If you stack up the orange bars until 1/22, you get 444 cases. Now add up all the grey bars. They add up to ~12,000 cases. So when Wuhan thought it had 444 cases, it had 27 times more. If France thinks it has 1,400 cases, it might well have tens of thousands
The same math applies to Paris. With ~30 cases inside the city, the true number of cases is likely to be in the hundreds, maybe thousands. With 300 cases in the Ile-de-France region, the total cases in the region might already exceed tens of thousands.
Spain and Madrid
Spain has very similar numbers as France (1,200 cases vs. 1,400, and both have 30 deaths). That means the same rules are valid: Spain has probably upwards of 20k true cases already.
In the Comunidad de Madrid region, with 600 official cases and 17 deaths, the true number of cases is likely between 10,000 and 60,000.
If you read these data and tell yourself: “Impossible, this can’t be true”, just think this: With this number of cases, Wuhan was already in lockdown.
With the number of cases in countries like the US, Spain, France, Iran, Germany, Japan or Switzerland, Wuhan was already in lockdown.
And if you’re telling yourself: “Well, Hubei is just one region”, let me remind you that it has nearly 60 million people, bigger than Spain and about the size of France.
2. What Will Happen When These Coronavirus Cases Materialize?
So the coronavirus is already here. It’s hidden, and it’s growing exponentially.
What will happen in our countries when it hits? It’s easy to know, because we already have several places where it’s happening. The best examples are Hubei and Italy.
The World Health Organization (WHO) quotes 3.4% as the fatality rate (% people who contract the coronavirus and then die). This number is out of context so let me explain it.
It really depends on the country and the moment: between 0.6% in South Korea and 4.4% in Iran. So what is it? We can use a trick to figure it out.
The two ways you can calculate the fatality rate is Deaths/Total Cases and Death/Closed Cases. The first one is likely to be an underestimate, because lots of open cases can still end up in death. The second is an overestimate, because it’s likely that deaths are closed quicker than recoveries.
What I did was look at how both evolve over time. Both of these numbers will converge to the same result once all cases are closed, so if you project past trends to the future, you can make a guess on what the final fatality rate will be.
This is what you see in the data. China’s fatality rate is now between 3.6% and 6.1%. If you project that in the future, it looks like it converges towards ~3.8%-4%. This is double the current estimate, and 30 times worse than the flu.
It is made up of two completely different realities though: Hubei and the rest of China.
Hubei’s fatality rate will probably converge towards 4.8%. Meanwhile, for the rest of China, it will likely converge to ~0.9%
I also charted the numbers for Iran, Italy and South Korea, the only countries with enough deaths to make this somewhat relevant.
Iran’s and Italy’s Deaths / Total Cases are both converging towards the 3%-4% range. My guess is their numbers will end up around that figure too.
South Korea is the most interesting example, because these 2 numbers are completely disconnected: deaths / total cases is only 0.6%, but deaths / closed cases is a whopping 48%. My take on it is that the country is just extremely cautious: they’re testing everybody (with so many open cases, the death rate seems low), and leaving the cases open for longer (so they close cases quickly when the patient is dead). What is relevant is that deaths/cases has hovered around 0.5% since the beginning, suggesting it will stay there.
The last relevant example is the Diamond Princess cruise: with 706 cases, 6 deaths and 100 recoveries, the fatality rate will be between 1% and 6.5%.
This is what you can conclude:
Put in another way: Countries that act fast can reduce the number of deaths by ten. And that’s just counting the fatality rate. Acting fast also drastically reduces the cases, making this even more of a no-brainer.
Countries that act fast reduce the number of deaths at least by 10x.
So what does a country need to be prepared?
What Will Be the Pressure on the System
Around 20% of cases require hospitalization, 5% of cases require the Intensive Care Unit (ICU), and around 1% require very intensive help, with items such as ventilators or ECMO (extra-corporeal oxygenation).
The problem is that items such as ventilators and ECMO can’t be produced or bought easily. A few years ago, the US had a total of 250 ECMO machines, for example.
So if you suddenly have 100,000 people infected, many of them will want to go get tested. Around 20,000 will require hospitalization, 5,000 will need the ICU, and 1,000 will need machines that we don’t have enough of today. And that’s just with 100,000 cases.
That is without taking into account issues such as masks. A country like the US has only 1% of the masks it needs to cover the needs of its healthcare workers (12M N95, 30M surgical vs. 3.5B needed). If a lot of cases appear at once, there will be masks for only 2 weeks.
Countries like Japan, South Korea, Hong Kong or Singapore, as well as Chinese regions outside of Hubei, have been prepared and given the care that patients need.
But the rest of Western countries are rather going in the direction of Hubei and Italy. So what is happening there?
What an Overwhelmed Healthcare System Looks Like
The stories that happened in Hubei and those in Italy are starting to become eerily similar. Hubei built two hospitals in ten days, but even then, it was completely overwhelmed.
Both complained that patients inundated their hospitals. They had to be taken care of anywhere: in hallways, in waiting rooms…
Healthcare workers spend hours in a single piece of protective gear, because there’s not enough of them. As a result, they can’t leave the infected areas for hours. When they do, they crumble, dehydrated and exhausted. Shifts don’t exist anymore. People are driven back from retirement to cover needs. People who have no idea about nursing are trained overnight to fulfill critical roles. Everybody is on call, always.
That is, until they become sick. Which happens a lot, because they’re in constant exposure to the virus, without enough protective gear. When that happens, they need to be in quarantine for 14 days, during which they can’t help. Best case scenario, 2 weeks are lost. Worst case, they’re dead.
The worst is in the ICUs, when patients need to share ventilators or ECMOs. These are in fact impossible to share, so the healthcare workers must determine what patient will use it. That really means, which one lives and which one dies.
“After a few days, we have to choose. […] Not everyone can be intubated. We decide based on age and state of health.” —Christian Salaroli, Italian MD.
All of this is what drives a system to have a fatality rate of ~4% instead of ~0.5%. If you want your city or your country to be part of the 4%, don’t do anything today.
3. What Should You Do?
Flatten the Curve
This is a pandemic now. It can’t be eliminated. But what we can do is reduce its impact.
Some countries have been exemplary at this. The best one is Taiwan, which is extremely connected with China and yet still has as of today fewer than 50 cases. This recent paper explain all the measures they took early on, which were focused on containment.
They have been able to contain it, but most countries lacked this expertise and didn’t. Now, they’re playing a different game: mitigation. They need to make this virus as inoffensive as possible.
If we reduce the infections as much as possible, our healthcare system will be able to handle cases much better, driving the fatality rate down. And, if we spread this over time, we will reach a point where the rest of society can be vaccinated, eliminating the risk altogether. So our goal is not to eliminate coronavirus contagions. It’s to postpone them.
The more we postpone cases, the better the healthcare system can function, the lower the mortality rate, and the higher the share of the population that will be vaccinated before it gets infected.
How do we flatten the curve?
There is one very simple thing that we can do and that works: social distancing.
If you go back to the Wuhan graph, you will remember that as soon as there was a lockdown, cases went down. That’s because people didn’t interact with each other, and the virus didn’t spread.
The current scientific consensus is that this virus can be spread within 2 meters (6 feet) if somebody coughs. Otherwise, the droplets fall to the ground and don’t infect you.
The worst infection then becomes through surfaces: The virus survives for hours or days on different surfaces. If it behaves like the flu, it can survive for weeks on metal, ceramics and plastics. That means things like doorknobs, tables, or elevator buttons can be terrible infection vectors.
The only way to truly reduce that is with social distancing: Keeping people home as much as possible, for as long as possible until this recedes.
This has already been proven in the past. Namely, in the 1918 flu pandemic.
Learnings from the 1918 Flu Pandemic
You can see how Philadelphia didn’t act quickly, and had a massive peak in death rates. Compare that with St Louis, which did.
Then look at Denver, which enacted measures and then loosened them. They had a double peak, with the 2nd one higher than the first.
If you generalize, this is what you find:
This chart shows, for the 1918 flu in the US, how many more deaths there were per city depending on how fast measures were taken. For example, a city like St Louis took measures 6 days before Pittsburg, and had less than half the deaths per citizen. On average, taking measures 20 days earlier halved the death rate.
Italy has finally figured this out. They first locked down Lombardy on Sunday, and one day later, on Monday, they realized their mistake and decided they had to lock down the entire country.
Hopefully, we will see results in the coming days. However, it will take one to two weeks to see. Remember the Wuhan graph: there was a delay of 12 days between the moment when the lockdown was announced and the moment when official cases (orange) started going down.
How Can Politicians Contribute to Social Distancing?
If you’re a politician in a region affected by the coronavirus, you should immediately follow Italy’s example and order a lockdown.
This is what they ordered:
This is the least I would order. If you want to be safe, do it Wuhan style. People might complain now, but they’ll thank you later.
How Can Business Leaders Contribute to Social Distancing?
If you’re a business leader and you want to know what you should do, the best resource for you is Staying Home Club.
It is a list of social distancing policies that have been enacted by US tech companies—so far, 85.
They range from allowed to required Work From Home, and restricted visits, travel, or events.
There are more things that every company must determine, such as what to do with hourly workers, whether to keep the office open or not, how to conduct interviews, what to do with the cafeterias… If you want to know how my company handled some of these, along with a model announcement to your employees, here is the one my company used (view only version here).
It is very possible that so far you’ve agreed with everything I’ve said, and were just wondering since the beginning when to make each decision. Put in another way, what triggers should we have for each measure.
Risk-Based Model for Triggers
To solve this, I’ve created a model.
It enables you to assess the likely number of cases in your area, the probability that your employees are already infected, how that evolves over time, and how that should tell you whether to remain open.
It tells us things like:
The model uses labels such as “company” and “employee”, but the same model be used for anything else: schools, mass transit… So if you have only 50 employees in Paris, but all of them are going to take the RER, coming across thousands of other people, suddenly the likelihood that at least one of them will get infected is much higher and you should close your office immediately.
Are You Part of a Group of Leaders?
This math is selfish. It looks at every company’s risk individually, taking as much risk as we want until the inevitable hammer of the coronavirus closes our offices.
But if you’re part of a league of business leaders or politicians, your calculations are not for just one company, but for the whole. The math becomes: What’s the likelihood that any of our companies is infected? If you’re a group of 50 companies of 250 employees on average, in the SF Bay Area, there’s a 35% chance that at least one of the companies has an employee infected, and 97% chance that will be true next week. I added a tab in the model to play with that.
Conclusion: The Cost of Waiting
It might feel scary to make a decision today, but you shouldn’t think about it this way.
This theoretical model shows different communities: one doesn’t take social distancing measures, one takes them on Day n of an outbreak, the other one on Day n+1. All the numbers are completely fictitious (I chose them to resemble what happened in Hubei, with ~6k daily new cases at the worst). They’re just there to illustrate how important a single day can be in something that grows exponentially. You can see that the one-day delay peaks later and higher, but then daily cases converge to zero.
But what about cumulative cases?
This content was originally published here.