Coronavirus

[EVNow]

Open seating office is the worst for Covid containment.

Early Release - Coronavirus Disease Outbreak in Call Center, South Korea : COVID19

 

[SageBrush]

I don't know for sure, but in the video by Cuomo presenting the test results, which was posted here, he said they are not included with the numbers which are coming from nursery homes. If that is different for NYC numbers on that website, for example, I wouldn't know. But that is the best info I have at this point.

My understanding is that Cuomo includes out of hospital deaths in his daily reports because he relies on NYDOH numbers. NYCDOH is a laggard because they do their own due diligence of the death certificates.

Going back to the question of IFR, once you know the date of the serology study you can take the cumulative deaths 7 - 10 days later. I think 10 days is appropriate but there is room for reasonable disagreement.

 

[Norbert]

My understanding is that Cuomo includes out of hospital deaths in his daily reports because he relies on NYDOH numbers. NYCDOH is a laggard because they do their own due diligence of the death certificates.

Both nursery homes and at-your-own-home deaths are out of hospital. In that video, he was saying that the at-your-own-home numbers are generally not available as data. That's a different distinction, so it would require more info about NYCDOH to resolve that question.

Going back to the question of IFR, once you know the date of the serology study you can take the cumulative deaths 7 - 10 days later. I think 10 days is appropriate but there is room for reasonable disagreement.

I don't know how you arrived at that number, but if it is based on medians rather than averages (as your previous sources), then I would tend to disagree. As you may know, I have generally used 14 days for the time from test to death, but at the very end of a curve, an abrupt end, that might lead to incorrect estimates for the future. We don't know yet if in NYC we are close to a very abrupt end already. (The same is true for the very beginning of a curve.)

 

[Daniel in SD]

I thought these 400 samples are used to establish the specificity of the test with a certain confidence interval.
And that the specificity of the test is then used to estimate the confidence interval of an actual study using that test (a study with maybe 1,000 samples as an example).
Is that not the case?

No that's right. Except that some studies seem to be ignoring the confidence interval of the specificity.
The specificity itself doesn't affect the confidence interval. If the specificity were exactly 99% (95%CI 99%-99%) then you would know exactly how many false positives there were in your study and subtract them. You would also have to know sensitivity exactly too. The tricky math is incorporating the uncertainty in the specificity and sensitivity.

 

[Norbert]

Open seating office is the worst for Covid containment.

Early Release - Coronavirus Disease Outbreak in Call Center, South Korea : COVID19

What I notice is that in the upper room, cases (if that are the blue marks) are distributed quite evenly. To me, that would suggest that some people are more easily infected than others.

 

[HG Wells]

What I notice is that in the upper room, cases (if that are the blue marks) are distributed quite evenly. To me, that would suggest that some people are more easily infected than others.

Or, the source case sits near the AC return duct.

 

[Daniel in SD]

Yes, that's what they are all doing. Hopefully the reviewers point this out and make them change. Definitely Santa Clara study will undergo revisions since it has been commented upon by some academic heavyweights (apart from stats profs).

The Santa Clara study should be dumped completely.
Ray Bhattacharya’s wife recruited people for the study on a school listserv (Ray Bhattacharya claims he did not know). She told people the test was FDA approved (it is not) and that it would provide "peace of mind." It's really quite shameful. His wife is a doctor, she should really know better.
Coronavirus: Email From Stanford Professor’s Wife Claimed His Antibody Study Would Prove If You Were Immune

 

[Norbert]

No that's right. Except that some studies seem to be ignoring the confidence interval of the specificity.
The specificity itself doesn't affect the confidence interval. If the specificity were exactly 99% (95%CI 99%-99%) then you would know exactly how many false positives there were in your study and subtract them. You would also have to know sensitivity exactly too. The tricky math is incorporating the uncertainty in the specificity and sensitivity.

I'm not sure if there is a point in discussing it to this detail, but anyway: I would guess that the specificity is not exactly constant for any specific product, that it varies within a certain interval. But then you also have an additional error interval as the result of the sample size that you use to measure the specificity. Even if you use a very large sample size to measure specificity as precisely as possible, multiple studies each using a sample size of 500 would still encounter variations (I would guess).

When you do such a measurement, you probably get a single average number, and then you probably calculate an interval based on sample size. But you don't know what the amount of additional variations among different test kits is, if you know what I mean.

 

[Norbert]

Or, the source case sits near the AC return duct.

If the upper room has its own AC, yes. (I guess each room has multiple outlets.)

 

[ReddyLeaf]

It's amazing to see the can-do spirit of Market capitalism at work here at TMC. You are obviously inspired by the same Muse that gave me the idea for my product Moron Spray. The conjoined effects of these radical products could constitute a remarkable breakthrough. I hose them down and you blow Sunshine up there a$$es. It could be enough to dissolve even the current record levels of idiocy. Hopefully before anyone takes Trump literally and starts injecting themselves with Lysol, household bleach, Tide, Etc

 

[bkp_duke]

I'm getting the impression that many "researchers" don't actually understand statistics, they only know how to plug a couple numbers into a confidence interval calculator webpage.

Many of us don't. When I was working at a university, I would take my data to a medical statistician to go over with and help me determine the proper stats to run, etc. It's a proper science in and of itself. Unfortunately many researchers don't have access to a medical statistician to help them with the calculations.

 

[bkp_duke]

The Washington post has a story about younger people suffering severe strokes and testing positive for the virus. Can the doctors in this thread weigh in with their thoughts ?

I don't have much for this, sorry. We see this in bacterial infections, for sure. I'm wanting to see more data and reports before I postulate much of a hypothesis.

 

[sparcs]

Anthony Fauci says US should double its testing over next several weeks - CNNPolitics

Fauci, a key member of the White House coronavirus task force, estimated that the US is conducting approximately 1.5 to 2 million Covid-19 tests per week and that "we probably should get up to twice that as we get into the next several weeks, and I think we will."

last 7 days added up from the US historical rounds up to 1.5 million tests per week so that seems accurate. US Historical Data

So Fauci is wanting a test rate of 3 to 4 million per week which converts to 429k to 571k per day. Let's split the difference and say he is wanting 500k tests per day within the next several weeks.

 

[ReddyLeaf]

He is systematically following the advice of a random number generator.

That implies a method.

 

[dhanson865]

Different timeline. Germany started earlier

Can you ELI5? Walk me through that logic a few steps?

 

[AlanSubie4Life]

(94% specificity, 4.x% to 7.x% 95 CI).

I of course meant 94%, 92.x% to 95.x%, 95CI). Or whatever.

I would guess that the specificity is not exactly constant for any specific product, that it varies within a certain interval.

Yeah. As was mentioned, also depends on the sample in use and whether it reflects the population. False positives might happen for a specific reason - for example they are picking up an antibody which is similar. Depending on what metrics were used to screen the pre-COVID samples, that could mean that there is a higher/lower rate of these false positives when you actually run the test, vs. the calibration sample.

The upshot is really just to make sure the test is well-calibrated with a tight confidence interval. And then use it on a population with prevalence which is significantly higher than the noise floor of the test.

 

[SageBrush]

Sorry, typo corrected in Bold

No,

As N increases for the test performance study the range of test performance for a given confidence narrows.
So e.g., presuming that you want a 95% confidence of your test performance interval, if N = 100 then an interval might be [0.9 - 1.0] while N = 400 would narrow the range, say to [0.93 - 0.97]

 

[EVNow]

The Washington post has a story about younger people suffering severe strokes and testing positive for the virus. Can the doctors in this thread weigh in with their thoughts ?

Not a doctor … but I wonder whether the virus attacks the nervous system. The other notable thing is sudden loss of taste/smell. This could either be some inflammation near the related nerves or actual attack on the nerve cells. The other possibility is the thickening of blood / clots I've seen mentioned.

 

[SMAlset]

What I notice is that in the upper room, cases (if that are the blue marks) are distributed quite evenly. To me, that would suggest that some people are more easily infected than others.

My thought seeing the floor plan was were the occupants at the white desks tested or are the blue desk occupants only those that came down with symptoms so got tested and then recorded? Could the people at the white desks be asymptomatic and still carriers?

I hate open floor plans and remember a discussion I had with my husband when some of their offices were going that route as how do you stop people from spreading a cold or flu? You can say stay home but surely it gets passed in the early stages. Had no thoughts at that time to a pandemic spreading. Have no idea how you safely repopulate such an office set up when temperature checks won’t identify everyone who is in the early stages and a spreader.

 

[Doggydogworld]

Germany is listed here as having basically the same number of recovered cases as the US and about 1/10th the number of deaths with about 1/6th the number of cases. What gives? Difference in criteria for "recovered" or more successful medical care?

View attachment 535816

Many states don't track recoveries. Some do only partially. At one point 95% of TX recoveries were from my county. It's still well over 50%. We only have about 5% of total cases. COVID Tracker stopped showing national stats for hospitalizations, recoveries, etc. for this reason. They still show it at the state level, with lots of N/As.