WASH YOUR HANDS. STAY CALM. HELP EACH OTHER.

PRACTICE PHYSICAL DISTANCING. STAY AT HOME IF YOU CAN.

IT WILL BE OK IN THE END. IF ITS NOT OK NOW, ITS JUST NOT THE END.

Description of Analysis, Data Sources and Assumptions

GraphSnip

I have collected some of the numbers relating to the COVID-19 pandemic and created some graphs in an attempt to visualize what is going on, and how different parts of the world are coping. I use reputable data sources, but the analysis is mine and the comments represent only my opinions. Feel free to disagree. Leave a comment!

To help understand what these graphs represent, consider the analogy of a car:

Lets say you have a parked car that you want to race. When the car is stopped its speed and acceleration are both zero. This is like a region of the world before any COVID-19 infections have started.

When the car starts to move, the speed starts to increase, which is acceleration. This is like a region of the world where COVID-19 infections have started. The speed of the car is like the rate of new infections (measured as new cases per day).The faster the car increases speed, the greater the acceleration rate. This is like the growth of the number of new infections from one day to the next.

When the car reaches its maximum speed and starts to maintain a steady speed, the acceleration goes to zero but the car is still moving. This is like a region that has more or less the same number of new infections each day. There are still new infections, just not more new infections each day than the day before. 

To “flatten the curve”, the car needs to slow down to a steady rate such that, active infections (total cases less recoveries and deaths) reach a steady count that does not exceed the capacity of the health care system. This happens when that steady number of active infections does not exceed health care system capacity at the time stability is attained, or with a period of deceleration, where the speed decreases until the capacity of the health care system is no longer exceeded by the number of active infections. Deceleration can be fast or slow. When deceleration happens in a region, the count of new infections every day begins to decrease, finally coming to near zero new infections every day. That is the goal – zero new infections each day, but the curve can be considered flattened when the number of active infections does not exceed the capacity of the health care system. This analysis reports the number of active infections (see Recovery charts), but does not attempt to assess if the capacity of the health care system has been exceeded.

To describe the state of some of the regions as of the end of March:

  • The Chinese car is slowing down and nearly stopped.
  • The Italian car is nearly at its top speed, but still gaining speed.
  • The USA car is accelerating rapidly, not yet at its top speed.
  • The BC car is at a relatively constant speed.
  • The Costa Rican car has just left the starting line.

Fairly simple to understand right?

To further this analogy, imagine that the car from each region has a different size engine (horsepower). This is like the population density of each region (people per square kilometer). And we want to compare each region’s car against China’s car. This seems useful because China was first region to be impacted by COVID-19 and we all saw the response and its impact on the news. By comparing a region with China we can measure how that region is doing, and what might yet happen in that region. In order to make a meaningful comparison I did two things:

  • I put each car on the starting line at the same time. This is done by moving the regional data backwards in time to match up the case 1 date with China’s case 1.
  • I adjusted the horsepower of the Chinese car to match the horsepower of the regional car. This is done by scaling the reported case counts from China proportionally to the population density of the region.

OK, that was a bit more complicated, but the idea is to make the cars start at the same time, with the same size engine, then see how the regional car does against what we know already happened with the Chinese car. Its a simple idea. Not perfect, but simple.

The comments found below the graph for each region provide the speed (new cases each day), and acceleration or deceleration (change in the number of new cases) and reference this “car” analogy to make it easier to understand what the numbers mean.

I chose a particular set of regions to analyze:

  • China – as the starting point for this virus and the first worldwide experience of this outbreak. China seemed like a good region to compare to.
  • Canada and BC – because I live in BC, which is a province of Canada
  • Netherlands – because I have family there
  • Costa Rica – because I have family there too
  • Italy – because of the tremendous impact this is having there
  • USA – because anything of this scale happening in the USA will undoubtedly impact the rest of the world
  • Sweden – because they are less strict about locking down, depending on their people to take the right measures without needing the enforcement.
  • World – because its the whole world!

I only use reputable data sources as listed below.

  • World Health Organization – for Case and Death counts in China and the world. Updated with a situation report once per day. I note that the Canada numbers reported here are a day or two behind Canada CDC and are likely a day or two behind most areas of the world because it takes time to collect and report. There was a jump in reported cases mid February which reflects a change in the way the WHO collected and reported data.
  • Canadian Public Health – for Case and Death counts in Canada. Updated several times per day.
  • USA Center for Disease Control – for Case and Death counts in USA. This site is updated every weekday, but not always on weekends, which seems strange to me.
  • British Columbia Center for Disease Control – for Case, Death, and Recovery counts in BC. Updated several times a day.
  • CTV National News at 11 – for Recovery counts for Canada and the World. I note that CTV is consistently reporting numbers higher than the WHO for Cases and Deaths in Canada and the World. Perhaps they have earlier access to WHO information or have access to potential cases that I do not.Perhaps they are surveying the CDC in each country.  I use published WHO data for Cases and Deaths, but could only find World and Canadian Recovery counts on CTV National News
  • RIVM Netherlands for Case and Death counts in the Netherlands.
  • Johns Hopkins GIS App for recovery counts if I cannot find them on region specific CDC sites

Since some of these sites do not publish their own historical data, and since I only started collecting systematically in early March, I used the Way Back web site to collect some of the historical data. Not every day was captured by Way Back so there are some gaps in the data which make some of the graphs a bit “choppy” in January and February, but I did get historical data at least weekly using that approach.

Note that I collect data at the end of the day in BC, and verify or adjust them with early morning reports the next day. I update this page in the morning PDT. This analysis is only as good as the data behind it. Given that not all symptomatic people are actually tested, there is an inherent flaw in the data. Impossible to know what that is or even estimate it.

These are the major assumptions and formulas I used in these graphs:

  • I calculated the number of active infections by subtracting Deaths and Recoveries from Cases.
  • Formula: Infections = Cases – (Deaths + Recoveries)
  • I measured the number of days from case 1 in China to case 1 in other regions. I use this to slide the regional data backwards in time to align with case 1 in China. It ranges between 22 and 65 days difference in the date of case 1.
    Formula: Adjusted_Case_1_Regional_Date = Date_of_Case_In_Region – (Case_1_Date_In_Region – Case_1_Date_In_China)
  • Given that different regions have different population densities, which would likely impact the infection rate, I looked up populations and land mass areas to create population density numbers (people per square km). I calculated the population density ratios between China and other regions and used that to scale the Chinese data down in a linear fashion. This is a major simplifying assumption which undoubtedly introduces error, but I believe it brings the analysis closer to the truth than simply ignoring population density. I used Google to find populations and land masses.
    Formula: Adjusted_China_Case_Count = ( China_Case_Count ) / ( (China_Population / China_Land_Mass) / (Regional_Population / Regional_Land_Mass) ).
  • Note that different data sets are reported in different time zones so the data may be off by a day or so in some cases.

The following data has been used to display or adjust the regional data in order to compare it to China.

Using this basis I created the following charts. They are updated once a day (usually in the morning PDT) and should be normally one day behind today’s date.

Quick Links

07-Apr-20 * New Today *

  1. All regions published data today. 
  2. The most interesting graphs are the recovery graphs. Should have more of them in the next couple of days. BC is flattening!
  3. Taking most numbers from Johns Hopkins as they appear to be more current than WHO.

All Regions

07-Apr-20 Comment: This clearly shows the impact on the Italian population. USA, Netherlands, and Sweden are the next highest per capita. China’s case count, while high in actual numbers (80,000+), is relatively low per capita. Worldwide mortality rate for infected people ranges from 0.4% in Costa Rica to 12.6% in Italy.

World Case Counts

Comment: In the early part of the graph its mostly happening in China. In mid March is where it hit the world

07-Apr-20 Comment: The World reported 84,189 new cases in the last day, 60% less than the prior day which had 208,149 new cases. The number of new cases is not too high over the last 4 days (an average of 15.36 new cases per million people per day). In the last 4 days, the speed of the World’s car is not too high and it is slowly accelerating. In the World the number of active infections is increasing over the last 4 days (an average increase of 89,746 active infections each day), with a current total of 1,044,177 active infections. The recovery curve is starting to swing up to follow the case curve with the expected two week delay between diagnosis and recovery, but it is not yet tracking well to new cases.

China Case Counts

Comment: This shows the early exponential rise in cases in China in January and February, followed by the flattening out caused by the massive response.

07-Apr-20 Comment: China reported 66 new cases in the last day, 12% less than the prior day which had 75 new cases. The number of new cases is almost stopped over the last 4 days (an average of .05 new cases per million people per day). In the last 4 days, the speed of China’s car is almost stopped and it is very slowly decelerating.

Netherlands compared to China

Comment:  Slower initial rise in the Netherlands which is great to see!

07-Apr-20 Comment: Netherlands reported 906 new cases in the last day, 5% less than the prior day which had 952 new cases. The number of new cases is high over the last 4 days (an average of 59.07 new cases per million people per day). In the last 4 days, the speed of Netherlands’s car is high and it is slowly accelerating. However, the acceleration slowed slightly over the last two days.

Italy compared to China

Comment: This shows an initial outbreak in Italy very similar to that in China, with a slower initial spread.

07-Apr-20 Comment: Italy reported 6,638 new cases in the last day, 54% more than the prior day which had 4,316 new cases. The number of new cases is very high over the last 4 days (an average of 82.36 new cases per million people per day). In the last 4 days, the speed of Italy’s car is very high and it is slowly accelerating.

Sweden compared to China

Comment:  The initial rate of spread is slow compared to China.

07-Apr-20 Comment: Sweden reported 863 new cases in the last day, 123% more than the prior day which had 387 new cases. The number of new cases is high over the last 4 days (an average of 51.44 new cases per million people per day). In the last 4 days, the speed of Sweden’s car is high and it is slowly accelerating.

Costa Rica compared to China

Comment:  The initial rate of spread is slow, very similar to early days in China, but there is a 65 day difference in the date of the first case.

07-Apr-20 Comment: Costa Rica reported 16 new cases in the last day, 69% less than the prior day which had 51 new cases. The number of new cases is low over the last 4 days (an average of 5.85 new cases per million people per day). In the last 4 days, the speed of Costa Rica’s car is low and it is slowly accelerating. It is early in the infection cycle here, but looks promising.

USA compared to China

Comment: This would seem to show that early efforts delayed the initial spread in the USA, given more warning than China. However, it also could be an indication of a lack of testing or reporting. But the subsequent steep curve reflects the somewhat disjointed/delayed containment efforts in the USA once the virus had a foothold.

07-Apr-20 Comment: The USA reported 31,500 new cases in the last day, 47% less than the prior day which had 59,897 new cases. The number of new cases is very high over the last 4 days (an average of 122.68 new cases per million people per day). In the last 4 days, the speed of the USA’s car is very high and it is rapidly accelerating. The USA now has more than four and one half times the number of reported cases as China.

Canada compared to China

Comment: Similar to the USA, early efforts seem to have delayed the initial spread given that North America had more warning than China. 

07-Apr-20 Comment: Canada reported 1,230 new cases in the last day, 6% more than the prior day which had 1,155 new cases. The number of new cases is high over the last 4 days (an average of 35.87 new cases per million people per day). In the last 4 days, the speed of Canada’s car is high and it is slowly accelerating.

Canadian Recoveries

07-Apr-20 Comment: Canada reported 1,230 new cases in the last day, 6% more than the prior day which had 1,155 new cases. The number of new cases is high over the last 4 days (an average of 35.87 new cases per million people per day). In the last 4 days, the speed of Canada’s car is high and it is slowly accelerating. In Canada the number of active infections is increasing rapidly over the last 4 days (an average increase of 890 active infections each day), with a current total of 13,466 active infections. The recovery line had a sharp uptick in the last day, resulting in a downturn in the active infection line. This is good news. Stay inside!

BC compared to China

Comment: Similar to Canada and the USA, early efforts seem to have delayed the initial spread given that North America had more warning than China. The subsequent upward curve looks somewhat promising in that the spread is slower than that in China. 

07-Apr-20 Comment: BC reported 25 new cases in the last day, 32% less than the prior day which had 37 new cases. The number of new cases is low over the last 4 days (an average of 5.86 new cases per million people per day). In the last 4 days, the speed of BC’s car is low and it is slowly decelerating. This is good news. Stay inside – it’s working!

BC Recoveries

07-Apr-20 Comment: BC reported 25 new cases in the last day, 32% less than the prior day which had 37 new cases. The number of new cases is low over the last 4 days (an average of 5.86 new cases per million people per day). In the last 4 days, the speed of BC’s car is low and it is slowly decelerating. In BC the number of active infections is declining very slowly over the last 4 days (an average decrease of 6 active infections each day), with a current total of 443 active infections. BC is showing an excellent recovery cycle – the recovery line is tracking well to the infection line, with about ten days in between them. The active infection count is reasonably stable (and in fact declining slightly) which also implies that the demand on the BC health care system is reasonably stable. This is very good.

This Post Has 15 Comments

  1. Naadia Clayton

    Thanks Rob……very easy to comprehend 😀

  2. John Vanspronssen

    Very interesting Rob, glad to live in Canada not the USA

  3. Lisa Woudzia

    Thank you, Rob! Very informative and comprehensive!

  4. Paul

    Nice work Rob! Keep it updated!

    1. Rob vanSpronssen

      Thanks Paul. I’m trying to update it every morning (Pacific Time).

  5. Angela Termarsch

    Wow, now there is some interesting data! Thanks for the info and will check it to see how we are doing as the weeks go on..

  6. Wendy Laverock

    You really know your stuff Rob, this is interesting and well informing. Thanks and stay safe, love to you and Sherry and the boys💕💕

    1. Rob vanSpronssen

      Thanks Wendy. Our best to you and your family. Stay safe.

  7. John Ortynsky

    Thanks Rob,
    This helps put things in perspective.Im now working from home but Cherryl has to work with a much higher risk level so it’s very much top if mind for us.

  8. Tina van Koll

    Wow!! 😍 This is so interesting and detailed and well thought out. Thanks!!

Leave a Reply