March 12 update: the situation has deteriorated rapidly. See our latest writing.
This post was written by Tornus (who has no professional qualifications but has an extensive amateur interest in epidemiology) and Dr Smartypants, who is a physician and infectious disease epidemiologist with extensive experience working at the CDC and public health departments. We have endeavored to provide accurate, concise, and actionable information here. However, the best source of information about COVID is the official CDC page. You should trust the CDC more than us (or anyone else).
Over the last few days, we have learned enough to say with confidence that COVID-19 is a serious global pandemic and now is the time to take significant precautionary measures.
There is a great deal that we don’t know yet. It is quite possible that COVID-19 will be no worse than the seasonal flu and it is quite possible that COVID-19 will be as bad as the 1918 flu pandemic. We expect to have much more clarity about the situation by mid to late March.
Right now, there’s a fairly wide range of plausible scenarios. We know COVID-19 is not going to fizzle out like MERS or SARS did, and we know it isn’t going to end civilization. This is serious, but it isn’t the zombie apocalypse. What happens over the next year will probably fall somewhere between two scenarios:
The best likely outcome is that COVID-19 is like a typical flu season:
The worst likely scenario is that COVID-19 is roughly comparable to the 1918 flu pandemic:
We want to put the worst-case scenario in perspective. You should take this very seriously: it is entirely plausible that more people will die in 2020 than have ever died in any single year, and several people you know will die. But even in the worst case, it is likely that you and your immediate family will experience nothing worse than a bad cold. Governments will function normally, the lights will stay on, and your supermarket will remain fully stocked with food. Take this seriously, but do not panic: this is not the zombie apocalypse.
We currently have limited data about COVID-19 from China and a few other countries. Over the next few weeks, will will get data that is much higher quality, much more extensive, and much more relevant to predicting the course of the pandemic in North America and Europe. By mid to late March, we may learn that COVID-19 is less severe than we had feared and does not require extensive precautionary measures. However, we may also learn that this is going to be a very severe pandemic that requires extensive and lasting precautions.
We know from the 1918 flu that acting quickly is critically important to reducing the severity of a pandemic: a delay of as little as two weeks can substantially increase the final death rate. Therefore, given that we are now experiencing community transmission in North America, we believe it is prudent to take immediate precautionary measures.
If we’re lucky, we will be able to relax those precautions in a few week. And if we aren’t lucky, acting early will have saved lives.
Understanding the spread of new diseases is incredibly complicated, and we’re going to greatly simplify some very complex topics. But for the nerds among you, here’s some of the data that informs our thinking. You can skip the rest of this post unless you want to nerd out with us.
When judging the impact of a disease, we want to know two things in particular: how easily it spreads and how deadly it is.
Epidemiologists often refer to the R (reproduction) value of a disease. R is how many people each infected person spreads the disease to. So a disease with an R < 1 will tend to die out, while a disease with an R > 1 will tend to spread to an increasing number of people. Many factors affect R, including how many people have already been infected and what precautions are in place to prevent spreading.
A related question involves asymptomatic spreading. One reason MERS and SARS didn’t spread widely is that they very quickly caused severe symptoms, so sick people were easy to identify and tended not to circulate in the community.
Severity is a complicated topic, but the simplest measure of severity is case mortality: if 100 people catch the disease, how many will die? Case mortality typically varies with age: older individuals are usually but not always impacted much more severely than young and healthy people.
Two good points of reference are the seasonal flu and the 1918 flu pandemic.
Flu varies from year to year, but in a typical year, seasonal influenza:
Our best guess is that the 1918 flu pandemic:
Right now, all of our data about COVID-19 is extremely limited: expect substantial changes as we learn more.
What we know so far about infectivity isn’t encouraging. Current estimates of R range from 2.2 to 2.6, considerably higher than even the 1918 flu pandemic. In addition, there is some evidence that during the first week of infection many people are contagious but experience only mild symptoms. This complicates containment efforts and means that contagious individuals are likely to circulate in the community.
Case mortality estimates currently range from 1.4% to 3%.
There is a great deal of uncertainty about pretty much everything. Some of the limitations of our current data include:
There is some reason for optimism:
There are also some reasons for concern:
The CDC’s excellent information hub for COVID-19.
A good summary of typical seasonal flu patterns. Calculates an annual average of 389,000 deaths from flu, 2/3 of them among people 65 and older.
Excellent in-depth discussion of the 1918 flu pandemic. Cites a 33% global infection rate, with 500 million people infected and 50 million deaths.
General discussion of the 1918 flu pandemic (including a discussion of the reasons why it was inaccurately called the Spanish Flu). Cites a 27% global infection rate, with 500 million people infected and a death toll between 40 and 50 million. Note that Wikipedia cites two waves of flu, although the CDC identifies three distinct waves.
February 28 article discussing recent data about COVID-19. Cites case mortality rates of 1.4% - 2.0% and speculates about the true rate being considerably less than 1%. Cites an R0 of 2.2.
A detailed look at efforts to contain the 1918 flu pandemic. Cites a CEPID (cumulative excess P&I death rate) of 719/100,000 in Philadelphia (which waited 16 days after the first case to implement strong containment measures) versus 314/100,000 in St Louis (which waited 2 days).