Jungle Jim
Graduate Poster
- Joined
- Dec 16, 2008
- Messages
- 1,274
-
Security incident: ISF was recently accessed by intruders. Please change your password, and change it anywhere else you used it. Read more
Cont: 2019-nCoV / Corona virus Pt 3
- Thread starter Darat
- Start date
-
- Tags
- coronavirus diseases
- Status
The Atheist
Psychopath
- Joined
- Jul 3, 2006
- Messages
- 38,099
I saw that and it's potentially excellent news. The numbers look pretty good, but we'll have clearer results in days.
It seems to pretty common around the world that many Covid deaths aren't being recorded as such. I doubt we'll ever know the true number.
As to the German numbers - everybody know how good we are at following orders, and everybody who met us knows how much social distancing we practice voluntarily compared to Spain, Italy and France...
The Poms are among the more stand-offish people as well, but it didn't help them.
Living in hope is all, and preparing for success, if it happens. It's what Bill Gates is doing with building facilities manufacturing unknown treatments.
If one of them works, we're going to need several billion doses and pre-preparing a million for the most vulnerable in your country seems fairly sensible.
Mostly fair points, though I haven't seen any actual evidence that the test used in MA has changed at all during the recorded period.
I agree that the amount of testing is insufficient, even though the state has been touting its high number of tests (relative to other states and nations) all along. I don't know if the limiting factor on testing is still the number of tests available (as it was early on, before March 24th), or the number of people who can be swabbed per day by the available medical personnel, or the number of people able to get medical authorization to be tested. These factors can be interdependent; for example, if test center personnel were sitting idle because doctors weren't ordering a lot of tests, one might expect the medical authorities to loosen the criteria for authorization, and so forth.
The variations in tests completed per day could be driven by changes in the input stream, rather than by e.g. backlogs at labs. I know drive-in test centers have been opening at specific locations for one-day periods; how many such centers are open per day and how much "business" they do at a given location could vary, explaining the day to day variations in numbers of test results reported. If fewer drive-in or other test locations are being opened on weekend days, with the expected processing and results reporting delays, that could explain the recent Mon-Tue lulls.
There being any significant aspect of the disease process behind the stair-step pattern is therefore questionable, but the long-term changes in rate of increase of new confirmed cases per day are still well-evidenced, and are consistent with the curves now playing out in all different phases in many places.
The Atheist
Psychopath
- Joined
- Jul 3, 2006
- Messages
- 38,099
Interesting news off USS Theodore Roosevelt. They're testing the entire crew and after 95% of that, found 600 infections, 60% of which were asymptomatic. That's sure to include a number that were pre-symptomatic, given the long incubation period, but it is more evidence that somewhere around half of all cases are asymptomatic.
https://www.msn.com/en-us/news/us/c...aircraft-carrier-are-symptom-free/ar-BB12Jojz
https://www.msn.com/en-us/news/us/c...aircraft-carrier-are-symptom-free/ar-BB12Jojz
Interesting, and looking forward to learning the pre-symptomatic prevalence.
I have to go back and re-find a reference, but it's been reported in Boston that of the 146 non-symptomatic cases found at the Pine Street Inn (homeless shelter) nearly two weeks ago, only one (!) went on to require hospitalization, and "some" (unspecified number) developed "mild cold symptoms."
Well, asymptomatic at the time of testing, but symptoms could still follow?
Or is/was everyone tested asymptomatic already beyond the two weeks?
marting
Illuminator
- Joined
- Sep 18, 2003
- Messages
- 4,315
Yep. Very important difference. Normally, whether a person is symptomatic is noted at the time the test is given. It's rapidly spreading and positive tests occur days before symptoms. 60% asymptomatic will be much lower once the spreading stops and 10 days or so go by.
Last edited:
sir drinks-a-lot
Philosopher
I've heard the incubation period of Covid-19 as being from 5-14 days. I wonder how likely it is that it could be longer. Even 14 days sounded long to me, but I'm not a virologist by any stretch, and a quickly found source online suggests some viruses have significantly longer incubation periods.
Testing: specificity, sensitivity, false positives and negatives, Bayes
If you know your BayesWP, you already know that the numbers of false positives and false negatives, in a covid-19 test, have important implications for conclusions you draw from the results.
In medical tests, I learned recently that there are two metrics regarding the "falses", specificity (roughly, does this test detect covid-19 and only covid-19?), and sensitivity (roughly, how completely does this test detect covid-19).
This fairly recent STAT article explains some testing-related challenges (I think STAT is a reliable source): "Covid-19 testing: overcoming challenges in the next phase of the epidemic" (link).
A covid-19 test that is both 100% specific and 100% sensitive may exist in some alternative universe; in ours, not so much.
When you're desparately short on tests - still true throughout the US, sadly - it likely doesn't merit wasting brain cycles thinking about rates of "falses", for this or that covid-19 test, done here or there.
However, if you want to understand the reported test results - from Vò, Diamond Princess, Theodore Roosevelt, Iceland, or even Telluride - you do. Apparently even whoever is in charge of China's borders does too: 14 day mandatory quarantine for all international arrivals, no exceptions (a similar rule may apply in Australia too, especially since the Ruby Princess fiasco).
If the US CDC - and equivalents elsewhere - are on the ball, one criterion for giving approval to a covid-19 test would be reliable data on that test's specificity and sensitivity, and ensuring that that data is made public.
If you know your BayesWP, you already know that the numbers of false positives and false negatives, in a covid-19 test, have important implications for conclusions you draw from the results.
In medical tests, I learned recently that there are two metrics regarding the "falses", specificity (roughly, does this test detect covid-19 and only covid-19?), and sensitivity (roughly, how completely does this test detect covid-19).
This fairly recent STAT article explains some testing-related challenges (I think STAT is a reliable source): "Covid-19 testing: overcoming challenges in the next phase of the epidemic" (link).
A covid-19 test that is both 100% specific and 100% sensitive may exist in some alternative universe; in ours, not so much.
When you're desparately short on tests - still true throughout the US, sadly - it likely doesn't merit wasting brain cycles thinking about rates of "falses", for this or that covid-19 test, done here or there.
However, if you want to understand the reported test results - from Vò, Diamond Princess, Theodore Roosevelt, Iceland, or even Telluride - you do. Apparently even whoever is in charge of China's borders does too: 14 day mandatory quarantine for all international arrivals, no exceptions (a similar rule may apply in Australia too, especially since the Ruby Princess fiasco).
If the US CDC - and equivalents elsewhere - are on the ball, one criterion for giving approval to a covid-19 test would be reliable data on that test's specificity and sensitivity, and ensuring that that data is made public.
I know nothing about what covid-19 tests were, or were not, used in MA.
However, around the time the CDC's stranglehold on testing was loosened, there were reports of many different entities developing, and deploying, covid-19 tests. Some even mentioned either specificity and/or sensitivity (University of Washington sticks in my mind for some reason).
Yes, all current tests are PCR-based (not serological)*, but those used by LabCorp, say, are almost certainly not the same as what UMass is using (I have no idea if any UMass campus has developed its own covid-19 test, but you get the idea).
*Telluride - which was a failure - and Iceland (more?) aside
Fifth published by fiveThirtyEight earlier today*: "Experts Think The U.S. COVID-19 Death Toll Will Hit 50,000 By The End Of April"
Some snippets (the survey on which the article is based was "conducted April 13 and 14"):
"The expert consensus is that the U.S. will have reported around 47,000 COVID-19 deaths by May 1, with a 90 percent chance of having between 32,000 and 82,000.
This is very close to the latest forecast from the Institute for Health Metrics and Evaluation at the University of Washington, which projects roughly 51,000 deaths by May 1, with an uncertainty range from about 28,000 to 112,000."
Frightening that by today (17 April) the US death toll has blown past 28k, and by tomorrow will almost certainly have exceeded 32k.
"For five weeks, surveyed experts were asked what factors led to the difference between their best-case and worst-case estimates in longer-term death forecasts. The answers have had a common refrain: They can’t perfectly predict what stay-at-home orders will be enacted, how long they will last, and how well people will follow those orders. That last issue became particularly salient this week as protesters in Michigan, Ohio and North Carolina organized in opposition to their state’s stay-at-home orders."
I rather doubt that anyone's model could accurately include the effects of those sorts of protests.
"At the time the survey was conducted, three states — New York, New Jersey and Michigan — had reported more than 1,000 COVID-19 deaths. As of Thursday, Louisiana, Massachusetts and Illinois have also passed that threshold.
The experts were asked how many states will report more than 1,000 deaths by May 1. They estimated that the most likely number was between eight and 11 states, with a nearly 1 in 3 chance that at least 12 states would report that many deaths."
*I skipped the fourth; not very informative I feel. For the record: "Experts Think We’re Flattening The Coronavirus Curve, But Hospitalizations Haven’t Peaked Yet".
This is very critical for policies on epidemic control. This paper on epidemic control strategies;
https://www.medrxiv.org/content/10.1101/2020.03.10.20033738v3
suggests that if > 30% of cases are asymptomatic then a policy based on contact tracing will be ineffective.
This sort of analysis and these figures are important for armchair strategists who think the solution to the pandemic is simple; all that is needed is to carry out contact tracing. In the real world it may be more complicated.
Cheetah
Master Poster
We are also at the point where you have to decide how much more the economy can take. We have much less of a buffer than first world countries.
From Wits university.
South Africa needs to end the lockdown: here’s a blueprint for its replacement
From Wits university.
South Africa needs to end the lockdown: here’s a blueprint for its replacement
Dr.Sid
Philosopher
A third of the sailors serving with France's aircraft carrier Charles de Gaulle - 668 out of nearly 2,000 - are infected with coronavirus.
https://www.bbc.com/news/world-europe-52308073
Local news (in Czech) actually had fresher report with more then 1000 infected, that's over 50% of the crew.
https://www.bbc.com/news/world-europe-52308073
Local news (in Czech) actually had fresher report with more then 1000 infected, that's over 50% of the crew.
Loss Leader
I would save the receptionist., Moderator
This thread has been closed due to its many duplications with other science and medicine threads about Covid-19. Members are directed to a new, single thread HERE. Members may freely quote from this thread and import discussions here into the new catchall science thread. Thank you.
Posted By: Loss Leader
- Status