Coronavirus - worldwide spread and control

Status
Not open for further replies.
Just a follow up to what I said. I don’t know much about modelling so I may be getting the wrong end of the stick. But the modelling seems quite simple to me and also in particular

“Our model rests on the assumption that only a very small proportion of the population is at risk of hospitalisable illness”. But this assumption hasn’t been proven?

Don’t you first need to make sure as assumption is true before making models and releasing a paper on it? Or is that not how epidemiology works?
 
Last edited:
I agree @Adam pwme we need serology testing to know.

But I also wish instead of the govt spending so much time on talking about antibody testing for the past two months whenever they’re asked a question about testing, maybe we could actually first start testing people who already are presenting with symptoms of Covid-19, who are unable to get tested, who are being turned back home from hospital and not tested, doctors and nurses who are getting exposed but not tested, and so can be passing it onto other vulnerable patients. Instead I constantly see a focus on testing the population to check for degree of herd immunity. Surely that is secondary and some things are far more important.

Today this paper has generated A LOT of interest on Twitter with people saying it will have policy implications and Gupta saying she disagrees with the revised Imperial paper. I’m sure it will interest our govt as well. I’m also worried as to how people who already might not have taken coronavirus seriously, may behave now, on the basis of a paper which uses unproven assumptions, (edited to add: especially when they are told less than 1 in a thousand infected people get hospitalised according to the model).

I’m worried about both social and policy implications: with the govt putting forward an even greater focus on antibody testing rather than PCR testing for the infected, and people thinking it’s not serious so they don’t need to social distance. The paper and Gupta recommend “urgent” need for serology antibody testing of the population now. The research group has released a tweet saying they did this to stimulate discussion about “immunity”. Honestly? I’m really getting fed up of hearing about immunity, herd or otherwise, while people are dying and ill in the UK without testing, and their contacts aren’t being traced either.

 
Last edited:
Last edited:
Roy S said:
I seem to have an ability to sleep through those. It must be a superpower.

Since Fauci was not at that one,


Click to expand...


Here is a summary of the conference yesterday: https://www.politico.com/news/2020/03/23/trump-coronavirus-lockdown-skepticism-143800

As I understand it now, policy would include federal aid, including building hospitals in "hotspots", and possibly a vague form of tracing, but no lockdown and 'restarting the economy' by easter. With regards to the tracing, something was only mumbled vaguely about it on post-code type basis. There were heavy allusions to herd immunity and Trump would be ok with that - here's a quote

He also repeated a similarly criticized comparison of coronavirus to the number of Americans killed in car crashes annually. Experts have called both attempts to draw parallels with coronavirus instances of false equivalency.

Then Trump asserted that economic downturns could be fatal in their own right.

“People get tremendous anxiety and depression, and you have suicides over things like this, when you have terrible economies. You have death,“ the president argued, adding - without evidence - that “probably ... definitely, it would be in far greater numbers than the numbers we“re talking about with regard to the virus.“
Click to expand...

It seems to be the economic cost of this will be far worse. We are far behind Italy with far lighter restrictions.
 
Last edited by a moderator:
lunarainbows said:
I think this is the model from Oxford which says over 50% of the population of UK are infected. Does anyone want to have a look at it?
Click to expand...

I had a brief look. I don't think this is consistent with what we know. The curves we have seen would be different if we are getting near to population saturation. My impression is that these modellers are completely incompetent. The understand maths but not how to apply it to reality. This is an extremely common problem with complex system dynamics in science.

In other words forget it. I doubt it will have a lasting effect on the politics. By next week we will know it was wrong.
 
lunarainbows said:
I agree @Adam pwme we need serology testing to know.

But I also wish instead of the govt spending so much time on talking about antibody testing for the past two months whenever they’re asked a question about testing, maybe we could actually first start testing people who already are presenting with symptoms of Covid-19, who are unable to get tested, who are being turned back home from hospital and not tested, doctors and nurses who are getting exposed but not tested, and so can be passing it onto other vulnerable patients. Instead I constantly see a focus on testing the population to check for degree of herd immunity. Surely that is secondary and some things are far more important.

Today this paper has generated A LOT of interest on Twitter with people saying it will have policy implications and Gupta saying she disagrees with the revised Imperial paper. I’m sure it will interest our govt as well. I’m also worried as to how people who already might not have taken coronavirus seriously, may behave now, on the basis of a paper which uses unproven assumptions, (edited to add: especially when they are told less than 1 in a thousand infected people get hospitalised according to the model).

I’m worried about both social and policy implications: with the govt putting forward an even greater focus on antibody testing rather than PCR testing for the infected, and people thinking it’s not serious so they don’t need to social distance. The paper and Gupta recommend “urgent” need for serology antibody testing of the population now. The research group has released a tweet saying they did this to stimulate discussion about “immunity”. Honestly? I’m really getting fed up of hearing about immunity, herd or otherwise, while people are dying and ill in the UK without testing, and their contacts aren’t being traced either.

Click to expand...


Rushed reply.

Good idea i.e. rather than the Government owning up i.e. South Korea has put in place mass testing (for virus - PCR) and we haven't. The Government are "re-directing/re-focusing" the discussion to a test which doesn't currently exist i.e. an antibody test that confirms exposure and recovery.

George Orwell's work was broadcast on the BBC (Radio 4 etc.) recently ---"battles which were reported and never happened" and "battles where hundreds died and were never reported".
 
lunarainbows said:
Today this paper has generated A LOT of interest on Twitter with people saying it will have policy implications and Gupta saying she disagrees with the revised Imperial paper. I’m sure it will interest our govt as well. I’m also worried as to how people who already might not have taken coronavirus seriously, may behave now, on the basis of a paper which uses unproven assumptions, (edited to add: especially when they are told less than 1 in a thousand infected people get hospitalised according to the model).
Click to expand...

The paper is here
https://www.dropbox.com/s/oxmu2rwsnhi9j9c/Draft-COVID-19-Model (13).pdf?dl=0
 
Adam pwme said:
Coronavirus may have infected half of UK population — Oxford study
Click to expand...
I agree it seems unlikely.

On 19/03 the UK only had about 100 confirmed deaths, similar to South-Korea now (as of 24/03 they report 120 deaths). In South-Korea they have tested 348,582 people of whom 9,037 tested positive. So that's 2,6%. I assume there are all sorts of ways I which this estimate will be biased (not everyone who has been infected will have a positive PCR test for example) but the difference with 50% of the population seems too large.

lunarainbows said:
“Our model rests on the assumption that only a very small proportion of the population is at risk of hospitalisable illness”. But this assumption hasn’t been proven?

Don’t you first need to make sure as assumption is true before making models and releasing a paper on it? Or is that not how epidemiology works?
Click to expand...
I also find it a bit strange. What exactly is this model trying to do? If you assume only a very small proportion of the population is at risk of hospitalisable illness, yes then you'll need less Intensive care units - obviously.

Predicting how many people have been in contact with the virus based on an assumption of how many people are at risk of being hospitalised seems like things upside down. Wouldn't one normally estimate the infection and fatality rate first based on the available data and then use a model of how this might impact things in the future under different scenario's?

Trying to read and understand these papers hasn't exactly made me a fan of 'modelling studies'...
 
Last edited:
lunarainbows said:
Just a follow up to what I said. I don’t know much about modelling so I may be getting the wrong end of the stick. But the modelling seems quite simple to me and also in particular

“Our model rests on the assumption that only a very small proportion of the population is at risk of hospitalisable illness”. But this assumption hasn’t been proven?

Don’t you first need to make sure as assumption is true before making models and releasing a paper on it? Or is that not how epidemiology works?
Click to expand...

Really you should think of models as ways to animate a set of assumptions and see how they play out. The predictive power is often really poor and you need to look at the sensitivity of the model to the assumptions. The more sensitive to the assumptions the worse the predictive power.

But sometimes people doing modelling forget this.
 
Adrian said:
There is another modelling paper that was released yesterday in the lancet looking at interventions in Singapore. I don't think its saying anything different from other modelling papers.

One thing of interest though I'm not sure if it is in this paper is that from what I gather Singapore is not in lockdown but seems to have got some form of control on things,

https://www.thelancet.com/journals/...avirus20&utm_source=twitter&utm_medium=social
Click to expand...
You're correct, Singapore is not on lockdown. My best friend lives there.
 
Michiel Tack said:
I agree it seems unlikely.

On 19/03 the UK only had about 100 confirmed deaths, similar to South-Korea now (as of 24/03 they report 120 deaths). In South-Korea they have tested 348,582 people of whom 9,037 tested positive. So that's 2,6%. I assume there are all sorts of ways I which this estimate will be biased (not everyone who has been infected will have a positive PCR test for example) but the difference with 50% of the population seems too large.


I also find it a bit strange. What exactly is this model trying to do? If you assume only a very small proportion of the population is at risk of hospitalisable illness, yes then you'll need less Intensive care units - obviously.

Predicting how many people have been in contact with the virus based on an assumption of how many people are at risk of being hospitalised seems like things upside down. Wouldn't one normally estimate the infection and fatality rate first based on the available data and then use a model of how this might impact things in the future under different scenario's?

Trying to read and understand these papers hasn't exactly made me a fan of 'modelling studies'...
Click to expand...

I'll be much less insightful than @Adrian and talk longer to say it!

Basically I agree with what you are saying @Michiel Tack first what is the purpose of your model/predictions? In this case to examine strategies and outcomes e.g. South Korea tested, tested, tested (i.e. to identify people who were ill and infectious) and then traced their contacts and the outcome was --- (reduce R naught, don't overwhelm your ventilator capacity, lower % deaths) best we've seen.

Italy appears to have allowed the virus to become established (the South Korean's didn't) and it had an older population. It's healthcare system has been overwhelmed and the % of fatalities exceed those in South Korea (since there are insufficient ventilators).

So basically you input the variables (rate of transmission R naught) and the limiting resource (ventilators) and you examine various strategies. As for not using the best available data; that is mind boggling! As for publishing a study which does not use the best available data ---.

I agree about the combination of poor modelling and a poorly informed audience (us) --- not good (unless you are content with the outcome--- defend the indefensible!).
 
I wish they would make it clear to people that even if they 'feel fine' that they might still be infected and subsequently passing the virus onto other people by not taking the appropriate measures.

When I think of the AIDS tv public health announcements back in the 80's that scared the hell out of everyone...........this on tv last night from the CMO just didn't do it

 
Status
Not open for further replies.
Back
Top Bottom