Corona Virus Data Analysis

PHYS291 efo033 assignment

Spring 2020

PHYS291 Corona project by Espen Johannessen Folkedal



corona_header width=



1. Introduction
2. Codes
3. Plots and analysis
4. Deaths to infected ratio
5. Discussion/conclusion
6. References




1. Introduction

This spring the whole world turned upside down, when COVID19 was spread from Asia. Most countries shut down everything that was non-essential, and millions of people had to work from home. Sweden, on the other hand, did not. Schools were still open, people could go to open bars, and they did not have as many precautions as other countries. They have collected much crisism for this.

In this assignment, I want to look at data from Norway and Sweden, and compare them to one another. I want to see if the countries different approaches will have an effect on how many people got infected and how many died.

I collected data from the website:
https://github.com/CSSEGISandData/COVID-19/tree/master/csse_covid_19_data/csse_covid_19_time_series
Which is data from Johns Hopkins University (https://systems.jhu.edu/). They collect data from all over the world.

2. Codes

The codes I have made for this can be found on this website:
https://folk.uib.no/efo033/codes/ You will also find the datasets i have downloaded.

The codes have been run in root. I will specify in which order further down. The download files helped me download the files from the website to the computer. Afterwards, i used conf2data.C and death2data.C to transfer the data to a usable type (float).

3. Plots and analysis

3.1 Infected-plots


With the code plot_land.C i can plot the number of infected people for any country I want to. To do that, I have to run conf2data.C first, then run:
.L plot_land.C
plot_land("Sweden");
Then, my plot for Sweden comes:
Sweden_infected
Fig. 1: Shows the number of infected in Sweden from day 0 = 22nd of January.


Day 0 equals to the 22nd of January.I did the same for Norway:
Norway_infected
Fig. 2: Shows number of infected in Norway from day 0 = 22nd of January.


When you compare these data, it is important to know that Sweden has 10 million inhabitants, while Norway only has 5. I then divided the data with the population in millions. I then run plot_land("Sverige",10) and plot_land("Norway",5) in the same root session:


SwedenandNorway_infected
Fig.3: Shows Norway and Sweden with normalized data. Sweden is the triangle-shaped points, Norway is the line.


Sweden is the triangle-shaped points, while Norway is the line. You can see that the number of infected per million where higher for Norway at the start, but at around day 75 (7th of April), the Norway-curve starts to flatten out. This is a couple of weeks after the government decided to shut down most of the non-essential businesses, schools, etc... After this, Norways curve is quite flat, wich means that it's only a few new cases per day. The Swedish graph, on the other hand, continues linearly the whole time. This may be a result of the different precautions the countries made.

What we also should consider is that the number of tested people is not constant, and it is increasing almost every day. Therefore, the data in the early days might not be realistic, but it is what we have. We should consider that it is some unrecorded nubmers. But even then, we can clearly see that Sweden has a much higher number of infected per million than Norway at this stage.


3.2 Cases per day


With the code cases_day.C I can plot how many cases were registered each day for a country. To do that, I ran conf2data.C first, then:
.L cases_day.C
cases_day("Sweden")
The result was:
Sweden_daily
Fig. 4: Shows how many positive cases were registered each day in Sweden since 22nd of January


The same graph for Norway is:
Norway_daily
Fig. 5: Shows how many positive cases were registered each day in Norway since 22nd of January


We can clearly see that Sweden's infected rate is higher than Norways, and it's consistant at a high rate. Norways graph tells another story. When the outbreak started, the infected rate rose high, but at around day 75 (April 7th) the rate decreased. After ca day 105 (May 7th) there was no days with more than 50 cases registered. Sweden's numbers are consistently between 200 and 800 in the same period.

Norways highest number of cases is around 400 in one day, while Sweden has one day with around 2200 cases, and several with over 600. Clinical Effectiveness Research Group from UiO [1] is saying that Norway has tested more people than Sweden, which indicates that the difference might be even higher. Nettavisen, a news magazine from Norway, says that Sweden has increased the testing of COVID-19 the latest days, which correlates with the higher number for the last days [2].

3.3 Deahts

The number of registered cases might have some unrecorded numbers, because everyone is not tested every day. The number of deaths, however, is thought of to be more accurate.
To plot the number of deaths I used death2data.C, and then the file plot_deaht.C. For Sweden, the result looked like this:
Sweden_deahts
Fig. 6: Plot of deaths due to COVID-19 in Sweden since 22nd of January


Norway's plot looks like this:
Sweden_deahts
Fig. 7: Plot of deaths due to COVID-19 in Norway since 22nd of January


The normalized plot for both of them:
Sweden_Norway_deahts
Fig. 8: Normalized of deaths due to COVID-19 in Sweden and Norway since 22nd of January. Sweden has triangle shaped points.

Sweden again is the triangle shaped points, while Norway is the line. The plot shows clearly that Sweden have a linear increase in deaths, while Norway has managed to flatten out the curve to around 50 deaths per million inhabitant.

4. Death to infected ratio

A simple way to measure the mortality is to divide the number of deaths to the number of infected people. The last data that was sampled for Norway showed 8576 positive COVID-19 tests, and 239 deaths. For Sweden, there has been 45924 positive tests, and 4717 deaths. This gives the mortality rates:

MSweden = 4717 / 45924 = 10.3 %

MNorway = 239 / 8576 = 2.8 %


Sweden does have a significantly higher mortality than Norway, which the Clinical Effectiveness Research Group from UiO also concluded with [1]. They say that it's not possible yet to know if this has a direct coherence with the different ways the countries has handled this pandemic. Since the dataset is relative small, even one death can change the mortality rate quite much.

5. Discussion/conclusion

As the plots show, Sweden does have a much higher number of infected people per million inhabitants. The CERG-group [1] actually says that Norway has testet more people than Sweden, which even further strengthens the theory. It's very early at this stage to conclude with anything, but we can see a clear coherence with low emergency measures and the number of infected people between these two countries.

The plots of deaths due to COVID-19 also speaks to our favor. As Figure 8 shows, Sweden has about 10 times more deaths per million inhabitants. These data are more reliable, because you can't test everyone in the whole country of COVID-19, and therefore the death numbers are much more easy to track.

In every analysis of datasets there will be some discrepancies. One of them are the lack of a standarized way of testing. Every country has their own way of testing, i.e who should be tested, when, etc. Norway has recently started to test everyone who has small symptoms. That means that many more people are getting tested than before. This even further solidifies that Norways infected numbers are smaller than Sweden's.

At time of writing (14th of June, 2020), Norway is starting to open up for sports, gyms, bars, etc., so it will be very interesting to follow the development in both countries. The loosening of the emergency measures might have an effect on the spread of the virus. We saw a very discrete decrease of cases per day the weeks following the "shutdown" in the middle of march. With the society starting to come back to normal, people will naturally come closer tho each other, which may increase the spread of COVID-19.

6. Referanser

[1]

Kalager, Mette, et al. "Forekomst og Dødelighet av COVID-19 og sesonginfluensa i Sverige, Norge og Danmark", Klinisk Effektforskning, Universitetet i Oslo og Oslo Universitetssykehus, Oslo, Norge. Institutt for Medisinsk Biologi, Universitet i Tromsø, Tromsø, Norge


[2]

Lepperød, Trond. Nettavisen.no : https://www.nettavisen.no/nyheter/sverige-snart-50000-smittetilfeller---trapper-opp-provetakingen/3423980160.html . Downloaded 14.06.2020