How to decide when to take precautionary action about coronavirus (COVID-19) and what to do: an analytical model

I’ve been tracking and analyzing COVID-19 obsessively since it came to my attention in early January.

Coronavirus has caused the largest ongoing lockdown in human history — 60+ million people or as many as 760 million depending which source you trust. In tandem, the WHO and leading experts warn of high risk of pandemic, meaning epidemics will likely erupt in other countries.

Global demand for surgical masks has surged 100x, causing mass shortages, and some have been panic buying food and even hazmat suits.

Precautionary action makes sense if you believe a virulent epidemic will reach you, but how should one know if the virus is about to erupt nearby, and when one should take action?

Below, I present my general analysis, my precautions and two models, one for Singapore and another for the US.

My background

I’m a serial entrepreneur with a background in biology and am able to read scientific articles but I am not a medical expert.

That said, my strength for the past 14 years has been identifying trends early, using growth models to evaluate potential outcomes, and taking swift action.

COVID-19 is nasty and it will likely come to your country

Since emerging in China, COVID-19 has been found in people in dozens of countries. Also, researchers have shown that, as opposed to SARS or MERS, 50% of infections occur during a long asymptomatic incubation period of as many as 14+ days.

Even worse, the virus seems to survive for days on surfaces, making it harder to find the newly infected.

And COVID-19 is virulent. Experts report the following impact profile for the first ~70,000 COVID-19 cases: ~81% mild, ~14% severe (requiring hospitalization) and ~5% critical, with a 2%-3% Case Fatality Rate (CFR).

But not only is it coming to your country. It’s likely to cause an epidemic because of its high R0 (“R naught”), which represents the number of people each infected person will go on to infect and is expressed in the following formula:

R0 = transmission risk (p) x contact rate (γ) x duration (D).

If R0 is greater than 1, each infected person will infect more than 1 person, causing an epidemic. According to experts coronavirus’ R0 is between 1.4–4 and many believe it is higher than 2. This also implies a COVID-19 epidemic would erupt even based solely on the asymptomatic period.

For further discussion of the growth fundamentals, R0 and more please see Ed Baker’s fantastic post.

Here is an example assuming an overall R0 of 2.6 and therefore an “asymptomatic R0” of 1.3:

Assuming a first infected cohort of 10 people and an incubation period of 10 days, by the time the first batch of 10 people is symptomatic and easier to identify, there will be 13 new asymptomatic people who may be at large.

Given COVID-19’s survival on surfaces, to curb infections a country would have to round up every person who was in the proximity of any place an asymptomatic infected person went, even after the infected person had left.

This seems infeasible, even for China.

Also, all said areas would have to be thoroughly disinfected; China seems to be trying this and has even gone so far as to disinfect and/or destroy potentially tainted money.

Why all these lockdowns in China?

Without a vaccine on the horizon, and with the vicious growth channels above, China can only try to drive R0 under 1 by reducing transmission and contact.

Also, I imagine that at some point without city lockdown, R0 worsens as containment efforts are harder to coordinate and hospitals are overwhelmed.

In that scenario, the local healthcare system becomes an infection source and collapses. This hypothesis is supported by the fact that China locked down major cities every time the confirmed number of infected crossed a mere ~500.

The big question is: how many people were actually infected when cities were locked down with an officially reported number of 500 confirmed cases?

Because many infected people experience mild symptoms, it’s hard to know the true number, but we can imagine two possibilities.

Possibility 1: the infection had spread to many more than 500 when the confirmed number was 500 — implying that it’s very hard to know how many people are infected before a country is forced to take extreme measures. In this case R0 may be lower, but detection is harder.

Possibility 2: the total infections were quite close to the confirmed infections and the true R0 is greater.

My guess is that Possibility 1 is more accurate and the virus was widely spread before authorities started testing people. Remember that even severe cases may only surface as COVID-19 after the virus is widespread since early serious cases are few and may be misdiagnosed as flu or pneumonia.

Whichever turns out to be more accurate, I sadly predict that the Chinese lockdowns will continue for months, until the authorities believe no more than a small number of infected people are at-large when free movement is restored.

Keep in mind the following for all locked-down Chinese cities

• Thousands of infected dead may be in their homes, from which they need to be evacuated
• All infected residences must be thoroughly disinfected
• Diagnostics are in short supply, making it impossible to know who is actually infected
• Based on official confirmed diagnoses, infection is still ongoing, meaning it’s likely that there are still more than 500 asymptomatic infected at large
• The healthcare system is overrun with coronavirus

What about my country?

We know infected people have arrived in dozens of countries but we do not know how many asymptomatic people are at large.

Given the asymptomatic nature of contagion and the estimated R0, confirmed infection numbers will seem very small at first.

Specifically, if an average growth cycle takes 10+ days, numbers will stay small for 4–6 growth cycles, over 2–3 months. But because of the exponential growth of the virus, once it’s noticed it may be too late.

Therefore, as long as there are any asymptomatic cases in your country an outbreak may be hard to contain.

Singapore: A model country that is so far failing to control the virus

Singapore is the gold standard to watch because it is the first dense, developed, effectively governed country that is dealing with COVID-19. The country currently reports 75 confirmed infections and shares excellent data.

Disturbingly, 34 of the cases are Local; that is cases with no known link to other infections or to China.

Based on the numbers from the WHO and other reliable entities, Singapore infections are growing at a rate of 1.6x every 5 days and and 2.6x every 10 days.

Given this consistent growth, there must be many more infected at large. Singapore is playing a game of whack a mole to find these people and drive the R0 under 1.

Assuming a continued growth of 2.6x over a 10 day growth cycle and of 1.6x over a 5 day growth cycle, we will see Singapore at 1100+ infections by March 15th. In that scenario Singapore may be forced to enforce a lockdown, just like Wuhan.

When I would take action if I were in Singapore

• If by February 26th there are over 190 confirmed infections in Singapore I would take 2nd measures as noted below.
• If one week later, by March 4th, there are over 381 confirmed infections, I would take 3rd measures as noted below and self-quarantine.

The United States: how COVID-19 could play out

The US currently has 15 confirmed COVID-19 infections but what we want to know is the number of infected at large.

Let’s try to estimate this number.

Around 3M Chinese visit the US annually not counting other foreigners or American citizens who travel to China and return.

Before US Borders were closed to China travelers the virus was already prevalent for at least ~14 days during which infected people traveled freely.

That implies at least 115,068 visitors who had recently been to China entered the US from Jan 9th — Jan 23rd. To guess how many may have been infected I noted that China’s initial quarantine area encompasses 60M people which is 1/23rd of China’s population.

Therefore ~5,000 suspected persons arrived in the US from affected areas. Assuming the infection rate was around 0.5%, we can guess there were around 25 infected people at large by January 23rd. But which 25 of the 5,000??

Note: some infected may continue to come in from China but only if they are US permanent residents or citizens (and they are forcibly quarantined in military installations). The US does not currently limit arrivals from Thailand or Singapore however, as opposed to Israel, which was the first to announce such limits today.

In an assertive scenario the US government would track down all 5,000 suspected infections above as well as anybody they interacted with and put them all in quarantine.

I have heard no reports in that vein and believe it’s likely not the case. On the other hand, there have been reports of dangerous mistakes, such as incorrectly removing an infected person from a hospital in San Diego.

Even worse, we can’t tell how many people are now infected but don’t have a reason to suspect they have it. They may have been around an asymptomatic infected person who came from Wuhan or touched a contaminated surface.

If in fact COVID-19’s “asymptomatic R0” is over 1, a single infected person would start an epidemic cascade. I think it is highly probably this is happening now but that we are in the early stages of the exponential curve and therefore the problem is invisible.

Furthermore, because most cases are mild, infected persons may be misdiagnosed or not seek treatment. The virus may go through 5–10 growth cycles before we realize a city outbreak is underway.

Once it’s clear there is an outbreak in say Columbus, OH, it’s hard to say how many people the local infected persons have already spread the disease to and it will be hard to find them all.

That’s one reason why the CDC instructed medical professionals across the nation to report cases of flu-like symptoms to monitoring centers and also released a diagnostic test for COVID-19 that it’s distributing to 115 labs. However, the test was faulty and for now samples are still going to Atlanta, which means results will be delayed.

Sadly, a faster acting, reliable test that could be delivered to thousands of labs will take months to develop.

My take is that if the number of total infected (not merely confirmed infected) in a city exceeds a few hundred, that city is in trouble.

With 25–100 infected people at large who may be anywhere in the US we should be very concerned.

When I would take action

• If my city had more than 15 confirmed cases, I would take 2nd measures as noted below
• If my city had more than 20 confirmed cases, which grew to 30 cases or more in less than 10 days, I would take 3rd measures as noted below

A note on developing countries

Developing countries will have fewer resources to correctly identify infected people. Therefore they will report lower nominal infection growth numbers, understating the number of total infections by only focusing on confirmed infections.

But at a certain point, the numbers will skyrocket and quickly collapse the local healthcare system. This is the WHO’s nightmare scenario.

Precautions: 1st, 2nd and 3rd Measures

I’m currently based in San Francisco, a mere few blocks from UCSF, where two confirmed COVID-19 patients, diagnosed in Santa Clara, were transferred for treatment.

1st Measures (I’ve already taken these)

• Buy six weeks worth of food in case self-quarantine is necessary
• Cease attending local group events such as gyms
• Cease using public transportation which I would otherwise do every other day. Instead I use my car or Lyft/Uber
• Purchase $350 worth of masks for self and wife. Haven’t found a good solution for daughter
• Avoid air travel to East Asia
• Minimize air travel and when traveling, carry Lysol wipes to disinfect seating area, avoid using the toilet, wear a mask for the duration of travel. Avoid eating. Drink only from my own bottle.

2nd Measures

• Expand food stocks to 4 months
• Cease all air travel
• Cease eating out at restaurants

3rd Measures

• Self-quarantine for 2 weeks to observe whether things are static or worsen. If infected numbers are static, revert to 2nd Measures, otherwise keep self-quarantine.

Conclusion

Based on its impact profile and nefarious growth channels I am sadly convinced that the COVID-19 epidemic is likely not going to be confined to China.

Instead, it may be gathering steam and could become the worst pandemic in 100 years.

We should input available data into our models and watch things closely to identify when to take precautionary action.

If you found this article useful, please share it with other concerned people.

Also, if you have your own analysis or suggested improvements— please reach out!