Coronavirus outbreak

[vickster]

Well, yes, obviously as it exceeds the population 3x over...

Indeed but you provided the data in response to @Pale Rider question on how many people and stated 160 million

 

[roubaixtuesday]

Indeed but you provided the data in response to @Pale Rider question on how many people and stated 160 million

Well, we both agree now anyway

 

[matticus]

Ruby: I have to give you credit for flagging the problem back in August, when I admit that I was just living life (safely!) and trying to be optimistic:

Currently our case numbers have been rising week on week for some time.

I hear on the radio that govt are going to run an advertising campaign to get people back to offices due to the effect of low foot fall on city centres.

But surely, if cases are rising, we need to take action to reduce transmission, not actions which will increase it, particularly with the return of schools too. I genuinely don't understand govt strategy.

In hindsight, the rising figures through summer/autumn should have triggered Government action long before schools started back.

However, I'd like to question your recent statement:

It's obvious that more social interaction = more viral transmission.
So trivially, "Eat out to help Out" subsidised increased transmission of the virus.

Equally, quantification of this is difficult at best, and probably impossible, as the Warwick study attempt notwithstanding there are huge uncertainties.

It was, however, part of a pattern of behaviour of attempting to play off economic benefits with viral disbenefits that was widely challenged at the time and culminated in the catastrophic decision to keep shops restaurants and schools open until the U turn one day into the new school term.

The bold phrase: surely this is inherently part of all governments' jobs during this pandemic. In every affected country. It's simply not feasible to have total lockdown - the economy tanks (which has huge long-term effects on society+healthcare), and many practical issues crop-up in the short term (e.g. essential workers and the infrastructure they need).

But how you can make blanket criticism of this? You're not the only one, so maybe I'm on the wrong side here; but I am genuinely confused!

 

[roubaixtuesday]

But how you can make blanket criticism of this? You're not the only one, so maybe I'm on the wrong side here; but I am genuinely confused!

Yeah, that's a fair question.

First, a general point. It's very clear that those countries who have minimised viral transmission have also minimsed economic impact. So, in general it must be false to say that there is a play-off between health and economics - all the evidence shows that minimising the virus actually minimises economic harm. Because if we don't we end up locked down with massive economic harm.

What does that mean in any specific circumstance? Well, I'd say, if cases are rising, we should always respond by taking action to reduce transmission. Exponential growth means we know it will always run out of control eventually, so better to act sooner rather than later. More resource to test, track and trace, or more restrictions in essence.

During last summer, cases were rising. We should have taken action to stop that. At least, we should have stopped taking action that we positively expected to exacerbate it!

 

[Ajax Bay]

So the 'trick' has to be identifying NPIs which have relatively effective 'R' reduction effect versus relatively less economic and social detriment. UK Cases bottomed out in early July (at a very low level, 7-day averages, date of specimen) and rose inexorably for 3 months thereafter (as @rt revisited).
9 Jul = 587
24 Aug = 1181 (doubling in 46 days)
2 Sep = 2369 (doubling in 11 days)
18 Sep = 4756 (doubling in 16 days)
28 Sep = 9556 (doubling in 10 days, and continues its exponential rise at that index till 5 Oct)
So 'exponential' in the 6 weeks: 23 Aug - 5 Oct. Perhaps we might infer that the effect of the tertiary residential (multiple occupancy) education term starts in September (and loads of testing) were a significant contributor together with associated leakage out to the local community,
In October the curve becomes linear. Not clear why though: steady increase 6 Oct to 10 Nov - implication is that R only a tad over 1.
And then rapid drop off which we can credit to the lockish down month 5 Nov - 1 Dec.
Lock down Dates
Before the daily cases steep rise through December to its peak in early January, the 'minimum' was on 27 Nov - still well in the meat of that 4 week lockish down. Why did incidence start rising? We could assume the prevalence of the B.1.1.7 variant (with its +40% transmissibility) was starting to have an effect counteracting/overmatching the 'R' control (reduction) that the lockish down had achieved.

In the relevant 4 month period 20 Jun - 26 Oct, testing (PCR) quadrupled (78k --> 312k, steady increase week on week). I suggest this is a possible significant contributor to the very slow rise in incidence in July and August (and the actual prevalence increase was minimal, eating out or not).

 

[roubaixtuesday]

In the relevant 4 month period 20 Jun - 26 Oct, testing (PCR) quadrupled (78k --> 312k, steady increase week on week). I suggest this is a possible significant contributor to the very slow rise in incidence in July and August (and the actual prevalence increase was minimal, eating out or not).

I suggest it doesn't come close to explaining it

Tests 1st July 109k.
Tests End Aug 187k

(7 day averages from ourworldindata)

This will likely have had some impact on positives, but no way explains it; test positivity rose over the period !

You're comparing test figures between June and October and claiming they're relevant to a case rise in July and August.

 

[mjr]

So the 'trick' has to be identifying NPIs which have relatively effective 'R' reduction effect versus relatively less economic and social detriment.

I agree with that...

UK Cases bottomed out in early July [...]

But why the fark are you then looking at cases not R estimates? It's now far enough in the past that we have better estimates using well-known time series methods like Kalman filters and corrections for things like testing capacity varying.

The graph looks pretty clear: both the Bank Holiday weekend finishing "Eat Out To Spread Covid" and Gove's "Go To Work, Save Pret, Spread Covid" were great boosts — for covid. If this really was a war, as Johnson sometimes suggests, both Sunak and Gove would probably be accused of aiding the enemy by now.

 

[Ajax Bay]

In the relevant 4 month period 20 Jun - 26 Oct, testing (PCR) quadrupled (78k --> 312k, steady increase week on week). I suggest this is a possible significant contributor to the very slow rise in incidence in July and August (and the actual prevalence increase was minimal, eating out or not).

I suggest it doesn't come close to explaining it
Tests 1st July 109k.
Tests End Aug 187k
(7 day averages from ourworldindata)
This will likely have had some impact on positives, but no way explains it; test positivity rose over the period !
You're comparing test figures between June and October and claiming they're relevant to a case rise in July and August.

I did not say increased testing "explains" the very slow incidence rise; I suggested it was a contributing factor to the increase.
The test figures rose steadily at a relatively constant rate from 20 Jun - 26 Oct. 1 Jul - 31 Aug is a sub-section of that.
gov.uk figures (7-day averages)
1 Jul = 112k tests, 572 cases
31 Aug = 185k tests, 1978 cases

 

[mjr]

In every affected country. It's simply not feasible to have total lockdown - the economy tanks (which has huge long-term effects on society+healthcare), and many practical issues crop-up in the short term (e.g. essential workers and the infrastructure they need).

But how you can make blanket criticism of this? You're not the only one, so maybe I'm on the wrong side here; but I am genuinely confused!

We can make blanket criticism of it because the expert guidance has been explicit for a while now: lockdowns should be swift, strict and short. If a government delays, that means more people die, both as a direct result of exponential spread and indirectly because the lockdown has to last longer to bring cases back down, which means more economic damage and other long-term effects.

Johnson's government had not yet learned this lesson by January, dithering for weeks to months each time, but hopefully maybe vaccination and seasonality means we don't discover whether they've learned it for lockdown 4. Macron's government doesn't seem to have learned this, trying to postpone France's current lockdown far too long since mid-Feb at least. Merkel's government appears to have farked up this time, too, listening to anti-lockdown idiots too long, possibly worried by elections, when it looks like the coming lockdown would have been best about a month ago.

But it's fair to point out that this is easier to say with hindsight. I'm not sure if anywhere has got this consistently correct. By definition, when a country needs to lockdown to recover, then testing, tracing and quarantine have failed, so multiple lockdowns means those public health measures failed multiple times and what chance is there that a country with a failing public health system can still trigger a lockdown quickly?

 

[roubaixtuesday]

I did not say increased testing "explains" the very slow incidence rise; I suggested it was a contributing factor to the increase.
The test figures rose steadily at a relatively constant rate from 20 Jun - 26 Oct. 1 Jul - 31 Aug is a sub-section of that.
gov.uk figures (7-day averages)
1 Jul = 112k tests, 572 cases
31 Aug = 185k tests, 1978 cases

Apologies, I misread your post.

I stand by the fact that all the evidence is that prevalence genuinely increased through July and August - and it was obvious at the time.

Because we're looking at exponentials, that it was at a low level isn't really all that relevant IMO.

 

[Eric Olthwaite]

We can make blanket criticism of it because the expert guidance has been explicit for a while now: lockdowns should be swift, strict and short.

Yes I think the success of the Welsh October lockdown demonstrated that beyond doubt.

 

[roubaixtuesday]

I agree with that...


But why the fark are you then looking at cases not R estimates? It's now far enough in the past that we have better estimates using well-known time series methods like Kalman filters and corrections for things like testing capacity varying.

The graph looks pretty clear: both the Bank Holiday weekend finishing "Eat Out To Spread Covid" and Gove's "Go To Work, Save Pret, Spread Covid" were great boosts — for covid. If this really was a war, as Johnson sometimes suggests, both Sunak and Gove would probably be accused of aiding the enemy by now.
View attachment 583749

I think you're torturing the data there TBH.

Specifically the huge spike is almost certainly the result of the testing shortage in Sept being resolved rather than a genuine spike in R.

I'm extremely skeptical R values derived from nationwide case data tell us much more than the general trend rather than impact of specific events or policies.

 

[roubaixtuesday]

Yes I think the success of the Welsh October lockdown demonstrated that beyond doubt.

The Welsh October lock down was highly effective in reducing cases.

Returning immediately to the same restrictions known to allow R>1 was highly effective in putting them straight back up again.

 

[matticus]

We can make blanket criticism of it

No, you have misunderstood my post. Maybe that is my fault, I do not know. Please see the 9 words that I highlighted in bold.

Ruby has correctly interpreted my post, and has answered quite adequately for me. I felt it was polite to point out your misunderstanding, which is the reason I'm ignoring your long - and no doubt well-constructed - reply.

 

[mjr]

Yes I think the success of the Welsh October lockdown demonstrated that beyond doubt.

That wasn't swift: it was over a month later than needed, so it broke the wave but didn't get cases down enough and, like England, Wales entered the third wave from a higher level.

But on the short/sharp aspect, it seems to have worked: it achieved a comparable reduction in about two weeks to what England's "lockdown lite" did in four.