Yes. In Denmark, I see a lot of, even in mainstream media, handling the number of tested positive as the same as number of infected.
Two days ago, Denmark enforced drastic measures and at the same time changed the testing strategy. Before, patients were tested on suspicion in order to try to contain. Now, only those with severe symptoms are tested. I would have expected the number of daily new positives to go down - and it did. But not by much. That is really alarming.
I think the phone app is optional.
JD
3136
So yeah, Premiere League shutting down for now.
jsnell
3137
There’s just no way that’s true. Iran is estimated to have roughly 2 million infected (including via methods that are robust to the “maybe there are a lot more mild cases than we think, which never get reported” theory). They’re digging mass graves.
draxen
3138
I have the exact opposite opinion (as per usual) :)
I think your concerns are valid though and this post is just a dump of my thinking rather than any kind of rebuttal to you…
The more I think about the UK strategy the more I feel confident it might be a good one. They’ve obviously been planning for exactly this scenario for a long time. The document I linked to earlier in the thread is from 2011 and it sets out many of the same key points of the strategy.
(I’ll link again in case anyone is interested)
https://www.gov.uk/government/publications/responding-to-a-uk-flu-pandemic
What will be interesting (scary!) is whether NHS capacity will be able to meet demand. Can they mobilize enough resources to keep the mortality rate low. If they’re overwhelmed the mortality rate skyrockets (ala Italy). This could be the greatest test the NHS has ever faced. What’s somewhat comforting is that by all accounts the NHS can, in times of emergency free up (and ramp up) considerable amounts of resources.
Much focus is given to testing but there are many other data streams for monitoring the current infection rate. e.g. the 111 service. They said in the press conference that they’re changing their test strategy:
(Paraphrasing) It is no longer needed for us to test every case. We will pivot our testing capacity to identifying people in hospitals who have got symptoms so we can pick them up early, treat them well and make sure they don’t pass on the virus to other people in hospital. Testing will be based on symptoms and severity.
Aceris
3139
Sure. But the entire strategy is predicated on keeping very deliberate control of R. The only way to measure that without being swamped in confounding factors is to test random samples. 1000 people a day. The point is not to detect the virus for treatment and containment - people will probably become symptomatic and self-isolate before they get the test results anyway. The point is to have actual solid data on the current infection rate, to enable the level of social distancing policy to be gradually increased. Otherwise you’re relying on a proxy measure, probably serious cases, which is going to lag by about 10 days or more, which is how Italy got to where they are.
Given their current estimate of the infected population (10000) it’s probably a bit early for that to be really useful, but getting the procedure in place now is going to pay dividends. Also you might get a nasty surprise and find out the infected number is way higher than you thought, in which case the NHS is going to be swamped in 5-10 days time and you need to go to full containment NOW to slam the brakes on the disease.
Also, having an accurate estimate of the current number of infected would be very helpful in persuading people that they know what they are doing.
draxen
3140
Yes, I see what you’re saying. Good point.
jsnell
3141
But just how do you think they’ll mobilize enough resources to handle 15-20M concurrent cases, of which 20% require hospitilization, in a couple of months?
The document I linked to earlier in the thread is from 2011 and it sets out many of the same key points of the strategy. (I’ll link again in case anyone is interested)
Thanks! I’d missed your earlier link. That document seems to be planning for 1-4% of symptomatic patients requiring hospital care, and a death rate of 2.5% for those with symptoms and no treatment. Do those parameters apply to Covid-19?
Oh, that plan is also predicting a vaccine in 4-6 months from now. That seems rather optimistic compared to any other estimates I’ve seen.
The modeling for the effect of travel restrictions is interesting, and I guess it makes sense given the framework of assuming nothing can be contained, and the only thing you’re doing is moving the peak.
Aceris
3142
By not planning to have that many concurrent cases and not believing that 20% of cases require hospitalization.
Hm. Does it have to be random?
You’re trying to compute the probability of a person being infected. Call that p(A).
But you are really only viewing p(A | they need medical care), if you’re only testing people needing medical care.
As long as p(person needs medical care | A) is known, it seems computable even from a non-random sample (or rather, a random sample with a known selection bias).
Aceris
3144
I dont think we do know p(person needs medical care | A).
Also people turning up to hospitals is a lagging indicator of infection. By the time that number is too high all of those people have also infected people.
Well, right now, random sampling would need to involve an extreme number of people to find anyone from the .01% of the population who have it, so it’s not clear it’s even useful yet.
It might also be possible to cancel out the unknown value if you’re just trying to estimate the spread coefficient.
jsnell
3146
But that’s just the thing: the plan appears to be for them to have that many concurrent cases. It’s the only way to make the math work out based on what they’ve said.
I didn’t think the ratio of cases requiring hospital care was up for debate at this point, at least at the order of magnitude level. Sure, maybe it’s 15%.
Aceris
3147
Sure, but if the plan is to build herd immunity then that number needs to go way way up.
I’ve been running the numbers and protecting the vulnerable is so, so important if you want to build herd immunity. I’m surprised and a little worried that we haven’t heard more about that. What I would want to see in the near future (early next week( is:
- Some policy to protect the vulnerable (with real financial support behind it).
- Some plan to verify their estimates of hospitalisation rates, mortality rates transmission rates, policy effects by measuring the true infection rate.
BAck of the envelope you need to get that need for hospitalisation down to 1% to avoid being overwhelmed as you build herd immunity, which means over65s are basically going to have to self isolate for months.
I do not understand the math between the herd immunity plan unless this really is “take it on the chin” or they think there is a very high rate of undetected asymptomatic and mild cases. That’s why I’d like to see clearer evidence around that.
Yes, I just meant we’re not there yet. We will be. Maybe by the time we can do random population testing we’ll have a better handle on the cases needing hospitalisation anyway.
This thread has some good modelling info on the basic stuff.
And I don’t disagree with you; hopefully all that is in the soon-to-come next phase.
Agreed. I can only assume this is what they believe. Fingers crossed.
“These aluminum tubes could only be used for centrifuges.”
There is always an expert ready to give you the answer you want. The question is, what do the other experts say?
Aceris
3152
That thread is really useful. It’s very clear that the optimum strategy (assuming you are building herd immunity) is to build up the infected number rapidly and then implement extreme social distancing, which you then gradually taper as the immune population grows. Effectively taking the flatter peak, making the approach to the peak steeper, and making the peak itself wider.
Yes. The alternative (effectively containment) would be to impose lockdown for all time - I don’t think you ever can end that strategy without ‘taking it on the chin’ at some later date. (Worth doing a bit of this to push to summer months where health system less burdened).
Caveats on the simple modelling assumptions etc.
