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

The issue, of course, is that we have no way of knowing whether the jury found the statistics talk the prosecution provided to be irrelevant or not.

We don't. They didn't find her fully for everything though. So by inference it does look as though just her very presence on shift wasn't enough.

Right though that we will never know for sure.

Without that evidence though, would the case have been strong enough to even be brought to trial, I wonder?

I do not see an image.

@TheChickenOrTheMiniEgg no I see it's not appeared. I'll just have to link to the reddit about it and someone posted it-

https://www.reddit.com/r/lucyletby/comments/1qusqse/doctos_rota/

Really? Even before Shoo Lee came out the woodwork? Do you work in healthcare then?

I thought you were sick of statistics.

Does the key mean presence at deaths and collapses on the indictment, or for convicted cases?

What about other collapses and deaths that were ruled non-suspicious because Letby wasn’t there?

And what does “presence” mean, especially with regard to the insulin cases?

Yes. To all questions. I assumed there had been mis-reporting because it’s such a preposterous notion.

OK so are you disbelieving they did it that way then?

Interesting. So you don't believe it was AE at all? It's so rare and we only have limited accidental cases to go by. Why would it be a preposterous notion?

Why are you talking about air embolism? Did you not understand my post?

I thought you were sick of statistics.

Ha yeah so did I. Actually hadn't seen that chart before though. I dunno I would have all those same questions. Just from that sample it looks very suspicious though.

I must be!

The charts, with or without doctors' names, are meaningless unless you establish:

1. objective and transparent criteria for including events

2. staff working patterns

Saying they look suspicious without this information is exactly the kind of misuse of statistics we have all been talking about.

This is an example of exactly what we’ve been discussing. There are many interesting points about this in the Royal Statistical Society’s report, previously linked and again, below.

It references many points discussed here and is actually an interesting and accessible read. But if you want to go directly to a relevant piece of information, have a look at page 18. Let us know what you think.

https://rss.org.uk/RSS/media/File-library/Policy/2022/Report_Healthcare_serial_killer_or_coincidence_statistical_issues_in_investigation_of_suspected_medical_misconduct_Sept_2022_FINAL.pdf?ext=.pdf

I think this is a lot of people's problem with statisticians. You provide them everything you have and they just ask for more and more until the point no one can answer further. It's like that email Dewi Evans had with one of them. Back and forth for ages, gave him all the info and the statistician still never got back to him. Didn't like where it was going that's why! No wonder the police dumped the one they had.

A lot of people’s problem with statisticians.

That’s quite an ironic statement in this conversation. 😀

Should you be accused of murder, would you not hope that any data used to prosecute you was rigorously and accurately analysed? By a statistician?

But surely nobody should have created this chart unless they had transparent, objective criteria for what events to include?

And I wouldn't expect you to have the information on work patterns. But the police? For a homicide investigation? It's not asking much.

There is something statisticians say which will probably annoy you but needs to be said. Bad data isn't better than no data. Either you have enough information to conduct a statistical analysis, or you don't conduct a statistical analysis (or imply that you have done so)

You must be what?

You do understand the difference between air via a NG tube and an intravenous air embolism?

What if someone doesn't have all the data they require though? Seriously, read that Dewi Evans email and you'll see how they just ask for more and more impossible data.

And no I'd think I could prove I didn't murder someone without having to rely on statistics. I don't think I'd need a statistician at all.

Yeah I get you now, read it all a bit too quick earlier and assumed you meant AE. How many times was she accused of air down the NG tube? And what about all the other methods?

But if you’re going to refer to data and say “That looks really suspicious, doesn’t it? That can’t be a co-coincidence”, then it’s perfectly reasonable to expect a statistician to ask for the metadata and calculate the probabilities.

And you over and over appeal to statistics (“what are the chances..?”) without apparently even understanding that’s what you are doing.

No, of course statistics cannot prove or disprove a murder, but when they are being mis-used to over-state probabilities, then they need to be challenged.

But the cases they looked at were initially picked by Letby's accusers. That creates a risk of unconscious bias, where they're more likely to judge a case suspicious if Letby was around, and non-suspicious if she was not. The only valid way to remove that risk is to have outside experts who don't know the suspect's name review all cases (or a representative random sample), flag what they think is suspicious, and only then analyze whether the suspect's association with flagged events is infeasible by chance. Without such blinding which was not done any "common denominator" conclusions are invalid.

Evans who is not a statistician and had limited data to work with attempted a statistical analysis and asked someone more used to statistics to look it over.

Obviously that individual couldn't just pretend it was all okay. Why should Evans be good at statistics? That's not a criticism - it just isn't his field.

The British Medical Journal published a good clear short article recently on problems with this kind of approach. Amy Wilson, who is working with Lucy Letby's defence, was one of the authors

www.bmj.com/content/392/bmj.s226

The convictions for Baby C and Baby P were murder by injecting air via a NG tube. They hedged their bets for Baby I and convicted on air via the NG tube and/or intravenously.

You know perfectly well what I think about the other supposed methods, we have discussed them repeatedly.

However let me remind you that we’re having this conversation specifically because you accused me of “coming to the case months after it ended and jump on the MOJ bandwagon desperate for it to be one”, said you were surprised that if I had “followed the trial in real time I had not thought the evidence pointed to overwhelming guilt”. I read the news reports throughout the trial, and shook my head at the poor reporting, because surely it wasn’t possible that anyone would think it was possible to kill by injecting air via a NG tube.

Be honest now, you hadn’t understood the difference between air embolism and air in the stomach, had you?