top of page

Science and Law: Adhering to the Limitations of Scientific Protocols

Updated: Nov 16, 2023

The Insulin evidence

As is very clear, in the case of Lucy Letby, the prosecution stated that the insulin test result was valid. Based upon the verdict of the jury, it is clear they agree the test results were valid. It is not for any person to claim that this is not so, without being able to establish a basis as to why they believe their perspective should override the determinations of the jury.

Science is an iterative process, as such it is formed from asking fairly simple questions, and upon obtaining an answer proceeding to answer further questions. In the case of the insulin poisoning claims one would actually ask a simple first question:

Overall can I trust the test results as they stand?

This question is then broken down into further constituents, so as to actually validate the test results. The first thing one does is to validate the test results based on the parameters set out by the manufacturer.

The test specifications sheet states as follows:

“fasting individuals gave the following results for the 5th-95th percentile, 17.8 - 173 pmol/L.”

Importantly, the infants in this case were not fasting, they were receiving huge amounts of glucose. It is widely understood that glucose causes insulin secretion (Komatsu et al., 2013; Wee et al., 2021; Ye et al., 2022), as such where there is a blood test taken when glucose was given in the period immediately prior, one might expect to see fairly high levels of insulin. It is not typical that one tests for insulin in a non-fasting state.

Cartoon of Cell and Insulin Secretion
Figure 1. Molecular pathway demonstrating the role glucose plays in the release of insulin (Wee et al., 2021)

The Science and Law of Referring to Manufacturer’s Instructions

Based on the manufacturer’s specifications sheet, it is clear that the concentration of 4657 pmol/L is far outside what would be considered a reliable test result. Nevertheless, the manufacturer states that the cut off for the actual detection of insulin in the test is 6945 pmol/L or 1000 µM/ml. (*Note this assay relies on the old standard of converting insulin from µM/ml to pmol/L, the agreed standard is 1000 µM/ml = 6000 pmol/L). So based on that finding one can assume that the insulin concentration as observed in the cases at CoCH was measurable. The caveat for this is whether the concentration provided is in pmol/L as opposed to µM/ml. If the insulin concentration is given as 4657 µM/ml then this would be way beyond the detection limitation of the test, as the manufacturer reports the cutoff as being 1000 µM/ml.

In the first instance one can conclude:

if the concentration given for both tests is in µM/ml then they both exceed the detection cut-off and the tests are not valid. The insulin test results would be: Child F: 4657 µM/ml and Child L: 1099 µM/ml, where the cutoff is 1000 µM/ml

Conversely, if the concentration of insulin is in pmol/L then they are below the cut-off and are technically valid, but they would fall above the 95th percentile, which raises questions as to their reliability and validity, as this means that these concentrations would be observed in fewer than 5% of all tests taken.

So then on to the next question. Given that we should assume that the concentration of insulin given for the first child is valid, at 4657 pmol/L, then what corresponding concentration of C-peptide would we predict to obtain assuming the concentration is endogenous?

The Hook Effect

From there one identifies the insulin:C-peptide ratio in neonates, based on peer reviewed research (Mitanchez-Mokhtari D, et al., 2004), where it was determined that in preterm neonates who had normal blood glucose levels the ratio of insulin : C-peptide was 1 pmol/L : 21.3 pmol/L. From here, one simply applies that same ratio and you can calculate a predicted C-peptide, of ~99,277 pmol/L. For the first child that concentration would be impacted by the Hook Effect, which is observed at concentrations of C-peptide greater than 60,000 pmol/L, as such would result in a false negative.

Graph depicting Hook Effect
Figure 2. The Hook Effect results in high C-peptide concentrations being reported as a false negative

Scientifically, one does not need to do anymore than that as an analysis, given that once they observed the very high insulin concentration there was a need to dilute the samples as it was clear they would otherwise be impacted by the hook effect.

Many people have become fixated on the insulin concentration, when the actual claim is that the low C-peptide concentration is proof positive of insulin poisoning. In the case of Child F, that claim is untrue, because one cannot actually reliably test for C-peptide concentrations that would correspond to high insulin levels because the concentration of C-peptide that one would expect to see exceeds the bounds of the C-peptide blood test. It is likely, that this fact is why the lab explicitly states that one cannot rely on those assays to determine exogenous insulin administration because the C-peptide ranges do not go high enough to correlate with test parameters in the insulin test.

As far as I can see, in the case of the first infant the blood test is simply not valid. And should not have been relied upon, because, to repeat, one could never get a C-peptide test that would yield the corresponding concentration of C-peptide based on the insulin concentration of 4657 pmol/L, and the reported predicted ratio of insulin : c-peptide, as reported in the literature.

In the case of the second infant, the test result of 1100 pmol/L is still greater than the 95th percentile, which makes it somewhat suspect. However, in the case of this baby, there appear to be a number of physiological factors that suggest the lower C-peptide concentration is not particularly exceptional. In particular, there were significant health concerns for the mother who was in hospital for an extensive period prior to and following birth. It seems to be self evident, that in the first baby the test is not valid due to the C-peptide hook effect applying. In the second case, the issues surrounding maternal health need to be properly investigated before relying on a test result which represents less than 5% of all blood tests for insulin, which were conducted by the manufacturer, when developing the assay.

In both cases, the fact that the blood tests, if in pmol/L, are beyond the 95th percentile raises immediate doubts, and would require further validation. Though if those test results were given in µM/ml then they are totally invalid and cannot be relied upon.

Challenging Science and Law: Looking for Patterns Beyond what the Data Permits

It is very dangerous to assume that there is an actual link between the blood tests identified in the case. It is an assumption of extreme proportions to take two blood tests, from completely different children, born in different circumstances, months apart and conclude they are linked. The desire to find the same cause for discordant insulin : c-peptide ratios, is an act that cannot be performed until the tests are at first scrutinised for individual validity.

A big error in this case, and many that I have viewed, where medical doctors are expert witnesses is the fact that there is a marginal investigative technique adopted. By that I mean, where one is uncertain of a given finding it warrants further research and investigation. What one actually observes is an effort to create some sort of pattern with data such that no further investigation is required. There is no evidence that insulin issues between the first and second baby are linked. Yet time and again people have sought to link these two completely separate events. What is the basis for doing this? Well, that is impossible to know. It appears that people wish to lump these events together, without actually knowing how many other blood tests taken at CoCH in that same period looked similar to those described here.


It is perhaps time that everyone stepped back a little bit, and asked themselves what exactly are you trying to understand? What questions do you need to ask Science and Law in order to determine what the blood test means? It seems very simple to me, before we do anything in biology we must assume that we adhere to limitations of the protocols we seek to follow. In this case, we can see that this was not done. It is, rightly, the desire of the medical doctor to diagnose, and it appears that there is a willingness, with some in the profession, to extend way beyond speculation to achieve this outcome. Though, it cannot be overstated, that this is not a remotely scientific, rigorous, or reliable approach to take. At some stage one can simply say, there is insufficient evidence to proceed. Of course, the issue for many is the lack of concrete answer, still, the legal system cannot entertain this level of speculation and apparent prejudice, masquerading as scientific investigation.

At the point that the testing facility publishes on its own website that the blood tests cannot be used to determine exogenous insulin, we cannot suggest that they identify such an outcome. No more than one can say any other blood test is predictive of a physiological state where it explicitly states it cannot measure that state. If we are going to ignore the instructions from the manufacturers of the test kits we rely upon for diagnosis, one wonders what is the difference between medicine and witchcraft? I am serious. The speculation surrounding the health of the mothers etc, is in my view a step too far, and a serious ethical issue concerning privacy. Though, this door was opened when it was implicitly suggested, by the expert witnesses, that one could override the manufacturer’s reported limitations surrounding one’s ability to determine exogenous insulin from their blood tests.

What is clear in this case is that medical doctors rely heavily on subjective, biased, experiential assumptions in the course of carrying out their duties. We should not be surprised by that, indeed, many patients seem to be uncomfortable when a medical doctor does not know something without at first looking it up. However, this practice of relying on such experiential knowledge should never be permitted to apply to something as technical as forensic scientific investigations. We cannot afford to get it wrong in forensic sciences, because the result is someone’s freedom. What we are seeing here is the application of medicine, a noisy, error-prone practice, which is heavily reliant on patient feedback and responses, being applied to a highly technical, methodical practice of forensic science, where it is essential that one‘s assessment is not clouded by one’s subjective experiences.

The reason that subjective reasoning cannot be relied upon in forensic science is very simple. In a living human being, if we obtained the blood test results as described in these cases we could, in real-time, ask further questions based on subjective reasoning. We might eventually get to some sort of conclusion, only after making several false assumptions, that were corrected by our ability to conduct further testing in the moment. In the forensic setting, we have simply a snapshot of the moment. We cannot create hypotheses about those blood tests once it is established that we cannot use them in isolation to infer exogenous insulin administration. We cannot apply subjective knowledge here, for obvious reasons, because we cannot conduct any further testing to validate our assumptions. What happened, in this case, is that lacking any basis to apply subjective knowledge the prosecution experts determined that identifying another similar case would be sufficient to override the need for a structured, objective examination of the facts.

Many people have accepted, on its face, wholly unscientific, irrational, and specious claims put forth by the prosecution experts. It is understandable as to why they may do this, as it flows that the closest experience one might have of a rigorous scientific investigation is an experience with a medical doctor. Though this is the first, and most enduring error that serves to plague our criminal justice system, and likely maintains an untold number of people in prison for crimes they likely did not commit. A medical doctor has a diagnostic error rate of anywhere up to 45% of all diagnoses. And in reality, that may be okay. It may be absolutely fine for a doctor to have to go through 4-6 different conditions until they reach an accurate diagnosis. The continual feedback from the patient, either through further examinations, shifting test results, or the presentation and changes in symptoms, is a crucial factor in the diagnostic process undertaken by a doctor. It appears to be pretty effective, I am not challenging this approach.

Medicine is not a discipline whose benefits can be applied after the fact

There is a stark reality, and unless we wake up to it, we are permitting repeated atrocities to be handed down to people based on what can only be described as opinion, by expert witnesses. The reality is that medical doctors are applying guesswork, speculation, individual prejudices, and total defiance of tested protocols in our criminal justice system for the purposes of convicting people of crimes. In many of the cases, doctors, as expert witnesses, arbitrarily adopt certain findings while rejecting out of hand those findings that are more removed from their own prejudice.

The excerpts below are taken from a case concerning the surviving sibling of an child, EF, who died at 18 months old with apparent neurological findings that were interpreted as non-accidental injury. The death of EF was referenced in the proceedings relating to the surviving sibling [Re DC (a child) (non-accidental injury) Southwark London Borough Council v GH and others (KL intervening) [2020] EWFC 112], where it was alleged that the expert witnesses found that EF’s injuries were inflicted via a shaking episode, combined with blunt force trauma to the head. The following statements were made to justify these claims:

“In his report Professor Al-Sarraj considered that the subdural haematoma had to be considered traumatic in origin due to a lack of evidence for other explanations.”

Further support for this ”traumatic” injury to the head was an expert who testified to a potential injury to the leg.

“Professor Mangham was asked to look at a potential fracture to EF's femur. He concluded that although he could not be certain whether the abnormality in the distal metaphyseal / diaphyseal junction represented a healing classic metaphyseal lesion (CML), on the balance of probabilities it was more likely to represent a healing CML than not. This was because it was difficult to find an alternative explanation for the abnormality.”

The child apparently exhibited a bone abnormality, which was not indicative of a fracture, however the inability of the expert to ascertain any other cause was sufficient for him to speculate that it was a bone fracture and not another abnormality with a clear origin. Worse still, continuing in the same case, it becomes clear that even when presented with evidence which apparently counters a traumatic origin the tested evidence is freely discounted by the medical doctor as more than likely a false positive:

”Dr Keenan reported that the results of EF's results of the PFA screening test were in the range seen in the severe platelet function disorders, Glanzmann's thrombasthenia and Bernard Soulier Disorder. Both conditions were very rare and seen in 1 in 1,000,000 of the general population. The definitive test to diagnose or exclude both disorders was not performed as no blood clotting tests could be performed after death. As these conditions were very rare, it followed that even with this single abnormal result it was still probable that these conditions were not present and that there was no explanation for the bleeding seen in EF.”

It is widely understood that in infants, very low platelet count can lead to intracranial haemorrhage, though in this case failure to conduct appropriate testing to rule out this possibility was apparently a sufficient basis to rule out the possibility that the child had a bleeding disorder. The doctor further went on to assert that the insufficient blood testing that was performed should essentially be dismissed as a false positive, as at a totally separate time in the child’s life when he had surgery there was no issue with bleeding:

“In his oral evidence, Dr Keenan confirmed that platelets tests were not particularly reliable and produced false positives. It was more probable that the result was a false positive. He was more cautious about the conclusion that as there had been no obvious problems during invasive surgery: this was further evidence of a reduced chance of a bleeding disorder.“

To be very clear, this is not science, it is almost the opposite of science. It is an effort to selectively reason about the condition of a deceased child, where there is a refusal to accept their own findings as an indication that further investigation is required. Where a medical doctor is stating that he basically finds no basis to substantiate their perspective then they are required to either tolerate that they have no results to report or they conduct further investigation both in terms of literature reviews, to explore other conditions, and further testing of whatever samples were collected at autopsy. Though what is happening in the courts in the UK is something radical. The standard of forensic science in the UK is frighteningly poor and yet we are being led to believe that by inviting the medical profession to conduct forensic science, without any basic research science background then we can forego the need to invest in any forensic science services.

The Canary in the Coal Mine

The Lucy Letby case is not an anomaly, it is a reflection of a system that has become emboldened to liberally distort our understanding of basic science. One wonders how many people are sitting in prison on the back of expert witnesses dismissing out of hand the limitations of test results. We shall not know, but we can be sure that the number is most certainly greater than 1.

To summarise, the pathologist, Prof Patrick Barnes, who testified against Louise Woodward subsequently stated that if he had to testify in her case today, that he would never testify that she was the cause of the death of the child under her care. In an interview, concerning over-zealous allegations of child abuse by medical doctors, Prof Barnes made the following statement:

“It is critically important, particularly for children and babies six months and younger, because at that age they could actually have conditions that have yet been diagnosed that stem from birth process, that are delayed effects of trauma at birth or other conditions passed from the mother to the baby and so forth. Plus, for the babies that we lose, or even if they survive, number one, we need to diagnose those conditions, because even if the baby survived, that baby may suffer another episode due to that medical condition. We didn't diagnose it, we didn't treat it, or we lost the baby, let's say, and the mother and father who may have been thinking that the caretaker did this to the baby want to have other children. And if we didn't diagnose that condition that maybe the next child may have, then we have done a disservice to family planning for that family.., [A]fter participating in a number [of cases] for [the] Innocence Project in this country, and the Innocence's cases convicted individuals getting new trials and everything and looking at the entire environment on all of this, I just felt that it was important to be more involved and hopefully, possibly become maybe not so much a leader, but [a] modeler in this particular area…”

We can only hope that more of the expert witnesses who submit bad, erroneous, and speculative scientific claims to our criminal courts, in the service of ensuring convictions, whether they should apply or not, can rise to a point where they can reject their own faulty practice. Indeed, the liberty of scores of people is likely dependent on such events occurring. Such a time cannot come soon enough.

I would like to dedicate this post to all of the medical doctors, nurses, psychologists, therapists, teachers, and social workers who have in myriad ways come forward following the conviction of Lucy Letby. To see so many individuals state their great concern with the evidence in this case is the first step to recalibrating the scales of our justice system. Though, I believe that the problem we face is by far greater than this one single case. We cannot ignore that for individuals like Dr Dewi Evans, Dr Sandra Bohin, and Dr Andreas Marnerides, their legacy extends back to our family courts, where we shall never know the impact of their work as experts.


Komatsu M, Takei M, Ishii H, Sato Y. Glucose-stimulated insulin secretion: A newer perspective. J Diabetes Investig. 2013 Nov 27;4(6):511-6. doi: 10.1111/jdi.12094. Epub 2013 May 15. PMID: 24843702; PMCID: PMC4020243.

Mitanchez-Mokhtari D, Lahlou N, Kieffer F, Magny JF, Roger M, Voyer M. Both relative insulin resistance and defective islet beta-cell processing of proinsulin are responsible for transient hyperglycemia in extremely preterm infants. Pediatrics. 2004 Mar;113(3 Pt 1):537-41. doi: 10.1542/peds.113.3.537. PMID: 14993546.

Wee J, Pak S, Kim T, Hong GS, Lee JS, Nan J, Kim H, Lee MO, Park KS, Oh U. Tentonin 3/TMEM150C regulates glucose-stimulated insulin secretion in pancreatic β-cells. Cell Rep. 2021 Nov 30;37(9):110067. doi: 10.1016/j.celrep.2021.110067. PMID: 34852221.

Ye, Y., Barghouth, M., Dou, H. et al. A critical role of the mechanosensor PIEZO1 in glucose-induced insulin secretion in pancreatic β-cells. Nat Commun 13, 4237 (2022).

Re DC (a child) (non-accidental injury) Southwark London Borough

Council v GH and others (KL intervening) [2020] EWFC 112


Recent Posts

See All


I think we do see at least a portion of the impact dewi has had on the family court system and alas also as a medical expert witness

So far to me at least it doesn’t look very good

Not many such experts are suggested by court they are potentially a blatant lier


Nov 11, 2023

The findings in this blog post and the previous one regarding the limits of the c-peptide test used are brilliant!

I noticed that the average c-peptide / insulin ratio > 20 observed for neonates (both premature and full-term) in the paper cited (Mitanchez-Mokhtari et al. 2004) is much higher than the range of 5.0 - 10.0 stated by one of the doctors in the testimony regarding child F (as reported in the media). Is the 5.0 – 10.0 range stated during the trial perhaps the normal range for adults, ignoring the findings in the above cited paper which give a much higher average c-peptide / insulin ratio for neonates?

By my calculations, there is at least one pre-term neonate in…


Correction: in my previous message, in the last sentence of the first paragraph "£2M a year" should have read "£2M a month".


pat chanse
pat chanse
Nov 08, 2023

It cannot be emphasised enough that a medical doctor is not a Pathologist, or better still , a Forensic Pathologist. Evans should never have been allowed to overturn the post mortem result. At least not without the Pathologist admitting error.


This case, and the other potential miscarriage of justice cases that can only be speculated about at present, reflect the kinds of problem that were an inevitable consequence of the exceedingly blinkered, deceitful and financially inept decision of the UK Government in 2010 to close down the Forensic Science Service. It was blinkered because the only reason for the closure was to save the cost of the service to the public purse. It was deceitful because the decision was made without any consultation with the important stakeholders: the Government's Chief Scientific Officer, the Forensic Science Regulator, the Director of Public Prosecutions, the Attorney General, the Lord Chief Justice, etc. In fact the only relevant consultee was the Association of Chief…

bottom of page