Thread by @RogueShoeButler

1/25 An important question, one that everyone should know the answer to!

I am not an expert on epidemiology, but I can share what I've read. This is not super-complicated, but it requires 20x more patience than most things. Still, everyone can understand it—and should.
#COVID19

https://twitter.com/LRarey/status/1245427493101809670

2/25 First, some background on the mathematics of epidemics. This is the part that requires patience.

3/25 This is the basic picture of an epidemic. At any time, there is a number of people who are susceptible (blue line), a number infected (red line), and a number recovered (green line).

4/25 When an outbreak starts, the number of infected starts off growing exponentially, slows, then decays exponentially. The growth happens because the rate of spread increases in proportion to the number infected so far. It slows and decays as the number susceptible decreases.

5/25 This is called the SIR model. It is the simplest mathematical description of infectious spread. Incubation times, fatalities, and other quantities make the math more complicated, but the essential pattern is the same (except for "second waves").
maa.org/press/periodic…

6/25 People often use the word "exponentially" loosely. Here we must use it precisely. Exponential growth means growth whose rate is proportional to the current size of the thing that is growing.

7/25 A well-known example of exponential growth is compound interest. The money you gain per year is a constant percentage of the money invested so far: the interest rate. At an interest rate of 2.5%, your money doubles every 28 years, whether you started with $100 or $1,000,000.

8/25 So, it does not make sense when people say that COVID-19 is "exponentially more contagious" than something else. That's just loose use of the word.

9/25 Eventually the number of people who have been infected and recovered, assumed now to be immune and incapable of infecting others, grows enough that they serve to insulate the remaining susceptible from the remaining infected—and the pathogen dies out.
en.wikipedia.org/wiki/Herd_immu…

10/25 To assess the danger of an infectious outbreak, we need to estimate how many people will be infected before herd immunity sets in, and what percentage of those infected will die.

11/25 The number who will ultimately be infected depends largely on the rate of transmission. This is measured by R0, the number of people one person will infect when the rest of the population is susceptible. R0 plays a role similar to "interest rate".
en.wikipedia.org/wiki/Basic_rep…

12/25 Very small changes in R0 result in huge changes to the number who are ultimately infected. If you go to @gabeeegoooh's Epidemic Calculator and move the R0 slider around, you can see the huge swings in the size of an outbreak starting with 1 person.
gabgoh.github.io/COVID/index.ht…

13/25 The number whom the infection will kill is the number ultimately infected × the "case fatality rate" (CFR). The Epidemic Calculator provides a slider for case fatality rate, if you'd like to see the effects on total fatalities.

14/25 Now I hope you can see that the numbers of cases and deaths on any one day early in an outbreak don't provide much information. Early in a vast, lethal pandemic, these numbers will be small; they will also be small early in a harmless outbreak.

https://twitter.com/LRarey/status/1245426056284319751

15/25 The relevant information is the parameters that determine the shape of the whole epidemic from start to finish.

16/25 Now, the data. The basis for declaring a public health emergency is the measurements of #COVID19 cases in Wuhan in December and January, included in this report. See "Transmission dynamics", starting on page 9, for R0; CFR is in the next section.
who.int/docs/default-s…

17/25 The measurements include the percentage of cases that ended in death and the progression of the curve seen so far (see "epidemic curve" in the report). Testing many people *over time*, with and without symptoms, gives you points along the curve; from those, you estimate R0.

18/25 The WHO report estimates the R0 for #COVID19 at between 2.0 and 2.5. This is quite high: in the early stages, each person infects on average 2.0 to 2.5 other people. Other estimates (see the Calculator) were comparable.

CFR varied from 0.7% to 5.8% depending on the region.

19/25 For comparison, the US seasonal flu of 2017/2018 had R0=1.53 and CFR=~0.14%.

That resulted in 45,000,000 infections and 61,000 deaths.
journals.lww.com/imd/Fulltext/2…
cdc.gov/flu/about/burd…

20/25 Here's a comparison of #COVID19 against other major infectious diseases.
triplebyte.com/blog/modeling-…

21/25 Something that is often missed: because small differences in R0 and CFR result in huge swings of total infections, the inevitable uncertainty in those measurements implies a correspondingly huge range of uncertainty for the estimated number of deaths.

22/25 The range of uncertainty for number of deaths assuming no countermeasures are taken (no social distancing, etc.) goes from 10,000s to 10,000,000s (in US). If the former, no worse than flu; if the latter, major catastrophe.

23/25 This article describes uncertainty about #COVID deaths in the US assuming some #SocialDistancing.
fivethirtyeight.com/features/exper…

24/25 Beyond numerical data, I think the decision was informed by experience reflected in this famous quotation by former Utah Gov. Micheal Leavitt:
"Everything you say in advance of a pandemic is alarmist; anything you do after it starts is inadequate."
books.google.com/books?id=va-pJ…

25/25 Did that answer your question?