust of what Lord Leggatt and Lady Rose have said about this being an unwarranted and unnecessary requirement (see their judgment at paras 71-78).

(3) The impact on the NHS

Mr Maskrey also fleetingly submitted that a consequence of treating the relevant event as the death would be that this would unacceptably increase the burden of legal liability on the NHS. That is not the type of socio-economic policy argument that the courts are well-equipped to assess and, at least as presented, it cannot outweigh the reasons of principle and legal policy that I have set out for allowing these appeals. It should also be borne in mind that primary victims (or, if they have died, their estates and dependants) are already likely to have claims for the medical negligence in question although, of course, allowing claims by secondary victims will inevitably increase the overall quantum of compensation that the NHS may be liable to pay for any particular negligent act or omission.

(4) The judgment of Lord Leggatt and Lady Rose

I have read with admiration the judgment of Lord Leggatt and Lady Rose. It will be apparent that the fundamental disagreement between us is that, in my respectful view, it would be an unwarranted backward step to insist that there must be an accident (in the sense of an event external to the primary victim) in order for there to be recovery for negligently caused psychiatric illness by secondary victims. Turning the clock back in this way would require, as Lord Leggatt and Lady Rose acknowledge (see paras 121-122), the overruling of Walters and a departure from the reasoning in almost all of the reported medical negligence cases in this area. Indeed, at para 123, they have left open for another day whether there can be liability even where there has been a medical accident (ie medical negligence comprising or causing an event external to the primary victim as in the examples set out in paras 185 and 205 above). In future, and subject to possible rare exceptions, the approach of Lord Leggatt and Lady Rose will mean that recovery for negligently caused psychiatric illness by secondary victims will be closed off in medical negligence cases.

For all these reasons, I would allow the appeals in these three conjoined cases.

LORD CARLOWAY ( with whom Lord Sales agrees ):

I agree with the judgment of Lord Leggatt and Lady Rose as based on English common law. However, the decision is likely to have considerable persuasive influence on Scots law in similar circumstances. It is therefore appropriate to explain that there is a difference in the two systems in relation to the right to claim damages caused by the death of another person. In contrast to the general rule in English law that "the death of a human being cannot be complained of as an injury" (judgment para 2), Scots law has, from ancient times, allowed claims by close relatives (spouses, ascendants and descendants) in respect of the death of another. This may have derived in part from the action of assythment, under which there was compensation payable for the homicide of, or injury to, a close relative. By the end of the eighteenth century it, at least in relation to non-criminal acts, had become part of the general law of reparation, notably quasi delictual negligence under the action injuriarum ( McKendrick v Sinclair 1972 SC (HL) 25 , Lord Reid at 53; Eisten v North British Railway Co (1870) 8 M 980, LP (Inglis) at 984). A close relative could claim both solatium (an amount representing pain and suffering of the relative) and loss of support from the wrongdoer. This was not a derivative action but an independent right vesting in the surviving close relatives. As with the Fatal Accidents Act 1846, and its successors, in England & Wales, this area is now governed in Scotland by statute. The Damages (Scotland) Act 1976 abolished assythment (section 8), but continued to permit claims by relatives for "Loss of Society" and support (section 1; see now Damages (Scotland) Act 2011, section 4).

Nothing turns on this speciality. Had Scots law been applied, the same result in relation to the present claimants would have been reached. These claims are not for solatium, loss of society or loss of support consequent on the death, but compensation for psychiatric sequelae occurring as secondary or indirect injury to the surviving relatives. As a generality such damage is irrecoverable because the person injured is not within the area of danger which the wrongdoer has in contemplation ( Bourhill v Young 1942 SC (HL) 78 , Lord Thankerton at 83, Lord Russell at 85-86). Exceptions to this generality have been made in the trilogy of cases, to which reference has been made, culminating in Frost v Chief Constable of South Yorkshire [1999] 2 AC 455 . These apply only to close relatives who were at the scene of an accident or who came across its immediate aftermath. It was not suggested that these exceptions should be reviewed.

At the risk of adding unnecessarily to the principal judgment, the key feature of these exceptional cases, in which recovery is permitted, is that the claimant is present at the scene of an accident or its immediate aftermath. There must be an accident to be witnessed. At the core of any claim is the need for the claimant to demonstrate that the defendant breached a duty which was owed to him or her. This depends upon both reasonable foreseeability of damage and proximity between the parties ( Taylor v A Novo (UK) [2014] QB 150 , Lord Dyson MR at paras 24 to 29).

As it was put in Young v MacVean 2016 SC 135 , Lord Brodie, delivering the opinion of the court, at para [5]):

"Not every adverse consequence of an act or omission, which, from some perspective or another, can be described as wrongful ... gives rise to a claim for damages. That is so independent of questions of foreseeability and causal connection. The law sets limits beyond which adverse consequences will be regarded as too remote from the relevant wrong to give rise, on the one hand, to a right of action and, on the other, to an obligation to make reparation. In order for the relevant right and the correlative obligation to arise, the loss must be caused by the wrong and it must have been reasonably foreseeable that the wrong would cause the loss but, in addition, there must be what is usually described as a relationship of 'proximity' between the person who suffers injury and loss and the wrongdoer. Thus, a particular wrong may cause loss to a number of persons but only those who can establish the requisite relationship of proximity with the wrongdoer will fall into the class of victims who have a claim for damages against the wrongdoer. Where the relevant wrong is a careless act or omission then the issue can be framed in terms of whether the victim was within the ambit of such duties of care as were owed by the wrongdoer. For there to be a duty of care owed by the wrongdoer to a particular injured person there must be a relationship of sufficient proximity between them. If the relationship is too remote then there is no duty and therefore no liability in the event of injury, even although injury has been caused by the wrongdoer's act or omission."

The issue in the present cases becomes one of whether the doctor owed the patient's relatives a duty to prevent them from suffering harm as a result of witnessing the death of their relative; itself caused by medical negligence. Again, for the reasons given in the principal opinion, no such duty exists on the facts of these cases. I too would dismiss the appeals.