Lucy Letby thread

[Words]

Starting this as don’t think we have a new one.

The key card data was selectively used and was flawed in any case (one of the doors wasn’t recording entries and exits). The statistician that the police initially hired told them that they were approaching their statistical analysis incorrectly… so they dropped her.

@PaterPower yeah but they fixed it by the second trial. Why does it matter anyway? She was placed there by colleagues, her own notes and involvement in making memory boxes etc. for many of the babies. Statisticians seem to work from a model of "x is extremely unlikely but there's still a minute chance of x happening somewhere in the world at some point so it's not murder, which isn't really helpful. I'm sure by the time they got rid of her they'd already realised it very much was murder, so she simply wasn't needed anymore. Why is everyone so obsessed with statisticians solving this case? They can't solve it, they have zero clue whether she killed those babies or not.

The idea that she could have spiked bags of insulin (in packaging designed to show evidence of tampering if interfered with) ahead of time, (and know which baby would get what bag), is just preposterous. As is the idea she could have added insulin via the feed lines, (particularly unobserved), by injecting it into the tiny diameter tubes.

There would've been evidence of tampering, she was just lucky the bags were discarded. She literally asked the police if they had the bags! And she knew already that they didn't.

And then there’s the refuting of insulin poisoning as cause of death. Nobody, even Evans, is still sticking by that theory.

I've not heard anything about Evans changing his mind on this-where did you hear that?

The fact that multiple experienced coroners somehow didn’t spot the foul play but a long retired non-specialist (who was considered to be a poor expert witness by another judge, who took the time to flag that up) was able to, merely from reading paperwork and not ever having looked at one of the babies, should stretch anyone’s credulity.

The coroners weren't looking at it forensically-they were doing what Shoo Lee is doing and trying to explain it by natural causes. The liver injury was always unexplained, as were some of the other post mortems. Of course Dewi Evans knew that was murder. Sandie Bohin agreed, as did a forensic pathologist-Dewi must have some influence to get everyone else on board with his theories.

It matters because we'll never know how important any given piece of inaccurate evidence was to the jury.

There would've been evidence of tampering, she was just lucky the bags were discarded. She literally asked the police if they had the bags! And she knew already that they didn't.

@Firefly1987

Perfectly normal question to ask.

Statisticians seem to work from a model of "x is extremely unlikely but there's still a minute chance of x happening somewhere in the world

No: there is a high chance of x happening somewhere, even if the chance of it happening in any given place is minute, because there are so many places. So the logic of "x happened here; the chance of it happening here was minute; therefore, it did not happen by chance" is incorrect.

Also, the prosecution's case was "there is NO OTHER POSSIBLE explanation other than murder, even if we don't know how it was done".

The mere existence of a possible other explanation fundamentally undermines the logic of their case.

Why is everyone so obsessed with statisticians solving this case? They can't solve it, they have zero clue whether she killed those babies or not.

But they do have a clue whether statements of the form "X is unlikely by chance" are valid or not. How to test such statements is the focus of their field. That’s why the police at first went to a statistician to test such a statement, though they dropped her later at CPS request.

Well that's even worse then lol. Basically it's possible for anything to happen naturally so there can't be any nurse serial killers because statisticians will always explain it away. I can see why the police got tired of that pretty quickly and sacked the statistician.

Yeah that's why the defence have their chance to try and claim it was natural causes. It's not the prosecution's fault they didn't.

It's concerning that you're ok with the state prosecution putting forward a case that simply isn't factual by saying it was the only possible explanation, that should never be allowed to happen, and making out that the onus is then on the defence to rebut it. That is not how it's supposed to work.

Of course if they had brought in the recommendations about expert witnesses from the Law Commission’s 2011 report then this may never have happened at all, but they didn't.

I think this may well be the case that ensures that those recommendations do end up being implemented.

When it is possible for something to happen naturally, and likely that it will happen somewhere, you can't prove it shouldn't have happened with statistics. That's all they are saying. So you need some sound proof that isn't statistical. That's what Lucy Letby's case lacks.

It's the job of the expert witnesses on both sides to give an honest account of the possibility of natural causes, if they are called.

No, you’re approaching this from completely the wrong angle and entirely misunderstanding what the statisticians mean.

The role of a statistician is to identify whether observations that appear significant are in fact random, and vice versa. They do this by obtaining a large sample of data with which to compare the potentially meaningful observations.

For avoidance of doubt, these observations are the things that you believe are meaningful because of “common sense”, but may actually not be meaningful at all.

To be able to determine this it’s necessary to obtain that large sample of data - all the deaths and collapses, at COCH and elsewhere, and all the people who were present before and during each of them, and then run statistical analysis. The analysis will show whether or not there is any substance to the claim that LL being “always there” was accurate. It cannot, and is not intended to, “solve” the case. Only to show whether or not there is a statistically significant correlation.

That wider sample has not been obtained. And even if it had, the data was ruled inadmissible by the judge. Because, presumably, the judge is ignorant about statistics, despite previous cases and the RSS’s public concern.

So, it is impossible to say whether or not LL was really always there.

You can claim that “common sense” tells the full story until you are blue in the face, but you will never be right.

The prosecution had evidence it wasn't natural causes so why on earth would they be taking a case to trial (for millions of £) and then arguing there are other possible explanations. If that was the case it likely wouldn't have made it to trial in the first place.

I mean if she is guilty then it IS factual that here are no other possible explanations? Or are you talking about the gospel of the world renowned experts=possible natural causes? They weren't on the scene back then.

I don't know that you can't prove it at all? Are you saying they would've said you can't prove Harold Shipman guilty with statistics either?

We'll be arguing forever about whether the case was statistical or not I think.

Well yes, it shouldn't have made it to trial in the first place because they hadn't actually thoroughly investigated the other possible explanations.

It would be similar to if let's say a police force thought that 2 people could have committed a crime and they just investigated one of them and not the other, and then put one on trial with very weak circumstantial evidence, for all they know if they had investigated the other individual there could have been a ton of much stronger circumstantial evidence against the other person, but they just put the first person on trial and made out that no one else could have done it.

They're not supposed to do things like that obviously, for any murder for example there could be circumstantial evidence against several people, so you cant just decide you have the suspect just because you have weak circumstantial evidence against one person when you know that there are still others to be investigated.

Or are you talking about the gospel of the world renowned experts=possible natural causes? They weren't on the scene back then

And if the police had followed the advice they had been given they would have had a panel of experts, but they didn't, they probably wouldn't have been the same ones but they still would have had a more robust panel than the people they did have.

Once again, statistics isn’t about proving someone guilty. It’s establishing whether a pattern is statistically significant.

I don’t know enough about the Shipman case but I expect the pattern of deaths associated with him was statistically significant.

We'll be arguing forever about whether the case was statistical or not I think.

You are the one claiming statistical significance (“she was always there”, “what are the chances”) without data. You are making this about statistics without understanding what you are doing.

We'll be arguing forever about whether the case was statistical or not I think.

Which just serves to illustrate the limits of your understanding and/or your dogged persistence in perpetuating a narrative which is now clearly false. You yourself repeatedly use and quote the terminology which proves the statistical element of the prosecution case. Always there, over and over…etc.

The prosecution had evidence it wasn't natural causes so why on earth would they be taking a case to trial (for millions of £) and then arguing there are other possible explanations.

They paid an “expert”, who was prepared to tell the police and testify in court that he could rule out natural causes and that there were no other explanations than deliberate harm. That was their evidence. There was no other evidence. He even said this in the cases of babies who were 23 and 25 weeks gestation, who had multiple confirmed medical issues.

Firefly we've talked about the Shipman case many times. Statistical evidence and unexpected deaths prompted the investigation. The fact there was a lot of direct evidence and a clear (financial) motive removed any reasonable doubt that he was a murderer.

There's a paper called "Shipman's statistical legacy" that's technically not free so I'll not link to it but I found a copy on github. It talks about the difficulty of assessing whether a GP should be investigated based on the number of deaths (or "excess" deaths). Apparently these vary wildly. They took data from 1009 doctors between 1993 and 1999. Twelve of these would have hit the "alarm threshold". Shipman would have triggered this alarm but he was not the highest on that chart. In fact the authors of the paper suggest that this type of statistical analysis could only be used as a screening tool to prompt further investigation.
So no, you can't find a "Harold Shipman" guilty based just on statistics.

The only reason Shipman was caught is that because it kept on and on happening. If he had only murdered once he wouldn't have been caught. Admittedly Lucy Letby is accused of far far fewer crimes. But babies kept dying. And the medical people began to be suspicious. I think she's guilty.

Have you read any of the few posts just before yours? You are making the same mistakes as Firefly in assuming that the pattern you can see is verifiably statistically significant.

You can believe there were murders based on the medical evidence (not sure how, but it’s arguable).

However you can’t say that the deaths were statistically significantly linked to Letby because that has not been demonstrated. Your common-sense view is irrelevant.

So how do you explain all the other deaths on that unit that they couldn’t (much as they’d have liked to) pin on Lucy Letby?

They came up with the preposterous idea that she’d spiked bags ahead of time, to try and explain that one away, and even THEN, couldn’t assign them all to her.

The unit was dirty, chronically understaffed and taking on babies who were far sicker / more vulnerable than they were rated to be caring for. The consultants didn’t come out of the trial looking at all good. Rounds just twice a week, and reportedly dismissive and bullying the nurses on duty (not just Letby either).

One of the doctors (who initially accused her) has been caught out by the Thirlwell enquiry, when an email submitted into their evidence has him acknowledging that Letby had asked him for assistance with a particular baby.

He told the police (and, I believe, the criminal trial) that she’d not done so.