Lucy Letby case expert witness was under fitness to practise investigation

[ITMA2000]

Wow! Pot, kettle, black? Need a new investigation?
Lucy Letby case expert witness was under fitness to practise investigation during trial | Lucy Letby | The Guardian

I can understand why Evans wouldn't have all of this information, but it's not much to ask of a six-year long, multi-million pound police investigation.

Evans didn't have enough data to conduct the kind of analysis he wanted. That's all. If he and Elston chose to email back and forth about it, that doesn't affect the misuse of statistics by the prosecution.

You can’t possibly genuinely think that the statisticians are “big mad”, whatever the hell that is supposed to mean. Nobody can be that ridiculous, surely?

All they have to do is moan constantly about how badly the stats were done, they won't ever analyse the stats themselves and give us those numbers.

What does this even mean? if they don’t have the data, how can they analyse it? That’s the whole point of their objections. Partial data has been used to spew out suppositions.

The statisticians are not offended, or whatever you are implying.This is their job, and the data set you listed is a perfectly reasonable ask. If I’m investigating an incident at work this is the kind of information I need, and more, to be able to understand what has happened, why, and what needs to be done to prevent a recurrence. I don’t think this is unusual.

The statisticians working with McDonald have given a full statistical report to the CCRC. Evans and Elston's emails are just a sideshow.

Elston does some interesting presentations from the data he has but so far as I know he isn't submitting anything to the CCRC.

I'd love to see the full report too, but I fully understand that they won't share at this point.

This case gravely impacts many families and the children they’ve lost. It gravely impacts the woman convicted and her family. It impacts past and future patients and the confidence in their care at the hospitals concerned and in fact all NHS hospitals.

Only if she's innocent, which you seem to be assuming is a given. Otherwise Lucy's own actions gravely impacted many families and their children.

I would wager he knows Dewi Evans can't possibly know the answers to all those questions. I'd also wager that if DE came back with for example the level of competencies for all staff (ie band 6 nurse or whatever) that still wouldn't be enough. He'd then want detailed analysis of how competent they all were. Something that's practically impossible. I'm sure a similar thing probably happened with the police and their statistician and that's why they dumped her. If you just keep asking for more and more data it doesn't get anyone anywhere.

Even if you think she’s guilty then you should be just as concerned about the safety of the conviction in order to prevent the issues that allow for these concerns and doubts.

The conviction is very very safe.

Great. So everyone should be happy then! So why all the complaints over statistics if they've got one (or multiple) for the defence willing to put in a report.

@Firefly1987 the defence had an opportunity to instruct a statistical expert and they decided not to. One offered his services to them. In the end he was warned by the police to stop talking about the case. He was actually on here.

There's no way a competent statistician would analyse the probability of Lucy Letby being present without answers to those very basic questions. It's not a flex, a power play or a mind game. Even if he didn't respect Evans, even if he was out to annoy him or thwart him, he needed that information (and more)

That's the thing about science. Even if Peter Elston is a pain, Shoo Lee is a supervillain, Neena Modi is a conspirator or whatever their detractors want to claim, their remit is to produce verifiable, replicable findings, guided by science.

When you read Elston's replies there, no matter what you think of his character or what games you think he might be playing, you can see that he is asking the questions anyone versed in statistics would ask. You can invent stories about why if you like, but the only important reason is that this information is scientifically necessary. Anyone else undertaking the analysis would ask these questions too.

People are complaining about how statistics were used in the original prosecution and conviction, not (so far as I'm aware) that the defence doesn't now have the statistical data it needed.

So what does that tell you?

I think that was Richard Gill? Was he actually on here? Neena Modi also made contact with the defence during the trial and didn't hear back. I have no idea though whether the defence could really have instructed them at that point.

As we have discussed above, the judge did not permit questioning of the construction of the case at the trial, and ruled out use of some of the materials that would have made this possible.

It would also have been a double edged sword, in my opinion, to introduce a statistician because a lot of people are irritated by statistics when they don't conform to their intuition.

Yeah, that's why she didn't call one to help her case, because she thought it'd irritate the jury 😆I've heard everything now.

I just can't help noticing that lots of people don't want to know about statistics when they challenge their existing opinions!

Coincidentally, there's a new video on YouTube today about how Judge Goss blocked avenues that could have led to statistical analysis by the defence, for anyone interested in that point

Well everyone and everything sure was up against our super nurse wasn't it? Now you're claiming calling a statistician would harm her case. It sure would but not for the reasons you seem to have plucked out of thin air.

And who said anything about the juries existing opinions? As if anyone WANTED to believe she was guilty?

If the statisticians came up with any figures of their own instead of just constant criticism about how it's been done previously then I'd listen.

It's not a theory I'm particularly attached to. The fact that they couldn't draw on the relevant data or discuss the construction of the case would certainly have been important. The other is just a thought. You are welcome to disregard it - that doesn't change anything.

Statisticians need data to come up with figures. But you are indignant about them requesting data. So why should you expect to see figures?

Evans compared death rates on Letby and non-Letby shifts:

"There were 15 deaths during Letby’s 163 shifts. There were 2 deaths during the (798 – 163) 635 shifts when she was not on duty."

First, his numbers are off. There were 13 deaths and she was on duty for 10 of 13. So, 10 on her shifts and 3 on others. Since his numerators are off, it's hard to trust his denominators (163 and 635), but let's say they're right.

Evans assumes you'd expect the same rate of deaths per shift for Letby and non-Letby shifts if she's innocent. But that's only true if the two sets of shifts don't differ in any ways that affect the death rate. Evans does nothing to check if they do. E.g. her shifts had 3 indisputably natural deaths (ones she wasn't charged with), which is 4x the rate in non-Letby shifts (3/163 vs 3/635). If you take Evans' original numbers it'd be 15x (8/163 vs 2/635). If there were no relevant group differences, you'd expect no difference in natural death rate.

Did Letby/non-Letby shifts have the same fraction of night shifts (when more babies die)? If she worked more nights than average, then non-Letby shifts will have correspondingly fewer nights, enhancing the difference.

Did Letby/non-Letby shifts have the same fraction of shifts with a high-risk baby? E.g. we know her shifts had several multiplets -- what fraction of non-Letby shifts did? Were the distributions of birth weights, APGAR scores, prematurity levels, and other known risk factors the same in the two groups?

Did Letby/non-Letby shifts have similar staff coverage with a similar mix of experience?

These are basic questions, not nitpicks. You have to compare like with like. Group differences in death rate will reflect all group differences. You can't say the difference is due to Letby unless she was the only difference.

Only if she's innocent, which you seem to be assuming is a given. Otherwise Lucy's own actions gravely impacted many families and their children.

No, what a shortsighted view, none of us know whether she is innocent or not. We are commenting on the concerns regarding the safety of the conviction. The case having potential issues as great as these discussed impacts the families who have lost children, of course it does. How can you possibly think that it doesn’t? You would need only a tiny amount of empathy to understand that.

You’re somewhat making a point for us again here, regarding the statistics. It’s complex to source clean, relevant data, and then extrapolate safe conclusions from it. That is indeed the whole point.

But they know that's poor care. No one has suggested there's a serial killer. Which is just more proof that no would need to invent one to cover for a poorly performing unit. Also maternity units are different to NICU. This is a national scandal that should concern all of us you're right but it's very different to the Letby case.

Gosh that layed out 15 deaths to 2 makes me see why the prosecution are so sure she’s guilty

those deaths when she is around are phenomenal - even after all the possible reasons - plus the poor performing unit would be a constant across all deaths

Maternity inquiry tends to be used as shorthand, but Baroness Amos is investigating the full range of maternity and neonatal care, including neonatal units and NICUs. The crisis isn't confined to delivery suites

https://www.matneoinv.org.uk/updates/independent-investigation-into-maternity-and-neonatal-services-in-england-interim-report/

Depends on the questions Elston asked - and more. Evans is also counting four deaths which happened at other hospitals - presumably either because Lucy Letby was working at Chester at the time, or because she was on shift when they deteriorated. But when babies deteriorated before her shift and died on it, they counted against her too. Lies, damned lies and statistics at their finest

There is also the lottery fallacy. When you have enough pairs of random events (a nurse works particular shifts / babies die on particular shifts, a man chooses particular lottery numbers / the National Lottery selects random numbered balls) some will end up matching. Very few nurses will end up at most deaths - but because there are thousand nurses out there, a tiny number will. Very few people will choose the same numbers as the National Lottery - but because thousands of people play, a tiny number will.

You are showing easily how statistics can mislead.

Yes, but what about the alleged cases that weren’t included because she wasn’t there?

The medical expert came up with at least one method of harm that was so outlandish that many people (including practising neonatologists) who read the trial reporting were alarmed and like me thought, hang on a minute, that isn’t a thing.

Well it wouldn't "be a thing" because healthcare workers tend not to attack babies. Obviously it'd be methods of harm a lot of experts had never heard of and be incredibly rare, thankfully.

He also cannot show the objective criteria for determining whether an incident was suspicious. Therefore his analysis is not reproducible and is unscientific.

Seems like that's the nature of medicine. Shoo Lee also can't rule out deliberate harm as it happens. Big surprise people aren't half as critical of him.

His colleagues in the case rubber stamped his opinions rather than independently coming up with their own.

So? Presumably they could disagree with them all the same.

1. The rest of the prosecution case was built on a web of so called circumstantial evidence such as handover notes and facebook searches which is actually meaningless when you look at it. E.g. the facebook searches do not show any particular interest in the indictment cases, and neither do the retained handover notes. Taking these irrelevant things together does not give them more weight.

Yeah sure, a 10 month trial and all it consisted of was facebook searches and handover notes. Not anything about establishing she was in the room at the point of all the collapses? Even when she had no reason to be in there? Are you sure? I'd say that had a lot more to do with her being found guilty than handover notes and facebook searches.

The problem with Evans's invented method of killing (air in the nasogastric tube) is that if it killed babies, they would die of air in the stomach by accident all the time. Because they get far more air in through ventilation (C-pap) than you could inject with a syringe.