# Coronavirus : Update 14 from the Epicenter

Let’s kick off with an adjustment to Update 13 in which I calculated the total number of people in Italy with immunity based (1.7M) on the projected final death toll (24k). On the positive side, the last two days numbers show a steeper curve than I thought, which should mean slightly lower final numbers :

On the other hand, I didn’t consider the fact that not every Covid-19 death may not be counted in the official numbers. As in most countries, deaths outside of hospitals are generally not attributed to the coronavirus and that may bump the official numbers significantly. I can’t find any official estimates, but there is plenty of anecdotal evidence that we may be talking about a 50-100% increase over the reported numbers.

“There are significant numbers of people who have died but whose death hasn’t been attributed to the coronavirus because they died at home or in a nursing home and so they weren’t swabbed. There were 164 deaths in Bergamo in the first two weeks of March this year, of which 31 were attributed to the coronavirus. That compares with 56 deaths over the same period last year.”

Giorgio Gori, mayor of Bergamo

A confirmation of the exact figure here would be far more important in determining immunity in the general population than slight adjustments to curve projections, and as we’ll see later on in this post, it has a significant (but counter-intuitive) effect on the results of any antibody testing / certification that we perform over the next weeks.

The UK’s Daily Telegraph has a useful article that outlines the 4 possible lockdown exit strategies :

**Intermittent social distancing**

We discussed this in Update 6, using this picture from Imperial College, and this remains the most likely scenario. The length of the periods will depend on adherence to social distancing protocols, border control to avoid an influx of new infections, and possible mitigation of symptoms using treatments that we will discuss in a future update :

**Seek and destroy**

This strategy involves waiting until the virus is at a very low level, lifting restrictions and combating the remaining cases through aggressive contact tracing, testing, isolation, and precision quarantine zones. The method has been adopted successfully in Singapore and South Korea, which have coped far better than most other countries, and have done so without major lockdowns. However, the bigger and more democratic the country, the more difficult such as policy will be to implement and enforce.

**Wait for vaccine or treatment**

This is probably the only definitive solution, however there are a couple of major issues :

- Despite unprecedented international cooperation, novel manufacturing techniques from companies like Centillion Technologies, and promises from billionaires like Bill Gates to
*“write checks so that progress is science limited rather than cash limited”*we’re probably still 18 months away from being able to immunize the world. That time-frame could be shortened if governments are willing to cut corners on the normal safety testing procedures - The lack of antibody build-up in younger survivors I reported in Update 13 does not bode well for the effectiveness of a future vaccine which typically provoke a weaker immune response compared to live virus.

**Allow the immune out**

Shielding the vulnerable, and allowing the immune out to carry on with their lives and get back to work, could be a way to break the deadlock and allow major parts of society to return to normal.

The UK Government has promised that antibody tests – which show whether someone has had the virus and is now immune – will be available within weeks, and Britons could soon be issued “immunity certificates” that would allow them to leave the lockdown.

If a large proportion – somewhere around 60 per cent – of the population become immune, the lockdown will not serve that much more purpose as herd immunity will be at a level that the disease would struggle to spread. So, lockdown over.

More than 60 companies have antibody tests on the market and approved for use in the EU.

Antigen detection tests detect viral components present during the infection in samples like nasopharyngeal secretions. Antibody tests detect the antibodies that later appear in serum as part of the immune response against the virus.

https://www.ecdc.europa.eu/sites/default/files/documents/Overview-rapid-test-situation-for-COVID-19-diagnosis-EU-EEA.pdf

But before we all rush out to get tested, it’s *critical *to understand that despite the accuracy of these technologies, you are much more likely to be negative if you test positive. The reason behind this is statistical, and counter-intuitive, but exactly correct. Let’s dig deeper. I’m using @taaltree’s tweets as a guide, and so he can kick us off :

**Sensitivity 93.8%:**Probability person tests positive if they are infected**Specificity 95.6%:**Probability person tests negative given that they are negative

93.8% and 95.6% sound good. They mean that only 6.2% of positive people test negative and 4.4% of negative people test positive. So far so good, but what does it actually mean to test positive?

Since the sensitivity is high, you’d assume that you’re positive and immune to COVID-19. But sensitivity measures the probability of testing positive given that you are positive. We want the probability of *being *positive given that you *test *negative. The conditionality is reversed, and (unlike multiplication) you can’t just swap the factors around. Taal explains using Bayes rule:

The probability you are positive (are+) irrespective of how you test, is the same as the proportion of positives in the population. In Update 13 we calculated that to be 3% for Italy, but we’re going to bump that to 5% given our intuitions about the true numbers of deaths.

The probability you test positive must be decomposed into the people who test positive and are positive plus the people who test positive and aren’t. We have all the necessary numbers for this probability expression and get a denominator of 0.0887 :

If we plug all that into Bayes rule, then we get a probability that you *are* positive if you test positive of 53%. That means it’s more likely that you are negative if you test positive!!!

…and it’s only that high because we used 5% for total infected population. If we trust the official numbers and use 3%, then the probability of being positive if you test positive drops to 40%. So for anyone rushing out to get a test so they can paaaaaarty should think twice. In fact they should get a second test if the first one is positive – just like you do if you test positive for HIV.

For everyone reading this who calls bullshit, I’d like to use a series of pictures from @pesasfi that illustrate the concept more clearly :

And I’d like to finish with a quote from an inspirational ultra athlete, who urges his subscribers to practice the following rather than ‘social distancing’ :

Physical distancing while maintaining your important social connections

@RichRoll

The power of figures ?.

In the age of ‘Closed Loop Analytics’ we still think winning a lottery is a good chance while we do not realize the danger around us. A wiseman should argue it is all about evolution, but we wouldn’t be here if we had not protected the weakest (and the dummies…).

Just for your information: Police on helicopter found out a group of people barbecuing on the roof of a building yesterday in the name of “ drinking will save us all”…

Great stuff but I now need an aspirin for my newest headache.

I found an interesting graphic showing how Covid-19 actually works.

https://www.azuravesta.com/covid-19-pandemic?cmpid=nl_morning24

Andrea