Monday, March 23, 2020

Who Has Exponential Growth

The NY Times says the best way to look for changes in the trend is via a log plot. OK, here were some case rates as of yesterday:

A straight line here means exponential growth (unless the line is flat, in which case the number of cases is a constant, viz no growth).

So Italy is experiencing something a bit less than exponential growth, and the US something a bit more. 

Don't forget, the problem with charts like this one is that it's easy to forget that exponential growth is bad enough. Slight reductions in it aren't much to be assuaged by, unless you are looking deep for a reason to hope.


Ned said...

We need a simple metric that can tell us whether testing is "catching up" to the spread of Covid-19 in the USA.

Here's one: are the counts of daily tests rising faster or slower than the counts of daily new cases and daily new deaths? The rate for each one will be the slope of the natural-log-transformed data over the preceding 7 days. I'll use the data here:

Here are the trends for the past 7 days:
* New cases: 0.31
* New deaths: 0.27
* New tests: 0.24

Thus, over the past 7 days, testing has been growing more slowly than infections and deaths.

Another metric that would be useful is the percent of new tests that are positive. If this metric is below 5%, testing is managing to catch most cases. If this metric is at or above 10%, many new cases are likely going undetected.

Here is the mean percent positive for tests in the past 3 days:
* Pct positive: 16.1%

Thus, over the past 3 days, it is likely that a substantial number of new cases are going undetected.

I'll update these metrics periodically to see whether we (the USA) are making progress on testing. Today, the numbers don't look good.

David Appell said...

Ned, this is a good analysis, thanks.

Do you happen to know why Trump rejected the WHO tests a few weeks ago?

Ned said...

My understanding is that the WHO normally provides tests to countries that don't have their own virology testing infrastructure. But the US, and other nations with large developed health sectors, usually produce their own tests. I don't think the WHO "offered" tests to the US because the WHO and everyone else simply assumed the US would be able to produce its own tests.

In the past this has always worked fine. The US has always been able to develop and manufacture its own tests, e.g. for the H1N1 flu in 2009, which was handled rapidly and efficiently. What's inexplicable is why the CDC/FDA system broke down this time.

If one had known at the time that the CDC and FDA would abjectly fail at providing a test in a timely fashion, then it would have made sense to ask the WHO for help. The whole situation is embarrassing to us as a nation.

See this Snopes piece about the claim that the US "rejected" a WHO test.

It doesn't discuss the WHO situation, but in all other respects, this Atlantic article is one of the best things you'll read about the lack-of-testing fiasco:

How the Coronavirus Became an American Catastrophe

Also worth following is Trevor Bedford, a virologist at the Fred Hutchinson Research Center in Seattle. Here's a thread by him from last week, about the outlook and what should be done:

(1) Massive rollout of testing, including at-home and drive-through testing (much more than we are trying to do now)

(2) Contact-tracing via cellphone data to target testing at people who have been in physical proximity to carriers, even unknowingly

(3) Serological assays to identify individuals who have been exposed to and recovered from the virus, and who may now be immune.

Ned said...

Over the weekend I posted some "predictions" for the # of new deaths each day in the USA. Here's an update to see how those predictions did on the first two days:

Sun 3/22: 85 [actual = 111]
Mon 3/23: 110 [actual = 140]
Tue 3/24: 142
Wed 3/25: 183
Thu 3/26: 236
Fri 3/27: 305
Sat 3/28: 394

Data from Worldometers. It looks like the number of deaths is running almost exactly one day ahead of my predictions so far. Usually though there is a fair bit of noise in these numbers so I expect them to be a bit more scattered.

David in Cal said...

Dr. Fauci denied that anyone screwed up on the testing. I don't agree. Some other countries were able to do a lot of tests before the US was. One clear screwup IMO was the CDC prohibition of independently developed tests. Whether the screwups were due to Trump, to his appointees, or to permanent employees of FDA or CDC, I do not know.

A second screwup was Trump's attitude. He did a good thing by closing China very early, but then he didn't understand how much more needed to be done.

OTOH once he did appreciate the nature of a highly contagious disease, he and his team did a great deal very quickly. Perhaps his background in hotels accustomed him to dealing aggressively and effectively with crises. I do not take for granted all his team has accomplished. See

List: 74 actions taken by Trump to fight virus and bolster economy

David in Cal said...

Ned - thanks for that nice analysis. In a sense, testing symptomatic people is less important than testing asymptomatic people. Even without testing, we know to keep symptomatic people away from others. When we have the capacity to test asymptomatic people who've been in contact with infected people, that should make a bid difference. As you point out, Ned, that day my be far off.


Ned said...

once he did appreciate the nature of a highly contagious disease, he and his team did a great deal very quickly. Perhaps his background in hotels accustomed him to dealing aggressively and effectively with crises.

LOL. Surely you're joking. I see lots of areas in which the Administration's response has been lethargic, weak, confused, and lackluster, and I don't see any significant areas in which they have been exceptional or outstanding beyond the basics of what one would expect from any Administration.

You said you've been watching the press briefings, so I suppose it's natural that you've absorbed a lot of self-praise by Trump and a lot of sycophantic congratulating of him by his own appointees or people who need his approval for their jobs. It's a lot like life in the former Soviet Union, TBH.

The fact that you think this mediocrity is praiseworthy is rather embarrassing.

David in Cal said...

Less than 3 weeks ago, we were doing a few hundred tests per day. Yesterday we did 66,000.

Ned said...

That would have been a pretty good number of tests in mid-February. It's not nearly adequate now.

Again, look at the numbers. Over the past week, the number of positive cases has grown faster than the number of tests. We are losing ground, David.

Extrapolating the week's rate of increase, there would be 85,000 tests today. If it's above 100,000, that would be a good start. To actually get this under control now, we need to be doing hundreds of thousands of tests per day.

That's the consequence of wasting the month of February. What would previously have been an adequate number of tests is now woefully inadequate.