Kathryn Leeming

[Kathryn Leeming @KatLeeming]

@BristOliver For the home testing kits its the same swab - throat first then nose - so only one experimental unit sent back.

[Richard Everitt @bayesian_stats]

I'm advertising for a postdoc in developing new ABC / particle MCMC and related methods to apply to (potentially misspecified) environmental models. Apply here by 2nd September https://t.co/YNaacQBAK9 @LandscapeDecis1 @NERCscience @warwickstats

[Kathryn Leeming @KatLeeming]

@BristOliver I'm a fan of the reproducible tweet plot format so I feel the need to add: dates = july

[Kathryn Leeming @KatLeeming]

@joannepenn @Kit_Yates_Maths @BBCPanorama @FayeKirklandGP The modelling wasn't about lockdown vs no lockdown, it was taking different steps (eg people with symptoms staying home), vs doing nothing. Once people were voluntarily changing their behaviour we were out of the "do nothing" scenario.

[Kathryn Leeming @KatLeeming]

@BristOliver Interesting mix here of taking CIs into account (RHS of slide) and not (LHS)

[Kathryn Leeming @KatLeeming]

@BristOliver I'm a 1. I also have been rejected at interview by a different member so maybe that's a +inf.

[Kathryn Leeming @KatLeeming]

@BristOliver Very selective screen grabs there, he goes on to say "Was it really Covid-19? I don't think so. The first confirmed case of coronavirus in the UK was in late January" Ie either he was infected earlier than the currently known UK cases, or he's had it since without knowing.

[Kathryn Leeming @KatLeeming]

@BristOliver It brings up a new q: is the misuse of a + worse than the omission of a ^?

[Kathryn Leeming @KatLeeming]

Here's what I found when I looked into what England's Covid19 daily deaths in hospital actually mean, and how comparable they are day-to-day.

[Significance @signmagazine]

From 2-23 April in England, says @KatLeeming, the majority of announced #Covid19 deaths on any particular day occurred more than 24 hours beforehand. "The difference between the date of death and the date of announcement can be several days, or even weeks" https://t.co/ekF7DwbxwW

[Royal Statistical Society @RoyalStatSoc]

.@FullFact on what we know, and what we don’t, about the true coronavirus death toll, with stats help from RSS stats ambassador @anthonybmasters
https://t.co/7lT5xgKkJx

[Kathryn Leeming @KatLeeming]

The level of detail in this is incredible.

[Office for National Statistics (ONS) @ONS]

We've also published an interactive map where you can search deaths relating to COVID-19 by postcode https://t.co/JBSczRj0tz
#COVID19 #coronavirus

[Kathryn Leeming @KatLeeming]

@BristOliver @phil_messenger HuffPost has 475 deaths in England, 445 is the comparable figure for Eng (+ve test, in hospital) as 30 of those had COVID on death cert but didn't have a positive test.

[Kathryn Leeming @KatLeeming]

From story: Different figures for deaths, explained https://t.co/BTg5bBixur via @BBCNews

[Kathryn Leeming @KatLeeming]

@BBCRealityCheck "The ones Health Secretary Matt Hancock just announced are deaths in the last 24 hours of people who had tested positive for coronavirus" - no they're not: 1. they're *reported* deaths 2. they were reported 48-24hrs ago.

[Kathryn Leeming @KatLeeming]

@pjie2 @drjameslmurray @d_spiegel In England the weekend effect seems to have (at least) two causes: fewer trusts reporting at the weekend, and fewer long announcing lag deaths are reported at the weekend.

[Kathryn Leeming @KatLeeming]

@dpm_79 @uniquenamearyan @DHSCgovuk @CMO_England @PHE_uk @NHSEngland @scotgovhealth @CMOWales @PublicHealthW @healthdpt @publichealthni The total will match, but the shape does not, eg date of highest deaths is later in the announced figures. The point is that we can't take these numbers and simply compare day-to-day, and the values don't reflect the number people that have died in a 24hr period.

[Kathryn Leeming @KatLeeming]

@jas88 @JoPickering468 @DHSCgovuk @CMO_England @PHE_uk @NHSEngland @scotgovhealth @CMOWales @PublicHealthW @healthdpt @publichealthni Yes, but I think the grey bar should apply to the whole chart, numbers are being revised upwards on all dates, sometimes by a surprisingly high amount.

[Kathryn Leeming @KatLeeming]

@BristOliver Hard to tell as obviously the bars aren't full particularly on RHS, but deaths announced within 3 days do look to be dropping (whereas those announced within in a day don't)...

[Kathryn Leeming @KatLeeming]

@BristOliver A lot of deaths have been announced this week that occurred more than a week ago (more of these than usual this week) - today 236 in Eng were from 7+ days ago compared to 125 of those announced last Sat, lag distribution may be making things look worse 🤞

[Kathryn Leeming @KatLeeming]

@kevins1966 @SamanthaJanePez @DHSCgovuk @CMO_England @PHE_uk @NHSEngland @scotgovhealth @CMOWales @PublicHealthW @healthdpt @publichealthni I downloaded the individual daily spreadsheets from the source website (NHS England) and created the plot from those.