Lucy Letby

This piece took a long time to write. JC is no criminal lawyer, has no particular axe to grind, and found himself constantly wondering whether he had been afflicted by his own tunnel vision. Had he too fallen into the oubliette?

The JC’s amateur guide to systems theory™ →Go uptown “What makes you think she’s a witch?” Index: Click ᐅ to expand: Systems theory The Devil’s Advocate Tell me more Sign up for our newsletter — or just get in touch: for ½ a weekly 🍺 you get to consult JC. Ask about it here.

“We’ve found a witch. May we burn her?”

— Monty Python and the Holy Grail

Preface

Ms. Letby may have committed these awful crimes, or some of them, or she may be innocent. No one, but Ms. Letby, can know for sure. That is the dilemma: how to treat such uncertainty fairly. The question is whether the criminal justice system has done so. In light of its recent performances — Andy Malkinson and the Post Office sub-postmasters refer — we should hesitate before saying, “Well, she’s had her day in court and has been convicted” as if that is really the end of it.

The case raises fascinating issues that should interest lawyers in any discipline — and informed citizens in any walk of life, for that matter. Given JC’s general aspiration to try to understand the world as a complex system that behaves, and incentivises individuals to behave, in ways we would not expect, it seems like a fair topic for conversation at least.

On herd minds, groupthink and narrative bias

Herculio: ’Tis neither malice, spite, nor virtue
Whose ledger swells, or plucks, the seedy fruits of progress —
But mainly accident.
Lest thee with surety know aught else —
Withhold thy assignations.

Triago: Pish upon thee, Nuncle. Pish!
Dost thou mean to say
Things peel this way
Through doughty misadventure?

Herculio: Peradventure —

Triago: Pish abeam!
Has thou no more to say than that?
Wouldst thou on this shaky surmise
Withhold rebuke?

Herculio: Perchance, per case, mayhap dear Triago
’Twas but a fluke?

Triago: O! This nuisant planet weighs upon my soul!

Herculio: If ’tis this and nought beside
That flies you to a vernal rage
Our fickle globe in its manifold confound’ry
Lies prettily indeed
For thy alignment.

Lucy Letby is back in the news in 2024. On 24 May, the Court of Appeal denied her leave to appeal against her convictions. On 2 July, she was found guilty on retrial for the attempted murder of “Child K”.

All of this came against a backdrop wherein, on 13 May, New Yorker magazine published Rachel Aviv’s “A British Nurse Was Found Guilty of Killing Seven Babies. Did she do it?”, a 13,000-word investigative piece questioning Ms. Letby’s original convictions. This, in turn, followed a dogged campaign by a rag-tag bunch of amateur statisticians, scientists, forensic specialists and plain old keyboard warriors who were convinced of an injustice.

Since then, the mainstream media has leapt into the fray, most expressing rising discomfort about the safety of her convictions and some — notably the Daily Mail — backing both sides of the argument, running pieces in the same week by Liz Hull, who sat through the trial and is certain that Ms. Letby is guilty, and Peter Hitchens who has read the New Yorker article and is certain she is not.

Having been convicted, in two separate trials, of multiple infant murders, the public’s view is — or perhaps was — clear: justice has been done. The sooner Ms. Letby’s name fades from the commonplace the better. As media consensus evolves the received view may fragment.

For as we look closer we see the picture is complicated. The physical evidence implicating Ms. Letby is all circumstantial: eyewitness evidence does little more than put her at the scene of the alleged incidents, and this may not say very much: it was a place she was, after all, contractually obliged to be.

Nor did anyone see Ms. Letby do any of the acts of which she has been accused. The medical evidence indicating foul play is in all cases technical and in most heavily conjectural. Some of the stronger parts of it — arguments about just how strong are heated — relate to babies who survived.

The evidence that resonates most strongly with the public — the shift rota and Ms. Letby’s alleged “confessional note” in particular — is not strong at all.

Nor is the evidence that any infants were the victims of foul play at all, overwhelming. They were, after all, premature infants in an intensive care unit. If they were not at high risk of mortality, they would not have been there.

The case for conviction thus emerges from a preponderance of small pieces of evidence that all point towards Ms. Letby, but weakly. There is no smoking gun. The most compelling proposition seems to be Ms. Letby’s opportunity: she alone was rostered on and onsite during all the alleged incidents.

The circumstantial evidence of the shift rota is founded on improbability. What are the odds that the one person present at all the incidents did not have something to do with them? Neither prosecution nor defence framed the argument in terms of probabilities, and none brought any statistical analysis to bear on the question. This, as several statisticians have noted, is a pity. Especially given the criminal law’s chequered history with statistical reasoning.

A binary status

Now she is in prison for the whole of her life, all appeals exhausted and all pending retrials completed, just two narratives remain. The simple one is that Ms. Letby is the personification of unspeakable evil: a serial murderer of defenceless, premature infants. The more unsettling one is that she is the victim of a breathtaking, and no less unspeakable, miscarriage of justice.

Advocates on both sides grow increasingly entrenched.

But on the very same probabilities, this is a most unlikely place for the public debate to have landed.

Serial murderers of any kind are vanishingly rare in Britain. Wikipedia lists fifty-five, since 1600. But so are miscarriages of justice. Wikipedia lists fifty-four, since 1255.

Both narratives are highly improbable. Of all the explanations we might offer for the unusual spike of incidents in 2015 and 2016, as a matter of “prior probabilities” they are the least plausible: all else being equial, it is highly unlikely there was a serial murderer at work in the Countess of Chester Hospital. But, one having been convicted, all else being equal it is highly unlikely her conviction was a miscarriage of justice. This is why tempers run high. Both positions seem untenable.

They leave untouched a wide range of more probable, but less satisfactory explanations. We like our narratives to explain the world to us. We do not find useful ones that say, “Well, we just don’t know”. We prefer conclusions, however unlikely they may be.

So our institutions persuade us to deprive Ms. Letby of that wide middle in which she was an ordinary person, with pluses and minuses, virtues and failings: neither angel nor devil but who just had the misfortune to be involved in these tragic events.

Especially where, as here, there is no direct evidence, the criminal justice system forces us into an odd kind of anecdotal reductionism: we must focus on discrete actions and isolated pieces of highly technical evidence: literally, microscopic observations: implied levels of insulin and c-peptide; unusual skin blotching patterns and so on. We expected to thread these atoms together and be able to deduce from them a compelling picture of emergent human, moral actions.

We are pattern-matching machines. We use whatever morsels we find to build a narrative. Once we have a theory, we dive deeper into the details to prove it out. We strive to fit the remaining facts into our model. We are at risk of tunnel vision, when our immersion in those atomic details crowds out a wider perspective.

Only later, if at all, do we ask the “big picture” question: leaving aside the microscopic details how plausible is it that this was murder?

Case for the prosecution

Lucy Letby’s case, on its face, seems a lot more like Lindy Chamberlain’s than David Bain’s. It is not beyond question that none of the infants — all extremely premature and in an intensive care unit, after all —met with foul play. For each event taken separately, there is a plausible non-criminal explanation. Indeed, Ms. Letby was not immediately suspected of involvement in any of the deaths.

The case for Ms. Letby’s innocence has been well-stated enough. The New Yorker piece describes the statistical problems. Peter Elston and Michael McConville’s We need to talk about Lucy Letby podcast goes into great detail about the credibility of the technical medical case — it is not very credible — and the excellent TriedbyStats website does a tremendous job of conveying just how unintuitive the statistical implications of the evidence are.

Of course, individuals committed to the proposition that there has been a miscarriage of justice may suffer their own form of “defender’s” tunnel vision — but in the main, their arguments are clear, well-grounded and compelling.

Yet this was no kangaroo court: the British justice system is rightly a thing of international renown. Over centuries, the common law has developed principles that avoid injustice. Clearly, they are not perfect, but the presumption of innocence, the burden and standard of proof, rules of criminal procedure regulating admissible evidence, pre-trial disclosure and acceptable methods of examination and cross-examination are specifically designed to ensure fairness. Ms. Letby certainly got her day in court — ten months worth of them — and spent 14 days giving evidence during her trial, at which she was represented by a King’s Counsel who left few stones unturned.

Still, there ought to be clear, well-grounded and compelling arguments to displace the presumption that Ms. Letby was simply unlucky, implicated by circumstantial evidence but fundamentally just in the wrong place at the wrong time, and instead set her out as a demonstrable serial murderer.

We might expect tunnel-visioned innocence campaigners to overlook inconvenient facts — half of confirmation bias is blindness to what you do not want to see — but not those who are persuaded of Ms. Letby’s guilt. They must surely be able to clearly articulate these “clinchers”.

And a strident segment of the commentariat is persuaded of her guilt. A Reddit community of yeoman amateur investigators numbers 11,000 members and is replete with theories. Journalists who covered the trial in detail have published films, podcasts and have forthcoming books about the “Killer Nurse” (notably Liz Hull from the Mail and Judith Moritz from the BBC). The prosecution’s main expert witness, Dr Dewi Evans, has gone to unusual lengths to dismiss the concerns of innocence campaigners whom he feels to be unfairly targeting him.

But if these people can articulate these damning facts, in their voluminous reporting to date, they haven’t, particularly well.

The timid voice of protest grows louder.

Confirmation bias

“Angel” and “devil” narratives become self-fulfilling: once you’ve adopted one, you can panel-beat almost any subsequent information to suit your view.

The little definitive peripheral evidence there is about Ms. Letby gets coloured dramatically by whomever is looking at it. For example, her social media activity. Here is BBC Panorama reporter Judith Moritz, in a piece to camera:

“Sparky, full of fun, popular — she looks like the life and soul of the party in these photos. I don’t know what Britain’s most prolific child killer should look like — I’m pretty sure it’s not this, though.

And then, a few moments later:

“She comes across as — mousy; a bit normal — you can’t really marry that with the enormity of what she’s been accused of.”

If and only if we are persuaded of her guilt, should an outwardly vivacious personality and busy social life mark her out as a monster. In that case it corroborates and amplifies her wickedness.

But it cannot also be evidence of her wickedness.

But if we are prepared to give Ms. Letby the benefit of the doubt, this is perfectly normal behaviour. It tells us nothing: it places Ms. Letby in that ordinary space, with the rest of us, within a standard deviation of the mean. There have been very, very few serial killer nurses. We have no idea what, as a rule, they are like. This information does not help us. This tells us that Ms. Letby is not unusual.

That Ms. Letby searched online for the parents of the deceased is consistent with either breathtaking malevolence — if you take it that she is a serial killer — or affecting compassion — if you take it that she is not. By itself, it is evidence of neither. We all search online for individuals we meet in real life — even people we know we probably shouldn’t: this is perfectly normal behaviour. We are curious, imperfect animals.

Doubt

Whatever the reason for it, a lack of direct evidence of the defendant’s foul play — of any foul play — means there will always be some doubt. No one saw anything. Inference is needed. The question is whether the inference is justified. Has the prosecution done enough to remove all reasonable doubt?

In some types of crime — burglary, say — we should expect no direct evidence: competent burglars target unmonitored environments where there are no witnesses.

But burglars, we fancy, would steer well clear of the intensive care unit in a neonatal ward. There could hardly be a worse environment to get away with crime undetected. Entries and exits are monitored and audited. Medicine is logged, controlled, secured and signed in and out. Medical experts conduct routine rounds and check on patients around the clock. Biochemistry is complex and its signals are delicate. There are specialists at hand with deep expertise and sophisticated machinery who can detect the merest traces of the unusual or sinister.

So the lack of such evidence is a curious feature of the “healthcare serial murder” cases. This one is no exception. No-one saw anyone actually doing any harm.

Avoiding eyewitnesses, let alone forensic detection, would require great skill, caution and planning. You would expect a malign perpetrator to refine and perfect a careful technique — a modus operandi — and stick with it.

You would not expect a skilful murderer to make opportunistic attacks, to use multiple, unrelated methods or to improvise on the spot to suit the circumstances.

Certainly not one under active suspicion. The hospital’s lead consultants raised concerns with management about Ms. Letby as early as October 2015.

Yet this is what Ms. Letby is supposed to have done, and still left no implicating evidence. Over 13 months, Ms. Letby is alleged to have variously injected air intravenously, injected it via nasogastric tube, caused blunt force trauma to internal organs (quite how, is not clear), overfed with milk, dislodged breathing tubes, poisoned with insulin, and physically throttled.

Ms Letby was described in the court as premeditated, calculating and cunning, using a number of different methods, thereby misleading clinicians into believing the collapses had a natural cause.

Perhaps — or maybe the collapses did just have a natural cause. There is one other germane feature of an intensive care unit: visitors tend, by unfortunate necessity, to be very, very sick. They have an unusually high probability of dying. If they did not, they would not be there.

The other explanation for this cunning calculatedness is that it was nothing of the kind.

Onus

And that is a heavy onus. Between prosecution and defence it is not a straight fight: the prosecution must prove its case beyond reasonable doubt. If it is incumbent on a defendant to prove anything — technically it isn’t, but practically it would be a brave defendant who did not — it is simply that there is some credible doubt.

When a seriously ill patient dies in intensive care, even unexpectedly, and nothing beyond mere presence implicates any perpetrator, there is credible doubt.

For each of the Countess of Chester collapses taken in isolation there is reasonable doubt that the victim was the subject of foul play, let alone that Ms. Letby was responsible for it. Indeed, foul play was not initially suspected in any of them.

How can a series individual cases which have reasonable doubt turn into a collection of cases where there is none?

Probability

Once is happenstance. Twice is coincidence. Three times is enemy action.

—Ian Fleming, Goldfinger

The answer is probability.

It is sometimes said that Ms. Letby was “not tried by statistics”, the implication being that probabilities did therefore not influence her conviction. But this is not right: a case depending on circumstantial evidence is necessarily about probabilities. The “probative value” of circumstantial evidence is: “Does this make it more or less likely that an event, which no one witnessed, happened?”

True, statistics are not always an appropriate lens to assess probabilities, especially where data are limited or collected in unusual circumstances. But the shift data are statistics. The Crown may not have framed its arguments about them that way, but its arguments were statistical: the shift data have no probative value other than as statistics.

Now: flip a fair coin once and heads is as likely as tails — that’s “happenstance”. Flip it twice and, while two heads are less likely, they still carry a one-in-four chance — that’s “coincidence”.

But flip it twenty-five times and you have a less than one in thirty million chance of getting twenty-five heads. If you get this result, it is most unlikely your coin is “fair”.

This is why the shift data is so important: it is the equivalent of a string of twenty-five coin flips that all came up heads. It can dispel those “credible doubts” in individual cases: okay, one collapse may have been innocent. Two, a coincidence. But twenty five?

If there is a single clinching argument in the evidence presented against Ms. Letby, this was surely it.

But it only holds if the “flips” really were consecutive: If there were fifty flips, and only twenty-five came up heads, that would not be unusual at all.

And here we come to the statistical problems with the shift data. It was a small selection from a much bigger sample of over 700 shifts in the “suspect period”. It records only the events for which Ms. Letby was charged. These all occurred, Q.E.D., while she was on duty. There is necessarily, therefore, an X by her name for every incident. The chart tells us simply that Ms. Letby was on duty every time she was on duty.

It has been widely criticised, perhaps most eloquently by Rachel Aviv in her New Yorker article. Aviv raises the prospect of the prosecutor’s fallacy, also known as the “Texas sharpshooter”. Are all the incidents included? Did any relevant events happen when Ms. Letby was not on duty? What even counts as a “relevant event”? Has inconvenient data that does not fit the narrative been omitted?

There were ten other deaths in the suspect period for which Ms. Letby was not charged. We do not know why she was not charged, nor whether she was on duty for these events. (A few commentators claim to have seen evidence that Ms. Letby was on duty for every death in the suspect period — but that evidence is not made public, and importantly was not presented at trial.)

Had the Crown submitted a comprehensive table including every shift in the suspect period, and marked every incident on it, whether Ms. Letby was implicated or not, we might have a better story. But they did not.

The Police have hinted “further charges may be forthcoming” as they continue to review thousands of historical cases dating back to 2012, but you would expect they would have found something after nine years.

On tonewood and prosecutor’s tunnel vision

Triago: Ho, Ho.
Let not thy witty fool, nor his foolish wit
Besmirch the fruited science of th’ academy.
“A little learning is a dangerous thing” —

Nuncle: Yet not half so dangerous as a lot.

Triago: — So sayeth Pope, you know.

Nuncle: But not the one in Rome:

Queen: Good ser knight: art thou drunk upon the Pierian spring:
A hypoxic draft that suffocates the brain,
So deep no shaft of light can bring it round again?

Triago: My conjecture comports a grain of truth
As pure and true and golden—

Nuncle: — but yet no more roundly causative
Than are the month-past flappings of a Latin papillon
Upon a brewing Filipino typhoon.

JC has, elsewhere waxed long and lyrical about the collection of cognitive biases called prosecutor’s tunnel vision. These biases tend to show up where clinching evidence does not: if there were any clinchers, things would be clinched. Therefore those prosecuting — and, for that matter, defending — start to fixate on finer and finer technical details to win isolated arguments. Once one side goes down a rabbit hole, the other is obliged to follow. Insulin assay tests, entry card swipe data, the significance of skin discolourations — all have been cited as smoking guns for prosecution and defence when they are transparently nothing of the sort.

The fact is, there are very few political, social, and especially personal problems that arise because of insufficient information. Nonetheless, as incomprehensible problems mount, as the concept of progress fades, as meaning itself becomes suspect, the Technopolist stands firm in believing that what the world needs is yet more information.

—Neil Postman

Information is fractal. The more you look at it, the more you can create: it subdivides infinitely — there is no bottom — and the more raw information you create, the more possible arguments you can have about it. This is the implication of information theory: the more data points there are, the more unique patterns you can make from them. If we take it that a theory is simply “a pattern of consistent data drawn from available information from which we can draw a valid inference” then the more information you have, the more plausible alternative theories of the case you can make.

In other words, you cannot win an argument by simply descending further into the weeds. Those who have drunkenly debated evangelical Christians, resolute atheists, real ale connoisseurs, or tonewood freaks — JC has done all of these — will know this. The thing about descending into the oubliette is that as the points atomise, the arguments grow more heated and the less difference they can possibly make.

This is the “tonewood” debate in a nutshell: sure, in theory the harmonic resonance of nitrocellulose varnish could affect the sound of an electric guitar, but not so as a mere human could possibly hear it in perfect conditions let alone in the moshpit at the Roxy on a sweaty night in 1976.

Whatever did make Steve Jones’ Les Paul sound like a screaming chainsaw, that is, it wasn’t the varnish.

Zeroing in to draw further inferences will not help. The only way to do it is by zooming out.

Oh come on ref

It is a common enough trajectory: we start with an instinct that a popular, diligent, young, female nurse from a stable background is the last person you would expect to be a serial murderer.

We might quietly chide ourselves about our unconscious bias in favour of people like us. We might remind ourselves that natural sympathy with our in-group is not scientific, let alone legal evidence. The law requires one to set that kind of prejudice — literally, prejudgment— aside and consider the evidence as it accumulates in the abstract and on its merits.

But hold on: if circumstantial evidence “changes the prior probability of an event”, it adjusts the inferences we can reasonably draw from baseline facts. Inasmuch as Ms. Letby’s background, socialisation and mental health reflect the prior improbability that she would be a murderer, and they are not displaced by direct evidence that she is a murderer, then our intuitions are useful. They are all we have. If we ignore them, our pattern-making instincts may lead us down rabbit-holes where we wind up chasing down tendentious circumstantial minutiae at the expense of basic common sense.

This is how miscarriages of justice happen.

The Northern Territory Coroner’s report to Lindy Chamberlain’s trial is perhaps a high watermark of this kind of thing. A week after her disappearance, Azaria’s torn and bloodstained jumpsuit was discovered in the desert, 4 km from the campsite. Rather than seeing this as corroborating Ms. Chamberlain’s story the coroner arrived at precisely the opposite conclusion:

The evidence in relation to the clothing is consistent with an attempt to simulate a dingo attack on a child by person or persons who recovered the buried body, removed the clothing, damaged it by cutting, rubbed it in vegetation and deposited the clothes for later recovery. Such deposition is indicative that the deposition was made with the knowledge that dingos were in the area.

In addition, there is no evidence to positively support the involvement of a dingo in the taking of the child, the carrying of the body some four Kilometers and removing the body from the area where the clothes were found.

Prior probabilities, again: not only did Lindy Chamberlain have no opportunity to fabricate this evidence and make the eight-kilometre round trip to plant it, and not only did no-one witness her doing so, it was in any case vanishingly unlikely that a sane, well-adjusted and happily married mother of three children would do any of the horrendous things of which she was accused.

Before even entertaining the possibility of foul play, the coroner should have asked himself, and satisfactorily answered, a far simpler question: why on Earth would any normal, well-adjusted mother do that?

Now, barristers will remind us that motive is never required for a criminal conviction, and this is true. But. If there is no direct evidence of wrongdoing and there is a plausible innocent explanation — in Ms. Letby’s case, natural causes — we should be able to plausibly answer that question.

Are we really to believe that an experienced nurse with no history of mental illness, criminality, sociopathy or familial dysfunction, let alone prior medical misadventure — could suddenly transform herself from a trusted senior medical professional into a killer of such sophistication as to go undetected for eighteen months? How could she do this? Why would she do it? Why, in the absence of compelling, hard evidence displacing the usual presumption that this is not what normally socialised people do, would the preferred theory of the case be serial murder, and not “an unusual but not remarkable cluster of accidents”?

Miscarriages of justice and the perfect crime

It may seem outrageous that there is such a thing as a “perfect” crime, wherein the villain leaves no trace of his dastardly deed and thereby walks away scot-free — but as a general proposition you cannot convict someone of a crime for which there is no evidence. Even if she definitely did it. Here, again, is our old friend form over substance: this time the the requirement for form following the uncomfortable implication of epistemology that we cannot know substance. Form is all we have. Here is David Hume:

“We have no other notion of cause and effect, but that of certain objects, which have been always conjoined together, and which in all past instances have been found inseparable. We cannot penetrate into the reason of the conjunction. We only observe the thing itself, and always find that from the constant conjunction the objects acquire a union in the imagination.”

If Enlightenment philosophy is a bit much for after-dinner reading, Terry the Cook from Fawlty Towers puts it more colourfully:

“What the eye don’t see, the chef gets away with.”

If you want to put someone away, you must prove it. The defendant,by contrast, does not have to prove anything. (There is some tension here between crystalline legal theory and the practical reality that it is in the defendant’s interest that someone casts doubt into jurors’ minds. Since the Crown plainly won’t do that, unless it trusts the jury to confect plausible doubts by itself, the defence must plant them there.

In the vernacular, the burden of proof is on the prosecution. The standard of proof that the prosecution must attain is “beyond reasonable doubt”. Not “probably”, not “highly likely”: there must be no plausible alternative explanation for the events.

Here is where the difference between direct and circumstantial evidence is stark. If there is reliable witness evidence that the defendant did it, or did not do it, then the burden and standard can be satisfied easily. There is not much call, or even scope, for doubt.

But in a case where there is no direct evidence that the defendant did anything — no eyewitness testimony, no CCTV, no uncoerced confession nothing — then the circumstantial evidence suggesting that she did do something is going to have to do an awful lot of work. It must, create an overwhelming probability that there was a murder, that the defendant was responsible, and that it was not a non-culpable death, or that it was but the defendant was not responsible for it.

Resources

Academic

Brave Clarice—healthcare serial killers, patterns, motives, and solutions, Rahma and Esraa Menshawey, 1 December 2022.

Online

• That New Yorker Article (choose your own snapshot from the Wayback Machine)
• Mephitis - The Lucy Letby Case May In Time Be The Story Of The Century
• TriedbyStats
• Richard Gill statistics
• The Insulin ‘Evidence’ in the Lucy Letby Trial. Criminal Justice in England: Disagreeable Facts

Podcasts

• We need to talk about Lucy Letby (Peter Elston and Michael McConville podcast)
• The Trial of Lucy Letby (Daily Mail podcast)

The difficult cases

There are excellent podcasts on each of the difficult cases mentioned above.

• A Perfect Storm: The True Story of the Chamberlains
• Conviction: The Christchurch Civic Creche Case
• Black Hands: A Family Mass Murder

See also

• Miscarriages of justice
• Texas sharpshooter
• Prosecutor’s tunnel vision