IGX

Sangoma wrote: ↑Sun Aug 09, 2020 11:34 pm

Bennyonesix1 wrote: ↑Sat Aug 08, 2020 6:01 pm PCR tests by themselves are useless. Even more so with the sig delay in reporting.

They should prob take a PCR and antibody test at the same time.

Any test taken out of context from pre-test probability is of very limited value. I did explain it earlier in this or another thread, when the issiue of testing came up. It wasn't met with enthusiasm.

Sorry.

Fairly detailed analysis of mask material re: droplets.

https://advances.sciencemag.org/content ... d3083.full

Testing speed an accuracy are going to be an issue in cold weather even if schools are open. My 5 year old has the sniffles and some fatigue consistent with a couple hours in a pool over the weekend and a five mile run tag-a-long on his bike. Unfortunately that's also consistent with COVID in kids, so it keeps him and my youngest out of daycare until a doc says it's not COVID. That will take approximately five full days.

Come winter, most kids have sniffles without fever. If each instance requires a week out of school, it'll be incredibly disruptive for working parents.

Just repeating my post early in this thread, for those remotely interested in testing math.

Imagine there is a disease XYZ for which there is a test. This test is reliable, and only 5% of the results are false positive. Say, you send the parent for testing, and the test is positive. The patient is asking: "Doc, so what's the probability that I have the disease?" Think about it before reading the answer in the next post.

The standard answer among my registrars is almost invariably 95%. Sure, 5% are positive then the rest should be true positives, right?

Wrong. Testing cannot be taken out of context with the incidence of the disease. Bear with me, the next paragraph requires a bit of concentration.

Let's say the incidence of the disease XYZ, for which the patient was tested, is 1:1000. let's say we initiate a screening campaign and test the whole town. How many patients in 1000 tested will come back positive?.... Fifty (5% percent false positive, out of 1000). How many will have the disease? One. So the answer to this guy's question - the probability of him having the disease - is 1:50, or 2%.

More detailed explanation of pre- and post-test probabilities on Wiki: https://en.wikipedia.org/wiki/Pre-_and_ ... robability

Sangoma, you give a great argument for why we need to test a lot more, in order to know what the distribution is.

Let’s say we are doing thousands of random tests per day, and currently have a positive test rate of 20% with your test. What is the probability that someone who tests positive is positive?

Depends on the prevalence. Let's say 1% of US population is currently infected and the test gives out 5% false positives. It means 10 out of 1000 will carry the disease, and 50 will be falsely positive. The odds are 10:50 or 1:5. Answering your question bluntly it is possible that out of 20% positives 4% will be truly positive.

It is not that straightforward, because prevalence differs between groups. Among persons with four suggestive symptoms - for example fever, malaise, shortness of breath and cough - prevalence will be much higher, and therefore positive tests will have much higher predictive value. Add to these ground glass opacity on chest CT, and the test becomes a formality.

Sangoma wrote: ↑Tue Aug 11, 2020 12:43 pm Depends on the prevalence.

So how do you determine the prevalence?

You test large samples and then (for example) use a maximum likelihood estimate to come up with the prevalence.

You need to do broad randomized testing to estimate the prevalence, is the key point. As you note, just testing sick people doesn't give you the data you need.

nafod wrote: ↑Tue Aug 11, 2020 1:19 pm

Sangoma wrote: ↑Tue Aug 11, 2020 12:43 pm Depends on the prevalence.

So how do you determine the prevalence?

You test large samples and then (for example) use a maximum likelihood estimate to come up with the prevalence.

You need to do broad randomized testing to estimate the prevalence, is the key point. As you note, just testing sick people doesn't give you the data you need.

Jesus dude this is why we don't do a lot of flu testing. The flu stats are based off diags and extrapolations. And no doc uses the flu test to diag the flu.

It's basic stuff that everyone and especially the experts knew before March. Then they stopped knowing it apparently.

This whole time people who know better have been using the wrong tool for the job. And basing policy on useless results. And they refuse to change.

One has to ask why?

Seriously, why were generations worth of knowledge and best practices thrown out the window?

If one watches this like we watch other respiratory viruses the pattern is normal and expected: CLI% in ER then a week later Hosps rise then a week after that deaths increase.

nafod wrote: ↑Tue Aug 11, 2020 1:19 pm

Sangoma wrote: ↑Tue Aug 11, 2020 12:43 pm Depends on the prevalence.

So how do you determine the prevalence?

You test large samples and then (for example) use a maximum likelihood estimate to come up with the prevalence.

You need to do broad randomized testing to estimate the prevalence, is the key point. As you note, just testing sick people doesn't give you the data you need.

In Australia as soon as you turn 50 they send you the test for occult fecal blood, as part of a colon cancer screening. This test is sensitive, but not specific, so the majority are false positive. Obviously, you cannot draw conclusions about the prevalence of colon cancer based on this test.

What I am getting at is that using indiscriminate testing for covid is not going to give you meaningful estimates of prevalence.

How to solve this dilemma I am not sure. One is to trace contacts of symptomatic patients and use mathematical models based on the data of spread that's available.

Sangoma wrote: ↑Wed Aug 12, 2020 1:29 am

nafod wrote: ↑Tue Aug 11, 2020 1:19 pm

Sangoma wrote: ↑Tue Aug 11, 2020 12:43 pm Depends on the prevalence.

So how do you determine the prevalence?

You test large samples and then (for example) use a maximum likelihood estimate to come up with the prevalence.

You need to do broad randomized testing to estimate the prevalence, is the key point. As you note, just testing sick people doesn't give you the data you need.

In Australia as soon as you turn 50 they send you the test for occult fecal blood, as part of a colon cancer screening. This test is sensitive, but not specific, so the majority are false positive. Obviously, you cannot draw conclusions about the prevalence of colon cancer based on this test.

What I am getting at is that using indiscriminate testing for covid is not going to give you meaningful estimates of prevalence.

How to solve this dilemma I am not sure. One is to trace contacts of symptomatic patients and use mathematical models based on the data of spread that's available.

I think that violates US privacy laws. I know they were doing something like that in SK but it's not viable here.

Law works in strange ways. After 9/11 - the event that killed 3000 people - they pushed the Patriot Act without much consideration for individual freedoms. According to Snowden everybody is watched as much as needed. Yet with 150,000 plus deaths attributable to Covid privacy is problem.

In Australia they tried to create an app that helps trace contacts - mobile data is recorded and stored forever anyway. It has been a failure. I am sure Silicon Valley minds could come up with something like this. In terms of privacy they could make a law that the data is used only for medical purposes. In Australia if a patient dies under anaesthesia the anaesthetist sends the report to the dedicated Committee. After they review the case and come up with the recommendations the paperwork is destroyed. They started doing it after they refused to release the papers to the police for investigation. I am sure something like this could be done in relation to peoples movement for tracing purposes.

I heard about a very simple and cheap test yesterday. I believe it's referred to as a "spit test". It seems like it works similarly to a pregnancy test. Deposit some saliva on a strip of some kind and you get an indicator that says you have COVID to the point where you're infectious or not.

There are apparently CDC reporting requirements that complicate things but imagine a world where we could test ourselves every time we had the sniffles or wanted to be in a group. Seems like it would help open the world a lot without a excess complexity. I hope it's true.

Sangoma wrote: ↑Wed Aug 12, 2020 1:29 am What I am getting at is that using indiscriminate testing for covid is not going to give you meaningful estimates of prevalence.

Randomized broad testing is exactly how you get estimates of prevalence. You can account for the sensitivity and specificity of the test in the estimate too.

Google on "estimate prevalence with imperfect test". Tons of info.

nafod wrote: ↑Wed Aug 12, 2020 1:00 pm

Sangoma wrote: ↑Wed Aug 12, 2020 1:29 am What I am getting at is that using indiscriminate testing for covid is not going to give you meaningful estimates of prevalence.

Randomized broad testing is exactly how you get estimates of prevalence. You can account for the sensitivity and specificity of the test in the estimate too.

Google on "estimate prevalence with imperfect test". Tons of info.

Setting that up in a way that would yield meaningful data would be difficult in practice.

My people (the krauts) break down PCR testing and why it is the wrong tool for the job. It's really good. The first 2min are intro etc. 15min of info. It's excellent.

I don't think we understand what makes a pandemic turn into an endemic illness. But we just saw it happen.

I came to a realization this morning.

We're not getting through this until Big Pharma gets paid for making a vaccine.

Their regulatory capture of the US and world health system means they can make this happen.

I wish they'd just take money to go away like the mafia and leave us alone but for some reason they've decided on a vaccine.

The smartest thing for politicians to do is figure out a way to get the money to them while limiting the amount of vaccinations and also putting in stops to prevent their doing this again.

The secondary bacterial infection as the thing killing ppl re covid is looking very likely. It's excellent news wrt to future treatment.

https://www.nature.com/articles/s42256-020-0206-1

https://www.nature.com/articles/s42256-020-0207-0

More possible good news. Tcell immunity could be measured without massive cost. We'd know if the 80% number is correct and who doesn't need a vaccine.



"sandeep chakraborty
@sanchak74
This is how one measures Tcells response - different from Bcells antibodies (which varies) - by measuring (a constant) protein IFN-gamma

Pre-existing Tcells exposed to Cov2 (or Cov2-like) will secrete IFN-gamma when it meets a viral antigen.

Should be reasonably easy and cheap"

https://www.technologynetworks.com/diag ... ssion=true

It'd be nice if catching a specific cold would infer some Covid immunity. If they could identify the right virus...

Bennyonesix1 wrote: ↑Thu Aug 13, 2020 3:19 pm I came to a realization this morning.

We're not getting through this until Big Pharma gets paid for making a vaccine.

Their regulatory capture of the US and world health system means they can make this happen.

I wish they'd just take money to go away like the mafia and leave us alone but for some reason they've decided on a vaccine.

The smartest thing for politicians to do is figure out a way to get the money to them while limiting the amount of vaccinations and also putting in stops to prevent their doing this again.

Big Pharma is probably right on this one. There’s a lot of risk here (many companies will try, few will succeed), the margins on vaccines are low, It won’t be a long term profit generator, and governments tend not to keep their word to companies when it comes to public health.

nafod wrote: ↑Thu Aug 13, 2020 4:16 pm It'd be nice if catching a specific cold would infer some Covid immunity. If they could identify the right virus...

No dude. It looks like there's a wide spectrum of colds and flus that convey tcell level "immunity". Estimates centering around 80% of us already have it.

Think about it. The population is 80% immune and the vulnerable are not healthy enough to take the vaccine. How do you reduce Herd Immunity via vaccine? Massive repeated rounds of vax on ppl who don't need it. Gates knew this shit long ago. There's no point to it other than $.

Turdacious wrote: ↑Thu Aug 13, 2020 4:17 pm

Bennyonesix1 wrote: ↑Thu Aug 13, 2020 3:19 pm I came to a realization this morning.

We're not getting through this until Big Pharma gets paid for making a vaccine.

Their regulatory capture of the US and world health system means they can make this happen.

I wish they'd just take money to go away like the mafia and leave us alone but for some reason they've decided on a vaccine.

The smartest thing for politicians to do is figure out a way to get the money to them while limiting the amount of vaccinations and also putting in stops to prevent their doing this again.

Big Pharma is probably right on this one. There’s a lot of risk here (many companies will try, few will succeed), the margins on vaccines are low, It won’t be a long term profit generator, and governments tend not to keep their word to companies when it comes to public health.

https://www.spiegel.de/international/wo ... ssion=true

Bennyonesix1 wrote: ↑Thu Aug 13, 2020 4:32 pm

Turdacious wrote: ↑Thu Aug 13, 2020 4:17 pm

Bennyonesix1 wrote: ↑Thu Aug 13, 2020 3:19 pm I came to a realization this morning.

We're not getting through this until Big Pharma gets paid for making a vaccine.

Their regulatory capture of the US and world health system means they can make this happen.

I wish they'd just take money to go away like the mafia and leave us alone but for some reason they've decided on a vaccine.

The smartest thing for politicians to do is figure out a way to get the money to them while limiting the amount of vaccinations and also putting in stops to prevent their doing this again.

Big Pharma is probably right on this one. There’s a lot of risk here (many companies will try, few will succeed), the margins on vaccines are low, It won’t be a long term profit generator, and governments tend not to keep their word to companies when it comes to public health.

https://www.spiegel.de/international/wo ... ssion=true

That kind of makes my point.