The update of analysis on the impact of social distancing and BCG vaccination on COVID-19 (as of Apr. 12th)

This article is an update of the previous analysis.

My basic idea is that if we combined the hypothesis of early quarantine (Eisuke's report) and BCG vaccination (Jun's report), it may explain the situation of coronavirus spreading well.  

Also, I would like to refer to the analysis conducted by Ryuji-san (Ryuji's report), which implies that COVID-19 with high mortality likely to have started from Italy and spreads to Europe and US. In my understanding, this analysis shows the importance of spread timing issue that each country has different timing for coronavirus to spread.

In this article, I would like to update the previous analysis to verify the combined hypothesis while taking the spread timing issue into account.

My philosophy on this analysis is as follows.
・The analysis should be verifiable in order for any of us to verify its validity and evolve these hypotheses.
・Data sources should be open and easy to access, and the process of analysis should be kept as simple as possible.

Firstly, I update the previous analysis. (Fig. 1)

The definition of each parameter and the data source are the same as the previous article. What I should mention here is that I picked up the 20 countries which are the top 20 in # of confirmed cases as of Mar. 28th for comparison, and I added Russia to this list because it's been ranked in top 20 recently (As of Apr. 12th)

The key finding that the correlation between the BCG vaccination policy and days of doubling in # of total deaths after social distancing seems to be valid.

As I mentioned in the previous article, we can well recognize the current situation of coronavirus spreading in each country in terms of the two key factors of "the timing of introducing national social distancing policy" and "BCG vaccination policy".

We can categorize the countries by these two factors. Firstly, I divide these countries into two by whether # of confirmed deaths of that day is over 50 (Later social distancing) or not (Earlier social distancing) Secondly, I divide these countries into two further by the BCG vaccination policy. (Mandatory or Not)
Category 1 (Later social distancing and NOT Mandatory BCG vaccination)
・・・Italy, Spain, France, UK, Switzerland
Category 2(Earlier social distancing and NOT Mandatory BCG vaccination)
・・・USA, Germany, Netherlands, Austria, Belgium, Canada, Norway, Australia, Israel
Category 3(Later social distancing and Mandatory BCG vaccination)
・・・Iran, Brazil, Japan?
Category 4(Earlier social distancing and Mandatory BCG vaccination)
・・・China(Hubei), China(w/o Hubei), South Korea, Portugal, Russia, Japan?

I categorize Japan in Category 4 in the previous article because Japan closed public schools from 2nd Mar. Unfortunately, social distancing in Japan might not be effective until Japan declared a state of emergency on 6th Apr. In that sense, it should be appropriate that we categorize Japan in Category 3.

I further investigate the coronavirus spreading pattern of each category by setting the key indicator as "doubling time" of # of deaths per capita and that of confirmed cases per capita. "Doubling time" means days of doubling in each parameter. 

I also set a "starting date" that means # of deaths per capita (per 100M) becomes a certain number. (In this analysis, this param is set as 15)

The reasons why I take "doubling time" as key indicator and why I set a "starting date" are as follows.
・Doubling time is comparable between countries provided that measurement method has not been changed in each country.
・The starting date of coronavirus spreading is different from countries to countries, it might be inappropriate to compare these params. on the same day. (for example, Mar. 22nd) We can compare these params on the same days from the "starting date". (for example, 10 days after from each "starting date")

[How to calculate the params]
▼Days after the "starting date"
"Starting date" of each country is the day when # of deaths per capita exceeds 15. Days in the graph shows # of days after the "starting date."
▼Doubling time in # of total deaths per capita
This means days in which # of total deaths becomes double. The larger it is, the slower the spread speed is. Each figure on a day is calculated using that of  -5 days and that of + 5 days.
It is calculated as 10 days divided by Log2(Y/X) where X is # of deaths per capita on -5 days (5 days before that day) and Y is # of deaths per capita on +5 days (5 days after that day).

This graph shows the doubling time in # of deaths per capita on days after "starting date" ( when # of deaths per capita exceeds a certain number)　（Fig.2)

This graph shows the doubling time in # of confirmed cases per capita on days after "starting date" ( when # of deaths per capita exceeds a certain number)（Fig.3)

I put these figures in another form where X-axis is doubling time in # of confirmed cases per capita and Y-axis is doubling time in # of deaths per capita. Each line shows the change from day 1 from the "starting date" until Apr. 12th.（Fig.4)

It is a bit busy, so I show this graph by category.

[Category 1 (Later social distancing and NOT Mandatory BCG vaccination)]（Fig.5)

Findings
・Both of doubling time in # of deaths and that in # of confirmed cases are gradually increasing in Spain, Italy and Switzerland. On the other hand, these figures struggle in France and UK.  
・Even though, doubling time in # of deaths increases, it takes about 30 days from the "starting time" to become over 10 days in doubling time in # of deaths.

[Category 2(Earlier social distancing and NOT Mandatory BCG vaccination)]（Fig.6)

Findings
・Due to the earlier social distancing, doubling time in # of confirmed cases is increasing, whereas # of deaths is around 3 to 8 (relatively low)

[Category 3(Later social distancing and Mandatory BCG vaccination)]（Fig.7)
I categorize Japan in category 3, because of ineffectiveness of quarantine until the state of emergency on Apr. 6th.

Findings
・Doubling time of # of deaths increases up to over 10 days  in Japan and Iran. (Japan in a shorter period and Iran in a longer period to become over 10 days in doubling time in  # of deaths)
・Japan is somewhat struggling with doubling time in confirmed cases recently.

[Category 4(Earlier social distancing and Mandatory BCG vaccination)](Fig.8)

The broader point of view on X-axis is like this.(Fig. 9)

Findings
・Doubling time of # of deaths increases up to over 10 days in around 15 days from the "starting date" in China and Korea.
・Doubling time in # of confirmed cases increases rapidly after around 15 days from the "starting date" in China and Korea.

Conclusion
・The combination of hypotheses of early quarantine (Eisuke's report) and BCG vaccination (Jun's report) explains the situation of coronavirus spreading well as of today. (See. Fig.1 )
・The earlier quarantine measurement becomes effective, the higher doubling time in # of confirmed cases. (See Fig.6  and Fig. 9)
・In the countries where BCG vaccination is mandatory, the higher doubling time in # of deaths realizes in early days from "starting date" (when # of deaths per capita exceeds a certain number)(See Fig.7  and Fig. 8)

Further discussion and analysis might be as follows.
・Medical verification should be conducted before any actions based on this hypothesis are taken.
・I only pick up the top 20 countries in terms of # of confirmed cases, but we can analyze more countries in the same way shown above.
・I don't include the medical capacity in each country, which could impact on # of deaths in a rapid spreading situation.
・I don't take into account the difference of BCG stains by country as mentioned in Jun-san's report. If we take it into account, we might get further implication on the impact of the difference of the BCG stains.
・It might be important when the mandatory BCG vaccination policy was introduced. The later it is, the more elderly people who haven't taken it, which in turn leads to higher risks to become critical.
・The situation in each country might depend not only on how early the national social distance policy is introduced but also on how strict it is. The further investigation can address this point.

Bunshiro Ochiai
I declare no conflicts of interest associated with this report.