Coronavirus 2019-nCoV

[Kung Wu]

Proximity to the outbreaks in Italy probably ought to be taken into account, which is why comparing Sweden to other Scandinavian countries might make more sense.

Norway and Denmark are half the population, I think. Maybe I should combine those two and see what happens?

[Kung Wu]

Proximity to the outbreaks in Italy probably ought to be taken into account, which is why comparing Sweden to other Scandinavian countries might make more sense.

Denmark + Norway:
Combined population 11.1 million people.
First 10 confirmed cases 03/03 (Denmark), 02/29 (Norway) -- use 03/01?
Lockdown on March 13 (Denmark), March 12 (Norway)
Total deaths as of 04/21 = 370 + 182 = 552.
0.005% of their country's population.

[Kung Wu]

Netherlands:
Population 17 million people.
First 10 confirmed cases 03/02.
Lockdown on March 12.
Total deaths as of 04/21 = 3,916.
0.023% of their country's population.

[SHOCKvalue]

Netherlands:
Population 17 million people.
First 10 confirmed cases 03/02.
Lockdown on March 12.
Total deaths as of 04/21 = 3,916.
0.023% of their country's population.

Stop analyzing the Coronavirus statistics on a per capita basis! How dare you! Fear boner-ing has no room for this sort of context.

[wufan]

To be fair and consistent, probable flu deaths are also counted in flu death counts in any given year as well. So this is not out of the ordinary and ME's make probable cause of death assumptions for a lot of people.

Absolutely! A couple of problems with moving from counting confirmed to counting probable for Covid:

1. Early counts did not include this. There will be an artificial inflation in the death rate that needs to be addressed in the numbers. Basically, the rate of change just shifted.

2. Flu deaths AND infections are estimated. In the current circumstance, only the numerator is being inflated which will greatly increase the already flawed mortality rate.

ShockingButTrue posted the flu deaths and the confirmed cases earlier in this thread. Using JUST those two numbers gives the flu a 10% mortality rate. It was flawed when he posted it for the flu, and it will be flawed if anyone uses it for Covid.

[wsushox1]

Some interesting news tonight. Santa Clara County ME is going back, with permission, and looking at suspicious deaths (pneumonia) in Santa Clara County. The ME found Coronavirus RNA in a patient death from February 6th. 23 days before the first “official” COVID-19 death in WA.

I don’t know if that is particularly illuminating, or surprising, and if it changes the narrative too much. Most credible epidemiologists have already stated an empirically supported introduction in mid-late January - which this death would further support.

source: https://www.sfchronicle.com/health/a...n-15217316.php

[wsushox1]

Lastly, for those that have the time or care, a critique of the Santa Clara County and LA County seroprevalence tests:

[wufan]

Both of the above posts are interesting. The work being done is being put out there much more quickly than under any normal circumstance.

[pie n eye]

Both of the above posts are interesting. The work being done is being put out there much more quickly than under any normal circumstance.

At least by my internal logic, which is neither peer reviewed nor remotely credible, given what we think we know about how quickly the virus spreads combined with the fact that we’re finding earlier cases of Covid would seem to support the multiple studies that say we’re underestimating the number of people who have already been infected.

I don’t know enough (nothing) about statistics to even begin to verify their work or to even understand the tweets from the guy who says he found an error in their calculations. However, I read through the thread and didn’t see an explanation of how the error found would meaningfully change the results or conclusion. Maybe it was obvious to statisticians in the audience.

But, the guy out of Stanford did say in his interview on Uncommon Knowledge that while everything in the study might not be perfect, we need information and it is good enough to be confident that the disease is more widespread that the numbers say. Exactly how much more widespread we will continue to discover as more studies occur, the tests improve, and others can review the existing studies.

[pie n eye]

Did more digging on the twitter thread to try and understand the implication of the errors.

Correct me if I’m wrong but they’re basically saying that we don’t know how accurate the tests are and in turn question the accuracy of the studies conclusion that more people have been infected than actually reported by a magnitude of 40-80 (appx, don’t remember exact number).

Furthermore, the study should more heavily weight the possibility that all the positives in the study were actually false positives which would mean that the low end should actually be 0, meaning that we already know how many people have been infected and there are no unreported cases where people had it with either mild or no symptoms?

So, again from my layman’s understanding, don’t take these 40-80 magnitude numbers as gospel. The true answer may be found within that range but, given their review of the study, the number might be lower.

[wsushox1]

Did more digging on the twitter thread to try and understand the implication of the errors.

Correct me if I’m wrong but they’re basically saying that we don’t know how accurate the tests are and in turn question the accuracy of the studies conclusion that more people have been infected than actually reported by a magnitude of 40-80 (appx, don’t remember exact number).

Furthermore, the study should more heavily weight the possibility that all the positives in the study were actually false positives which would mean that the low end should actually be 0, meaning that we already know how many people have been infected and there are no unreported cases where people had it with either mild or no symptoms?

So, again from my layman’s understanding, don’t take these 40-80 magnitude numbers as gospel. The true answer may be found within that range but, given their review of the study, the number might be lower.

What the thread is stating, is that they calculated their confidence interval (an interval where you can be 95% sure the mean falls in) incorrectly. If they had calculated it correctly, it is very possible the confidence interval would extend down close to 0%.

That does not mean that there are not unreported cases, because there most certainly are, but what it does do is extend the confidence interval into a wider range. This is not particularly good. I am 95% confident in saying that the true rate of covid-19 infections is between 0 and 30% of a given population. But the difference between 0 and 30 is a lot. In this case, the difference between 0 and 6% on the upper bound is quite a bit.

We are no doubt undercounting infections, but you can not extrapolate the information in that study widely. For example, I saw some well known people on Twitter doing that and coming up with an Infection Fatality Rate (different than Case Fatality Rate), to get to an IFR of less than .1%. More than .1% of NYC has already died due to COVID-19 - with more dying each day - so we know the IFR is likely to be higher than that. Just one flaw in extrapolating one study in a certain population to any given population.

[Shockm]

More testing is being done in Ford County is why it is a hot spot right now. Food packing and nursing homes are the hot spots at the current time according to this article. It isn't necessarily an increase in cases but an increase in testing.


"One of the hot spots in the state right now is Ford County in Southwest Kansas. The Ford County Health Department Monday received notification of 42 new cases. They say the reason they are seeing such an increase is that the health department investigation staff is being aggressive in testing of close contacts, even those that are asymptomatic. There is not necessarily an increase in cases in Ford County, but rather they are simply testing more people and actively looking for any evidence of community spread. They are identifying as many current cases as they possibly can as opposed to “passive surveillance.” Ford County now has the fourth most cases of the virus in the state at 180.

Coronavirus clusters are being reported at three meat processing plants in Kansas. The Kansas Department of Health and Environment reports that two meat processing plants in Ford County and one plant in Lyon County have been confirmed as cluster zones. State officials say the plants are enhancing their cleaning efforts and making changes to staffing and scheduling as a precaution. Thirty-nine COVID-19 clusters have been reported in Kansas, including 15 at private companies.

Dozens of new coronavirus cases are linked to a Sedgwick County nursing home. Officials have confirmed 29 additional COVID-19 cases at the Clearwater Nursing and Rehabilitation Center. There have been 42 cases, including two deaths, reported at the Clearwater facility. All nursing home residents have been tested for COVID-19, and officials are still waiting for some of the test results."

[ShockerFever]

What the thread is stating, is that they calculated their confidence interval (an interval where you can be 95% sure the mean falls in) incorrectly. If they had calculated it correctly, it is very possible the confidence interval would extend down close to 0%.

That does not mean that there are not unreported cases, because there most certainly are, but what it does do is extend the confidence interval into a wider range. This is not particularly good. I am 95% confident in saying that the true rate of covid-19 infections is between 0 and 30% of a given population. But the difference between 0 and 30 is a lot. In this case, the difference between 0 and 6% on the upper bound is quite a bit.

We are no doubt undercounting infections, but you can not extrapolate the information in that study widely. For example, I saw some well known people on Twitter doing that and coming up with an Infection Fatality Rate (different than Case Fatality Rate), to get to an IFR of less than .1%. More than .1% of NYC has already died due to COVID-19 - with more dying each day - so we know the IFR is likely to be higher than that. Just one flaw in extrapolating one study in a certain population to any given population.

I'm not sure I trust a lot of the data coming out of NYC.

[ShockerFever]

So now all of the new talk is centered around the "Second Wave" coming in the fall and winter. So with that said, shouldn't we be seeing a spike in cases and deaths in southern hemisphere countries right now or is only the United States and northern hemisphere nations prone to it?

[wsushox1]

I'm not sure I trust a lot of the data coming out of NYC.

Why?