Coronavirus : Update 11 from the Epicenter
Dear friends, when I started writing these accounts it was to provide you with a picture of the real dynamics of the epidemic while the UK and US governments were busy downplaying it’s predicted impact on those countries.
Now (just two weeks later!!!) Spain, France, Germany, the UK and the US are all in the grip of exponential infections, lockdowns and impending economic collapse. I’m going to start focusing on post-lockdown predictions as Italy is seeing some light at the end of the tunnel. How should we get the country working again without restarting the epidemic? What can we learn from China?
First though, you may have seen the news this week that Google has started releasing our anonymized movements as COVID-19 reports – presumably to shame us into #stayingat home. I have prepared a summary page of our countries that shows some interesting differences. You can see how government policies affect the curves, and how the whole world hit the parks (they were quickly closed in Italy). As usual, the country-level numbers mask the nuances in each state, county and city. You can drill down into the reports and see what’s happening in your local area if you like :
Information about the pandemic is everywhere, but I’m going to keep publishing my graphs for the sake of consistency. If you’ve been following along then at least you can see it through to the end in the same format.
The death to cases curves below show the maturity of the epidemic in a country. A spike at the beginning indicates that a country’s first sign of the virus were fatalities. We then see a drop as previously undetected cases are rapidly uncovered, and then a steady rise as the infected start dying.
Italy’s curve is destabilizing (a good sign that we’ll come back to later). The UK and Germany’s fatality rates are out accelerating their infection rates which places them deep into the crisis. The US is just at the beginning of their mortality explosion – just as their stockpiles of masks and ventilators are drying up…
I have a couple of new graphs to share. Let me know if they’re useful. First, I used the techniques shared by Minute Physics :
As you can see in our graph, none of countries we are following has ‘stalled’ and ‘fallen off the graph’ like China and Singapore did in the video, however Italy is no longer exponential – it is (hopefully) ‘stalling’ and will soon begin to fall.
I did the same for deaths, and you can see that Italy is again starting to turn the corner :
The final charts for today show rate of change. This is useful because it’s very difficult to understand what’s happening at the tip of an exponential curve. I therefore plotted the change in case and death numbers over the last few days for each country. Don’t misinterpret what you see here – the US looks like it’s in great shape, but a growth rate of 15% a day when your numbers are in excess of 250k is not good… but the explosion is slowing :
In order to understand the relative trajectory of each country, it may also be useful to see yesterday’s ranking tables ordered by different criteria :
…and finally, at the end of Update 9 I threw out the idea that there may be a link between air pollution and Covid-19 transmission. I used these two graphs to show apparent correlations in the data :
I thought I was being a little disingenuous because it seems probable that both distributions are related to a third common factor – population density. I expected the old adage “correlation is not causation” to hold true here, however I received a comment from Simone Naso (SedApta Group) showing that this analysis has already been carried out and a positive, direct correlation was identified! The study is in Italian, but here is the link if you’re interested, and here is the corresponding graph from their study :
…and here is the graph showing the linear relationship between the concentration of airborne particles (PM10) and Covid-19 infections :
Thanks Simone! Stay safe everyone!
Great insights Alastair.
Would be interesting to apply some of your tools and analysis to the part of the crisis we are together helping at Siemens – supply of critical goods (PPE, medical devices, medicines, and ultimately vaccines). I wonder how some of that data (if available) correlates to death rates… or length of the curve… will for sure be interesting to project what lies ahead, by watching China & N Korea curves & behavior, and applying some educated guesses. Perhaps data from past pandemic may help.
Billy, it’s reeeeally complicated. Even if I was capable of serious analysis, there are just so many unknowns and moving targets. The biggest unknown is the actual number of cases (most countries are only testing the very sick) and the biggest moving target is the effect of the different stages of social distancing and lockdown. Each move is happening within hours of the previous ones and further changes the trajectory of the epidemic, but with a 14-day lag. It’s hard to answer the question “will 200k or 2M people lose their lives?” That’s the difference between the number of people who die in car accidents and a global conflict. On that canvas it’s pretty impossible to answer the question “what will the effect of wearing a mask in public have on mortality rates due to a reduction in viral loading?” This said, I guess it’s truly a case of every little helping. As you point out, once this is under control, the big question is now what? That’s what I’ll try to think about next.
While the headlines are alarming, health experts cautioned that it was too early to gauge the severity of the outbreak. There are too many unknowns: Where did it start? How easily does it spread? How does it compare to other coronaviruses, like SARS?