BK-S (Children) (Expert Evidence and Probability)

This is an appeal against findings of fact made by His Honour Judge Levey sitting in the Family Court at Portsmouth on 11 June 2014. At the conclusion of the hearing of the appeal the court dismissed it with reasons to follow. These are my reasons for concurring in that decision. The welfare hearing that was listed to follow on from the appeal remained undisturbed by the proceedings before this court.

The findings were made in public law children proceedings brought by Hampshire County Council in relation to four children of a mother by three different fathers. The findings concerned one of the children, a young boy who at the time of the events was six months old. I shall refer to him in this judgment as Z.

Most of the background facts are not in dispute. The key issue in the case was the identification of one or more perpetrators of harm caused to Z from the theoretical pool. Z had been discovered to have the drug Olanzapine in his body on three occasions. Olanzapine is an anti-psychotic drug used for the treatment of schizophrenia and bi-polar disorder. It is only available on prescription and was not prescribed to any relevant person. It has some sedative effects and can alter consciousness. In large doses it can be fatal to a child but recovery from moderate doses is relatively swift leaving no identified lasting injury. It is available in tablets, capsules and ‘melts’ which dissolve when placed on the tongue or in the mouth. It can be administered orally or rectally and in solution in other fluids such as water or milk. It can be injected but there was no evidence of any needle marks on Z.

Z was admitted to hospital on three occasions: 15 July 2013, 21 July 2013 and 5 August 2013. The first admission was thought to be for gastro-enteritis. The second and third admissions were for an undiagnosed illness with an overlay of altered consciousness. A serum test was taken on the second occasion and a blood test was taken on the third occasion. The results of the blood test were available on 13 August 2013 and showed a concentration of Olanzapine in Z’s blood as at 5 August 2013 of 257 micrograms per litre (ug/L). On 14 August 2013 Z was re-tested and a smaller concentration of Olanzapine at 43 ug/L was found. The accuracy of these two results is not in question. All of the children were removed from home as a consequence of what was discovered.

The examination by the court focussed on who had been in contact with Z and accordingly who might have had the opportunity to administer the drug. As part of that examination, those responsible for Z’s care at the relevant time were identified and the contact that Z’s father had had with him was scrutinised. The pool of potential perpetrators was agreed i.e. those who had the opportunity to administer the drug were Z’s mother, father and paternal grandmother, the latter because she shared father’s contact with Z. There was no evidence of accidental ingestion and everyone in the pool denied both responsibility and having any access to the drug. The judge decided that Z’s mother was the sole perpetrator and that is the finding which is appealed to this court.

Z was in the primary care of his mother at all material times and had contact with his father and his paternal grandmother from 12 noon to 4 pm on 7 July, 14 July, 21 July, 4 August and 11 August 2013. Z was taken into care on 13 August 2013 before the final test result was administered and as soon as the 5 August 2013 test result became known to the local authority. The test result taken on 14 August 2013 did not become available until 28 May 2014.

An expert toxicologist Professor Johnston was instructed to analyse the test results and to give an opinion on the window of opportunity. His evidence assumed significant importance during the hearing such that he was re-called for further cross examination as the evidence unfolded, in particular once the existence of the 14 August 2013 test result became known. Elements of his evidence are not in dispute. They include the following a) that ingestion of the drug would take longer if it was in solid rather than solution form, b) peak symptoms would occur two to five hours after administration, c) peak efficacy and peak symptomology would coincide with the highest level of concentration of the drug in the body and d) the symptoms that would be expected to be observed in a baby or young child who had ingested the drug would be somnolence, lethargy and agitation.

It is also not in dispute that Olanzapine dissipates and is removed from the body over time. The speed at which it dissipates and is removed varies inter alia with age, physiology and metabolism. That can be predicted by what is known as the ‘half life’ of the drug. That was the subject of important scientific evidence that was heard by the judge to which I shall return in due course. In babies the presence of an enzyme CYP1AZ affects the speed at which the drug is dissipated / removed.

Timings and the proximity of care / contact to predicted administration were key to the task the court had to perform. The judge began with a summary of the basis upon which a prediction of the timing of the administration of the drug could be made. He accepted the evidence of Professor Johnston in the following terms:

“Professor Johnston also gave evidence as to the half life of Olanzapine: in other words the time it takes for the concentration of Olanzapine in the blood to reduce by 50%. He described a process by which a drug once in the system reduces by 50% and then by a further 50% and so on. ”

Although it was common ground that the half life of the drug would vary from individual to individual, the question was whether a safe bracket or range of half life predictions i.e. the range of timings for dissipation and removal could be identified so that it could be used to interpolate a time for administration from the test results and timings that had been recorded.

The principal ground of appeal is that ‘the judge wrongly attributed a set time for the excretion of the drug Olanzapine from the body of the child and wrongly conducted a calculation for the time of the likely dose which purported to identify the mother as the sole perpetrator of its administration’. There were other grounds of appeal going to the weight of evidential factors that it is said were relevant to the identity of the perpetrator but before this court Mr Feehan realistically accepted that if he did not succeed on his primary ground, those factors however weighed would not be sufficient to set aside the judge’s finding.

The difficulty in this case is that Professor Johnston’s written materials and oral opinions were based upon a partial knowledge of the test results i.e. he could not have known and did not know of the result of the 14 August 2013 blood test until it was revealed to him during the hearing. The revelation of that result was no doubt the dominant reason for the Professor being recalled to give further evidence. Up until the point that the third test was disclosed it is fair to say that, as Mr Feehan strongly submits, Professor Johnston’s evidence was that there were a wide range of possibilities for perpetration that arose out of the scientific analysis of the half life and timings material and he was not prepared to be drawn into expressing a conclusion. I put to one side the obvious point that it was not for the expert to be drawn into a debate about the probability of one or more of the perpetrators having the opportunity to administer the drug: that was the question for the judge. What Professor Johnston could do was to give an opinion on what was the window of opportunity.

What Professor Johnston had been able to advise when he was first called was that by reference to a research paper referred to in cross examination as the Tanoshima paper, [Acute Olanzapine Overdose in a Toddler: A Case Report, Tanoshima et al, Ther Drug Monit Vol 35 No 5, October 2013] the half life of the drug in children would be quite a bit shorter than in adults. For a child his opinion was that “13 hours sounds reasonable” but with the caveat that the research only related to one child and “we do not know”.

When Professor Johnston was recalled on 28 May 2014, the following oral evidence was adduced:

“Q. [..] There are two reported studies. One that says a half life is 11.6 hours in a 28 month old child. The other one is 13.72 hours for a child of 17 months.

A.

Yes.

[…]

Q.

Can we safely assume – and I mean with almost certainty – that the half life of [Z] would have been less than 21 hours?

A.

I think that would be a reasonable assumption.

Q. Yes. I think you also said in your previous evidence that it would be a reasonable assumption to take the 13.7 in the Tanoshima case as well?

A.

Yes.

Q. Would that be right? So if I were for instance to take 18 hours, that would be safe as well?

A. Yes.”

The judge accepted the evidence that was adduced in the following passage in his judgment:

“Professor Johnston agreed that to assume a half life between 21 and 13.7 hours would be likely, but that working on a half life of 18 hours in those circumstances would be safe.”

Professor Johnston prepared a schedule to illustrate interpolations in graphical form using half life assumptions between 12 and 24 hours. The schedule demonstrates a predicted administration time and date taking the test result of 43 ug/L on 14 August 2013 at 4.50 pm as the base line. The schedule was accepted by the judge and was referred to by him in his judgment.

The significance of that final test result and the interpretation that it permitted is this. Before the sample was taken on 14 August 2013, Z had last been in the care of his father and paternal grandmother at around 4 pm on 11 August 2013. Taking the outer parameters of the half life assumptions of 12 and 24 hours, the level of the drug in Z’s system at that time can be deduced from the schedule to be between 2749.6 ug/L (with a 12 hour half life) and 343.8ug/L (with a 24 hour half life). Professor Johnston specifically confirmed his calculations in oral evidence. His evidence about the effect of various levels of concentration in a child’s body was this:

It would have been much better if the expert had been permitted to answer open ended questions giving his own opinion but that is not the basis of the appeal before us, no-one took objection at the hearing and the parties had the research paper in front of them which informed the propositions that were put. It can therefore be seen that the judge had opinion evidence before him which he accepted (and which was not contradicted) which predicted the effect on a child of an administration during the father’s last contact with the child.

Balanced against that opinion evidence was the factual evidence that Z had not shown any signs of illness during the contact with his father and paternal grandmother and perhaps more importantly, was not reported by his mother to be showing signs of illness when he was returned to his mother’s care on 11 August 2013. That left the judge with two options: either Z’s mother failed to report Z’s symptoms at that time or the Olanzapine that was identified by the test taken on 14 August 2013 was administered after 11 August 2013.

The judge’s conclusion about the administration of the Olanzapine discovered in Z’s system by the test taken on 14 August 2013 was accordingly neither his own speculation nor an unwarranted calculation or deduction of his own. It was a proper inference drawn from the available factual evidence and the un-contradicted scientific opinion evidence. His conclusion, which was in the following terms, is accordingly unassailable:

“[63] As to toxicity he said that it was possible to say knowing that with a level of approximately 250 micrograms on 21^st July that [Z] was extremely poorly, given that even on the most conservative assessment, which I have just set out at 343, the level would have been significantly in excess of that, then one would expect that [Z] would have been extremely poorly. Of course, those levels, 343 is achieved by using the extreme half life of 24 hours and at the lesser half life (the “safe” level as above), at 18 hours, the reading would be 687.6 micrograms, and Professor Johnston’s view was that a level of over 300 would result in extreme symptoms. At higher levels still [Z] would be likely to be unconscious or possibly even dead as a result of the administration.

[64] So in those circumstances, given that he said that the appropriate half life, on a very cautious basis, was 18 hours, in those circumstances [Z], on his evidence, would have been extremely unwell, gravely ill, possibly deceased. […] his view excluded the likelihood of Olanzapine having been administered on the 11^th August. He also commented that it was highly unlikely that it was a residual reading from Olanzapine administered on the 5^th August and thus it was likely that the Olanzapine had been administered more recently.”

The overall context is that the judge decided that the symptoms reported on 21 July and 5 August 2013 were the consequence of the ingestion of Olanzapine. The judge relied upon medical evidence which impressed him to come to that conclusion. He carefully excluded any administration of the drug on 15 July 2013 because the symptoms reported on that day were inconsistent with the same. He considered the fact that Z’s symptoms worsened on 21 July 2013 while Z was in hospital with the consequence that either Z had been administered a further dose of the drug while in contact with his mother in hospital or he had experienced what was described as ‘intestinal stasis’ which can alter the presentation of the symptoms by the effect it has on absorption.

Given the material that he had the judge was able to make a safe finding as to perpetration relating to the period 11 to 14 August 2013. He drew inferences from that finding and other circumstances including the symptoms reported on 21 July 2013 to come to the conclusion that the mother was the perpetrator. This was not a case of two or more improbable propositions being inappropriately elided together to make a probable conclusion. The inherent improbability of a parent poisoning a child did nothing to dislodge the actual evidence that was available and relied upon by the judge.

The hearing at the end of which the findings were made was what is known as a ‘split hearing’ i.e. a hearing limited to a discrete issue of fact without a full analysis of the welfare context. Counsel for the parties before this court acknowledged that the decision to have a split hearing which was taken by a different judge when different advocates were involved cannot have been right given that the issue to be decided was perpetration in the context of an incident of harm, rather than whether the harm occurred.

It is unnecessary for this court to do other than refer to the clear guidance on the point that has been firmly and repeatedly given by this court but just as repeatedly ignored, see for example In the matter of S (A Child) [2014] EWCA Civ 25 at [27] to [31]. There is no discrete issue that would determine the proceedings in a case like this where harm has been suffered and the perpetrator of that harm is unknown. The social work assessments of those in the pool of potential perpetrators may cast important light on the allegations that are to be determined and upon the reliability of those in the pool and the other witnesses and materials that are available.

For these reasons I agreed that the appeal should be dismissed.

I agree.

I also agree.