England and Wales Court of Appeal (Civil Division) Decisions

IN THE SUPREME COURT OF JUDICATURE
COURT OF APPEAL (CIVIL DIVISION) ON
APPEAL FROM QUEEN'S BENCH DIVISION
MANCHESTER DISTRICT REGISTRY
(BRIAN LEVESON Q.C.)

		Royal Courts of Justice Strand, London, WC2A 2LL
		18th October 2002

Lord Justice Mance:

1. The claimant appeals against the judgment and order dated 19^th July 2000 of Mr Brian Leveson QC, as he then was, sitting as a deputy judge of the Queen’s Bench Division, by which he dismissed the claimant’s claim for damages for negligence against the respondent health authority. He did so on the grounds that, although the doctor treating the claimant had failed to act with due care and skill in three respects, it was not shown on the balance of probability that this had caused the cerebral oedema which led to the permanent brain injury in respect of which the claimant claimed.
2. The relevant events took place over the night of 12^th/13^th February 1991. The claimant, having been born on 23^rd February 1982, was then aged nearly nine. He had been ailing for some time, and his mother thought that he had contracted flu over the weekend of 9^th/10^th February 1991. His further deteriorating condition led to his GP referring him to the respondents’ Duchess of York Hospital, where he was seen by the Paediatric Registrar, Dr Economou at about 7.00 p.m. on 12^th February 1991. Dr Economou correctly diagnosed diabetic ketoacidosis (“DKA”). DKA is an extremely serious condition of biochemical derangement that can occur on initial presentation of diabetes in children. The malfunctioning of the kidneys leads to excess glucose and excess urine, and so to dehydration, which cannot be compensated by the sufferer’s increasing thirst. The resulting drop in water in the blood-stream is not matched by any fall in the quantities of glucose and sodium (salt) present. This means that the blood volume reduces and the remaining blood is of increased tonicity (concentration) and may be highly acidic. In Neil’s case a blood sample taken by BMStix (whereby a nurse in the ward can obtain a rough immediate result from a pin prick of capillary blood) showed a glucose level in the blood apparently raised to 44 millimoles per litre (mmol/l) compared with a normal laboratory measured range of 3.2 – 9.2 mmol/l, while urine tests gave strong positives for glucose and ketones. Neil’s acidosis was so severe (a pH of 6.976 compared with the normal 7.36-7.42) as to be “only just compatible with life”.
3. In a person suffering from DKA, osmotic effects occur as different parts of the body seek to equalise their concentrations of solute. The brain cells are to a considerable extent isolated from the vascular compartments containing the blood which feeds them, but will with osmosis be likely to display some degree of increased concentration. The judge further explained that “if water is added to the blood volume, reducing the concentration of solute, not only will salts be pulled from the body into the blood volume but also water will move into the extra vascular compartments and, potentially, the extra cellular and intra cellular compartments (which could include the brain)”. The treatment of DKA necessarily involves the administration of insulin and the patient’s re-hydration by infusion of fluids to reduce the blood’s tonicity. Cerebral oedema is known to occur as a result of the movement of fluid from the blood circulating around the brain and any other extra-cellular fluid into the brain cells. The risk of brain damage exists because the brain can only expand into the space around the spinal column, and is likely to suffer a lesion when it tries to expand there.
4. Neil was monitored throughout the night until 7.15 a.m. on 13^th February when he complained of a headache, and shortly afterwards lapsed into a coma. Despite treatment at these stages, to which no criticism is attached, and the involvement of the consultant, Dr Roberts, it was ascertained that Neil had tragically suffered cerebral oedema and brain damage.
5. The judge made three findings of fault against the respondent hospital:

i) Dr Economou should have involved Dr Roberts in Neil’s management at an early stage. The judge gave three reasons for this: (a) the seriousness of the condition; (b) the absence of any clear protocol for its treatment in the hospital; Dr Economou went back to some notes which he had, but did not check that they were up to date; and (c) the desirability of having two heads to interpret the various test and laboratory results that were likely to be obtained overnight, which could (as in the event occurred) appear inconsistent and, quite apart from this, could call for speedy decisions about changes in treatment. By itself, however, failure to involve the relevant consultant could not be relevant, unless the consultant would (or should) have managed the treatment differently: cf Bolitho v. Hackney Health Authority [1998] AC 232.

ii) The management of Neil’s condition was at fault, in a respect which Dr Roberts would have avoided or corrected, relating to the salinity of the fluid infusion undertaken to correct Neil’s dehydration. Dr Economou started, appropriately, with the infusion at 300 ml/hr of a standard size bag of normal saline (consisting of 0.9% sodium chloride), together with 12 mmol potassium chloride. By 8.00 p.m. 133 ml had been given together with 150 ml of plasma. The laboratory results of blood tests taken at 7.00 p.m. and available at 7.45 – 8.00 p.m. showed a blood glucose of 26 mmol/l, still way in excess of the normal though surprisingly different from the prior BMStix result. A further BMStix test at about 8.30 p.m. showed blood glucose at 22.3 mmol/l. At about that time the fluid infusion was altered. Instead of normal saline, a solution of one-fifth normal saline (i.e. 0.18%) with 4% dextrose saline and added potassium was set up to run at 200 ml/hr. Dr Economou’s evidence was that he was concerned by the “very significant drop” in the blood glucose, which he said was shown by comparison of the two BMStix figures, and by the risk that cerebral oedema might result from too rapid a reduction in blood glucose. However, he did not wait for a laboratory result which was due shortly from a 7.30 p.m. blood test and which in the event showed a glucose level of 32.1 mmol/l; nor did this affect his management once it was available. Still more confusingly, a further BMStix result at 9.00 p.m. produced a level of 12.8 mmol/l (a figure which the judge concluded cannot have reflected the real level). A later BMStix result at 10.00 p.m. and the laboratory result for 10.00 p.m. (probably available about 10.45 – 11.00 p.m.) gave results of 21.2 and 21.5 mmol/l, the 11.00 p.m. BMStix reading was 18.9 mmol/l and the BMStix and laboratory readings thereafter until at least 5.00 a.m. next morning, were all above 20 mmol/l. The judge held that the decision between 8.00 and 9.00 p.m. to change from isotonic normal saline to dextrose saline was outside the band of decisions that a reasonably competent doctor should have taken, and that a steady regimen should have been maintained, with any reduction only taking place after it was (reliably) established that a fall in blood glucose level had been achieved. Having said that, the judge recorded that Dr Economou had clearly been alive to issues of excess fluid and had later adjusted fluid levels several times, from 200 to 100 to 70 to 40 ml/hr; the salinity of the infusion was, however, only increased on one occasion and then temporarily, when Dr Economou at some time after 1.00 a.m. ordered an extra 8 mmol/l of sodium to be added to one bag. That only increased the salinity of the infusion (by now itself reduced to a rate of 100 ml/hr) from one-fifth to one-third normal, and the bag in question appears to have been used up by about 6.00 p.m.

iii) Dr Economou was also at fault in not arranging further laboratory tests, having regard to the fluctuations in results obtained, the change in management between 8.00 and 9.00 p.m. and the acknowledged risk of complications occurring at times when the youngsters in question appeared otherwise to be improving. The judge found it likely that, if such tests had been taken, they would have shown that the sodium level in Neil’s blood had dropped slightly. The known figures for serum sodium are 134 mmol/l at 7.00 and 7.30 p.m., 133 at 10.00 p.m., 132 at 11.50 p.m. and 129 at 7.30 a.m. (the judge commenting that this last figure “might – but only might – have been affected very slightly by the commencement of the Mannitol infusion but I do not accept that by the time the blood test was taken, it would have made a great deal of difference”).

6. The judge went on to reject the further criticism of Dr Economou (based on American papers), that he failed to realise specifically that it was essential to monitor the serum sodium level and to change the regimen to ensure that the sodium value increased or, at worst, remained the same. However, if Dr Economou had arranged further laboratory tests, as he should have done according to the judge’s third finding of fault, these would necessarily have revealed some decrease in the sodium level of Neil’s blood as stated in the previous paragraph. It is not clear to me what if any significance the judge thought that Dr Economou should have attached to that decrease, by itself, unless to draw his attention to the desirability of higher tonicity fluid, which brings one back to the second aspect of fault.
7. On the basis of these findings of fault, the judge considered the issue of causation. The issue, as it was argued before him, was whether, but for the faulty treatment which occurred, Neil would, on the balance of probability, have suffered the cerebral oedema, and thereby the brain injury, that he did. On this issue the judge heard the evidence of three experts, Dr Tasker and Professor Brook for the appellants and Dr Edge for the respondents. They had made reports, held a joint meeting on 12/13^th February 1999, following which they signed a joint statement, and produced a considerable number of relevant articles, some by Dr Tasker and Dr Edge.
8. The judge started with some introductory observations about the problems with which he was faced. The respondents’ contention was, and remains, that, although the “mechanism” of cerebral oedema is known, the cause is not. By “mechanism” in this connection was meant the fact, which was common ground, that cerebral oedema involves entry of water into blood cells by osmosis, which itself involves “osmolar” differences, that is differences in concentration, between brain cells and other adjoining parts, such as the blood, extra-cellular fluid and cerebro-spinal fluid (“CSF”). The appellant’s response is that the case is not concerned with proof to a scientific standard, but merely with proof on the balance of probabilities; that a good deal is known about the causation of cerebral oedema, and that, even in the absence of positive evidence as to the cause or as to precisely how particular circumstances result in cerebral oedema, a court may conclude as a matter of probability that changes in management would have made a difference.
9. The judge commented on this that:

   “The problem is that the less that is known about the cause, the more difficult it is to conclude on the balance of probability that avoidance of the circumstances would (as opposed to may or might) have resulted in a different, more favourable outcome. Best learning may mean that there is some evidence of an association between a particular treatment and an undesirable outcome in that some people avoid the outcome if treated in that particular way. But if others do not avoid the outcome if treated in that way, there has to be some basis for concluding that the case under consideration would, on balance, have fallen into the former rather than the latter category.”

   This comment may raise a question of relative risk (cf Loveday v. Renton and Wellcome Foundation Ltd. [1990] 1 Med LR 117, 163, per Stuart-Smith LJ). If comparison of the outcomes of two different treatments shows in the one case only a 1% occurrence of a particular condition and in the other a 3% occurrence, the relative risk of adopting the latter treatment is 3. The question may arise whether it is possible, all other things being equal, to conclude without more that a failure to adopt the former treatment was more probably than not the cause of the condition. The problems of a purely statistical approach of this sort are the more obvious if one assumes a relative risk of only 2, or 1½.

10. The appellant relied before the judge and before us on the contention (which the judge accepted) that the respondent was at fault in administering a reduced saline infusion from prior to 9.00 p.m. onwards. This was a contention which the judge accepted as having been based firmly on best practice at the time, which was in turn later embodied in guidelines for treatment of DKA drawn up, with Dr Edge’s assistance, and accepted by the British Diabetic Association. The judge was not however prepared to accept that it was evidence on the issue of causation, saying:

“Given that the management now recommended does not, as far as is known, have any disadvantages (on the basis that the body can excrete excess salt), it may well be – as Dr Edge argues – that it is merely the best that has been devised in the light of all that is known and is given in the knowledge that it will not harm albeit that it may do some good.”

The judge went on to refer to certain general statistics on the incidence of cerebral oedema to which I will return.

11. The judge then analysed the position and evidence of each of the experts called before him, from the perspective of how it had emerged in the proceedings and with reference to the various papers produced before him. Dr Tasker’s report had been carefully drafted; it recognised the exact incidence, aetiology and outcome of cerebral oedema in the course of DKA as “unknown”, and that the whole area was contentious. In doing this, Dr Tasker listed “several theories” that he said had been put forward in medical research literature as to why cerebral oedema develops as a complication of DKA, as well as concern in the clinical literature as to “the possibility that some aspect of fluid administration or glucose reduction therapy may contribute to the development of cerebral oedema”. He went on to say:

“However, it should be remembered that in severe cases of [DKA] cerebral oedema may well be a universal feature, which may even be present before treatment is initiated. As already suggested above, Neil’s disorientation in mental state before treatment started may well have indicated that he had a degree of cerebral oedema.”

12. Dr Tasker went on to consider papers by Rosenbloom et al, Duck & Wyatt and Harris to which I will come. He concluded his comments on Neil’s treatment as well as his summary of his opinion with the following sentence:

“A change in the tonicity (i.e. saline concentration) and volume of fluid given may have made a difference in Neil’s eventual outcome, more than it is likely not to have done so.”

The odd drafting suggests that the sentence may originally have been conceived without its last ten words.

13. The real difficulty about Dr Tasker’s evidence comes, however, when one turns to the joint experts’ statement, following their meeting, and to his oral evidence. The experts were set an agenda of 25 questions, preceded by a brief pre-amble of five short paragraphs, the last of which read:

“The experts are reminded that civil cases operate on balance of probabilities, that is more likely than not.”

The questions, with the answers given, included the following:

Question 10: What would [further electrolytes and glucose] tests have revealed and what if any change in management would have resulted? Had Neil’s management been altered accordingly, would he have suffered raised intracranial pressure/cranial oedema? Answer: It is possible only to speculate on the results of such tests, which were not performed. We cannot therefore say whether there is evidence either way to dictate the further outcome.

Question 11: Would any changes in management have avoided:- (1) a rise in intracranial pressure? (2) cerebral oedema? Answer: We cannot say whether changes in management would have avoided a rise in intra-cranial pressure or cerebral oedema.

Question 12: Were regular biochemistry tests warranted, if so at what intervals, what would they have shown, what change in treatment would this have caused and what effect would this have on any developing cerebral oedema? Answer: See Questions 9-11.

Question 13: What would laboratory tests of blood glucose, urea and electrolytes, performed between midnight on 12.02.91 and 7.30a.m. on 13.02.91 have shown and what changes, if any, in management would have been likely? What difference if any would that change in management have made? Answer: We cannot know what laboratory tests would have revealed and therefore we cannot know what difference, if any, a change in management would have made.

Question 17: What are the likely causes of Neil’s cerebral oedema? Answer: Nobody knows the cause of cerebral oedema

Question 24: Is it accepted that children who develop [DKA] as their first presentation with diabetes are more likely to develop cerebral oedema than those with known diabetes? Answer: Data published in October 1999 (Royal College of Paediatrics and Child Health, British Paediatric Surveillance Unit Annual Report 1998/1999) indicate that 11.9 per thousand newly diagnosed children develop cerebral oedema and 4.3 per 1000 in known diabetics.

Question 25: It is [sic] accepted that Neil was at high risk of developing cerebral oedema from the moment he presented at hospital? Answer: Because the causation cerebral oedema is unknown and a condition cannot be diagnosed until it occurs, we cannot answer the degree of risk of Neil developing cerebral oedema from the moment he presented at hospital. It is the contention of Dr Tasker and Professor Brook that, since no observations were made, there was no chance of intervention. Dr Edge doubts that intervention would have altered the outcome.

The risk of cerebral oedema is one of the reasons why the responsible consultant should have been given the opportunity to be in charge of the management of the patient under his care, either in the hospital or by transfer to another.”

14. The judge said that in his oral evidence: “Dr Tasker expressed himself in very different terms”. In particular, Dr Tasker explained that for something to be “known” scientifically, there should be a 95% confidence level (that is a likelihood of only 1 in 20 of its being wrong), that this was the first occasion he had ever taken part in a meeting of experts, and that, in agreeing to the joint experts’ statement, he had the required scientific level of confidence in mind (despite the preamble to the agenda and the form of some of the questions, e.g. questions 17, 24 and 25). As a matter of probability, however, his view was that, with knowledge of the sodium level in the context of the glucose and the fluids that had been administered, “something salvageable could have been done” and that, even if Neil had suffered a degree of cerebral oedema, it could have been limited or treated earlier with lesser consequences. There would have been a 90% chance of avoiding permanent brain damage at the start of the evening, at 3.00 a.m. at a guess still a 70-80% chance, but by 7.00 a.m. no chance.
15. The judge, evaluating Dr Tasker’s evidence, identified his very real concern that Dr Tasker had “persuaded himself that the issue of causation is less finely balanced than he initially believed”, and could not “accept that the way in which Dr Tasker expressed himself in his original report was the product of thinking that causation was proved to the level of 90% certainty or even that all the damage would necessarily have been prevented”. He said that he could therefore attach little weight to Dr Tasker’s present formulation. As to the joint experts’ statement, the judge observed that the answers “on the face of them certainly appear to reflect a lesser confidence in causation even than the cautious support in Dr Tasker’s report”. He returned to the joint experts’ statement at a number of points, but I do not read him as rejecting outright the explanation given by Dr Tasker (and Professor Brook) that they were focusing on knowledge in the scientific sense. However, the failure to read and digest the pre-amble and the general inconsistency of Dr Tasker’s various statements on the subject of causation were clearly important considerations, when the judge came to decide whether on the balance of probability the negligence he had found had resulted in Neil’s development of cerebral oedema. For my part, if the appellant is to obtain any assistance from Dr Tasker’s evidence on causation, it seems to me that it must be because of the independent force of the written materials and information produced by Dr Tasker, rather than because of any expression of opinion that Dr Tasker himself gave. I come back to such materials and information below.
16. Professor Brook’s written report contained, as the judge put it, “the pithy assertion” that “Regardless of his state of orientation on admission, the regimens used were inappropriate and, far from rescuing a remediable situation, were directly responsible for Neil’s present condition”. Professor Brook went on:

“The pathogenesis of cerebral oedema is poorly understood but rapid changes in blood osmolality probably contribute so [wrongly transcribed in the judgment as ”to”] dehydration, acidosis and hyperglycaemia should be corrected over several hours.

…. There are data to suggest that complications are more likely to develop if serum sodium concentrations fail to rise as glucose levels fall (Harris et al 1990).

Although there may be some dispute about the exact cause of cerebral oedema, its danger is well-recognised and when it occurs it is devastating. Its avoidance is standard practice.”

17. Professor Brook in his report then indicated his view that oedema would occur if too rapid a correction was allowed of any imbalance of solute concentrations in neighbouring body compartments. Hence, he suggested, the need when treating DKA for slow correction of dehydration, avoidance of hypotonic (low tonicity) intravenous fluids and adequate replacement of sodium and potassium levels. He expressed the view that Neil’s hyperglycaemia was corrected too quickly. In this connection, he referred, firstly, to the BM Stix of 12.8 at 9.00 p.m., and said, secondly, that Neil was given the hypotonic solution (dextrose saline) too quickly so that “water overload” occurred. In oral evidence, he volunteered only the second reason, and when counsel for the appellant mentioned the BM Stix of 12.8, said at once that he would not have believed the reading (an assessment of its reliability with which the judge agreed). The judge placed some weight on this as cutting at least part of the ground from under the Professor’s opinion. However, the other part, that water overload resulted from the infusion of a hypotonic solution, remained independently, and, as the judge recognised, it was not being challenged that “cerebral overload was caused by water overload”. The question is whether the water overload causing cerebral oedema in Neil’s brain cells can be linked with the tonicity of the infusion administered to reduce Neil’s dehydration. On this the judge said:

“The problem is that infusing water (albeit with sodium and potassium) is essential for the purpose of rehydration and unless the tonicity (saltiness) of the fluid is the determining factor (which is the effect of the Harris suggestion by reference to serum sodium levels and is discussed above), it is difficult to see what distinguishes those cases in which cerebral oedema results and [those in] which it does not.”

18. I will come later to what was being said in the papers by Harris, which were before the judge. What matters at this stage is the judge’s next passage:

“As Professor Brook said, some doctors use an inadequate protocol and “get away with it”” others use the correct protocol and do not “get away with it”. Why is that? The answer is that we do not know. Yet I have to be able to say on the balance of probability that had Neil been treated with a “correct” regimen, he would have “got away with it”.”

The judge’s question “Why is that?” may be said to be of particular relevance, if there was no real difference in the incidence of cerebral oedema according to whether or not the “correct” protocol was followed. I will consider below the evidence and the written material put before the judge bearing on the question whether it is possible to say one way or the other at present that this is the case. The judge went on to note that of five features noted in an editorial by Harris for predicting those with the highest risk for oedema, Neil was admitted with the first two, viz. an extended history over many days of poor or absent control (which relates to the dehydration consequent upon undiagnosed diabetes) and moderate to severe acidaemia. Professor Brook at the end of his cross-examination expressly confirmed that there were “data to suggest that the length of the history and the increasing amount of dehydration and all the rest of it makes him more at risk from cerebral oedema than he would have been if he had not had such a long history and been so ill”; that, had the sodium level of the infusion not been reduced, then “with the bounds of probability …. already discussed with Dr Tasker”, the cerebral oedema would have been prevented; and that different management would have made a difference to the outcome. After re-examination, he was questioned by the judge about the concluding statement in his original report: “If Neil had been treated with an acceptable standard of care, the devastating result of cerebral oedema might have been avoided. The management made a material contribution to his present contribution”. He said:

“What I meant …. was that ….. you can obey all the rules and sometimes slip, you can skate on thin ice and sometimes get away with it. My view was then, and is now, that he would have had a substantial chance of not having cerebral oedema if he had been treated with any single one of … any of the protocols that I have seen in any of the books. As it was, he was treated with a completely rogue protocol and the result was as we have seen.”

19. The judge focused on the reference to “a substantial chance”, but I think that the fair reading overall is that Professor Brook was saying that, in his view, the treatment administered did probably cause cerebral oedema which Neil would in his view otherwise have avoided. Professor Brook was, like Dr Tasker, also party to the joint experts’ statement, but gave the same explanation as to the basis on which he had answered it. I have already said that I do not read the judge as rejecting outright the explanation given by Dr Tasker and Professor Brook that they were focusing on knowledge in the scientific sense. The fact that Professor Brook, like Dr Tasker, failed to read and digest the pre-amble was unsatisfactory. Professor Brook’s evidence was not open to the other more general criticisms of inconsistency that Dr Tasker’s was, but it was considerably more positive than Dr Tasker’s as to the course of events likely to lead to cerebral oedema, identifying over-rapid correction of the condition of DKA as a probable contributor and the level of serum sodium as a marker to watch.
20. The oral evidence on causation that convinced the judge was that of the defendants’ expert, Dr Edge. He rejected her evidence that the treatment which Neil received fell within the range of appropriate treatment for his condition of DKA. But he accepted her evidence that it was not possible to say, even on the balance of probabilities, what was the cause of Neil’s cerebral oedema or that any different treatment would have had any significant effect in avoiding it. In reaching this conclusion, the judge analysed a number of papers, including one then still unpublished, written by Dr Edge. She has made a particular study of the whole topic. The judge summarised her oral evidence. She said that the use of one-fifth normal saline (comparing it in this context with the use of half normal saline) may be related to the causation on cerebral oedema, but said that “You’ve still got no idea whether they are”. The most that she would accept was that there was a possible link. She said that the fact that the published guidelines, which she prepared, recommended the use of half normal saline “doesn’t mean I believe that hypotonic fluids cause cerebral oedema”. Asked by the judge whether she did believe that, she replied:

“I think that in some instances they may but I don’t believe that is the cause of cerebral oedema.

….

Because, as I have said in several places, the mechanism will probably be many causes.

…

The use of hyptonic fluids, as we have said, may be a cause in some cases ….. But that doesn’t answer what the majority of children, why children develop cerebral oedema. There are many, many things we don’t understand”.

21. The judge accepted Dr Edge’s oral evidence, after saying that “there was nothing substantially inconsistent between what Dr Edge said in her report or in her publications on this topic, what had been concluded at the Joint Meeting of Experts (which had not surprised her) and what she said in evidence”. Before us, the appellant raises a number of criticisms in relation to Dr Edge’s evidence. The first is that Dr Edge’s report accepted the obviously suspect BMStix reading of 12.8 at 9.00 p.m. at face value and that, in her oral evidence (TR 2/4/87C), Dr Edge at one point even relied upon this reading as showing a fall in blood sugar so quick as to be a possible factor causing or contributing to the cerebral oedema. However, it is right to point out that she had said that, on seeing the 9.00 p.m. BMStix reading, she would have repeated it (i.e. to check its correctness) because it was open to question (cf TR 4/44G and 4.90D). Secondly, the appellant points out that Dr Edge seems to have written her original report and formed her views in the belief that the total fluids administered had been less than 4 litres per metre squared (of Neil’s body surface) per day. She wrote: “At no time, until after the event, did the plasma sodium level fall significantly, and his fluid input was not excessive (less than 4 ltrs/m²/day)”. This appears to pick up the contents of a paper by Duck & Wyatt (referred to earlier in her report), where the initial head-note says: “Only 4 of 40 cases [of DKA] occurred at fluid intakes < 4.0 ltrs/m²/day”, and the text reads: “The previously proposed limit of 4.0 ltrs/m²/day remains an important guideline to be considered while one is attempting to establish the overall rate of rehydration”. Counsel for the appellant put to her that the correct figure based on volumes of fluid taken from the records and an assumed body area of 0.74m² would be 4.77 ltrs/m²/day. As the judge said: “she did not resile from her figure, but said that even if the higher figure was accurate (although it may be that the figure should be less than 4.77 ltrs/m²/day but more than 4 ltrs/m²/day) it remained within the general range and was, in any event, far less than many regimens would administer”. Dr Edge also suggested a body area of 0.80 m². The figures for fluids administered were not however shown before us to be wrong, so that, even if one takes that body weight, the resulting figure for fluids administered would be 4.41 ltrs/m²/day.
22. Thirdly, the appellant observes that Dr Edge erred in assuming that the increase in the saline content of the infusion from about 1.00 a.m. was permanent, since she calculated its value on a daily basis. In fact, it was only for one bag. Bearing in mind that this bag lasted until 6.00 a.m., however, the point cannot be of any ultimate significance. Fourthly, the appellant points out that, in her oral evidence in chief, Dr Edge treated the main allegation as being failure after midnight to revert from a one-fifth saline solution infusion to a more concentrated saline solution (TR 2/4/26F-G), whereas the appellant’s primary case (accepted by the judge in relation to the issue of fault) has always been that the change to one-fifth saline solution between 8.00 and 9.00 p.m. should never have occurred. This is capable of linking up with the first point above. Fifthly, it is pointed out that the judge erred in saying that one aspect of the evidence relied on by the appellant in final speeches had not been put to Dr Edge (J40). That concerned a table of information regarding outcomes obtained from hospitals with regimes involving treatment of DKA with one-fifth and one-half normal saline infusions. The judge did not think that this evidence took the appellant’s case any further forward, and his error in thinking that it had not been put to Dr Edge cannot have been material.
23. Standing back from all these points, it seems to me that the first, second and fourth points are, both individually and cumulatively, points which the judge could have been expected to address more specifically, as possible reasons for questioning the basis on which Dr Edge was expressing her opinions. (The judge only touched on the first.) Had he done so, however, I do not think that, by themselves, they could have outweighed or can now undermine the judge’s positive assessment of Dr Edge’s oral evidence. In my view the outcome of this appeal must depend on a closer analysis of the substance of what Dr Edge was saying and whether her evidence discounting any relevant causative link whether general or specific can stand with her own and other published material, to which I next turn.
24. The judge went through the published material, in the course of considering the experts’ evidence. It is the experts’ common theme that cerebral oedema generally occurs during treatment at a time when the patient otherwise appears to be making a rapid recovery from DKA, although cases do from time to time occur prior to any treatment. Hence, the attention paid to the treatment of DKA. In a 1984 paper published by Krane et al, Subclinical brain swelling in children during treatment of [DKA], the authors said that cerebral oedema complicating DKA occurred “unpredictably” and that, although “five decades have passed since the original pathological description of the syndrome, yet the reason for its occurrence remains unclear. Osmotic disequilibrium between the brain and blood has been cited most frequently as the cause. This theory is supported by two animal studies ….”. Later in the paper, they wrote:

“….the infrequency of its occurrence makes clinical evaluation of new protocols difficult. We are unable to extrapolate from our data a specific alteration in the present management of [DKA]. It appears to be prudent to avoid rapid fluid administration, especially with hypotonic solutions, at a time when blood glucose and osmolality are rapidly falling. It is not known, however, whether modification of current therapy would prevent the rare clinical cases of cerebral oedema or alter the radiologic findings reported here. Indeed, it is not known whether the radiologic findings are the result of therapy or are actually present before therapy is begun.”

25. Writing again in 1987 on DKA, Krane repeated, with respect to the mechanism by which cerebral oedema occurs, that:

“The most frequently invoked theory is that an osmotic disequilibrium exists between the central nervous system and extracellular fluid during the therapy of DKA, which promotes movement of water along an osmotic gradient into the brain. As discussed earlier, during periods of systematic hypertonicity the tonicity of the brain increases in parallel, owing to the formation of unknown osmotically active molecules within the brain. Theory holds that these organic molecules dissipate or are metabolised more slowly than the rate at which serum tonicity declines, hence during therapy the tonicity of the brain exceeds the serum tonicity, and water movement into the brain is favoured.”

However, he continued by saying that the experimental evidence existing for this theory was incomplete, and that the contribution of osmotic disequilibrium to the development of cerebral oedema “therefore remains speculative”.

26. In 1987 Duck & Wyatt wrote a paper on Factors associated with brain herniation in the treatment of [DKA], in which they started by observing that “reports of children who develop brain herniation as a consequence of cerebral oedema during the first 24 hours of treatment of severe [DKA] are decreasing” and that “some articles and editorials have remarked that efforts to relate fatal brain herniation to specific aspects of treatment have been fruitless”. Their conclusions after analysing the history, date, rate and composition of fluid and insulin administration and time to onset of brain herniation in 9 new cases and 33 prior reports were, however, that (1) the overall rate of fluid administration was inversely correlated with the time of onset of herniation, (2) only 4 out of 40 cases occurred at fluid intakes less than or equal to 4 ltrs/m²/day, (3) during treatment “calculated” serum sodium concentrations fell significantly and were less than 130 mmol/l in 33% of cases at the time of herniation and (4) these data indicate that excessive secretion of vasopressin may exacerbate the brain edema, and that limitation of the rate of fluid administration may be prudent. On the other hand, the text starts by acknowledging that the cases in which “sudden and unpredictable brain herniation during the first 24 hours of reasonable treatment of severe DKA” occur “are not well understood”, and continues:

“Certain factors, including insulin administration, osmolar disequilibrium between intracellular and extra cellular brain water, and disordered cerebral vascular dilation and blood-brain barrier integrity, are thought to be necessary for cerebral edema to occur. Some degree of generalised cerebral edema probably occurs in every patient, but only a rare patient experiences the fatal consequences of brain herniation.

This retrospective review could not answer the question of cause, and no definite recommendation can be made to lessen the likelihood of precipitating brain herniation. Recent discussions have suggested that no identifiable therapeutic consideration is significantly associated with brain herniation, but this review of a large number of cases suggests that excessive rates of fluid administration and possibly excess vasopressin in secretion are factors that may increase the likelihood of brain herniation during therapy for severe DKA.

Only the overall rate of fluid administration was significantly inversely correlated with time to herniation. The previously proposed limit of 4.0 L/m² day remains an important guideline to be considered while one is attempting to establish the overall rate of rehydration. Four patients (10%) were reported to have received less than this proposed limit, suggesting that other factors are important besides fluid infusion rate.”

27. Secretion of excess vasopressin (an anti-diuretic hormone) is the body’s natural response to dehydration (TR3/27). The judge emphasised the use of the words “possibly” and “may”, and also observed that Duck & Wyatt had not studied children who had DKA, but did not develop cerebral oedema. Dr Tasker’s explanation of the use of the word “may” was the same as his explanation of his agreement to the joint experts’ statement. Scientists will always use such terms, unless the confidence level is as high as 95%. Nevertheless, the judge thought that the language was “far more cautious than would be used to express any view about a possible cause as opposed to a possible risk or aggravating factor”. He also thought that the advice regarding management was in “the language of potential increased risk rather than cause”.
28. The judge quoted extensively from and commented to some degree adversely on an editorial, by Harris, Fiordalisi & Finberg, which appeared in the same issue as Duck & Wyatt’s paper, headed Safe management of [DKA]. The adverse comment attached to the editors’ statement that Duck & Wyatt “conclude from low or falling serum sodium values, that excessive secretion of antidiuretic hormone may be the cause of cerebral oedema and herniation”. Dr Tasker agreed that Duck & Wyatt had not measured antidiuretic hormone (or vasopressin) levels. But Duck & Wyatt did refer to literature for the proposition that “Plasma vasopressin concentration is excessive and inappropriately regulated in patients with DKA” and went on to explain that “Inappropriately high vasopressin concentrations enhance free water retention, generate a more dilute serum and aggravate hyponatremia [low sodium level]” This consequence would be expected to enhance the movement of free water from the relatively hypotonic intravascular space into the relatively hypertonic brain tissues. This sequence was reported to occur in one patient and is consistent with the findings of this larger case review”. They also stated that a “falling “calculated” sodium level can easily be determined from routine laboratory monitoring and is likely correlated with dysregulation of arginine vasopressin in these patients”, and that excessive rates of fluid administration “and possibly excess vasopressin in secretion” are factors that may increase the likelihood of brain herniation during therapy for severe DKA”. Further, a later article, by Greene, Jefferson and Baum, Cerebral Oedema complicating [DKA], which the judge quoted in a different respect, also referred to Duck & Wyatt as having “further suggested that an abnormal fall in measured and/or calculated sodium following excessive fluid administration in the treatment of [DKA] induced inappropriately high antidiuretic hormone secretion”. I do not myself see how the editors, Harris, Fiordalisi & Finberg were vulnerable to the criticism that the judge directed at them. Again, however, this leaves unaffected the central aspects of the judge’s reasoning.
29. The judge also observed that the editors spoke only in terms of possibilities. They referred to the factors which “may be considered” to give “some predictability for identifying patients at highest risk for brain swelling”. They referred to brain swelling during treatment of severe DKA as “a tragedy that may now be prevented with modification of conventional fluid and electrolyte therapy”, “strongly” recommending in this context “avoidance of rapid, high-column administration of hypotonic fluids at a time when osmolality is high and a fall in osmolality is expected”. Further, the judge rightly pointed out that Greene, Jefferson & Baum in their 1990 article examined four children suffering cerebral oedema during treatment of DKA and concluded that they suggested that “all children are at risk of developing life-threatening cerebral oedema”, and that “There was no common theme among our patients to suggest any single therapeutic manoeuvre as the causative factor”, although they also concluded that “limited conclusions” could be drawn, making certain suggestions for treatment, including that “excessive swings of glucose, osmolarity and pH should be avoided”.
30. Another 1990 paper to which many subsequent references have been made in the literature, Rosenbloom et al, Cerebral Oedema complicating [DKA] in childhood (1990), involved a retrospective study of 17 young people under 21 who developed DKA. The authors concluded that “Contemporary notions regarding an etiologic role for rate of blood glucose correction or speed of hydration are not substantiated by this experience and review”, and evidently found little significance in Duck & Wyatt’s paper:

“This review is consistent with the recent study of Duck and Wyatt …., which was restricted to 42 patients with herniation, in failing to implicate rate of fall of blood glucose or the level attained. However, Duck and Wyatt found a weak but significant correlation between time of onset of herniation and rate of fluid administration, and they found that 90% of the patients were receiving [less than 4 ltrs/m²/day], an arbitrary figure they considered a critical cut off point for risk for cerebral edema ….. The clinical significance of a correlation between rate of fluid administration and time of herniation is obscure and unsupported by this study. The arbitrary safe rate of fluid administration is at a level likely to be exceeded by current regimens. Lacking have been case-control data that would indicate whether patients who developed cerebral edema were hydrated in any way that was different from those who did not. Because half the patients in this review received <100% of the currently recommended replacement rates, and 3 patients received oral fluids only, it is unlikely that this was a major factor in any but the most blatant circumstances of overhydration.”

31. Harris, Fiordalisi, Finberg and two others published in 1990 a retrospective study of 219 episodes of DKA, and Harris & Fiordalisi published in 1994 a further paper. These papers are central to the appellant’s case that causation has been sufficiently established. The aim of the first was “to determine the trend of the concentration of sodium in serum as glucose declined during treatment of uncomplicated episodes and of episodes with complications attributable to brain swelling”. The paper found that the concentration of sodium failed to rise as that of glucose declined in 82 of 164 uncomplicated episodes and in 18 of 20 complicated episodes - 13 of the 20 representing minor complications, such as headaches, and 7 representing major complications (death or near death). 58 cases of DKA in patients aged 1½ to 20 were then studied prospectively on a 48 hour treatment plan; in 55 of these cases serum sodium concentration rose in the first 12 hours and in none of them was there a major complication. The authors concluded that “failure of sodium concentration measured in serum to rise as glucose concentration falls is a marker for excessive administration of free water”, and suggested treatment involving the infusion of higher sodium salt concentrations over an expanded “repair period”, as a regime which “may be protective against near-death episodes and brain herniation during treatment”. In their text they put their views on causation and treatment as follows:

“What remains obscure are (1) the reason that some patients who sustain negative sodium trends succumb to severe complications, whereas others with negative sodium trends appear to do well, and (2) the reason that some children with reported positive sodium trends do poorly. A critical combination of pretreatment factors (age; volume of free water ingested; degree of initial brain swelling; degree of acidemia, dehydration, and hyperosmolality) coupled with management-related problems (administration of excessive volumes of solutions often hypotonic relative to the patient; a rapid decline in osmolality or a critically low osmolality during the first 24 hours of treatment, or both) probably converge, resulting in movement of water into the brain at a rate faster than it can be accommodated. The result would be an elevation of cerebrospinal fluid pressure to the point of symptoms or death.

We conclude that DKA is a state of intracellular starvation, hyperglycemic (and more rarely hypernatremic) hyperosmolality, systematic dehydration, and, not infrequently, brain swelling. Although the causes of cellular swelling may well be multiple, we suspect that all will prove to have a common pathophysiologic pathway, namely the movement of water driven across membranes by osmotic forces. This situation calls for vigilance in the repair of water, glucose, electrolyte, and acid-base disturbances from the outset of therapy, to avoid aggravating water shifts and any possible evolution of subclinical brain swelling to increased intracranial pressure and herniation. By expanding the treatment plan to 48 hours and using repair solutions of higher sodium salt concentrations, the physician can achieve the expected rise in the measured concentration of sodium in the majority of patients. This approach should be coupled with careful monitoring of mental status and circulation, along with frequent measurement of serum concentrations of glucose, sodium, potassium, total carbon dioxide, urea nitrogen, and blood pH. We believe that such care will minimise the risk of near-death episodes and death from herniation, and will help prevent even lesser degrees of brain swelling during the treatment of DKA.”

32. The judge observed that Dr Edge had pointed out that, although all those suffering cerebral oedema had a negative “calculated” sodium trend, 50% of those without cerebral oedema also had such a negative sodium trend, and said that it was therefore “difficult to incriminate negative sodium trend as a cause rather, again, than a potential risk factor”. The difficulty that I can see about that observation is that it seems to postulate the existence of one single cause which will always cause cerebral oedema if it is present. Clearly, as the appellant’s case accepts, the position is more complex. Some patients may be pre-disposed to suffer from cerebral oedema, if their sodium level is allowed to fall as their glucose level falls. Others may escape. The lack of salt, or to put the matter the other way round the excessive free water in their blood as a result of the lack of salt, may be sufficient to trigger cerebral oedema in some but not other patients. Other factors, such as weakening of the barriers between the vascular system and the brain cells may play their part. What Harris et al were doing was identifying a factor which they suggested could trigger cerebral oedema, in those cases where the patient’s condition otherwise pre-disposed him or her to that condition.
33. Harris & Fioridalisi’s 1994 paper, Physiologic management of [DKA], was a prospective study of 231 consecutive episodes of DKA in 149 patients aged 10 months to 20 years admitted during a 5-year period, and had as its objective “to determine whether gradual rehydration in moderate and severe [DKA] can safely prevent untoward declines in calculated effective osmololity … early in treatment and, hence, help prevent major central nervous system complications”. On 209 occasions the patient’s condition was managed throughout in accordance with guidelines, whereby plasma calculated sodium level was not allowed to fall. In the remaining 22 cases, the initial therapy did not comply with such guidelines. None of the patients on any of these occasions suffered cerebral oedema. The judge said that this caused him “to consider that this regimen might at least provide some clue at the least to [sic] the extent of any potential cerebral oedema”. However, on six occasions (five of them showing positive sodium trends), Mannitol (the effect of which is to withdraw water from the brain) was administered because of mental status changes, two of these occasions relating to the 209 cases; and Dr Tasker in cross-examination accepted that, but for the administration of Mannitol, it was possible that these six patients might have suffered cerebral oedema. He was asked to compare a ratio of 2 out of 209 with the average incidence of cerebral oedema reported in the 1998/9 British Paediatric Surveillance Unit Annual Report, being 11.9 per 1000 in newly diagnosed diabetic children and 4.3 per 1000 in known diabetics, making an average of around 7 per 1000, which it was put to him was in the same region as 2 per 209. His answer was that the statistics of 11.9 and 4.3 per 1000 were derived from a “wrong control group”, selected for different reasons at a different time and not a true epidemiological study. The judge in his judgment he said that “it is not immediately apparent that, on the basis of Dr Tasker’s concession [viz. that without the Mannitol some or all of the six might have gone on to suffer cerebral oedema], the outcome was so much better that probable cause could be inferred”.
34. Before us, counsel submitted on behalf of the appellant that it was not comparing like with like to compare cases where Mannitol had been administered and no cerebral oedema suffered with cases where cerebral oedema was suffered. The British Paediatric Surveillance Unit Annual Report did not give any figures for cases in which Mannitol had been administered and no cerebral oedema had followed. There may be force in this explanation, but it is not a point which Dr Tasker made or was asked about. Further, Dr Edge in her latest paper had drawn attention to the possible significance of the fact that there were still six children to whom Mannitol had been required to be administered. Dr Tasker further accepted that cerebral oedema could sometimes occur despite a rising calculated sodium level. It seems to me therefore that the judge had reason to be cautious about drawing any positive conclusions about the relationship between sodium levels and the incidence of cerebral oedema on the basis of Harris’ second paper.
35. Turning to Dr Edge, the judge started by analysing her writings on the incidence of cerebral oedema in DKA, a subject in which she had specialised. First, she wrote a paper with Dunger in 1994, Variations in the management of [DKA] in children. It examined 25 management protocols from throughout the UK and 22 questionnaires returned by 30 members of the British Society of Paediatric Endocrinology known to have an interest in diabetes. On this basis, the paper sought to relate variations in fluid, bicarbonate, insulin and potassium regimens to the incidence of cerebral oedema in each centre. 10 centres had experienced no cases of cerebral oedema, as against 8 which had experienced such cases. The paper tabulated “the only significant differences between the protocols between those centres” which had and had not seen cases of cerebral oedema. The paper summarised the relevant differences as follows:

“Those units which had experienced cerebral oedema tended to recommend larger volumes of blood plasma to resuscitate, larger volumes of maintenance fluids and to be more likely to change to [one/fifth normal] rather than [one-half normal] saline once the blood glucose had fallen during treatment.”

It was this table and the potential relevance of these differences that the judge wrongly thought had not been put to Dr Edge. The differences were put (at TR4/75-77), where there was full cross-examination to the effect that the presently relevant difference consisted of the change (at a very early stage) to a low saline infusion. The effect of Dr Edge’s evidence was that her paper written with Dunger in 1994 suggested no more than a possible difference in the incidence of cerebral oedema according to the protocols followed, that it would require more detailed studies to confirm this as any sort of probability, and that she was seeking to make such studies (although she expressed some doubt even as to their feasibility in view of the rarity of cerebral oedema during DKA).

36. Dr Edge’s paper went on to refer to Harris et al’s “large retrospective study”, saying that in it the only feature of management that “may” have been related to the development of cerebral oedema was a failure of the plasma sodium concentration to show the expected rise as the blood glucose concentration fell during the early phase of treatment. It acknowledged that “other authors have not found strong evidence for a role of free water overload”, citing Rosenbloom’s 1990 paper. Nevertheless the text concluded:

“The results of this audit, suggesting an association between fluid administration and the incidence of cerebral oedema, are retrospective and based on recall which may not be reliable. Also, it is not known to what extent the children actually received therapy according to each unit’s guidelines. In spite of these misgivings, this audit has been helpful in pointing out the areas of management which may be amenable to randomisation in a careful prospective study.”

37. It was following this paper that Dr Edge assisted in drawing up what became the 1995 guidelines approved by the British Society of Paediatric Endocrinology and later accepted by the British Diabetic Association, which provide for a change to a half normal saline infusion once glucose levels have dropped below 12 mmol/l. The guidelines are prefaced by a statement that they represent a “consensus”, but “cannot be considered perfect as complications still arise and the pathophysiology of cerebral oedema is not completely understood”. Then in 1996 Edge & Dunger published in Balliere’s Clinical Paediatrics their next, very full paper, entitled [DKA]: what is safe and effective treatment. The judge quoted its concluding summary:

“Diabetic ketoacidosis remains a life-threatening condition. There has been no change in the incidence over the last 10 years, and there is a greater risk of late recognition with the trend towards a younger age of presentation. Clear guidelines for treatment are necessary, although they must always be tailored to the individual. Management is usually straightforward, but complacency is unjustified, as there is still a significant mortality and morbidity, largely arising from the unpredictable complication of cerebral oedema. The pathophysiology of this devastating condition is still unknown, and there is a need for more research in this area. In the meantime, no guidelines can be considered completely safe. Good supervision from senior members of staff is essential, and there should be early concern if progress is not as predicted. Rapid intervention with Mannitol and hyperventilation is necessary if signs of cerebral oedema develop.”

He also quoted from the long section of the paper headed “Pathophysiology”. In that section the authors described in detail the effects of DKA; they pointed out that, once rehydration was commenced, blood plasma and cerebral spinal fluid (“CSF”) would become hypo-osmolar (less concentrated) with respect to brain, “favouring net water movement into brain cells” and added that “treatment that lowers plasma (and CSF) osmolality rapidly may be expected to increase the likelihood of brain swelling”. In a sentence quoted by the judge, they said that “These considerations of fluid balance provide insights but no clues which might explain the sporadic and unpredictable nature of cerebral oedema” (a sentence to be read presumably with a notional comma after “insights”). The authors then identified a number of theories as to aetiology advanced since the condition was first described in 1936. They said that it “was initially thought” that cerebral anoxia from the reduced blood volume might have caused it; that hypoxia from rapid bicarbonate infusion was postulated in 1980, but that no case series had been able to implicate bicarbonate infusion, which anyway was “now much less common” without incidence rates in cerebral oedema having changed. The authors mentioned as suggested contributing factors “a high initial plasma glucose concentration (Durr et al., 1992), excessive rates of intraveneous administration (Duck & Wyatt, 1988), and a fall in plasma sodium concentration (Harris et al, 1990; Harris and Fiordalisi, 1994)”, adding that “Animal studies have suggested that insulin itself is required for cerebral oedema to occur, since rats failed to develop the condition when hyperglycaemia was treated with saline alone (Arieff & Kleeman, 1973, Tornheim, 1981)”. After this, in another passage the first sentence of which was quoted by the judge, the authors wrote:

“There is accumulating evidence that the development of cerebral oedema may be related to the rate and quantity of fluid administration, and failure of sodium and hydrogen pump mechanisms in the brain cells. This, together with the accumulation of intracellular products of metabolism may lead to fluid shifts into the brain cells.”

38. The authors also referred to Duck & Wyatts’ suggestion that fluid rates should be less than 4 ltrs/m²/day, and to Harris’ suggestion that plasma sodium concentration should be maintained (although observing that Rosenbloom had “failed to find any evidence to support either proposition”). After a further reference to the differences in experience reported in Edge & Dunger’s own 1994 paper between centres using one-fifth and one-half normal saline, the authors commented:

“The implication that more concentrated rehydration fluids carry a lower risk requires confirmation.”

The authors concluded this section by observing that the many previous reports of collections of cases on a retrospective basis had as their main failing the fact that

“there has very rarely been a control group of subjects who presented with DKA but did not develop cerebral oedema, and so all the possible factors described above may also be present in uncomplicated DKA. …. There have been only three small case-control series reported in the literature, and these have all been retrospective analyses [citing, inter alia, Harris et al, 1990]. There have been no prospective case-control studies in this important area of research.”

39. In 1998 a further paper was published by Ford-Adams & Edge, entitled Coma in the child with diabetes, containing the following passage quoted by the judge:

“The pathophysiology of cerebral oedema is poorly understood. Most cases occur some hours into treatment so it has long been believed that cerebral oedema is a result of over-enthusiastic treatment leading to rapid shifts into brain cells causing oedema. A number of mechanisms have been suggested for this including accumulation of intracellular osmoles, defective glucose uptake in rehydration and increases in arginine vasopressin levels. However, there are also cases of documented cerebral oedema occurring before any treatment has been started. Features which are thought to increase the risk of cerebral oedema are:

Newly diagnosed diabetes

Age <5 years

Blood glucose >35 mmol/l

Blood pH<

Severe prolonged dehydration

Fall of plasma sodium concentration during treatment.”

40. Lastly, the judge referred quite shortly to Dr Edge’s paper, Cerebral oedema during treatment of [DKA]: Are we any nearer finding a cause?, which was at the time of trial unpublished although accepted for publication. In it she said that “The cause (or causes) of the cerebral oedema occurring in association with DKA is still somewhat of a mystery, and is the most fascinating aspect of this condition. …. [I]t has become apparent that there is no single cause, and multiple factors may be operating. Some attempt has been made to list the possible contributing factors in Table 1”. Table 1 read:

Table 1

Theories as to possible contributing factors to the development of cerebral oedema.

Mechanisms Clinical Factors

Hypoxia Degree of dehydration

Use of bicarbonate

Excitatory amino-acid receptors

Reductions in plasma osmolality Excessive fluid volumes

Rapid fall in blood glucose levels

Failure of plasma sodium levels

to rise during treatment.

Large doses of insulin

Accumulation of intracellular osmoles Duration of symptoms before

admission

Sodium-proton antiporter Degree of acidosis

Ketones”

41. The text of the paper dealt at this point with hypoxia in similar terms to the 1996 paper, giving it little prominence. Under the head “Excessive rate of rehydration”, it started: “The rate at which fluid is initially infused may be expected to relate to the speed of change of osmolality and electrolyte concentrations in the brain extracellular fluid, and thus to influence the development of cerebral oedema”. Referring to Duck & Wyatt’s findings regarding the coincidence of cerebral oedema cases with received fluid rates in excess of 4 ltrs/m²/day, the paper pointed to the absence of controls, and said that “it has been stated that contemporary management guidelines in the USA would recommend such fluid rates, and other case-studies have not found any relationship between fluid rate and cerebral oedema”. Dr Edge referred in the latter connection to Rosenbloom’s 1980 paper, but she added, for the first time, this note of caution:

“However, in Rosenbloom’s series, although half of the patients were said to have had inadequate fluid replacement, this was based on retrospective analysis of estimated fluid requirement from a “common” formula assuming 10% dehydration in all cases.”

42. I interpose another possible note of caution regarding Rosenbloom’s findings, in that Dr Tasker gave uncontradicted evidence that Rosenbloom’s own figures show that, of 52 patients for whom serum sodium levels are given both at the start of treatment and at the time when cerebral oedema develops, 43 had a serum sodium level that had either fallen or not risen (as against 9 whose sodium level had risen). After further references, Dr Edge’s recent paper concluded that “it seems likely that in at least some cases, the rate of rehydration might have some role in its [i.e. cerebral oedema’s] pathogenesis”.
43. Under the head of Plasma Osmolality and sodium levels, the paper started: “Change in osmolality of plasma and CSF might also be expected to produce changes in brain cell size”. It cited inter alia the conclusion of a 1990 paper by Bello & Sotos that “the best predictors of cerebral oedema were duration of symptoms and fall in serum osmolality”, and suggested that this also showed that “the fall in plasma sodium may be more specific for the development of cerebral oedema”. It cited first “two specific case reports which support the idea that a reduction in plasma sodium may be important”, and secondly Duck & Wyatt, before saying that “the most convincing evidence for an effect of plasma sodium concentration in the pathogenesis of cerebral oedema” comes from Harris et al’s 1990 paper. Dr Edge also cited Harris & Fiordalisi’s 1994 paper, and commented on the six children in whom behaviour change or increasing obtundation were treated with Mannitol (cf paragraph 34 above). She mentioned that there was further support for a relationship between falling plasma sodium concentration and cerebral oedema from a small study in very young children (by Hale et al in 1999), but added: “It is however possible that the falling sodium concentrations may be a consequence of incipient cerebral oedema owing to dysregulation of anti-diuretic hormone, ie. an effect, rather than a cause”, citing Rosenbloom’s 1990 work. Her paper also cited American experience, with use of normal saline for only the first hour or two, but a “remarkably similar” overall incidence of cerebral oedema, as “suggesting that the fluid type used may have a limited impact on cerebral oedema risk”, and said that this was “supported by an Australian study [by Mel & Werther in 1995] showing a similar incidence during two time periods when different fluid regimens were used”.
44. Dr Edge’s paper went on to consider the possible significance of insulin dosage and blood glucose concentration, but found nothing in the case studies to implicate the former, and no strong evidence for an effect of plasma glucose on the development of cerebral oedema. She concluded this section by saying:

“It is difficult to tease all these predisposing factors apart, since those who have the highest blood glucose levels are likely to be those who are the most severely dehydrated, receive the most fluids and have the greatest change in plasma osmolality and sodium during the first few hours.”

Dr Edge then examined in her paper “theories as to what may be happening within the brain cells to produce the final common pathway of cerebral oedema". She postulated as possible mechanisms: 1) Accumulation of intracellular osmoles; 2) the role of vasopressin and atrial natriuretic factor; 3) sodium-proton antiporter; 4) hypoxia revisited; and 5) role of ketones. Her conclusions were:

“The mechanisms which lead to the development of cerebral oedema in DKA are probably multiple and a complex interaction of many of the pathways discussed. Mechanisms of maintenance of brain cell volume during osmolal stress are obviously extremely important, but how these relate to the changes in fluid and electrolyte balance during the treatment of DKA is still very poorly understood. The problems with retrospective studies of cases from the published literature are several. Firstly, the patients have been treated at different times, in different places and by different clinicians. This problem may be impossible to overcome, since cerebral oedema is sufficiently uncommon that multi-centre studies are essential to examine predisposing factors adequately. Secondly, because of the retrospective nature of the studies, not all the data is available for analysis. Thirdly, and probably most important, in the majority of studies there is no control group of children with DKA who have been treated in a similar way but who did not develop cerebral oedema.

These shortcomings need to be overcome by a large case-control study of children who develop cerebral oedema during treatment of DKA compared with children with DKA who do not develop cerebral oedema. Such a study is currently being carried out using the resources of the Royal College of Paediatrics and Child Health Surveillance Unit in the UK, and results of the incidence and outcome of cerebral oedema have now been published (5).

It was recommended as early as 1971 that plasma osmolality should be reduced very slowly during treatment of DKA (86), but we still seem no further forward in preventing cerebral oedema. Despite recommendations to avoid large volumes of hypotonic fluid, to ensure that blood glucose falls slowly, to give insulin in small doses, and to ensure that plasma sodium levels do not fall, the risk of cerebral oedema does not appear to have reduced over the past two decades. Until the pathophysiology of cerebral oedema during DKA is further defined and careful prospective clinical studies are completed, we cannot be dogmatic about treatment recommendations.”

45. Under the first of these mechanisms, Dr Edge again identified the possibility that “treatment which lowers plasma (and thus CSF) osmolality rapidly may be expected to increase the likelihood of brain swelling”. Under the second, she indicated that it was uncertain whether vasopressin had a role. Under the third, she cited a hypothesis according to which this “exchanger” involved in cell volume regulation might play a role. Revisiting hypoxia, she mentioned a recent “revitalisation” of the “appealing” theory that it was relevant, but observed that there was very little evidence for hypoxia in most cases of DKA. Finally, she referred to “the most recent theory of all” as implicating the presence of ketones (apparently linked with acidosis) in the aetiology of cerebral oedema, by virtue of their capacity to alter the permeability of the blood brain barrier, and said that “there can be no doubt that this theory will be examined further”. I note in this regard that Neil on admission gave a strong positive reading for ketones (cf paragraph 2 above).
46. In her evidence (TR2/4/5C-D), Dr Edge agreed that it was too early to say what impact the use of the guidelines was having on the incidence of cerebral oedema. She said:

“I think we would have to wait 5 years of so before we would get any idea of whether the incidence of cerebral oedema in this country has been reduced by the use of these guidelines.”

All that can therefore be said is that, notwithstanding the general adoption of the guidelines, cases of cerebral oedema still occur, as indicated by the British Paediatric Surveillance Unit Annual Report which the judge cited.

47. I have reviewed the written material bearing on the causation of cerebral oedema during treatment for DKA at even greater length than the judge did in his clear and comprehensive judgment. I have done so, because of the emphasis which (not surprisingly, having regard to the judge’s assessment of the quality of the oral evidence before him) was placed upon the written material in the appellant’s submissions before us. The essence of those submissions is that the treatment in this case involved the administration of significant quantities of fluids (more in fact than Duck & Wyatt recommended) having (most significantly) a much lower salinity than good practice required. The effect was to infuse Neil with low tonicity fluids, and so to create necessarily a lower serum sodium level in the blood and a greater imbalance than normal between blood plasma and brain cell tonicity. It is an imbalance between blood plasma and brain cell tonicity which is a pre-condition to cerebral oedema, while a fall in serum sodium concentration has also been associated with its occurrence. So where an abnormally large imbalance in tonicity has been allowed to arise, accompanied by a slight falling of serum sodium levels overnight, and cerebral oedema has followed, the natural conclusion, in the appellant’s submission, is that the treatment has caused the cerebral oedema. The appellant argues that Rosenbloom’s 1990 paper should be discounted as a retrospective study open to the criticism made by Dr Edge in her most recent paper, and that Harris’ papers constitute compelling, prospective studies, consistent with the conclusions and table contained in Edge & Dunger’s own 1994 paper based on the answers to questionnaires examining the success rates attaching to treatment with higher and lower rates of saline infusion.
48. Mr Langstaff QC for the appellant submits that the judge placed too much weight on the fact that some cases of cerebral oedema occur even prior to treatment or despite treatment complying with the guidelines (see e.g. TR pp. 36-37) or despite a negative serum sodium trend (TR p.33). The published material establishes the likelihood that a number of factors play a part in the development of cerebral oedema. That some of them suffice in some cases does not mean that others are not the trigger in others. That those others may not, when present, always trigger cerebral oedema again does not mean that they do not do so in some or many cases. Another, linked criticism made by the appellant is that the judge approached the evidence as if it was necessary to resolve the issue of causation generally, when all that was required was to determine whether the respondents’ established fault was a triggering cause in this case of Neil’s cerebral oedema (or was the cause of cerebral oedema as severe as that which he suffered). Even looking at the literature generally, the appellant submits that the relevance of fluid management, tonicity and sodium levels is sufficiently shown as a matter of probability (as distinct from scientific proof), particularly by Harris’ and Dr Edge’s own papers; that the administration of “free water” (or water of low tonicity) must exacerbate the disparity in osmolality which is the pre-condition to brain swelling; and that Dr Edge herself accepted that “the use of hyptonic fluids, as we have said, may be a cause in some cases”. Her immediately following observation that “that doesn’t answer what [sic] the majority of children, why children develop cerebral oedema. There are many, many things we don’t understand” should not in the appellant’s submission divert attention from an obvious triggering cause in this case. The existence in Neil’s case of at least two other factors associated in the literature with the incidence of cerebral oedema, that is an extended history of illness and excessive acidaemia, may in the appellant’s submission have meant that he was more prone to cerebral oedema, but by themselves these were likely to be predisposing conditions (affecting for example the integrity of the membranes between the vascular system and brain cells and possibly already leading to a degree of cerebral oedema, even prior to treatment), whereas the essential trigger for brain swelling was an imbalance in tonicity allowing fluid to cross the membrane by osmosis, and such an imbalance would directly result from the substantial quantities of hypotonic fluid administered in this case to Neil.
49. The nature of the causative link that the appellant had to establish was the subject of submissions. It was not suggested, either below or before us, that any question could arise of an award purporting to represent a percentage assessment of the risk or chance that the cerebral oedema which Neil suffered in 1991 resulted from the respondents’ negligence (cf Hotson v. East Berkshire Health Authority [1987] AC 750). Mr Langstaff referred us, however, to Bonnington Castings Ltd. v. Wardlaw [1956] AC 613, Nicholson v. Atlas Steel Foundry and Engineering Co. Ltd. [1957] AC 613 and Gardiner v. Motherwell Machinery and Scrap Co. Ltd. [1961] 1 WLR 1424. These cases, particularly the first two, as explained in Fairchild v. Glenhaven Funeral Services Ltd. [2002] UKHL 22; [2002] 3 AER 305, establish that, where an employee is in the course of a single employment exposed to a noxious agent, such as dust, in circumstances where the greater the exposure the greater the risk of his contracting a disease, and a significant part or period of the total exposure is shown to have resulted from his employer’s negligence, the employer’s negligence may be regarded as having materially contributed to the disease for the purposes of grounding tortious liability (for the whole of the injury suffered). In cases within this category, the test is not whether the claimant has shown on a balance of probabilities that but for the employer’s negligence he would have avoided the disease: cf Fairchild per Lord Bingham at paras. 14-16 and Lord Rodger at para. 129.
50. McGhee v. National Coal Board [1973] 1 WLR, also explained in Fairchild, involved a further step. The workman had suffered dermatitis. He had been exposed during his work to coal dust, an exposure which had been prolonged unnecessarily by the defendants’ negligence in failing to supply washing facilities to enable the workman to wash before cycling home. But the medical evidence left entirely open whether the dermatitis resulted from the length of exposure (in which event the case would have been on all fours with Bonnington and Nicholson) or from a single abrasion which could have been sustained (a) at work or (b) while cycling home or at home. The defendants’ negligence lay only in allowing exposure during period (b). Nevertheless the defendants were held liable, because of the increased risk to which they had exposed the claimant.
51. Wilsher v. Essex Area Health Authority 1987] QB 730 (CA); [1988] AC 1074 was a medical negligence claim. The hospital negligently administered excess oxygen. That increased the risk that the claimant would suffer RLF, an incurable condition of the retina impairing vision or causing blindness, as he did. However, there were other factors, some of which could be shown to apply to the claimant, which without any negligence could also have led to RLF. In these circumstances the claim failed. In Fairchild, a majority of the House of Lords disapproved a dictum of Lord Bridge’s in Wilsher, which had suggested that McGhee was an example of the application of a conventional balance of probabilities test to the issue of causation. But it regarded Wilsher as correctly decided. Lords Bingham and Rodger approved in this regard reasons given by Sir Nicholas Browne-Wilkinson V-C, as he was, dissenting in the Court of Appeal in Wilsher, namely that there is a relevant distinction between disease resulting from exposure to a single noxious source, to which the claimant has been exposed (even though the defendant’s negligence only relates to part of the exposure), and disease which may have resulted from “any one of a number of noxious agents” (cf per Lord Bingham at para. 22 and Lord Rodger at para. 149). Lord Hoffmann did not regard this distinction as principled (cf para. 72 – and note also Mustill LJ in this court in Wilsher at pp.771-772), but supported the distinction between the results of McGhee and Wilsher on policy and pragmatic grounds related to the distinction in subject-matter between an employer’s liability claim and a national health authority’s duty to take medical care (cf paras. 69-70).
52. Fairchild itself was a case of an employee being tortiously exposed in several different employments, in circumstances where the relevant disease, mesothelioma, could have been caused in any one of such employments. The majority of the House of Lords extended the “increased risk” approach to causation, which had been adopted in McGhee, so as to render all the employers in Fairchild liable. It was unnecessary to determine what the position would be if (a) the exposure had been tortious in only one or some, but not all the relevant employments, or (b) different but in each case tortiously actionable noxious agents had been involved in circumstances where it could not be said which agent had caused the injury (cf per Lord Rodger at para. 170).
53. Before us Mr Langstaff does not suggest that either McGhee or Fairchild has any application. I am also not persuaded that Wardlaw and Nicholson are of any assistance to the appellant’s case. There was there no doubt about the identity of the noxious agent and that it caused the claimant’s disease on an accumulative basis, i.e. the greater the exposure to it, the greater the risk of the disease. Here, the respondents challenge root and branch the existence of any single causative agent; they contend, with justification, that the evidence and findings show that there are a number of different causes which were potentially in play, and that these may operate independently or cumulatively. In short, the circumstances resemble those in Wilsher. The claimant has to establish on a balance of probabilities that the administration of hypotonic fluids triggered Neil’s cerebral oedema.
54. Summarising the position, there were and are, on the facts and evidence in this case, other potential causes, which may cause cerebral oedema, even independently of the hypotonicity of the blood plasma. The substantial history of the appellant’s illness prior to admission and its severity are factors which on the evidence may not only have predisposed him to cerebral oedema, but which, on the current state of medical knowledge, may by themselves or in conjunction lead to cerebral oedema, without any excess infusion of hypotonic fluids (cf paragraphs 40, 44 and 45 above). The judge in these circumstances, and faced with contentious medical scientific issues and a paucity of statistical material to assist in their resolution, had to form a view as to the causation of Neil’s cerebral oedema, and as to whether it was possible on the balance of probabilities to attribute it to the cause assigned by the appellant. We must as an appellate court review his conclusions bearing in mind the general deference due to tribunals of first instance in respect of matters of fact, even when they depend on expert evidence. As Lord Bridge said in Wilsher at p.1091G:

“Where expert witnesses are radically at issue about complex technical questions within their own field and are examined and cross-examined at length about their conflicting theories, I believe that the judge’s advantage in seeing them and hearing them is scarcely less important than when he has to resolve some conflict of primary fact between lay witnesses in purely mundane matters.”

55. The judge was not of course bound to accept the evidence of any of the expert witnesses, if he considered as a matter of legal probability, rather than scientific certainty, that causation was established. But I do not read him as treating himself as so bound. He very properly sought to test the experts’ evidence by reference to the scientific writing put before him. He formed a clearly favourable view of Dr Edge and her testimony, in contrast to that of Dr Tasker and Professor Brook. I do not read his judgment as omitting to focus on the position in the instant case, or as seeking to resolve the general problems of causation of cerebral oedema rather than the particular case before him. The two were of course related, and the expert evidence addressed both. The reasons the judge gave for accepting Dr Edge’s views are open to some criticisms at certain points, but they very largely survive, particularly in the core areas. Such criticisms as may be made of the judge’s approach to Dr Edge and her testimony cannot I think by themselves outweigh or undermine the positive impression made by her oral testimony, supported in substantial measure by the published material. The judge regarded the papers on which the appellant places especial reliance as no more than pointers towards a possible aetiology, and they are pointers which are themselves not uncontentious and derive from very limited statistics and studies. Dr Edge was consistent in her views to this effect both in her published material and in her evidence. Her published material will have been directed primarily to issues of scientific certainty, but in her oral evidence she focused on the issues relevant to this case.
56. In all these circumstances, I have come to the conclusion, with some reluctance, and with great sympathy for Neil’s tragic condition, that we cannot come to any conclusion different from the judge’s conclusion that no relevant link has been established even as a matter of probability between the respondents’ negligence and the cerebral oedema that Neil suffered, and that this appeal cannot therefore succeed.

Lord Justice Schiemann :

57. I agree that this appeal ought to be dismissed although I have felt throughout an instinctive desire to find in favour of Neil’s claim.
58. He was treated with an infusion of one fifth normal saline and developed cerebral oedema.
59. Various theories exist as to the possible cause or causes of cerebral oedema. One of these theories is that (1) the excess water in the brain cells arrived there by a process of osmosis from the neighbouring blood, extra-cellular fluid and cerebro-spinal fluid and (2) this process is increased if the infusion is at less than normal salinity. However, the problem is so rare, the studies so few and the variables so many that no conclusion can be drawn from such statistics as there are, other than that they do not rule this out as a possibility.
60. Given (1) our lack of knowledge as to what causes cerebral oedema and (2) the fact that a physically plausible cause of oedema is that described in the preceding paragraph, and (3) that there are no significant advantages which flow from using an infusion of such low salinity, the medical profession in this country has advised against giving such an infusion. The judge found that therefore the doctor who administered that infusion was negligent in so doing.
61. Within hours of the infusion Neil developed cerebral oedema. So it is no wonder that Neil’s parents feel that the hospital is to blame for Neil’s cerebral oedema. The doctor did what he should not have done. Precisely that has happened which the medical profession feared might result.
62. The issue between the parties in this appeal is whether the judge was entitled to conclude that it had not been shown that the infusion was the probable cause of Neil’s cerebral oedema.
63. Now the mere fact that the cerebral oedema followed the negligence does not show that the negligence probably caused the cerebral oedema. If I have an egg for breakfast and then fall under a bus on the way to court no one would think that it was my eating the egg which caused the accident. That is easy to understand.
64. What is less easy to understand at first is how someone in the doctor’s position can escape liability. The medical profession is of the view that the course of action which he followed should not be followed. The reason the profession is of that view is because it is thought that giving this treatment might cause the damage which in the event occurred. Many would intuitively feel with the parents that in those circumstances if the damage occurs then the negligent doctor is responsible.
65. But things are not that simple. If we knew that giving the negligent treatment caused damage to x% of patients and we knew nothing else then we would say that there was a risk that it was responsible for the injury. But we do not know that. All the profession knows is that (1) it does not know that following negligent treatment ever harms anyone, but (2) that following the negligent treatment may harm some and has no significant advantages.
66. Whether in truth there is a risk that administering a low saline infusion in some cases is a cause of cerebral oedema is independent of man’s knowledge. It is an objective fact. One can illustrate this by a simple example. It was not always known that there is a risk that sexual intercourse might result in pregnancy. Now it is known that in a proportion of cases it does so result. The mechanism is also known. But the risk was there before the mechanism was known or the risk appreciated.
67. The claimant’s problems in the present case arise from the finding of the judge that in the present state of knowledge it is not known whether the giving of a low saline infusion is ever a cause of cerebral oedema. He accepted it as a possibility but not as a probability. That finding is one to which the judge on the evidence before him was entitled to come.
68. In order to succeed in his claim Neil must prove, at least, that administering a low saline solution significantly increases the risk of cerebral oedema for some groups of which he was a member. One of the many difficulties is that there have been very few prospective studies; it is difficult to find a statistically significant control group as between the members of which other things are equal and to some of whom a low saline solution is given and to others of whom an infusion of normal salinity is given.
69. It may be that, when knowledge has advanced further, it will be shown that the giving of a low saline infusion is a cause of cerebral oedema in a percentage of cases. Then people in Neil’s position may well succeed in a claim against a doctor who does this. It may on the other hand be that, when knowledge has advanced further, it will be shown that the giving of a low saline infusion never causes cerebral oedema. Then people in Neil’s position who claim will undoubtedly fail and indeed the treatment will no longer be stigmatised as negligent.
70. Meanwhile we are in a state of not knowing whether or not cerebral oedema is ever caused by the negligent treatment. This lack of certainty might not have prevented Neil from recovering if he had been able to show that the weight of scientific opinion was of the view that for people in Neil’s situation the giving of a low saline infusion would significantly increase the risk of cerebral oedema. However he was not - as my Lord has demonstrated in his review of the material which was put before the judge.
71. It is quite clear from the evidence that one can not say with any confidence that administering a low saline solution significantly increases the risk of cerebral oedema for any group.
72. In those circumstances, I agree that this appeal must fail.

Lord Justice Mantell:

73. I agree that this appeal must be dismissed for the reasons given by Mance LJ and by Schiemann LJ in his judgment, which I have had the opportunity to read in draft.