Data and analysis thread, started 12 November

[NoGoodPunsLeft]

Previous thread here:

Data and Analysis Thread, started Oct 29 www.mumsnet.com/Talk/coronavirus/4064113-Data-and-Analysis-Thread-started-Oct-29

Regular lurker but I frequent poster, didn't want to lose the threads.

@sirfredfredgeorge - actually I did think a model SHOULD be sensitive to adjacent areas, especially ones with so many roads, bridges and a park crossing between them, never mind a raft of shoppers, schoolkids, buses, trains etc.

Whoops, hit send too soon.
I missed hospitals, GPs etc.
I would say it's really bloody hard, except that physical mapping tools do exist and that's why they include longitude/ latitude postcode mapping. A major modelling tool like Imperial should really be using them if we're to rely upon the results.

A quick look for England at the number of councils increasing / decreasing in terms of case numbers. This takes the 7 days to 16th November (not far off being complete) as compared with previous 7 days. The light dotted red line shows where the cut off between decreasing and increasing was when I did this last week. Obviously we want that vertical red line to keep shifting further right on the graph.

The column is the cumulative decrease / increase in cases, and the blue line the percentage change by specific LA. In the decrease the percentage decreases are relatively small - so mainly LAs coming down from a high starting point; conversely there are some significant percentage increases, reflective more that many of the councils that have seen increase in cases are coming from a relatively lower starting point in terms of absolute case numbers.

There are distinct regional differences which I'll post in a bit

worth adding, and something I should have included, is that the councils showing week on week falls, or remaining flat, account for 46.3% off all Councils and 47.8% of the total population in England

Thanks @boys3. I think almost half can be tentatively celebrated, especially as the cases by specimen date looks to finally be speeding up plus total cases dropping, so this may look even better in another week. Of course by then we've only got 4 more days of lockdown left.

Do we have any rules yet for how councils will have their tier allocated? Also how much will be a regional approach? Or are they waiting to see what the cases do to decide?

maybe nosecco at this stage :) @Mrex

Regional breakdown to go with the graph - should be self explanatory.

Only 2 of 32 LAs in London showing week on week fall; as opposed to 34 of 39 in the North West and 17 of 21 in Yorks & Humber.

Even split in the East Midlands (19 up, 19 down and 2 unchanged), but a net reduction of just over 500 cases.

South West not far off an even split, and whilst overall cases increased this numbered less than 50.

North East overall positive trajectory; also reduction in cases relatively limited.

South East around 2/3 up, and adding over 2,500 cases net. Although digging deeper, all districts in Oxfordshire decreasing, and most districts (8) in Surrey also showing a fall; whereas all bar 1 in West Sussex increasing, all in East Sussex increasing; all 12 in Kent increasing

East also a bit Jekyll & Hyde, 18 down, 1 unchanged, 26 increasing, but over 1,000 cases added

so this may look even better in another week. Of course by then we've only got 4 more days of lockdown left.

I'd certainly hope we see more progress over the next couple of week - and of course even with lockdown ending we'd expect to see its impact carry through for perhaps a further 7-10 days. Which brings us on to:

Do we have any rules yet for how councils will have their tier allocated? Also how much will be a regional approach? Or are they waiting to see what the cases do to decide?

A cynic might assume that whilst rules may be set there may be a certain level of fluidity in their application. I'd anticipate some sort of tier system based on hospital admissions and capacity, cases rates and positivity levels etc . There will likely be much wailing and gnashing of teeth from a variety of self interested groups, and the benefits (along with those of lockdown) of whatever is eventually put in place may well get thrown away with a complete Xmas free for all.

Nosecco, not even a beer? Oh boys3, you need more faith.

you need more faith

@Mrex just watching the BBC News - nothing about Angela Merkel replacing Boris.

Maybe I could, at a push, stretch into Cava (cheaper own brand end of the market though). :)

I’m really struggling to understand how and why cases in Kent have risen so much given we’re on lockdown. Is it that people are ignoring restrictions or it is somewhere like schools being the driver, and why are cases worse now than they have ever been given we're on lockdown with higher restrictions than we’ve had in months?

In Swale I believe they're are large numbers of buses taking kids to and from the mainland, as to why overall, this is just lockdown lite and I suppose the virus was going to do whatever it was going to do anyway and is just going through the same pattern in the south as it did in the north a month or so ago?

I’ve NCed but been here for ages

RP has spotted that they’ve removed 150 deaths from the death stats, going back to 18th March! So the actual death number yesterday is artificially reduced by 150

Rates in North Kent were similar to those in Essex, Essex asked to go into tier 2, I think there is a good argument that Swale and Thanet should have too. However Kent is a massive county and Folkstone had extremely low rates (at one point lowest in England).....

Just been looking at the stats for Kent. Rates were rising in most parts before lockdown but as someone below has pointed out, Kent is a large county so looked at overall they weren't that huge, despite some areas within Kent being high. They were rising in pretty much all parts of Kent though, including the posher areas (Seven oaks, Tunbridge Wells) - I'd say this is an example of evidence that Tier 1 level restrictions weren't sufficient.

Swale stands out in as particularly high. The rise there was sharpest between the 6th and 12th September, which would be suggestive of a rush of socialising immediately before lockdown. Since then rates have been up and down a bit. The local press seems to be blaming this in a lack of mask wearing or maintaining distance www.kentonline.co.uk/sheerness/news/emergency-meeting-called-as-covid-19-cases-soar-237795/ But the article also mentions some outbreaks in the 3 prisons in the Isle of Sheppy - the map data would suggest this is part of the answer as East Sheppy has very, very high rates (and I think that's where the prisons are). Other possibilities would be schools, school buses or onwards transmission within households of cases caught pre-lockdown. It's also quite a deprived areas, not many professionals working from home and I would guess many needing to travel in to London for jobs via public transport (not much local work, unless they work in the prisons....), so the recent lockdown may not have had a huge impact.

Can anyone tell me on this data thread why the death rate is coming down but hospital admissions due to Covid are still going up? Normally there is a correlation between the two surely? I know hospitals are getting better at treating Covid but are there any new treatment protocols ( trials) improving the death rate?

Because of the delay between admission to hospital and death maybe?

So in a couple of weeks deaths should follow the admission pattern.

There are also deaths in care homes and in private homes, classification of those deaths happens on a different timescale to hospitals. It's not unusual for them to have much older cases added or removed following actual death investigations. Also not all hospitals seem to report deaths at weekends, so a weekend drop means very little and it's weekly totals that are more useful.

I think the death rates have plateaued rather than gone down. The number of deaths reported yesterday is artificially low because of a backdated adjustment (as mentioned by TheDinosaurTrain).

Hospital admissions, on the other hand, have decreased from the peak (11/11 - 1,711 for England). The latest number is 1,571 (18/11). I am only looking at England - Wales uses a different methodology so UK wide hospital admission numbers, unfortunately, are not very meaningful.

@teta

Can anyone tell me on this data thread why the death rate is coming down but hospital admissions due to Covid are still going up? Normally there is a correlation between the two surely? I know hospitals are getting better at treating Covid but are there any new treatment protocols ( trials) improving the death rate?

Yes that seems to be the case. Fewer very sick patients (due to treatments and age profile of admissions) and shorter stays.

On the previous thread we were all mulling over the ONS ad hoc data on teachers and everyone with a statistical brain was confused by it. The ONS has been reported for misuse of data and this Twitter thread is quite geeky, and illuminating! For those of you who want to revisit it:

twitter.com/SarahDRasmussen/status/1330294404758859779

I am sure this is the tip of the iceberg in terms of data fudging.

Interesting thread Piggy. She's not mincing her words!

It is not the role of the ONS to manufacture false evidence for purposes of state propaganda

Yes. Ouch indeed! It does seem dodgy.

I can't imagine there will be much of an outcome other than a mild telling off. Be interesting if they then do their actual research properly. Never been clear what they even did it for in the first place.

That's quite shocking @piggy. It's completely unacceptable to compare the two groups over different time periods when we know that Covid was increasing significantly during the latter half of October (the period when they were including cases from other groups but not teachers). I said at the time that the error bars were so large that the analysis was unable to prove anything. But on reading that Twitter thread I'd say I agree with her conclusion that it looks very strongly that they have deliberately created three groups of teachers (secondary, primary and unknown) in order that the headlines can focus on the lower (but statistically completely insignificant) rates in primary and secondary teachers whilst all three groups of teachers are individually too small to allow statistically significant differences between the important distinction of teachers Vs non teachers to be visible. Combined with the different time periods used, this is really unacceptable and I do hope they get held to account for it, and produce some more accurate analysis urgently.

I've just been watching the indie sage review on the 20th. Apparently there was a plateau in hospital admissions a couple of weeks ago, so that might account for the lower death rates thus weekend ( compared to last weekend) Hospital admissions went up after that, so it will be interesting to see what the figures are this week.