31 ms·

I saw something very bizarre yesterday, which seems like it could totally explain the surging case counts while death rate continues to fall, but I can’t quite

by zaroth 6y ago

I saw something very bizarre yesterday, which seems like it could totally explain the surging case counts while death rate continues to fall, but I can’t quite believe it’s really happening. Apparently since about mid-April the CDC decided that the COVID Case Count would include not just positive tests, but also “probable” cases. Probable cases are not like the Presumptive Positive cases we had early on where a state lab had a positive test but they wanted to confirm it at the CDC lab. The criteria for a “probable case” does not require a positive lab test at all, but simply a combination of symptoms (like a cough or fever) and contact with another person who themselves was positive or probably positive. Collin County (6th largest county in Texas) had a council meeting which included a presentation on this where they walk through the new criteria, which includes a slide showing how 1 positive lab test can result in the case count increasing by 17. Here’s an excerpt from that part of the council meeting; https://twitter.com/sav_says_/status/1278090647140995073?s=21 https://twitter.com/sav_says_/status/1278090647140995073?s=2... And in case you’re dubious, here’s the whole meeting, and you can scroll to about 15:25 to see this part; https://collincountytx.new.swagit.com/videos/62477 https://collincountytx.new.swagit.com/videos/62477

nostrademons 6y ago
The CDC case numbers are useless. Aside from the "probable" case issue, for a long time they were also including antibody tests in the total. Antibody tests measure the number of people who have ever had coronavirus, not the number of active infections, thus conflating two very different statistics. Most coronavirus trackers aggregate state & county level data, which doesn't suffer from the same blatant incompetence as the CDC data. It tells the same story, though: case counts are increasing exponentially, with a current doubling time of about a week, and the percent positive rate (a measure of how much testing is undercounting the true infection rate) is rapidly going up in many locations.
- zaroth 6y ago
  Even the NYT COVID tracker page has the following note; “In the tallies shown on this page, The Times is now including cases that have been identified by public health officials as probable coronavirus patients.”
graeme 6y ago
Deaths lag cases by about a month. Would expect to see the new cases result in deaths a month from now. It can take a while to die from covid, and also death statistics take a delay to compile.
- zaroth 6y ago
  There has been a clear divergence, even accounting for the expected lag in death rate. The daily case count and daily death count initially look like two of the same curve simply shifted — as you might expect. Since approximately mid-April that has no longer the case and they are fundamentally different curves. You can see it here if you scroll down to where the daily charts are one above the other; https://www.nytimes.com/interactive/2020/us/coronavirus-us-cases.html https://www.nytimes.com/interactive/2020/us/coronavirus-us-c... Something big changed (or several smaller somethings) mid-April to cause this divergence. If the case count measurement is no longer measuring actual positive cases, that could be a big part of the problem.
  - starkd 6y ago
    That could be because the newer cases are younger people, and the mortality rate is much lower. There are many people who get it who are completely asymptomatic.
  - slg 6y ago
    Testing has been on a rather steady increase and was truly awful in April. I know multiple people who had mild symptoms and couldn't get tested in April. We simply can't assume those case numbers from April are anywhere close to accurate. In all likelihood we had a huge case spike in April and we just don't have the test results to prove it. The April peak in new cases is therefore dampened to make it look like we had 3 months of roughly similar new case numbers when in actuality we had 3 months of declining cases. This is generally backed up by the percentage of positive tests[1] which saw a huge drop from April to the beginning of June before starting to increase again in late June. That spike isn't far enough in the past for the lagging death total to start increasing yet. Odds are the daily death increase is right around the corner. TL;DR - The new cases plot is not predictive of the new death plot because cases are a function of new tests while deaths happen regardless of testing. [1] - https://coronavirus.jhu.edu/testing/individual-states https://coronavirus.jhu.edu/testing/individual-states
guscost 6y ago
Reporting standards have definitely been changing over a similar time period, but this alone may not be sufficient to explain the recent jump in cases: https://www.washingtonpost.com/investigations/cdc-wants-states-to-count-probable-coronavirus-cases-and-deaths-but-most-arent-doing-it/2020/06/07/4aac9a58-9d0a-11ea-b60c-3be060a4f8e1_story.html https://www.washingtonpost.com/investigations/cdc-wants-stat... Of course this obvious factor is not mentioned anywhere in the headline coverage of the case numbers, which says a lot.
heavyset_go 6y ago
This is standard. You can fairly well deduce that someone died from CoVID from their presentation of symptoms, and a CT scan. A test is not necessary to make that determination, and similarly, it is a waste to test the dead when there are plenty of people who are alive and presenting symptoms today.
- guscost 6y ago
  This was not standard in [most of] the US two months ago.
  - heavyset_go 6y ago
    They certainly were doing this in the NYC area when tests were scarce and the dead were piling up. People were complaining about it without understanding the implications then, too.
    - guscost 6y ago
      Are you arguing that they have started doing this in other places because tests are scarce? Because the dead are piling up? Really?
      - heavyset_go 6y ago
        Please refer to a few months ago when the dead were piling up in NY faster than the infected could be tested for CoVID.
    - emn13 6y ago
      To further underline what you're saying is both true and OK; once they started doing that, death tallies started at least getting close to excess deaths, although (IIRC, and I might have misremembered) despite diagnosing covid without PCR, excess deaths always exceeded diagnosed cases - i.e. almost certainly the doctors etc were being fairly conservative. (Although they didn't try to diagnose people that died before even reaching the hospital, which were thousands - and would have contributed to excess deaths, but not covid numbers).
acqq 6y ago
> Apparently since about mid-April the CDC decided that the COVID Case Count would include not just positive tests, but also “probable” cases. This is easy to compare with the past data: https://www.worldometers.info/coronavirus/country/us/ https://www.worldometers.info/coronavirus/country/us/ Looking at the 7 day moving average: April 20: around 30K cases per day; May 20: around 24K cases per day; June 12: around 22K cases per day; July 2: 47K cases per day Doesn't fit at all with the "mid-April" hypothesis. The number of cases stayed for quite a while below 30K, only to surge very recently. Also see how the curves for different states differ, and note how it's completely different from the shape of the curves following the tests: https://covidgraph.com/usa/#daily https://covidgraph.com/usa/#daily Also: "Adm. Brett Giroir, the man Trump appointed to oversee testing, testified at a House hearing Thursday that "this is a real increase in cases" and not just attributable to increased testing. "There is no question that the more testing you get, the more you will uncover," Giroir said Thursday. "But we do believe this is a real increase in cases because of the percent positives are going up. So, this is real increases in cases." Giroir said the U.S. is not flattening the curve. "The curve is still going up," he testified." https://abcnews.go.com/US/coronavirus-updates-arizona-bar-allowed-employees-covid-19/story?id=71587798 https://abcnews.go.com/US/coronavirus-updates-arizona-bar-al...
- skinnymuch 6y ago
  I don’t think parent is right, but the 7 day avg test numbers are largely explained by just the tri-State NYC numbers. NYC, NJ, CT dropped more than the case avg by the June stat.

by zaroth 6y ago

nostrademons 6y ago

The CDC case numbers are useless. Aside from the "probable" case issue, for a long time they were also including antibody tests in the total. Antibody tests measure the number of people who have ever had coronavirus, not the number of active infections, thus conflating two very different statistics. Most coronavirus trackers aggregate state & county level data, which doesn't suffer from the same blatant incompetence as the CDC data. It tells the same story, though: case counts are increasing exponentially, with a current doubling time of about a week, and the percent positive rate (a measure of how much testing is undercounting the true infection rate) is rapidly going up in many locations.

zaroth 6y ago
Even the NYT COVID tracker page has the following note; “In the tallies shown on this page, The Times is now including cases that have been identified by public health officials as probable coronavirus patients.”

graeme 6y ago

Deaths lag cases by about a month. Would expect to see the new cases result in deaths a month from now. It can take a while to die from covid, and also death statistics take a delay to compile.

zaroth 6y ago
There has been a clear divergence, even accounting for the expected lag in death rate. The daily case count and daily death count initially look like two of the same curve simply shifted — as you might expect. Since approximately mid-April that has no longer the case and they are fundamentally different curves. You can see it here if you scroll down to where the daily charts are one above the other; https://www.nytimes.com/interactive/2020/us/coronavirus-us-cases.html https://www.nytimes.com/interactive/2020/us/coronavirus-us-c... Something big changed (or several smaller somethings) mid-April to cause this divergence. If the case count measurement is no longer measuring actual positive cases, that could be a big part of the problem.
- starkd 6y ago
  That could be because the newer cases are younger people, and the mortality rate is much lower. There are many people who get it who are completely asymptomatic.
- slg 6y ago
  Testing has been on a rather steady increase and was truly awful in April. I know multiple people who had mild symptoms and couldn't get tested in April. We simply can't assume those case numbers from April are anywhere close to accurate. In all likelihood we had a huge case spike in April and we just don't have the test results to prove it. The April peak in new cases is therefore dampened to make it look like we had 3 months of roughly similar new case numbers when in actuality we had 3 months of declining cases. This is generally backed up by the percentage of positive tests[1] which saw a huge drop from April to the beginning of June before starting to increase again in late June. That spike isn't far enough in the past for the lagging death total to start increasing yet. Odds are the daily death increase is right around the corner. TL;DR - The new cases plot is not predictive of the new death plot because cases are a function of new tests while deaths happen regardless of testing. [1] - https://coronavirus.jhu.edu/testing/individual-states https://coronavirus.jhu.edu/testing/individual-states

guscost 6y ago

Reporting standards have definitely been changing over a similar time period, but this alone may not be sufficient to explain the recent jump in cases: https://www.washingtonpost.com/investigations/cdc-wants-states-to-count-probable-coronavirus-cases-and-deaths-but-most-arent-doing-it/2020/06/07/4aac9a58-9d0a-11ea-b60c-3be060a4f8e1_story.html https://www.washingtonpost.com/investigations/cdc-wants-stat... Of course this obvious factor is not mentioned anywhere in the headline coverage of the case numbers, which says a lot.

heavyset_go 6y ago

This is standard. You can fairly well deduce that someone died from CoVID from their presentation of symptoms, and a CT scan. A test is not necessary to make that determination, and similarly, it is a waste to test the dead when there are plenty of people who are alive and presenting symptoms today.

guscost 6y ago
This was not standard in [most of] the US two months ago.
- heavyset_go 6y ago
  They certainly were doing this in the NYC area when tests were scarce and the dead were piling up. People were complaining about it without understanding the implications then, too.
  - guscost 6y ago
    Are you arguing that they have started doing this in other places because tests are scarce? Because the dead are piling up? Really?
    - heavyset_go 6y ago
      Please refer to a few months ago when the dead were piling up in NY faster than the infected could be tested for CoVID.
  - emn13 6y ago
    To further underline what you're saying is both true and OK; once they started doing that, death tallies started at least getting close to excess deaths, although (IIRC, and I might have misremembered) despite diagnosing covid without PCR, excess deaths always exceeded diagnosed cases - i.e. almost certainly the doctors etc were being fairly conservative. (Although they didn't try to diagnose people that died before even reaching the hospital, which were thousands - and would have contributed to excess deaths, but not covid numbers).

acqq 6y ago

> Apparently since about mid-April the CDC decided that the COVID Case Count would include not just positive tests, but also “probable” cases. This is easy to compare with the past data: https://www.worldometers.info/coronavirus/country/us/ https://www.worldometers.info/coronavirus/country/us/ Looking at the 7 day moving average: April 20: around 30K cases per day; May 20: around 24K cases per day; June 12: around 22K cases per day; July 2: 47K cases per day Doesn't fit at all with the "mid-April" hypothesis. The number of cases stayed for quite a while below 30K, only to surge very recently. Also see how the curves for different states differ, and note how it's completely different from the shape of the curves following the tests: https://covidgraph.com/usa/#daily https://covidgraph.com/usa/#daily Also: "Adm. Brett Giroir, the man Trump appointed to oversee testing, testified at a House hearing Thursday that "this is a real increase in cases" and not just attributable to increased testing. "There is no question that the more testing you get, the more you will uncover," Giroir said Thursday. "But we do believe this is a real increase in cases because of the percent positives are going up. So, this is real increases in cases." Giroir said the U.S. is not flattening the curve. "The curve is still going up," he testified." https://abcnews.go.com/US/coronavirus-updates-arizona-bar-allowed-employees-covid-19/story?id=71587798 https://abcnews.go.com/US/coronavirus-updates-arizona-bar-al...

skinnymuch 6y ago
I don’t think parent is right, but the 7 day avg test numbers are largely explained by just the tri-State NYC numbers. NYC, NJ, CT dropped more than the case avg by the June stat.