WHO - Confirmed cases doubled in 12 days

Status
Not open for further replies.
Everybody seems to be wondering the same thing: How much of the increase in confirmed cases is new infections, and how much is better detection due to more testing? If you've seen any data on this, please share, and I'll do the same if I find anything.
 
Which Roger said:
Everybody seems to be wondering the same thing: How much of the increase in confirmed cases is new infections, and how much is better detection due to more testing? If you've seen any data on this, please share, and I'll do the same if I find anything.
Click to expand...
Our county health website still showing 1 case in entire county. I know can't be true, they're just behind in reporting.
 
Rianne said:
Our county health website still showing 1 case in entire county. I know can't be true, they're just behind in reporting.
Click to expand...

And probably testing as well. That's been the real breakdown here.
 
If we assume that there are 5 times as many actual cases as those testing positive --that is, assume there are actually 1 million cases at present. And if we assume that the first case occurred in China in late November, 2019. Then, the doubling time is approximately 6 days => Log2 1 million = 19.93. Days passed ~ 120. 120/19.93 = 6 days/doubling. A medical study by Chinese doctors of the first 425 cases in Wuhan estimated the doubling time to be 7.4 days. https://www.nejm.org/doi/full/10.1056/NEJMoa2001316

So the case load will actually double by next week, but I would expect the number testing positive will go up by more than 2 because we are now able to test more people. Our main job at present is to lengthen that doubling period through social isolation so that we can "flatten the curve"


ETA: If you think the real number of cases is more than 5x the tested positive number, then just change the denominator in the equation above. Add 1 for each factor of two. So if you think there are 2 million current actual cases, then 120/(19.93+1) = 5.7 days/doubling.
 
Last edited:
The data for this country is inaccurate due to lack of testing.
Our state is still waiting to receive more testing kits, only 1400 people have been tested so far. The Public Health entity I used to work for has had ZERO ability to test anyone!
We are woefully behind the curve.
 
Gumby said:
If we assume that there are 5 times as many actual cases as those testing positive --that is, assume there are actually 1 million cases at present. And if we assume that the first case occurred in China in late November 20, 2019. Then, the doubling time is approximately 6 days => Log2 1 million = 19.93. Days passed ~ 120. 120/19.93 = 6 days/doubling. A medical study by Chinese doctors of the first 425 cases in Wuhan estimated the doubling time to be 7.4 days. https://www.nejm.org/doi/full/10.1056/NEJMoa2001316

So the case load will actually double by next week, but I would expect the number testing positive will go up by more than 2 because we are now able to test more people. Our main job at present is to lengthen that doubling period through social isolation so that we can "flatten the curve"
Click to expand...

I was going to post something similar. Just by inspection of some of the curves out there, the doubling rate was much higher in China. Considering the fast ramp-up in testing and public awareness (raising the numbers), doubling stretched out to 12 days is startlingly good.
 
The only really solid facts right now are numbers of symptomatic cases, hospitalizations and fatalities. Total number of positives (at least in the US) has little relationship to total number infected. I expect this to change dramatically over the next 2-4 weeks as we get more testing capacity online.

This is why draconian region-wide lockdowns such as we now have in California are necessary, despite the economic damage they cause. Until we can get a handle on testing and contact tracing we just have to assume everyone is infected.
 
Because we have data from countries that are farther along, and have tested much more extensively, it would seem to me that if you had a proxy (like difficulty breathing ER admission count), you could get a pretty accurate estimate of actual cases without widespread testing. Does such a proxy exist? All models are wrong, some are helpful tagline was what prompted this post.
 
I think we've exceeded the bounds of contact tracing. IMHO, the only way forward is to make virtually everyone stay in their house for two to three weeks, which should knock down the number of active transmissions enough so that we can actually go back to contact tracing and ring isolation.
 
This FT graphic shows the trajectory by country
.
.
 

Attachments

  • FT snip 03 20.JPG
    FT snip 03 20.JPG
    106.3 KB · Views: 158
I read this elsewhere, but think it's right: If you're interested in true infection rates, do a random sampling across the USA & see the results. A thousand or so would give a good handle I think. Probably need to repeat every month.
 
MichaelB said:
This FT graphic shows the trajectory by country
.
.
Click to expand...
Thanks. That graph will make larger countries look worse because they get to 100 cases faster; i.e., sooner in the spreading.
 
sengsational said:
Because we have data from countries that are farther along, and have tested much more extensively, it would seem to me that if you had a proxy (like difficulty breathing ER admission count), you could get a pretty accurate estimate of actual cases without widespread testing. Does such a proxy exist? All models are wrong, some are helpful tagline was what prompted this post.
Click to expand...

Yes, I would think the rate of confirmed cases is a proxy for the total cases, if you multiply it by 5 or so like Gumby did. But hospital admissions would be confounded by seasonal flu, which already has >36 million cases (and >22,000 deaths) in the U.S. alone this season. COVID-19 is still in the noise level compared to seasonal flu.
 
I have been watching the number of confirmed cases in Italy. For the last 4 days, the daily increase has been 13%/day. At that rate, the running total is doubling every 6 days.

Prior to that, the daily increase had been on the average 21%/day. It was doubling every 3.6 days. So, as terrible as doubling every 6 days looks, that's an improvement.

Italy now has 41,035 cases, with 3,405 deaths. European TV shows a long line of military trucks to take the dead from Bergamo to other towns to cremate. The morgues there are overflown, and the crematorium is already running 24/7.

PS. Italy started cordoning off the infected towns in northern Italy in February. The quarantine was expanded to entire northern Italy on 3/8, and the entire country on 3/9. On 3/11, all public places were ordered close, when it was observed that people still crowded restaurants and bars.
 
Last edited:
pacergal said:
The data for this country is inaccurate due to lack of testing.
Click to expand...
If the testing isn't random, more tests won't tell you the extent for the population as a whole, just for those tested
 
NW-Bound said:
The morgues there are overflown, and the crematorium is already running 24/7.
Click to expand...
Saw this also. Relatives live in Apennines above Bologna. Hate to be morbid, but cremation smoke, where does it go?
 
gerntz said:
I read this elsewhere, but think it's right: If you're interested in true infection rates, do a random sampling across the USA & see the results. A thousand or so would give a good handle I think. Probably need to repeat every month.
Click to expand...

Agreed 100%. I've been looking at charts, and I honestly don't know what to make of any of them.

Different countries have done a different amount of testing, and the ramp of their tests is likely different. And I don't know who they are testing. "new cases" probably tells us more about testing rates than infection rates.

As gerntz said, a totally random test across a fairly small population would provide a view of the cross section of America - just like the polls with a margin of error of 3%. But if they are testing people suspected of having it, how do we tie that back? Repeat that random testing each week to find trends.

AFAIK, none of that is happening. And it could be done in parallel of other testing, just keep the data separate for this purpose.

-ERD50
 
Rianne said:
Saw this also. Relatives live in Apennines above Bologna. Hate to be morbid, but cremation smoke, where does it go?
Click to expand...
Into the atmosphere around the crematorium.

"Smoke" is a relative term, though. Think "emissions." They are real and are supposed to be regulated.

I once lived close to a hospital that essentially cremated medical waste. Normally, you'd never know this was happening. (That removed spleen has to go somewhere.) Sometimes the device didn't get up to heat and it would smoke. I was glad to move away from there.
 
Imagine if we put a drop of ink into a pan of water. The ink spreads with time. Contagion is like that.

If we are interested in seeing how the fast the ink spreads, testing the water at the edge of the pan does not give us any info on the spreading rate.

The virus does not spontaneously appear uniformly across the country. Random testing across the country makes more sense near the end, when every town has been seeded.
 
NW-Bound said:
I have been watching the number of confirmed cases in Italy. For the last 4 days, the daily increase has been 13%/day. At that rate, the running total is doubling every 6 days.

Prior to that, the daily increase had been on the average 21%/day. It was doubling every 3.6 days. So, as terrible as doubling every 6 days looks, that's an improvement. ...
Click to expand...

I'm thinking deaths give us a clearer picture than "new cases" - I'm assuming each death can be tested/analyzed as being due to VV-19 more accurately, as there are fewer of them. But deaths are also a lagging indicator, it takes a while (10 days average?) for the disease to claim a victim.

So I did a super-rough analysis on Italy's death rate. I took the median value of deaths the past 3 days (427), and compared to the median of the 3 days from a week prior (189). And that actually is close to your doubling every 6 days, I got 2.26x over seven days.

-ERD50
 
NW-Bound said:
... Random testing across the country makes more sense near the end, when every town has been seeded.
Click to expand...

Yes, random testing country-wide may not be that helpful at first. But random testing within a specific area could be. IOW, get a cross section of that entire population, not a selected/suspected cross section.

If you took a radar gun to every car you suspected of speeding on a stretch of road, you could not calculate average speeds from that data. Similar thing.

But even a country-wide random test would start to show hot spots as the data is gathered.

-ERD50
 
For the individuals with some symptoms, it is desirable to know if it is the virus or the regular flu. We need more tests for that.

But for policy makers, all they need to know is the arrival rate at the hospitals. With the actual growth that they are seeing, they can see hospital beds, ICUs being filled a few weeks or months from now at that rate. And they know they need to put in drastic measures already.

There may be a lot more people with the disease and recovering well at home. That's fine, but you have to deal with the number of people needing hospitalization, and that's what you see right at your doorsteps. It may be only 1% or 10% of the total sick, but you do not have to deal with those lucky 99% or 90%.
 
Status
Not open for further replies.

Similar threads

B
  • Locked
Pepcid (famotidine) as a treatment
Replies
19
Views
2K
jim584672
J
S
  • Locked
Other symptoms that may signal COVID19
Replies
11
Views
1K
audreyh1
audreyh1

Latest posts

Back
Top Bottom