Mugweni v NHS London

Though as a matter of principle, the question in this as in any case of alleged negligence is whether on the balance of probabilities a defendant’s breach of duty has caused or materially contributed to damage to the claimant, this action for damages falls to be determined largely by a finding as to which of two events, each of which is quite unlikely to have occurred, caused brain damage to a child of 19 weeks of age, with the serious consequence that she has suffered from asymmetrical spastic quadriplegia, epilepsy, and some serious learning difficulties.

If the event is as the claimant contends, I have in addition to resolve the question whether (and if so, to what extent) the damage resulting from that event was caused or contributed to by the breach of duty alleged.

The child, Grace, is now a patient of 27 years old, having been born on 28^th. March 1983. She claims that her disabilities have been suffered as a consequence of the negligence of an anaesthetist, Dr. Hasbury, in connection with open heart surgery performed on her at Guy’s Hospital on 9^th. August 1983. It is said that after her sternum (which had been opened to permit access to her heart, so that a hole in the atrial wall of her heart could be sewn up) was closed he failed to recognise the signs that air was collecting in the pleural space, so as to cause a tension pneumothorax which led to a cardiac arrest.

The defendant, NHS London, argues that the relevant disabilities are the consequence of brain damage sustained as the result of Grace having undergone an operation which involved time on cardio-pulmonary bypass.

As will already be obvious, the critical events happened over 27 years ago. As a consequence, neither the cardio-thoracic surgeon, Mr. Yates, who performed the operation, and in respect of whom no criticism is made, nor Dr. Hasbury is available to give evidence even if they could now have recollected what occurred. Both have died. Nor has any evidence been available from those others (save one) who it must reasonably be assumed participated in the operation and immediate post-operative events – the perfusionist, a surgical registrar, trainee anaesthetist and/or anaesthetic assistant, and theatre nurses at least. The recollections such as they are of the one exception to this simply do not assist. As a further consequence, documentation which would have been illuminating is no longer available: in particular, the perfusion records, anaesthetic charts, ECG readings and other intra-operative records. All that remains are the clinical notes, including a typed operation note made by Mr. Yates, it must be assumed immediately after the operation.

Given the paucity of records, to establish what actually happened must necessarily involve a process of reconstruction. This is best approached, in my view, by focussing first on what is known to have happened, before considering the physiological and medical plausibility of explanations advanced as to why those events should have occurred, and the time-frames within which they did so. Many of the facts which underpin those explanations have had to be assumed, and may therefore not be accurate. Although it is in general a sound principle to evaluate the probabilities of each side’s case with regard not only to the evidence brought before the court but also to the evidence which it is within the power of one party to produce and the other to refute, the lack of evidence here cannot be regarded as the responsibility of either party. Neither party is entitled to benefit from the shortcomings of the evidence (except that the burden of proof has to be established on at least some evidential foundation and, if it is not then in that event, but that event only, the defendant may benefit from the lack of material). Although the claimant could have brought her case earlier, I have no sense here that adventitious advantage has been gained by her not so doing. I have to judge the case on the evidence as it is, making the assumptions that appear reasonable, but remembering always that it is not possible with the passage of time to give point accuracy to supposed facts where the truth will fall within a range (such as the time it took for such and such to have occurred).

Care must be taken with the impact of the balance of probabilities test. In a case such as this, the establishment of one supposed fact may be a matter of debate, and if it is to be established depend on whether another fact, also debatable, is established on a balance of probability. To routinely apply a test that the first fact is established if it is more likely than not, given that the second is established, is potentially to ignore the uncertainties which affect the establishment of the latter: a 60% likelihood that X is the case, and if X is taken to be established, a 60% likelihood that Y is then the case gives mathematically the result that Y is only 36% likely to have occurred. A mathematical approach of this nature is artificial: but the example indicates that what I must have regard to in applying the burden of proof is whether on the balance of probabilities the Claimant’s case is made out, and that will involve answering that question globally having regard to all the degrees of certainty and uncertainty reflected in the particular pieces of evidence that bear upon it, and the clinical experts’ assessment of them, rather than examining each step said to take me to that conclusion and applying the burden of proof separately at each point along the way. The evidence has to weighed, having regard to the whole.

Finally, by way of general introductory remark, most of the experts have struggled to understand why the brain damage which occurred should have happened in the circumstances in which it apparently did. There must have been some cause. A number of possible explanations have been considered, and one by one rejected, until only two remain as potential candidates. This process has been fairly recent. The puzzlement of many of the experts is reflected by their casting around for, or finding, other material which tends to support their supposition – though it may do so only thinly; and often by gradual evolution in their thinking. I do not criticise this process, which has continued on occasion even in the witness box in the light of cross-examination, because of the underlying difficulties in the case: it has been reflected in my own reaction to the evidence, which has swung on a number of occasions between the two remaining possibilities (and also in considering whether the burden of proof has been established), such that I can advance my eventual conclusions only with trepidation.

Sufficient records remain to show that shortly after her birth Grace was diagnosed with a heart murmur. She suffered some breathlessness. Her mother thought her floppy, especially by comparison with her two older half-siblings. She thought that Grace might have difficulty hearing. Vomiting, cyanosis, shortage of breath and the production of frothy white sputum developed such that she was admitted first to a local hospital and from there on referral to Guy’s, suffering from “left heart failure which is getting worse”. Episodes of this were becoming more frequent and alarming. She was by then a sick child. She had had episodes of facial weakness. At Guy’s she had an echocardiogram and catheterisation studies on the 7^th August 1983 which showed the presence of an atrial septal secundum defect (“ASD”, a congenital malformation consisting of a hole in the wall between the upper pumping chambers of the heart, allowing oxygenated blood returning from the lungs to the left atrium to intermingle with venous blood returning from the systemic circulation to the right atrium) and a patent ductus arteriosus (“PDA”, an open tube connecting the arterial and pulmonary circulations which has a useful function in utero but which normally closes naturally after birth).

Surgery to close off the PDA and ASD was performed by Mr. Yates, an experienced consultant cardio-thoracic surgeon, assisted by Dr. Sayin. Dr. Hasbury, a consultant anaesthetist, was responsible for the anaesthesia. Others, as described at paragraph 5 above, would almost certainly have been present.

The operation and its immediate aftermath are central to this case. The evidence of Dr. James, expert consultant anaesthetist, was that an open heart operation of the type performed would have taken approximately 4 to 4½ hours from its start to the end of anaesthesia. This was not challenged directly, though the initial evidence of Professor Hatch, expert consultant anaesthetist had been that it would take 2 hours (this fairly rapidly became 2-3 hours under cross examination, one of the reasons why I preferred the evidence of Dr. James on matters on which they differed). There would be around half an hour to 45 minutes of anaesthetic preparation, before the skin would have been draped. The drapes would probably not have been removed before the child was placed in the incubator following surgery (this is one of those findings which depends upon reconstruction from the expert’s recollections of what their, and general, practice was in 1983, so I reach it cautiously). The surgeon’s part in the operation would have begun with opening of the chest. The sternum would have been split, and the heart underlying it exposed. The PDA would have been ligated first, before the heart was put on bypass, so that as little time was spent on bypass as possible since it was and is well recognised that being on bypass carries particular risks. The heart would have then been put on bypass, the heart itself held in ventricular fibrillation so that it did not pump, and circulation maintained by a bypass machine. This maintains a steady flow, which differs from the pulsatile nature of normal circulation. Bypass would have lasted for between 30 and 40 minutes (Professor Hatch initially volunteered 15, but this rapidly became 30: and this point is of significant importance), though if anything I think nearer 30 than 40. The difference may depend on the speed of the particular surgeon. Mr. Firmin (consultant paediatric cardio-thoracic surgeon) first suggested 30 minutes or less. He was taken through the various steps along the way by Mr. Block QC for the Defendant. For each, he agreed the step and the probable range of time in minutes it would have taken. Added together, the least time in total these would have taken, analysed in this way, would have been just over 35 minutes, followed by either a gradual restart of the heart to wean it back into full pumping mode after the fibrillation current was withdrawn, or an immediate return to circulation (the choice depending entirely upon surgical preference, it was thought). If gradual, this could have added 5 minutes. After considering these timings Mr. Firmin accepted that the average length of time on bypass for such an operation in 1983 would have been 30 – 40 minutes. There may be a degree of unreality about breaking a familiar process into its component parts in order to assess the time which it might take overall, which I have to bear in mind: but I was left with the clear view that on the best available current estimates of the length of time it would have taken (Hatch 15 then 30; Firmin 20-30, then accepting 30-40; James 30 or so “at the quickest”, up to 40 if the surgeon was slower; Stark, paediatric cardiac surgeon, agreeing with Firmin’s eventual position though allowing a few more minutes) that more than 30 minutes – close to 35, though not in excess of it - was most probably spent on bypass in Grace’s case. This is a significant finding, as will appear below, and it must be recognised that though justified on the evidence it may well be in error, and cannot be regarded as certain.

When necessary during and immediately after the operation, though obviously not at those times when she was on bypass, Grace would have had her lungs ventilated. This would largely have been mechanical positive pressure respiration, provided by machine, but periods of hand ventilation would have been used. This would normally in 1983 have involved the use of a squeezable bag, connected via an Ayre’s T-piece to the ventilation tube, operated by the hands of the anaesthetist squeezing the bag. It is likely that at the point of coming off bypass, and again just before transfer to an incubator, manual ventilation would have been used and the machine disconnected. This finding – again, one which has a penumbra of uncertainty around it, and which assumes matters had progressed that far before the events about to be described – is also one of importance, when evaluating whether the anaesthetist should have detected what was about to occur.

I accept that before Grace came off bypass, her circulatory status would almost certainly have been checked (by both surgeon and anaesthetist). Haemostasis was assured. There would have been no bleeding. There was no reason for anyone to think that at that stage there was any compromise to the blood flow to her brain.

After Grace would have come off bypass, a period of time would have passed before she was to be placed in an incubator, during which her sternum was closed, and then the layers of skin over it, until the chest was closed. If that had been completed without incident, Grace would then have been placed in an incubator and quickly taken to the intensive care unit where mechanical ventilation could be resumed as necessary. However, at some stage before that process was completed (it may never therefore have run its full course) there occurred the cardiac arrest which led to the claim. If this process had run its full course, it would have taken no more than 20 minutes from the completion of sternal closure, and possibly less. (Professor Hatch’s evidence was that closure of the skin, after the sternum, would take between 10 and 20 minutes, and applying dressings after that would have taken around a further 5. James’ estimate was longer: 20 minutes for closure of the skin, and about 5 for dressings. Stark and Firmin agreed in their note of expert discussion that 10 – 20 minutes for chest closure was more likely than 20-30 – it would have been around 30 minutes in total from coming off bypass to being put in the incubator. The finding I have made above accepts that sternal closure would have taken around 10 minutes).

At some point – as to exactly when this was between the time of closure of the sternum and finishing applying the dressings there is no evidence – Grace suffered a “cardiac arrest”. Even this phrase is uncertain: Dr. James referred to 3 possible meanings: asystole (the heart stopping beating altogether); cardiac arrhythmia (such as fibrillation, with the heart twitching but pumping nothing); or a very low output, though still some ineffective pumping action. Dr. Shinebourne – a cardiologist – thought this latter did not do proper justice to the word “arrest”, and linguistically he is certainly right: though what matters is how that phrase was intended by those who used it to describe what had happened. That description comes contemporaneously from two sources: the surgeon’s note, in respect of which I was told practice generally was such that it would have been dictated immediately after the surgery, and a handwritten note in the clinical records.

Mr. Yates’ typewritten note reads:

“…..After venting all air from the heart and closing all incisions the heart regained sinus rhythm spontaneously on removing the fibrillation current. On discontinuing bypass the heart took over the circulation satisfactorily and post-op pressures revealed a right ventricular pressure of 35mmHg with no evidence of significant venous arterial Oxygen saturation step up across the right heart.

Having ascertained there was satisfactory haemostasis the pericardium was closed and the chest closed routinely with retrosternal and peri-cardial drains.

As the patient was being prepared to be moved to the incubator from the operating table cardiac arrest occurred and external cardiac massage was necessary.

The chest was re-opened under massage conditions and it was confirmed that there was no tamponade but the cause of the arrest was a right tension pneumothorax. The tension pneumothorax was relieved and a right pleural drain inserted by when the heart had regained normal sinus rhythm and a good circulatory status. At no time was there any significant period of circulatory arrest.

The chest was then re-closed routinely with retrosternal and pericardial drains.”

The handwritten note says:

“Proceed

Ligation of large PDA + suture closure of low 2^o ASD

Difficulty on coming off bypass because of R.sided pneumothorax + ET tube blocked with secretions (Patient arrested)…”

It contains a diagram showing the position of chest drains, under which was written words which I interpret as “needs isoprenaline”. That is a drug which has the effect of increasing both the rate at which a heart beats and its contractility, so that with each beat it may pump out more blood.

The day following an ultrasound was taken of the brain which was reported as normal. The same day (or, it may be, a day later: there is evidential support for each) Grace was noted to be suffering from epileptiform fits affecting the left arm and both legs. The rate of growth of Grace’s brain flattened. Although there was some evidence, produced by analysis of a chart plotting her head growth against normal, that the rate of growth of her head since birth had already dropped in percentile terms before the operation, from being on the 50^th centile to being no higher than the 9^th centile at operation, this flattening of growth was additional, referred to in evidence by one expert as indicating a “second hit”. Poor head growth may be an indication of damage to the brain.

The encephalopathy indicated by the fits has been shown by neuro-radiological imaging to be caused by damage to the watershed (or “border zone”) areas of the brain. These are the areas lying between the distribution of one cerebral artery and its peripheral vessels and the next. It is not difficult to understand that if blood flow to the brain drops for long enough, such that oxygenation of the tissues might be affected, these areas will be the first to be affected. An analogy thought by the experts to be useful is that of a garden sprinkler. The area of lawn between two sprinklers will be perfused so long as the water pressure is sufficient. If it drops, then the spread of water diminishes, and part of the lawn is parched and may die.

This pattern of injury is to be distinguished from the pattern shown where there is an acute hypoxic insult – as in a catastrophic heart attack, where there is sudden anoxia. In such a case, the deep grey matter will be affected (as may also be the borderzones, but damage limited to the borderzone area is not characteristic of this type of event). It is also to be distinguished from an haemorrhagic insult, as it is from tissue death caused by either a thrombosis or embolism – both these latter cause focal brain damage within the distribution of the blood vessels concerned, not more generalised damage caused by hypo-perfusion between the distributions.

The brain of a neonate is at a developmental stage. It is recognised that neonates may suffer from peri-ventricular leucomalacia (“PVL”) as a result of low blood perfusion. The damage is to cells which manifests itself in a particular location – hence the name. Dr. Rennie’s evidence, which I accept on this, was that in the process of brain maturation the white matter affected by PVL is in the course of its development in a watershed or borderzone area (Footnote: ¹): hence one would expect of a pre-term child that hypoperfusion would cause PVL rather than the injury described in Grace’s case as of a watershed type. Her thesis (to which I shall come) was that for around a month after term the neonate suffering from congenital heart disease also displays PVL as a manifestation of brain damage caused by under-perfusion, and that studies which show that the effects of apparently uneventful open heart surgery on neonates may well be such that they suffer from PVL demonstrate an under-perfusion injury, which if it occurred in an older infant (neonates being aged less than 28 days from birth, infants less than a year) would manifest itself probably as watershed damage of the type seen in Grace.

There was tentative agreement between the experts that during operation in 1983 there would have been no continuous monitoring of Grace’s blood gases and thus the level of oxygen saturation of her blood, but that there would have been monitors which would show her arterial and central venous blood pressures, and an electro-cardiograph. It is unclear what precise type of ventilator would have been in use, and whether or not it would have had a dial to show pressure (or, for instance, whether it may have been set to deliver a predetermined volume) but the evidence suggests this is more likely than not. Whatever the monitors, no records of readings from them now remain, though there is one reference (to blood pressure in the right atrium following the successful intervention after the cardiac arrest) to a recorded value, which appears consistent with recovery of the circulation at that point.

It is highly likely that many of Grace’s functional difficulties arose as a result of peri-operative events (this phrase covering either the consequences of being on cardio-pulmonary bypass, or suffering a tension pneumothorax which was unrelieved to the extent it caused cardiac arrest, or a combination of the two). However, it is undoubtedly the case that she also suffers, and suffered, from CHARGE syndrome, which is congenital. (This acronym stands for (C) colobomata (slits in the retinae and choroid of the eye), (H) heart malformations, (A) atresia (narrowing or blocking) of the choanae (channels at the back of the nose), (R) retardation of growth, (G) genital abnormalities, (E) ear abnormalities). It is well established that the syndrome has additional features, which may include specific behavioural problems including autistic like behaviour. In the event I did not hear from the geneticists who had reported to the court, because the parties agreed that the hearing loss, visual loss due to colobomata, autistic behaviour, and the delayed puberty from which Grace suffers are due to CHARGE, that 30% of her cognitive impairments and learning difficulties should be treated as attributable to CHARGE; but her spastic quadriplegia, epilepsy, any visual loss due to cortical damage, the effect of those three on her autistic behaviour, scoliosis if proven to be present and of clinical significance, and 70% of her cognitive impairments and learning difficulties should be treated as attributable to damage sustained peri-operatively.

I take, therefore, as a given that some of Grace’s present functional disabilities, which have a real significance for her everyday life, are a consequence of watershed brain damage, which classically is caused by under-perfusion of the brain.

In the foetus, this process would require around 30 minutes of damaging hypo-perfusion, following an hour of non-damaging hypoxia. On this, all the medical experts (and in particular, Drs. Rennie, Miles and Professor Mitchell) were agreed on paper. During the hour, described by some as a “priming period”, the resistance of the brain to injury is slowly and gradually overcome. A neonate and infant (as with older people) generally has the capacity to auto-regulate blood flow to the brain. Thus, if pressure of supply to the brain (arterial pressure) (on which flow and hence perfusion centrally depends, assuming that the resistance of the brain blood vessels is constant, though the evidence was that it also depends on volume and the extent to which oxygen is carried by the blood at the time) drops, the blood vessels in the brain will dilate, so that a flow of an equivalent amount of oxygenated blood to provide the necessary nutrients for brain function is maintained. Volume is increased with slower flow; and resistance is also reduced, maintaining sufficient oxygenation. Similarly, if pressure increases (as, in an adult, where there is significant physical exertion) the blood vessels will constrict, to the reverse effect. The automatic response is not simply to pressure. An increase in the concentration of carbon dioxide in the blood, for instance, will also cause dilation of the blood vessels, to increase cerebral flow.

The postulated half hour is that during which the damage occurs. Once a cell is damaged, it will die. In addition to the brain’s capacity more generally to auto-regulate blood supply, the cells will have reserves of nutrient, but in the absence of sufficiency of supply this is only some, though little, further defence.

Critically for the present case, the existence of watershed damage on its own shows there has not been an acute anoxia, as in the case of a damaging cardiac arrest. In such a case, the deep grey matter of the brain and basal ganglia would be affected. Here, they were not. Accordingly, however deep and long there may have been under-perfusion of Grace’s brain, from whichever cause, the damage was not caused by her cardiac arrest. This does not exclude the lack of flow during cardiac arrest and resuscitation contributing to the damage, by adding to a process of hypoxic-ischaemic damage which was established at the time, at least if Dr. James’ third meaning of “cardiac arrest” is accepted -though possibly in respect of his first or second too - since a lack of flow is by definition less than low flow for the time it persists. However, any decision whether there has probably been such damage must be heavily influenced by the surgeon’s own words “At no time was there any significant period of circulatory arrest” and the observation that once the pneumothorax was relieved by re-opening the chest the heart “regained normal circulatory rhythm and a good circulatory status”, both of which argue against it. Further, Dr. Rosenbloom (whose evidence on this I accept) told me that watershed damage is not progressive in extent even where hypoxic-ischaemia persists, unless the ischaemia worsens. To return to the analogy of the garden sprinkler: if low pressure causes the spray to fall short of an area, it will still perfuse the lawn within the area the spray still covers, and the reduced flow will not widen the belt of damaged lawn unless the pressure drops further, to the extent that the formerly (slightly) wet areas become dry.

Grace was not typical of those to whom the general parameters of an hour priming time and thirty minutes damaging period might apply. She was nearly five months old. She was a sick child. The evidence of Drs. Anslow and Kendall, neuroradiologists, was that on MRI scan in 2003 there was evidence of unusual features of her brain. There are nodules entering the ventricles. A CT scan taken on 22^nd. August 1983 (therefore less than 2 weeks after surgery) was reported as showing bilateral atrophy of the brain. There was disagreement whether the nodules were properly classed as heterotopia (developmental abnormalities), and whether the atrophy was consequent upon the hypoxic-ischaemic insult which caused the watershed damage or pre-existing (though the scan taken a day post-operatively was reported as normal, atrophy takes some time to develop if consequent on damage, and its presence might have been masked by the brain swelling which inevitably would have been present post-operatively). Insofar as necessary to do so, I would favour Dr.Kendall’s view that the atrophy represents the pre-operative state, because of the limited time post-operatively for the brain to atrophy, and the implications of the limited head growth where the circumference had dropped from being charted as at the 50^th centile at birth to a level which measured a few days after the surgery was between the 2^nd and 9^th centile – importantly, though, when plotted back to the growth line on the chart, this showed the head would have been around the 9^th centile at the time of the operation if there had been no growth at all since its date. The relevance of this, coupled with the medical history with which Grace presented for surgery, is that it emphasises that she was not a normal child at the date of operation. The point is underlined by noting that on the evidence before me it is rare for infants suffering from ASD to be treated operatively at such a young age. Her condition was unusually serious of its type, in her particular case.

Indeed, a feature which complicates analysis of the material placed before the court is that most children who have congenital heart disease either suffer from conditions which require urgent operative treatment within a week or two of birth if the baby is to survive (such as transposition of the great arteries, (“TGA”) where the pulmonary circulation usually supplied by the right ventricle is supplied instead by the left, and the systemic circulation by the right, or hypoplastic left heart where the pumping chamber of the left ventricle is atrophied) or are treated conservatively or palliatively until the child is a few years old, as is usual in ASD. But here, Grace’s condition was judged to be such that she needed operative treatment though not yet quite five months old.

The starting point – that which can with reasonable confidence be asserted as fact – is therefore that Grace, a sick and atypical child, suffered brain damage as a result of the hypo-perfusion of her brain. She did not suffer acute consequences from her cardiac arrest.

It is also certain that she suffered a tension pneumothorax.

The experts called to give evidence had initially and variously ascribed damage to the cardiac arrest itself, or (as suggested by Mr. Block QC to Professor Mitchell in his case) possibly simply to hypoxia rather than hypoxic-ischaemia (Footnote: ²), of which the critical additional component is poor flow; or (as Mr. Stark did) contemplated whether the heart stopped because of arrhythmia or for other reasons, but in the end the evidence was clear that the tension pneumothorax was the most probable cause of the arrest and potentially could (if it had lasted for long enough – for the half hour following the hour long priming period postulated for a normal child) have caused hypoxic-ischaemic brain damage.

The progress of a pneumothorax into a tension pneumothorax and from thence to causing cardiac arrest was accepted to take place in five stages. The anaesthetists agreed that stage 1 would be the entry of air into the pleural cavity from a damaged lung, to create a pneumothorax; stage 2 an increase of pressure within the pneumothorax as more air was forced into the pleural cavity from the lung; stage 3, the air under tension would compress the ipsilateral lung, then would displace the heart and mediastinum to the contralateral side, compressing the contralateral lung as well and compromising the circulation and pulmonary ventilation; stage 4, the compromise in the pulmonary ventilation and the circulation would cause (i) a fall in the oxygen saturation in the blood, (ii) a rise in expired carbon dioxide tension, (iii) probably then an increase in the heart rate, followed by a fall (iv) a fall in cardiac output, and (v) a reduction in the oxygen supply to vital organs such as the brain and the heart itself; leading to stage 5, a cardiac arrest.

It was common ground that whatever the cause of the pneumothorax (experts expressed a preference for either over-vigorous hand ventilation, perhaps to overcome a plug of mucus which may well have been present in Grace’s airways given her history, and the handwritten operation note, or for a surgical nick, undetectable to the surgeon at the time but creating the small hole through which the air might pass from the lung into the pleural cavity: it is unnecessary for me to choose between them, save that if it were caused by hand ventilation it is possible this would have happened some minutes into the process of chest closure, and would lessen still further the time available within which damaging hypoxic-ischaemia might have occurred) it could not have begun to occur until the sternal halves were brought together with a view to closure.

If the evidence that 30 minutes of damaging hypoxic-ischaemia is needed for watershed damage to occur is to be accepted, this gives insufficient time for the developing pneumothorax to have caused it (unless its effects persisted after the cardiac arrest was reversed). This is obvious given the finding I have already made that the process from sternal closure to incubator would have taken no more than 20 minutes: the drawing together of the sternum would occur roughly half way through its closure, so giving a maximum (and I stress that word) of 25 minutes. Moreover, the tension pneumothorax would not develop in an instant, as though phases 1 to 4 happened immediately. So the Defendant argues that I should exclude it as a probable cause of the damage, and since the cardiac arrest is itself to be excluded for the reasons given at paragraph 28 above, hold that causation by the postulated mechanism let alone by the alleged breach of duty has therefore not been established.

A pneumothorax in the course of open heart surgery was accepted to be either a “rare” or “very rare” occurrence. One causing a cardiac arrest perioperatively had not been encountered by any of the very experienced clinicians who gave evidence. The absence of experience of such an event is reflected by the complete lack of evidence in the literature to which I was referred of reports of a pneumothorax causing a hypo-perfusion injury to the brain. The establishment of such a consequence remains theoretical rather than empirical. Nonetheless, a tension pneumothorax undoubtedly existed here, and was most probably the cause of the cardiac arrest in Grace’s case.

The facts that there was an unusual event in the immediate post-operative period, with a cessation of blood flow, and that Grace suffered brain damage by a mechanism which involved under-oxygenation of her brain cells might nonetheless seem so linked, despite the argument that it simply could not have happened because there was insufficient time, as to imply that once it could be established that the pneumothorax was the cause of the one it (and its aftermath) would also be the cause of the other.

The absence of any coherent alternative explanation would strengthen this link. The Defendant however put forward such an explanation: that the damage was the result of being on cardio-pulmonary bypass. It was well-recognised by all the experts who gave evidence about it that there is significantly increased neurological morbidity after children have been on bypass. Indeed, when Dr. Richard Miles (a neurologist called by the claimant) gave evidence he appeared to think it likely that being on cardio-pulmonary bypass in Grace’s case would have led to under-perfusion of her brain: this was intrinsic to his explanation that being on bypass would have used up the reserves which Grace had, such that none of the priming period of the initially postulated “hour” would have been needed in addition to the half hour of damaging hypoxic ischaemia. Although for other reasons I generally prefer the evidence of Professor Rosenbloom and Dr. Rennie where it conflicts with Dr. Miles, I reject the suggestion that in saying this apparently for the first time in the witness box Dr. Miles was fashioning a case to suit the claim, by gaining vital minutes towards the postulated necessary half hour of damaging hypoxic-ischaemia.

Whereas the Defendant pointed to the difficulty of there being sufficient time within which the damage could occur if the claimant’s mechanism were accepted, the Claimant pointed to the weaknesses in the argument in favour of the Defendant’s mechanism: no literature supported a link between cardio-pulmonary bypass and watershed injury; such literature as there was as examined neurological sequelae to open heart surgery concerned children with much more serious presenting defects (TGA, or hypoplastic left heart syndrome), whose time on bypass was much longer than that spent by Grace, and those children had suffered PVL; and in any event the postulated hour of priming time could not be accommodated within the time scales in this case. Moreover, given that it should be taken as known that the injury was caused by hypo-perfusion, to conclude that bypass was the cause of Grace’s injuries was to conclude that despite what must be assumed to have been a careful setting and checking of the perfusion at the outset of going on bypass, nonetheless the pressures (on which flow to the brain depend) must have dropped below a level sufficient for her and stayed there. There was no hint of any observed drop in perfusion, or of any mechanical or other incident during bypass which might have a drop in perfusion as a result.

Although other mechanisms by which brain injury (or the cardiac arrest) might have occurred were explored prior to trial, it was common ground none could be accepted. I therefore say no more about them.

Thus were the battle lines drawn as trial approached. Both parties could legitimately say that there were serious problems in the acceptability of the other’s explanation. Each asked me to prefer the mechanism they advanced.

Before turning to the detail of these cases, as advanced in expert evidence and submission, I shall deal with the question of breach of duty. This overlaps factually with causation, since it is important to identify when prior to cardiac arrest, if at all, the adverse perfusion effects of a pneumothorax – which in humans are largely theoretical – should with that degree of reasonable skill and care to be expected of a competent anaesthetist in 1983 with the equipment available have been detected by him. The probability is the less significant, then the less detectable: if detectable only a minute or so before the cardiac arrest, then it is likely that I should find that the effects (if any) on perfusion of the brain arising from the undetected pneumothorax were not significant before then – and it would follow that much of the period of up to 25 minutes after bringing together the sternal parts for closure would have elapsed without significant underperfusion, making it less likely that this period was one in which damage occurred.

Terms such as “late” to describe the displacement of the heart and mediastinum, “significant” to describe perfusion, “rapidly” in relation to the development of the pneumothorax and their like were used in evidence. These are imprecise terms. Their use is a reflection of the fact that on the available material, the experts cannot be more precise, and must necessarily be impressionistic. That is how I have treated them.

A developing tension pneumothorax usually produces signs in a patient. Mr. James told me (and I accept) that the first sign will usually be a drop in measured oxygen saturations. That was not detectable here, since there was no continuous oxymetry, and I conclude that if blood gases had been taken (which is a possibility from time to time, and was the only way in 1983 at this hospital of monitoring saturation at the stage of chest closure) no significant result was indicated, for otherwise it would have been the surgeon’s responsibility to note it in his note, and I must assume that he discharged his responsibilities appropriately. No suggestion was made that blood gases should have been taken at any particular stage, such that there can be no proved negligence for failing to take them (if none was taken). I incline to the view that a blood gas would probably have been taken, at some stage during closure, but do so so hesitantly that I place no great weight on this view.

Falling oxygenation may produce cyanosis. The evidence was that this is never easy to detect, and more difficult under the lighting conditions in a theatre, especially where the child is – as was Grace – of mixed racial origin, and with what her mother describes as “mulatto colouring and skin tone”, and where a body is largely covered by surgical drapes. To miss any such sign would not necessarily be negligent.

If Grace was being ventilated mechanically, the dials (if there were such) might have indicated a greater resistance to inflow. The noises made by the machine might alter, as back pressure increased: and it was stressed to me that anaesthetists particularly of the “old school” who would have been prevalent in 1983 depended heavily on their unaided sense of hearing (as well as those of sight and touch). If by hand, it may be that the experienced anaesthetist might have detected a greater resistance to pressure which should have put him at least on enquiry. I cannot be so sure of either of these that a finding of negligence should follow, though my conclusion as to reading of any pressure gauge is inter-dependent on my conclusions as to the likely progress of the pneumothorax. As to the hand ventilation, some weight has to be placed on the observations in the handwritten note as to the difficulty of overcoming secretions. This must have been experienced by either a surgeon or anaesthetist clearing an airway at the time of closure or more probably regarded at the time as an explanation for what were sensed to be rather greater pressures than normal during hand ventilation. If so, it is asking too much of an anaesthetist that he should have considered the possibility of a pneumothorax (which as was emphasised to me was either “rare” (claimant’s evidence) or “very rare” (defendant’s).

The pneumothorax as it develops restricts the space within which the lung on its side of the chest can operate. The reduction in lung capacity, forced on it by external air pressure, means that it oxygenates less efficiently. The contralateral lung continues in operation, but the less well oxygenated blood returned from the affected lung mixes with the blood from the contralateral lung, such that there is some reduction in the oxygen carried by the blood. Hence the drop in saturation levels as an early sign. A consequence is that the heart rate rises, for a while. This too was not so likely to be picked up by even an alert anaesthetist that it would be negligent to fail to do so. The pulse might be detected by taking a carotid or radial pulse, but there would be no constant monitoring – and whether a pulse was taken at the wrist would depend on the anaesthetist’s personal practice as to whether (as would have been the case if Professor Hatch had been responsible for the anaesthetics) the arm was positioned out of the drapes up at the head end of the body (at which end the anaesthetist would have been positioned, behind a small curtain screen from the operative area of the chest) or down by the side, and covered by the drapes (as would have been the case had Dr. James been). There was no evidence of Dr. Hasbury’s preference. It is possible to check capillary filling by squeezing a finger, and seeing how quickly the skin appears to refill – something done more often in Dr. Hasbury’s time as a practising anaesthetist than occurs nowadays. However, in Grace’s case there is no way of knowing if the skin was available for this, because of the possible position of the hands, and in any event there would need to be some suspicion to require it in the first place.

Professor Hatch, though he did not accept that the signs I have mentioned would not have been present, accepted that they would be difficult to detect. Dr. James, whose evidence I accept on this, would not have condemned an anaesthetist as falling below the appropriate standards for failing to spot any or all of these signs.

But there were monitors of both arterial and central venous pressure.

Considerable evidence went to the physiology of brain perfusion. Arterial pressure, produced by the perfusion pump during bypass or the heart during normal circulation, is the main driver of blood flow. The vessels within the brain provide resistance, moderated where and to the extent that there is auto-regulation. The cerebral perfusion pressure is the difference between the pressure on inflow (arterial) and any pressure at outflow (venous). This may be due to resistance within the brain, though where arterial pressure is maintained but central venous pressure rises, cerebral blood flow will slow. The pressure drop across the brain has decreased.

An effect of the increase in intra-thoracic pressure caused by the pneumothorax is that greater pressure is put on the heart and vessels within the thorax. Blood cannot flow back so easily into the heart from the venous circulation. Central venous pressure rises. This reduces cerebral blood flow, assuming constant arterial pressure and an absence of autoregulation.

Thus a pneumothorax might be expected as it develops both to cause a reduction in the oxygen saturation of the blood, and (via its effect on central venous pressure) cerebral blood flow. It will also make it more difficult for the heart itself to act as an effective pump, thus lowering arterial pressure, and further reducing the pressure gradient and hence flow across the brain. This is likely therefore to increase the effects of poorer oxygenation and rising central venous pressure. Professor Mitchell described these effects as synergistic, though in submissions to me it was accepted that a better word might be combinative.

In general, I accept this picture. It was more clearly articulated in expert evidence called by the claimant, but not seriously disputed. It had support, moreover, from research conducted into the response of pigs to induced pneumothoraces, reported in a paper by Barton (Footnote: ³). Swine were chosen because it was supposed they would sufficiently replicate likely human response. Air was fed progressively into the chest cavities of pigs (whose lungs had not been punctured for this purpose) until circulation collapsed. He found that “with increased intrapleural pressure came a significant decrease in pulmonary arterial and bronchial arterial flow on the ipsilateral side, corresponding to a decrease in cardiac output. However, significantly for the present case, he found that “…mean arterial pressures and heart rates remained relatively stable until approximately 47% total lung capacity pneumothorax was reached. After this point was a progressive decrease in mean arterial pressures and increase in heart rate. Central venous pressure showed a progressive increase as mean intrapleural pressures increased with the progressive pneumothorax…..at the point of cardiovascular collapse was a dramatic decrease in carbon monoxide, heart rate and mean arterial pressure….”. In his abstract, though reporting in the body of the paper that the temporal nature of tension pneoumothorax itself is highly variable, and that it can develop within a few minutes following lung injury, particularly in patients on positive pressure ventilation, Barton suggests that “overt hypotension is a delayed finding that immediately precedes cardiorespiratory collapse”. This echoes the penultimate sentence of the text of the paper: “Hypotension has been shown to be a delayed finding that foreshadows cardiovascular collapse”.

In a paper by Paige (Footnote: ⁴), on which Mr Block QC placed some reliance, he reported one case in which the first sign of a right-sided pneumothorax (the same side, therefore, as was Grace’s) in a 6 year old child was an electrocardiographic change, detectable because in the child’s case there had been an additional monitoring lead attached: the oxygen saturation, heart rate, blood pressure, end tidal carbon dioxide and respiratory rate had remained stable, despite a pneumothorax having arisen which was sufficiently severe for a radiograph to show that it had pushed the heart and mediastinum well over to the left side. This paper has these implications: it may be that in some cases there is no observable alteration in pressure readings of arterial or venous circulation; or alternatively (the pneumothorax in that case having been detected before it could lead to circulatory collapse) the paper is otherwise consistent with Barton in suggesting that detectable signs of low perfusion may come late in the process.

As to the first, it may be difficult to extrapolate widely from a single case report, in which the focus was on detection rather than physiological effects, and it must be borne in mind that it seems unlikely the case would have reported unless it was itself unexpected. On balance, therefore, it does not disturb me from a conclusion that it is likely (though not completely certain) that at some stage before cardiac arrest, Grace’s arterial pressure would have gradually risen before falling abruptly; and that her central venous pressure would have risen. In Barton’s pigs the venous pressure more than doubled as the pressure from the pneumothoraces increased, though cardiac output was relatively well maintained till near the end (Figure 6 in his paper). I take the view it must have done so in Grace’s case, too.

But when? Barton’s concluding words indicate that if the animal model is replicated in humans (and “hypotension” is similarly understood) it may not be until just before cardiac arrest that it occurs. That may be sufficient indication of the effect of the pneumothorax (and suggest there can only have been at most a few minutes of hypotension, far less than the postulated 30). If I am concerned with its detectability his paper would indicate this would arise earlier.

As for humans, Professor Hatch thought that a drop in arterial pressure would be a “late sign”. He thought that venous pressure would begin to rise slowly as soon as intra-thoracic pressure started to rise. This would be gradual at first. He would have expected an anaesthetist to have noticed this, asked why it was happening and then looked for other signs: that “at least before the cardiac arrest” he would have had enough information to intervene. He refused to say more than at “some time before the cardiac arrest” he should have realised the chest might need to be re-opened.

Dr. James thought that probably the first sign would be a gentle rise in central venous pressure (he suggested that it might have been shown later if the line which measured pressure was inserted in the groin, rather than the neck, since intrathoracic pressure will directly affect the upper body: I reject this because I accept that both inferior and superior vena cavae connect before supplying the right atrium, and there is no reason to suppose that pressure would not equalise across the central venous system so quickly as to make any femoral lag unnoticeable if it existed at all). This rise would be closely allied with a fall in arterial pressure, and a rise in heart rate. Blood pressure changes (he asserted) happen late during tension pneumothorax.

Though tension pneumothorax is very rare immediately post-operatively, anaesthetists particularly in the early 1980s were also responsible for subsequent intensive care, during which they would deal with a large number of pneumothoraces. Dr. James told me, and I accept, that blood pressure changes happen late in the course of pneumothoraces. It would only be just before the cardiac arrest that there would have been any dramatic change. He added (in cross-examination) that if perfusion increasingly were affected, there would be signs of it, though if unaffected there would be no signs to see. The “preterminal” event is a huge drop in blood pressure, when there is mediastinal shift. In re-examination he thought a fall in arterial pressure and rise in central venous pressure would be late signs. He reached this conclusion for two reasons: first, that the anaesthetist would be particularly aware of the risks that the surgery might have adverse consequences. Cardiac tamponade is much more frequent than pneumothorax, yet the detectable symptoms are virtually identical. There is always a swelling of the heart whilst on bypass, requiring one to be careful whether the cause is indeed bypass, or some other danger. Second, here the leak of air must have been small. It takes time for pressure to build to such a stage that haemodynamic changes begin. Small changes in measured pressures, heart rate and possibly airway pressures are normal. As the pressure in the chest built up, the pressure drop between the inflated lung and the collection of air in the chest would lessen to the point of near equilibrium, slowing observable effects at that stage: he suspected that cardiac output would not fall substantially until just before or just after the arrest, and thus also no reason to suppose that cerebral perfusion would adversely have been affected.

Generally I accept the evidence of Dr. James in preference to that of Professor Hatch where they conflict. Professor Hatch was vague at a number of points in his evidence. In addition to the reasons which have already been highlighted, he had conducted no anaesthesia himself since the early 1990s, and plainly struggled to remember how things were likely to have been in 1983. He began his search for cause in Grace’s case by assuming that brain damage was caused by the cardiac arrest. I had the impression this coloured his views since. At one stage in his evidence, he appeared to be saying that the pneumothorax was probably caused by using high manual pressures to overcome a partial blockage, but at the same time said he was not suggesting this – which led to Mr. Block QC observing, with justification, that it was exactly what he was saying. He volunteered a view that the pneumothorax was more likely to have been caused by the anaesthetist than by the surgeon, but when he explained that he had familiarity with the anaesthetic literature but not so much with the surgical, he was driven to accept that he simply could not say whether one or other cause was the more likely. Thus he had volunteered a relative likelihood as between anaesthetic and surgical causes without any proper basis for the comparison. At one point in his evidence in chief he estimated that the whole episode (he was talking about the pneumothorax) would not have lasted very long – probably about 20 minutes. There was a pause. He added “to half an hour”. I cannot accept this, nor did it fit with the evidence he gave which I have summarised at paragraph 14.

The Defendant argues that if the central venous pressure had been rising in a way which should have been noted by Dr. Hasbury, for a period of more than a couple of minutes or so when the arterial and venous lines were disconnected as they would have been just before transfer to the incubator, and at a time when the ventilation would no longer be mechanical, it is surprising this was not noticed. It would not just have been Dr. Hasbury who failed to spot it. So too would the trainee anaesthetist, anaesthetic assistant, theatre nurse and one or other of the surgeon and surgical registrar if present, and possibly the perfusionist (though his work might have been done).

Although such submissions have to be viewed with care – they are to the effect that because negligence is generally less likely than is the exercise of due care, a court should incline to find that it is improbable, yet such an approach taken towards its extreme would suggest that few if any cases of negligence would ever be established – in this case I accept them, within limits. The period between sternal closure and transfer to incubator was not one in which clinicians could be relaxed about the signs. Tamponade (an accumulation of blood within the pericardial sac, putting pressure on the heart which if unrelieved leads to its arrest, and which inevitably compromises the circulation in ways identical to the postulated effects of a pneumothorax) is an ever-present risk after open heart surgery. It is likely that one or other or more of the operating team would have been alert to signs that it might be happening. If there had been such signs, for any appreciable length of time prior to arrest, they would probably have been spotted, and action taken to remedy a supposed tamponade (pneumothorax would at most be a secondary differential diagnosis, because of its comparative rarity).

The pneumothorax itself was unlikely in my view to have been caused by a large hole: this would have been detected surgically before closure. When a child comes off bypass, the lungs (which it has not been necessary to use whilst oxygenation is supplied by machine) are “re-recruited” as I was told; a large hole would have been obvious. Moreover, it would have required surgical closure, of which there would have been a note. There was none. This is therefore unlikely to have happened. The same is true of a hole of moderate size. The hole in the lung must, therefore, have been sufficiently small for it to close naturally, and for a chest drain once inserted (as it was afterwards) to relieve any pressures without the need for further steps. This argues for a pneumothorax which could raise intra-thoracic pressure only gradually.

Thus, on this point, I have to reconcile the fact that the pneumothorax developed within (at the very most) 25 minutes to the stage of arrest, probably gradually from a small leak, yet did so to the point of causing cardiac arrest, with the probable physiology (making due allowance for biological variability of response, as Paige recorded) which envisages increasing venous pressure and dropping arterial pressure as tension must have mounted within the chest, and with the view that it is unlikely that the monitor readings were clearly out of the ordinary until late in the process. Dr. James in evidence envisaged the readings might have required action for between 90 seconds to 2 minutes, at a time when it is not unreasonable to think that attention might have been elsewhere – lines disconnected, or being disconnected; baby prepared for moving to the incubator; hand ventilation distracting the anaesthetist. This is possible, and if so it might be asking too much of a clinician in that situation to have been aware of the sudden fall in arterial pressure and rise in central venous pressure presaging collapse, especially since it would not be clear if actual monitors were there to read at the time. However, I think it more likely that (reconciling these considerations) there was a slightly longer period of time within which a monitor was likely to have been displaying readings which vigilance should have indicated showed a need for urgent investigation. Putting a time on this is undoubtedly to select a figure which is wrong, and conveys a spurious accuracy if taken too precisely. However, I cannot think that the readings would have invited action for as long as 5 minutes. Before that any alterations in pressures would have been gradual, would not have indicated a tamponade to reasonably vigilant observers (and I cannot accept that all of the potential observers failed to be in this category), and the observable pressures must have been missed by Dr. Hasbury as the person centrally responsible when the attention of others was elsewhere, though briefly, in the period immediately leading up to transfer. It may have been that his manual ventilation (following machine ventilation) was that which very suddenly produced an additional surge of air into the chest which upset the broad equilibrium of pressures of which Dr. James spoke, to tip the pneumothorax from an insidious development to a potentially catastrophic one. I cannot easily accept, however, that the pressures would have been detectably worrying for as short a period as Dr. James suggested. The likely range is 2 – 3 minutes, though there can be no point precision about this.

Should a reasonably careful anaesthetist, exercising that degree of care and skill to be expected of such a professional, have appreciated that these signs called out for action? Given the ever-present risk of tamponade, the fact that it is the centre of the anaesthetist’s function to be concerned at this stage with ventilation and perfusion and hence with blood pressures and heart rate, and the evidence of all before me that if the signs were there they should not have been missed, I find that at a point shortly after the signs were detectable (measured here in seconds, not minutes) they should have been.

Accordingly, I hold that Dr. Hasbury was here in breach of his duty to Grace.

Whether any damage flowed from this breach is more problematic.

If the watershed damage was caused by low perfusion, it must follow that any hypo-perfusion injury prior to a point some 2 or so minutes prior to the cardiac arrest was not caused by any breach of duty (it is not suggested that there was any breach of duty in the pneumothorax arising in the first place). Within a few minutes, a time so short that the cardiac arrest caused no acute anoxic damage to the brain, circulation was restored. Here I have the surgeon’s note and the handwritten note to guide me. Dr. Powers QC submitted that the reference in the latter to isoprenaline suggests that a clinician thought that the heart needed more output, and an increase in heart rate. An alternative view was that the suggestion was precautionary rather than indicative of poor cardiac output and rate. Though this is perhaps less likely, no measurement of poor output is expressed in the notes. To the contrary, the surgeon’s typed note expressly records that the heart started beating again once the pneumothorax was relieved with “good circulatory status”. I must prefer this note, which expressly deals with the question of circulation, as indicating what happened immediately after the chest was re-opened and heart re-started. Further, the handwritten note suffers from the fact that it suggests that the cardiac arrest occurred “during bypass” which plainly it did not on any view (the heart was deliberately stopped at that time), and must thus be read in the light that at least some of its accuracy is only casual.

Accordingly, taking this together with the typed note’s statement that “at no time was there any significant period of circulatory arrest” (which I must accept at face value as a contemporaneous account, with no sufficient reason to doubt it) I conclude that there was no significant time after the cardiac arrest within which under-perfusion of the brain could be supposed to have continued. There are potential difficulties in applying cardiac massage where the position of the heart within the chest has shifted, because of the pressure placed upon it by a one-sided pneumothorax. Opinions given to me differed as to whether this was significant or not in maintaining circulation: again, I feel I must rely on the only note which deals specifically with this, and accept that whatever the difficulties may have been there was no significant period of circulatory arrest. Dr. Miles’ thesis recruited some of the post-arrest period to add to the time within which under-perfusion occurred, just as he argued that the undoubted low perfusion effects of cardio-pulmonary bypass (which he suggested were present in Grace’s case) had primed her for the period of hypotension as air collected in the chest and began to impose tension, so that any lengthy time was not needed then. For the reasons given, I cannot accept that the effects actually persisted after closure. They might have done - in one of the scientific papers it was postulated that such might occur – but for me to find this probable would be both to tread in uncertain waters so far as theoretical physiology is concerned, and to deny effect to the typed note to which I feel I must give full weight. If good circulatory status was regained, almost immediately, then by definition good perfusion across the brain was restored (assuming the blood was sufficiently oxygenated, and there seems no reason to doubt this).

The implication of my findings as to negligence on the case as to causation of the damage is that the damage must have occurred gradually, and to an extent that was not reasonably detectable until close to the arrest. For cerebral perfusion to be sufficiently low for damage to occur, the combination of flow (mainly), the volume of blood and its oxygenation would have to reduce to approximately half of normal (Drs. Rennie and Miles) – at around these levels, allowing for priming, damaging hypo-perfusion would begin. The closer in time to the arrest, the more likely it is this would start. Since it would not begin until the sternal parts were first brought together, and was not reasonably detectable with the exercise of reasonable anaesthetic care and skill until close to the time of cardiac arrest, it could not have begun (as a period of damagingly low perfusion) at a time much if at all before the 2-3 minutes of negligent failure to notice signs.

Given my findings as to the return of perfusion after the arrest there would thus be at most a matter of 10 minutes, and probably less, within which the watershed injury could have occurred. This is so far removed from the postulated 30 minutes that even allowing for that being a broad-brush figure there was simply not enough time for it to occur. It is so short that the period has more in common with an acute insult, such as loss of oxygenation consequent upon cardiac arrest – but the neuro-radiology rules that out.

If, despite this, the period of low perfusion was such that Grace’s injury did occur in this much shorter timescale, the negligence would have resulted in a loss of a couple of minutes of what (on this hypothesis) would have been a short period rapidly causing damage. Since a cardiac arrest in such circumstances involves lower perfusion still than the loss of flow which precipitates it, it is likely that it contributed to the damage to an extent which was not insignificant. It would however not be causative of any greater damage than this. At most, therefore, the negligence would have contributed some 3 minutes of a potential 9 – 10 minutes of damaging hypo-perfusion. This is however sufficient for it to be a more than immaterial contribution to the damage (assuming it to have been caused by a period of hypo-perfusion at this time).

What, then, was the probable cause of the injury which in fact occurred? The Claimant’s case suffers if I should accept that the postulated 30 minutes of damaging low perfusion were needed for the injury to occur.

The Defendant’s postulated alternative cause, that the injury was an effect of being on cardio-pulmonary bypass, is argued by the Claimant to be an insufficient explanation. It was either no explanation at all, or only part of the picture (the rest of which is the lead up to the arrest and its aftermath). If it is inherently unlikely, then the only matter standing in the way of accepting the claimant’s case overall is that of time, as to which the uncertainties are such that the claimant’s case is still the more likely.

The objections to the Defendant’s suggested cause as any part of the explanation appeared at the start of the hearing to be that:

The first of these arguments depends on the basis of the hypothesis. Drs. Miles and Rosenbloom agreed that the 30 minutes of damaging hypoxia would have to be preceded by “a period” of non-damaging hypoxia. Dr. Rennie and Professor Mitchell originally estimated this to be an hour, based on their experience of foetal injuries caused by poor oxygenation. When Dr. Rennie gave evidence in chief she explained why she had changed her view, such that she now thought this period would be very considerably shorter. The longer period would be necessary for auto-regulation to be overcome. This had, in Grace’s case, probably been impaired: she was a sick child prior to surgery. Once subject to significant hypo-perfusion, she would enter the damaging phase (Footnote: ⁵).

This evidence did not seem to me to be calculated to fit the facts to a preferred overall conclusion. Nor, though the same might have been more open to argument, was Dr. Rennie’s production of literature which she offered as some support for her thesis that there were cases reported in the literature where watershed injury had apparently occurred following open heart surgery (and was arguably causally related to it). Her evidence was in general careful, considered, and I had the sense she was trying to give me her best explanation of what had occurred from the standpoint of a neonatal neurologist.

Her counterpart as a neonatologist, Professor Mitchell, also gave evidence of authority and with care. Although much of the answer to this case must lie in the realm of theory, he struck me at times as being a little too inclined to the theoretical in his answers (for instance, that a rise in carbon dioxide levels causes dilation of the brain blood vessels is undoubtedly well accepted, but I confess some surprise that this mechanism designed to ensure continuing perfusion of the brain should in his view then give rise to a possible, theoretical, increase in cerebral pressure because of the increased volume of blood in the brain so as potentially to produce the opposite effect – it seems counter-intuitive that a bodily mechanism designed to produce one effect should actually do the opposite. This particular point was hardly central, but gave me pause). I had the sense that at times he was uncomfortable in his answers to a probing cross-examination; and that he had begun involvement in the case by assuming that the cause was the obvious cardiac arrest, and therefore had only belatedly given full consideration to the Defendant’s case on causation. He was not entirely convincing in his report in his use of the labels “hypoxia” and “hypoxic ischaemia”, which are as counsel pointed out two separate things. Though his evidence in my view deserves great respect, my inclination after having heard Dr. Rennie was marginally to prefer her evidence insofar as there was a choice to be made, and insofar as I have not otherwise resolved matters in dispute between them. That said, the decision to which I have come in this case is based on more than this preference as between two eminent experts.

Dr. Rennie had powerful support for the building blocks of her thesis. In what was accepted to be a seminal work by Barkovich (Footnote: ⁶) he said (Footnote: ⁷):

“Premature infants who suffer mild to moderate hypotension typically sustain injury to the periventricular white matter with sparing of the subcortical white matter and cerebral cortex. In contrast, term infants who suffer similar degrees of hypotension sustain injury in the watershed portions of the cerebral cortex and in the underlying subcortical and periventricular white matter. Classically, this change in injury pattern has been attributed to a changing location of the intervascular boundary zones (watershed regions)……….as a result of the immaturity of the premature brain and its vascular supply, the periventricular areas are the regions at highest risk when autoregulation is compromised. Periventricular white matter damage (periventricular leucomalacia [PVL]) is therefore a common finding on imaging studies of stressed premature infants…..Somewhere between the 34^th. and 36^th. postconceptual weeks, the pattern of injury begins to change as the regions at highest risk for injury extend peripherally to include the subcortical white matter and cerebral cortex in the interarterial boundary zones. These so-called watershed areas are almost always involved if mild to moderate hypotension of sufficient duration occurs after 36 weeks.”

It is common experience that term babies who suffer from congenital heart defects also exhibit PVL. It was accepted before me the cause was hypo-perfusion. Hence Dr. Rennie’s hypothesis that the brains of those suffering from CHD show a delay in maturation (there was a “maturation shift”, as it came to be characterised at trial), but by 4½ months of age the brain would react just as that of the normal term neonate would: hypo-perfusion would cause watershed damage.

Barkovich also noted (p.480) that where there is hypo-perfusion, the region that is damaged does not extend as the hypo-perfusion continues. The analogy with the lawn sprinkler holds good here.

Mahle (Footnote: ⁸) used MRI techniques to scan the brains of 24 term neonates with CHD prior to cardiac surgery. He found that whereas a few had PVL prior to surgery, when he scanned again after surgery over half now did so. The surgery itself had been apparently uneventful, yet ischaemic (Footnote: ⁹) lesions had resulted. Though most suffered PVL, one of the children had suffered a small cortical watershed infarct. (He also observed, in a comment I have not forgotten when considering the Claimant’s case, that studies in foetal and neonatal lambs had shown that during the phase of reperfusion after ischaemia a decrease in cerebral blood flow to white matter persists, despite blood flow recovery in all other brain areas.)

McQuillen (Footnote: ¹⁰) together with others (including Barkovich) in a team from California looked pre-operatively at 62 neonates due for open heart surgery. Peri-operative brain damage (for which being on bypass was significant as a risk factor) was seen in over half, with 35% acquiring the injury during or shortly after surgery. None suffered watershed injury: but the authors observed:

“..a predilection for WMI (white matter injury) in these term infants is highly unusual. The typical result of global injury in a term infant is predominantly damage located to the basal nuclei…or intervascular boundary watershed regions. WMI is more commonly observed after impaired oxygen delivery or infection in the premature infant. These patterns of injury are thought to result from differences in selectively vulnerable cell populations between premature and term infants’ brains…The continued susceptibility of the white matter in newborns with CHD suggests a relative immaturity of brain development……”

In a later review paper, McQuillen and Miller (Footnote: ¹¹) suggested (i) that the high incidence of white matter brain injury in newborns with congenital heart disease was due to a unique vulnerability and might be related to a delay in brain development and (ii) that intraoperative factors might interact with postoperative risk factors such that events during cardiopulmonary bypass might predispose the brain to injury from postoperative low cardiac output (Footnote: ¹²)

Other material was less convincing. Kinney (Footnote: ¹³) examined the neuropathology of 38 infants who had died after cardiac surgery, performed very early in life (before the 48^th. week post-gestation). PVL was the most significant injury in terms of severity and incidence, followed by a spectrum of grey matter lesions, including those to the cerebral cortex (where watershed injuries would manifest themselves). There was no significant association between the duration of deep hypothermic circulatory arrest (the method used in those cases: there was no such extreme cooling in Grace’s case) and the degree of severity of the overall brain injury.

The Claimant’s experts had, prior to the hearing beginning, pointed to a complete absence of watershed injury as an effect recorded in the literature as being manifest after bypass procedures in heart babies. Dr. Rennie showed that there were some reports of this, so the absolute nature of this objection did not hold good. The material she relied on was that produced in a paper by Chen, by Glauser, and by Yokochi together with those by Mahle and Kinney (to the extent mentioned above).

Chen (Footnote: ¹⁴) performed MRI scans of the brain post-operatively, between 3 and 14 days after operation. Any infarcts shown were categorised into one of three groups, depending upon radiologic appearance – acute, subacute, or chronic. If the MRI was obtained within 7 days, and the appearance were subacute or chronic, the infarct was judged to be pre-operative. Of 12 ischaemic infarcts revealed, 7 were grouped in a table said in the text to show perioperative leasions under the heading “Watershed”. One was acute (i.e. must have appeared since surgery). The author commented that “both arterial-occlusive and watershed infarcts were found, suggesting a multifactorial mechanism of stroke in patients with CHD that may include thromboembolism and hypoperfusion”. However, the text also confessed that there could be no definitive timing of the appearances shown, since scans were taken only post-operatively.

Glauser (Footnote: ¹⁵) looked at lesions which had arisen over 4½ years in 40 infants who had suffered from hypoplastic left heart syndrome. 45% had combinations of hypoxic-ishaemic lesions and intracranial haemorrhage. At p.993 she reports that “one baby had …an infarction in a watershed distribution between the anterior cerebral and middle cerebral arteries”. Most attention in evidence focussed on a Venn diagram drawn by the authors. This demonstrated that the lesions from which the infants suffered were not to be seen in isolation from one another. Little could perhaps therefore be gained from this sole reference.

Yokochi (Footnote: ¹⁶) looked at 13 children who had subcortical leukomalacia and border-zone infarction revealed by MRI, to see if causal factors could be identified. Two (one boy, one girl) had had heart surgery, in one case to repair a ventricular septal defect and patent ductus arteriosus, at 2-3 weeks of age; the other, who had Downs syndrome, to repair a ventricular septal and atrial septal defect. The former also had an interrupted aortic arch; the latter had suffered the effects of cardiac failure while on ventilator support. The text however suggested that in those two cases brain damage was “thought to be the result of a circulatory disturbance associated with heart failure (Footnote: ¹⁷) or cardiac surgery”.

This material does not in my view provide unequivocal support for the proposition that watershed damage has (contrary to the initial reaction of some of the Claimants experts) indeed been demonstrated as one of the consequences of surgery on bypass. Insofar as it goes, it hints towards it. There is certainly nothing inconsistent with Dr. Rennie’s theory. It has moreover to be recognised that the reports are recent because of the availability only recently of imaging techniques which permit it; that case/control studies cannot be conducted easily if at all in humans, for ethical reasons; and that it has been recognised for many years including 1983 (so I gathered from the evidence) that brain damage (generally so described) has resulted from time on bypass, when inevitably more primitive machinery and techniques were used than were at the times of the surgery in Grace’s case, such that there is no directly comparable study which is informative to the level which would most assist.

Dr. Miles as a neurologist also gave evidence which overlapped that of Dr. Rennie. He, too, volunteered in chief that of which there had been no sign in his written evidence: in his case, that cardio-pulmonary bypass in Grace’s case almost certainly exposed her to a risk of under-perfusion injury. This was precisely the Defendant’s case, allied to an argument based on the principle of scientific economy (“Occam’s razor”): if one cause was identified which was a sufficient explanation for the damage that occurred, it is unnecessary to look for another, second, cause. However, Dr. Miles viewed the hypoperfusion arising from bypass as having “primed” Grace for the further hypoxic insult which befell her during the course of the pneumothorax , the cardiac arrest, and their aftermath.

As to the injuries being restricted to the most complex forms of open heart surgery, in a paper by Fallon et al (Footnote: ¹⁸) 523 cardiac surgical discharge summaries were examined for signs of neurological events recorded between operation in 1990-1991 and time of discharge. Adverse neurological events were recorded as having arisen after operations which included those to repair ventricular septal defects, amongst the more simple of procedures. (There was also suggestion that the presence of pre-existing abnormality such as CHARGE syndrome may predispose to an adverse neurological event (see p.633, RH column). No reason was advanced that the physiological differences between the more complex and the simpler congenital heart surgery operations might make a difference: the feature of such operations seems not to be the nature of the underlying condition, but rather the fact of being on bypass. Moreover, the evidence was that Grace was unusually symptomatic: and was thus in a sub-group where mortality might be a little higher than the norm (Mr. Firmin suggested 1-2%; Stark that the risk if operated on in infancy as opposed to later was higher, observing “it may seem like a different pathology” where one had to operate earlier, and referred to a series of 8 cases seen by himself and a colleague, of which 3 died).

I am prepared to accept Dr. Rennie’s thesis as to “maturation shift”, as I am her evidence that autoregulation was so compromised in Grace’s case by her condition that the time on bypass would have been in theory sufficient to cause the injury which occurred.

Accordingly, by the end of the case a conclusion that apparently uneventful surgery involving cardio pulmonary bypass gave rise to an appreciable risk of neurological injury of the watershed type was well founded in the evidence; and the comparatively simple nature of the operation did not exclude it.

Dr. Powers argued that the injury was caused by low perfusion. Yet the whole process of bypass would be carefully calibrated and supervised to provide precisely the right level of perfusion. In the absence of any sign that something amiss had happened, I should therefore discount bypass as a cause of low perfusion.

I cannot accept this. It was common ground that significant neurological injury may be caused despite bypass procedures appearing to be entirely appropriate, and (as I have noted) Dr. Miles even relied on underperfusion having occurred in consequence of it for part of his thesis. The fact that no-one understands precisely how and why this occurs does not detract from the proposition that it does – though these features may naturally incline a court towards a preference for a demonstrated cause which would otherwise be sufficient explanation for what has occurred.

As to the Defendant maintaining that the Claimant’s postulated cause is not a sufficient explanation, because it is precluded by there being insufficient time, in closing Dr. Powers argued against acceptance of the “30 minute” period as a given, recognising that it must be an approximation and citing the aphorism that it is “better to be approximately right than precisely wrong”. He asked me to accept that a period based on experience with obstetrics was a poor guide, that there was no empirical basis for adopting this figure in the case of young infants; that the depth rather than the duration of low perfusion might matter more; that the time might be shortened because of the factor of the pneumothorax placing pressure on the heart, thus slowing venous return, coupled with cardiovascular failure; that what was critical was the net flow through the brain, which was determined not just by arterial pressure (“pressure in”) but central venous pressure (“back pressure”) such that when the latter rose as a consequence of an undetected pneumothorax, the flow would reduce, where the arterial pressure itself was reducing; that what mattered was cerebral perfusion pressure; and that despite the experts’ apparent agreement to a need for a 30 minute period, this might well be less, and should not be regarded as a gold standard against which to measure causation. He pointed to the fact that Dr. Rennie’s abandonment of around an hour as “priming” time demonstrated how fragile timing was (Footnote: ¹⁹).

I can readily accept that 30 minutes is a broad brush estimate, around which there is a margin – it cannot logically be that in every case of sufficient low perfusion there is no injury apparent after 29 minutes, but in every one injury after 31. However, no cross-examination attempted to unsettle this figure. It was agreed by any expert who felt competent to comment – indeed, it was necessarily the basis for Dr. Miles’ suggestion that post-arrest events should be counted into the running of the clock (he plainly felt the need to be able to show additional time if his preferred cause was to win acceptance). This was despite the move from the originally postulated prior hour of non-damaging hypoxia. If I were to accept these arguments, it would be to prefer the closing argument of counsel over the evidence in the case. As to the margin to be drawn around the 30 minutes, I cannot accept that it would necessarily all fall below that level, so that 30 should be seen as a maximum and not a central estimate – as Dr. Powers’ arguments seem to suggest. There is no proper basis for assuming it to be any more an under than an over-estimate.

I am not attracted by Dr. Miles’ suggestion that the “priming” period occurred during bypass, and was followed by a period of damaging hypoxia once the pneumothorax began. It ascribes the potentially damaging hypotension to two distinct causes, when one is inherently more likely. Next, it ignores the fact that the heart would have been re-started and good circulation established before commencing chest closure (and there is a record to this effect). If there were any hypoperfusion immediately before this stage, there seems no reason why it should have continued through this process – it would have been interrupted, at least for a period, by the restoration of blood flow whilst the pneumothorax insidiously (and, as I think most likely, slowly) developed. He did not grapple with this in advancing this case. His evidence seemed on occasions to be given for the first time, without mature reflection, in response to the question of the moment. I acknowledge that he, as with others, was grasping for an explanation as to cause: but in his case did so with less grounding in the available literature and the detailed facts of the case than was the case with Drs. Rennie and Rosenbloom. Save as I have indicated, on balance I prefer their evidence, for the reasons I have given.

Making all due allowance for margin, a period of 30 minutes is far removed from the time I have found available in this case. Despite many attendant uncertainties, the best picture is one of a maximum time of 25 minutes, within which the hypoxia would have had to arise, followed by a period around resuscitation and restoration of circulation, which I accept from the typed surgical note would have been short. Within that 25 minutes, I have already held it unlikely that there was any significant hypotension (so as to be detectable) until some 2-3 minutes before the arrest. The hypotension would have developed only gradually before that, given the likely size of the leak. The resultant period is on any view so far removed from 30 minutes as for reliance on that figure to preclude it. I have no better figure to adopt. Accordingly, I do not think it probable that the hypoperfusion injury to Grace’s brain was caused by the pneumothorax. Put simply, there was not enough time for this to have happened.

When allied to the evidence that there was no demonstrable acute consequence from the cardiac arrest, there is no evidence that that arrest caused any further damage – the case in respect of “breach causation” assumes that it did, as a natural continuation of an hypoxic process which was already underway. Given my conclusion that there was insufficient time for such a process to occur in connection with the pneumothorax, it must follow that there is no evidence of any process which was ongoing to which the short period of loss of cerebral flow occurring during cardiac arrest could have made a contribution.

For the sake of completeness in my review of the evidence, I should add that I was not significantly helped by the evidence of the cardiologists. Of those, I could place no reliance at all on Dr. Hunter (called by the Defendant, albeit by phone): his evidence was internally contradictory. Neither counsel sought to rely on it. (Footnote: ²⁰)

When the case began, it appeared that the claimant’s brain damage was caused by one of two mechanisms, each being improbable for particular reasons, yet being the only alternative causes since it was agreed all others could be excluded. As it progressed, the evidence against the claimant’s preferred cause of low perfusion caused by an undetected pneumothorax leading to a cardiac arrest grew stronger. The time within which the evidence showed it could reasonably have occurred was insufficient to permit it, on the best available medical view. The evidence that it might have been the defendant’s preferred cause as a non-negligent consequence of the operation itself, as one of those neurological injuries which it is well accepted happen during and around open heart surgery in neonates and infants, grew stronger, and the reasons for holding this to be an impermissible explanation – reviewed mainly at and from paragraph 76 above – became less compelling. Further, the argument that no literature demonstrated the causal relationship between bypass and watershed damage not only weakened, but was balanced (if not outweighed) by Mr. Block’s point that the Claimant had produced no literature or evidence to show that watershed damage was a recognised consequence of a tension pneumothorax – the case on this remained entirely theoretical and undemonstrated (Footnote: ²¹). In the end, I concluded that although the Defendant’s explanation was fragile, it was on the evidence realistically possible for the injury as suffered by Grace to have been a non-negligent consequence of surgery, and not realistically possible for it to have been caused by an undetected pneumothorax. Since I concluded that the pneumothorax (or, at least, cause for re-opening the chest as with suspected tamponade) should with the exercise of proper care have been detected, I have to ask further whether this and the cardiac arrest would probably have added to the damage. There is no evidence sufficient for me to make this finding. In conclusion, therefore, I find for the Defendant and dismiss the claim.