Discussion of the Sweden model

[variant]

Agreed, we were discussing some of those factors in the international comparison thread, and in the threads comparing the Australian response to the US.

It is also why it varies so much by state and county in the USA which has many different areas with differences in populations.

Also, it makes sense that different areas will have different plans or responses as the threat is fundamentally different in Montana vs New Jersey.

Sweden and Spain have different challenges.

 

[tall73]

Placing this news release in this thread to update, as it reflects on the earlier news about the antibody study.

https://www.thelocal.se/20200520/heres-what-swedens-first-coronavirus-antibody-tests-tell-us

Stockholm had the highest proportion of positive results, which was as expected given that the region has had the highest number of confirmed cases and deaths due to coronavirus. In Skåne, 4.2 percent of samples were positive, compared to 3.7 percent in Västra Götaland.

Although these samples were taken at the end of April, the Public Health Agency said: "The numbers reflect the state of the epidemic earlier in April, as it takes a few weeks for the body's immune system to develop antibodies."

"We aren't at seven percent [infection rate in Stockholm] now. It was seven percent around week 15, so that is quite a long time ago. These people were immune in week 18 [the week ending May 3rd], that means they fell ill at some point in week 14 or 15. We are somewhere around 20 percent plus in Stockholm now," Tegnell told journalists at the press conference.

 

[Occams Barber]

You’re referring to the IHME model. It was never a model. They thought we didn’t have enough information to model, so they did curve fitting. They used a normal curve, which goes up then down the same, though with some modifications based on Chinese data. Watching my state of NJ, they kept predicting that deaths would start going down rapidly starting the day after the model was done. In fact it didn’t work that way.

It was about the best that could be done. They didn’t predict just one number, but a range. The figures we’re seeing now are within the range. People seem to have quoted only the center of the range, without warning that there was a large degree of uncertainty.

The projections also end Aug 1, so the numbers are not totals, but only projections for the first phase.

Thanks @hedrick . I wasn't aware of the specific methodology used to predict the numbers.

OB

 

[hedrick]

Thanks @hedrick . I wasn't aware of the specific methodology used to predict the numbers.

OB

They've recently started doing some real modeling, though it's not clear whether it's used to predict deaths or just to estimate the number of people actually infected (as opposed to officially diagnosed).

There were lots of criticisms of their work by people who wanted more traditional models, but I'm not aware of anyone else who publishes the detailed state by state projections that they do on an updated basis.

 

[Occams Barber]

They've recently started doing some real modeling, though it's not clear whether it's used to predict deaths or just to estimate the number of people actually infected (as opposed to officially diagnosed).

There were lots of criticisms of their work by people who wanted more traditional models, but I'm not aware of anyone else who publishes the detailed state by state projections that they do on an updated basis.

I've found the IHME website.
IHME | COVID-19 Projections

The link takes you to some interesting projections including the one I've pasted in below. This graph is for projected US Covid 19 deaths up to 4 August. It predicts 143,357 US deaths within a range of roughly 115,000 to 207,000.

(note that you can change the various graphs and data sets to show a limited range of other countries or individual US states)

OB

 

[Radagast]

You’re referring to the IHME model. It was never a model. They thought we didn’t have enough information to model, so they did curve fitting. They used a normal curve

I'm pretty sure they didn't use a normal curve. Why would one use a normal curve?

It was about the best that could be done. They didn’t predict just one number, but a range. The figures we’re seeing now are within the range. People seem to have quoted only the center of the range, without warning that there was a large degree of uncertainty.

True.

 

[Occams Barber]

We should have done more, admits architect of Sweden's Covid-19 strategy

In a Guardian article published 3 June Anders Tegnell, the architect of Sweden's disastrous Corona virus response, has finally admitted that Sweden should have done more -

'Anders Tegnell, who has previously criticised other countries’ strict lockdowns as not sustainable in the long run, told Swedish Radio on Wednesday that there was “quite obviously a potential for improvement in what we have done” in Sweden.'
​

This understated admission has come in a week where Sweden's Covid 19 death rate, based on a 7 day rolling average to 2 June, was the highest in the world.

Now under political pressure, it appears likely that the Swedish government will be forced to set up a commission to investigate its reaction to Covid 19.

OB