COVID-19: European virus strains kill (and probably spread) faster than Asian ones, but European bodies seem less vulnerable; is this virus a welcome tool for white supremacists?

There are major differences in both fatality rate and the distribution of probability of death over time between South Korea and Switzerland. This clearly shows that either the virus strains in these two countries do not have the same impact on human bodies, or that Swiss bodies are less vulnerable to the virus, or (probably) both. This might be related to the reported higher mortality found in indigenous people in the Americas (who share the same origin with some Asian ethnic groups), and maybe also of African Americans. All these results are still preliminary, but the belief that people of European descent are less vulnerable to the virus is already widespread among white supremacists. It has certainly contributed to make the concept of herd immunity popular; some people consider that a higher mortality among non-whites could help them to achieve their ideological objectives. If there is indeed a difference in mortality according to ethnic origin, massive international pressure in favor of efficient containment of the virus especially in certain states in the Americas is required in order to prevent ethnic cleansing. Unfortunately, most of the academic community has decided to ignore the excellent South Korean data and to make the whole topic of functional differences between virus strains and different vulnerability according to ethnic origin a taboo.

Warning: this is work in progress. There are many possible ways to analyze and interpret the data used here. Please consider the results presented here as preliminary.

Abstract

The COVID-19 case fatality rate (fatality rate among people with symptoms due to COVID-19) and infection fatality rate (fatality rate among all people infected with COVID-19) have been controversial topics among experts. Estimates range from 0.1%, i.e. corresponding to a bad flu season, to 2% and above, i.e. similar or worse than the Spanish flu. Another important factor is the delay between infection and the peak of mortality. Both factors have generally been considered to be constant across all virus strains for people of all ethnic origin. This paper shows that this belief must be abandoned if we want to get a good understanding of the virus and the epidemic it causes. It is not only a question of containing the various outbreaks across the world. Understanding how far-right movements use this virus to achieve their ideological objectives is at least as important, and the topic of the present paper is directly related to various white supremacy movements, in particular in the Americas.

The chart below shows the number of confirmed COVID-19 cases in South Korea (light blue area), the hypothetical number of infections (dark blue line), the daily number of deaths (black bars) and the 7-day moving average of daily deaths (grey line). Data after April 9 is from Hubei (China) or hypothetical.

Based on the data from the chart above, we can calculate the daily probability of death (blue line below, with some artifacts due to the processing algorithm), which is smoothed for practical reasons (orange line). If we use the data represented by the orange line below to predict the number of daily deaths from the daily number of infections, we get the purple line above, which fits quite well to the actual daily number of deaths.

The chart above shows that the probability of death peaks 34 days after infection in South Korea. We can see a similar pattern in the earlier outbreak in mainland China, but because testing capacity was insufficient during the early phase of the outbreak, the Chinese data is not suitable to calculate the precise evolution of the daily probability of death. The long delay between infection and the peak of probability of death can explain why the virus was initially underestimated: initially, it seemed to cause relatively few deaths, and when more people started dying, a huge number was already infected.

From the not very accurate Chinese figures, and in the second half of March from the South Korean figures, we expected a similar evolution in Europe, but it did not happen. After the number of new cases dropped due to the lockdowns, the number of daily deaths dropped with only one week delay as compared to the drop in daily confirmed cases, which in turn follow the number of new infections with a delay of one week or more. This was unexpected. The chart below shows this graphically. If we apply the daily probability of death calculated from the South Korean figures to the Swiss daily confirmed cases (with an additional delay of 7 days), we get the grey line, which does not fit at all with the orange line showing the actual daily deaths. Switzerland was selected because the Swiss data is relatively reliable and uses the same age groups as South Korea.

In Switzerland, death occurs not only much earlier after infection and drops off earlier, fatality rates (after correcting for underascertainment) are also lower than in South Korea within the same age groups. This contradicts at least some aspects of the consensus among Western experts that all the virus strains are functionally identical and that bodies of all ethnic origins are affected in the same way by the virus.

The article below shows that much of the data we have got points to the fact that (some) Asian people are more vulnerable to this virus, and that the virus strain(s) which caused the outbreak in Wuhan was not very well adapted to European bodies. On the other hand, the Italian strain which emerged from it was better adapted to Western bodies and also replicates/spreads much faster. Without taking into consideration these factors, it is almost impossible to explain much of the data we have got. Calculating the daily probability of death is not complicated, and doing it with South Korean and Swiss data will immediately show a major discrepancy. Unfortunately, nobody within the academic community seems willing to calculate this and to find an explanation for the discrepancy between the distribution of daily deaths in Asia and in Europe.

More generally, Western experts have not been able to put together a coherent picture of this virus, based on all the data provided by reliable sources. This is certainly one of the reasons why conspiracy theories are experiencing exponential growth in many countries in Europe and North America. This article is part of a whole series to put things together, based on all the most reliable data sources and using relatively simple analysis methods, to bring some order into this information chaos.

Introduction

The COVID-19 case fatality rate CFR (fatality rate among infected people with symptoms) and infection fatality rate IFR (fatality rate among all people infected with the virus, with or without symptoms) have been controversial topics among experts. Estimates range from 0.1%, i.e. corresponding to a bad flu season, to 2% and above, i.e. similar or worse than the Spanish flu. The prevalence of asymptomatic cases plays an important role in this considerable range of uncertainty. Another important factor is the delay between infection and the peak of mortality. Both factors have generally been considered to be constant across all virus strains for people of all ethnic origin. This paper shows that this belief must be abandoned if we want to achieve a good understanding of the virus and of the epidemic it causes. This topic is especially relevant for understanding the role this virus plays in the discourse and plans of white supremacists and other far-right movements.

The method presented here to calculate case fatality rate CFR/infections fatality rate IFR in South Korea was originally developed in early April, but never published previously. The distribution of daily probability of death after infection was also calculated at that time. More data is available by now which could make solving this problem much easier. The method presented here is still interesting as a case study of real time data analysis in the middle of an outbreak. In late April I realized that the figures from South Korea did not fit with the European data at all. Not only were the case fatality rate CFR and the infection fatality rate IFR much lower in Europe, especially if we take into account the age distribution. What is more, in South Korea, the peak in death occurs 34 days after infection, whereas this happens much sooner in Switzerland, roughly 14–20 days after infection. Switzerland has been chosen as basis of comparison for Europe because it provides detailed data of good quality and uses the same age groups as South Korea. I present these results here even though they rely on outdated data because this is important, but the academic community does not seem to be aware of it. All the calculations will have to be done again with other algorithms and the most recent available data, but this will take time. Therefore, please consider this as work in progress.

The data

The main data used here is the number of confirmed new cases per day (“cases”) and the number of daily deaths (“deaths”). Cases for South Korea were downloaded from the COVID2019app.live Google Doc (1) until March 20. Cases and deaths after this date were downloaded from the South Korea CDC COVID-19 updates (2), which were also used to verify samples of the previous source. For China, in order to have a homogeneous dataset with the sharpest possible peak in new infections, we use data from Hubei province only, provided by the COVID2019app.live Google Doc for convenience and from the National Health Commission of the PRC (3), in the same way as for South Korea. For Hubei, we use only the evolution of deaths data in time, no cases data.

An important number of press reports, internal reports and testimonies through direct contact from various Asian countries was used to get a clear picture of how this data was collected, in particular with regards to testing, i.e. the procedure which provides data about new confirmed cases.

In South Korea, the various outbreaks they had were all successfully contained by mainly using testing-isolation and contact tracing (subsequently called testing-tracing) (4) (5) (6). This requires finding the huge majority of all cases, including as many cases as possible during the presymptomatic stage and also asymptomatic cases, by extensive testing and performing systematic contact tracing for every positive case. Even if a case is missed, if in turn it infects other people, this person will probably be found later on through the contact tracing of the people which got infected. Since this strategy has successfully contained all the outbreaks so far, we can safely assume that the number of confirmed cases is reasonably close to the real number of cases and even to the total number of infections. South Korea has always included asymptomatic cases into their count of confirmed cases. This allows us to measure directly infection fatality rate IFR. Generally speaking, South Korean data is of excellent quality and deserves to be used more extensively. For China, we must assume significant under-reporting for both cases and mortality during the first weeks of the epidemic; as soon as testing capacity was up to the task (second half of February), the data is of good quality, but much of the testing was done in catch-up mode. The results obtained here show that this assumption is reasonable.

Calculating the daily probability of death for South Korea

The following chart represents the raw data at the heart of the present paper. The number of confirmed new cases is represented with a different scale (on the left) than the daily deaths (right-hand scale). If we assume an IFR of 2%, a certain level of cases on the left scale corresponds to the same level of deaths on the right scale. This might help to get a grasp of the proportions. A curve representing a 7-day moving mean was added to help visualize the widely varying deaths figures. The value of 7 days for the moving means was chosen because official figures tend to be lower on weekends; a 7-day moving mean smoothens the weekly cyclical variations.

Figure 1: South Korea: daily new cases, daily deaths and 7-day moving means

The first thing to mention is that with these figures, as of April 9 0:00, South Korea’s naïve IFR is standing at 1.96%, i.e. well above the IFR value of 1% or below which is generally accepted by Western experts. The naïve IFR (nIFR) is obtained by dividing the number of deaths as of a certain date by the number of cases as of this same date. Since deaths tend to occur later in the course of illness than the diagnosis, nIFR will tend to increase over time. We will see below that under certain conditions, more reliable figures for the number of infected people can be obtained by antibody studies and more reliable figures for the number of deaths by looking at excess deaths. However, as explained above, we can assume that for South Korea, both the number of cases and the number of deaths are quite accurate.

The evolution shows that the mortality from the spike in cases peaking in the first days of March has not entirely subsided yet (as of April 9), but the peak is over. The question is now: after the mortality from the sharp peak has entirely subsided, at what value will IFR stand? This is the first question (“?#1” below). The second question is: after the sharp peak, there have been roughly 100 cases per day for a few weeks. How fast will mortality subside after this stops? This is the second question (“?#2” below).

Figure 2: The two missing data sets to calculate the total CFR/IFR

Once we will have found the total IFR for South Korea by estimating questions #1 and #2, we must keep in mind that there is no significant under-reporting; asymptomatic cases were found very efficiently and counted in the number of cases. We will therefore directly get a reasonably good approximation of the IFR: if somebody is infected, this is roughly the chance of dying from it. And this IFR will have been calculated for a country with a much lower share of elderly than what we have in Western countries and with an abnormally high number of cases in the age group 20–29 due to fast spreading in the Shincheonji sect which attracts many followers of this age. What is more, in South Korea, there was no overwhelming of the healthcare system, i.e. every patient got optimal healthcare. Figures for IFR which are provided by influential epidemiologists are generally below 1%. We will see that when answering the two questions above, we will find a value of 2.58% for South Korea. If we take the fatality rate in each age group and adapt it to the age distribution in Western countries (Europe and North America), we get an IFR above 5%. However, this is not what we actually observe in Western countries. It is time that epidemiologists have a look at this discrepancy.

How can we estimate the future mortality in South Korea to answer questions #1 and #2? On March 29, when the peak of mortality was finally visible, an overlap with Chinese mortality data for Hubei was performed for the first time, in order to use the Hubei data for the prediction of future mortality in South Korea, assuming that the daily number of deaths in Hubei and South Korea show a similar evolution. Since then, the overlap had to be adjusted by plus minus one day in the light of new South Korean data. The chart below shows in red the Hubei data. The number of deaths and the moving mean have been extrapolated accordingly into the future.

Figure 3: Completing the future mortality data from Hubei (China) data

In China, due to the extremely tough measures taken against the outbreak, after a spike of new infections at the end of January, the number of new infections decreased very fast. In South Korea, measures were much less radical, which has led to roughly one hundred new cases per day even after the initial outbreak was brought under control. After the peaks, whose amplitudes were adjusted to fit, the decrease for South Korea based on Hubei data has to be adapted to tend asymptotically to the hypothetic mortality rate corresponding to these 100 cases per day. All we know is that this mortality rate is somewhat above the 1.96% mentioned above. The value used here of 2.11% was determined empirically so that after the end of the Hubei data series, the naïve IFR (nIFR) remains stable at this precise value. This can be considered the answer to question #1. This is the reason why in the chart, the red line tends towards zero, whereas the grey line (7-day moving mean) tends towards 2.11 deaths per day (corresponding to 100 new cases per day).

In a second step, we will use the number of deaths extended into the future to calculate the probability of deaths for each single day for each single case after infection occurred. Based on media reports (7) (8), a hypothetical curve of daily new infections was generated, taking into account the following constraints:

• Patient #31, believed to be at the origin of the outbreak in the Shincheonji sect, developed a fever on Feb. 10, after being hospitalized for a minor car accident, in a hospital where several cases of COVID-19 were later found. We assume that she started to infect other members of the sect from Feb. 12 onward.
• On Feb. 18, patient #31 was tested positive, and the contact tracing team in charge realized that she could have infected other members of the sect, therefore measures to isolate members were taken, leading to a fast decrease in new infections.
• By Feb. 22, 433 cases had been diagnosed, which means that substantially more people must have been infected a couple of days before, since most infected people will not test positive from day one on.
• Since we assume that the probability of daily death is a smooth curve and does not consist of isolated spikes, the curve of daily new infections must be considerably more narrow than the curve of the peak of mortality, which are quite similar in shape for South Korea and China (with more irregularity for South Korea due to the smaller number of deaths).

The resulting generated hypothetical curve of daily new infections is shown in blue in the chart below. In a second step, the mortality data must be smoothed, otherwise the algorithms used to calculate the probability of deaths for each single case will give erratic results. For this purpose, three curves (thin green lines below) were added to fit the mortality curve. Two of them were manually fit to the two separate peaks in mortality. They are lognormal curves with σ=0.5, μ=0. They are fit to the deaths due to the sharp peak of cases in the second half of February. A third curve corresponds to the deaths due to the background of 100 cases per day; it tends asymptotically to 2.11. The three curves and their sum (thick green line) are represented in green below.

The shape of the three individual curves will have no impact on the result of the model. The only purpose is to have one smooth curve which is the sum of the three individual curves and is fitted to the number of deaths and tends asymptotically to 2.11. The parameters determining the two lognormal curves (position on the x-axis, width and amplitude) were in a first step temporarily adjusted to obtain a good visual fit with the daily deaths. The two following processing steps (see below) could easily introduce deformations to the final result which is the set of daily probabilities of death per case. In order to reduce this risk, since the purpose is to obtain a set of daily probabilities of deaths which fits the available data as closely as possible, the parameters of the lognormal curves were subsequently adjusted to minimize the error calculated by the sum of squares.

Figure 4: Approximating daily deaths with two lognormal and one arctangent curve

The data series of daily infections and the sum of the three green curves which were fitted to the daily number of deaths were then fed into an algorithm to calculate the probability of death for each single day for one single case on day zero. In order to understand this algorithm, we must consider that the number of deaths on each day can be approximated with a certain margin of error by the sum of the corresponding probability of deaths per day for each single case which was previously infected. This provides us with a system of equations which can be solved through multiple linear regression, thus finding the solution with the minimum sum of squares of the errors for each day after infection.

The problem is that the common implementation of multiple linear regression can provide both positive and negative coefficients. With the values illustrated in the chart above, here is a graphical representation of the coefficients returned by the ordinary least square (OLS) algorithm:

Figure 5: Attempt to calculate the daily probability of deaths through standard ordinary least square OLS algorithm

Obviously, these results correspond to an alternation of massive deaths and massive raising from the dead, with actually a probability of dying or raising from the dead more than once in a single day. Since no implementation of multiple linear regression is available which limits the resulting coefficients to positive values, an iterative algorithm using brute force was implemented in Java. It returns the following coefficients, which have got no scale due to the implementation:

Figure 6: Calculating the daily probability of death through iterative brute force implementation of least square regression, raw coefficients and smoothed through 8th degree polynomial regression.

Even though the resulting coefficients (blue line) are much more plausible than the results shown in the previous chart, the various spikes render the results unusable in this form. An eighth degree polynomial regression (orange line) was therefore used to smooth it.

All this sounds like a lot of trial and error, and that’s what it is. So how valid is the result which was obtained in this way? There is a very easy way to find out. Based on these abstract figures (i.e. without scale) for probability of death, the theoretical number of deaths for each day is calculated, based on the distribution of hypothetical new infections established above. Then the scale is adjusted in order to minimize the error calculated by the sum of squares over the whole time span. Here is the same chart as before, with the number of deaths (in purple) calculated based on the number of new infections (dark blue curve):

Figure 7: Daily deaths calculated from daily infections using daily probability of death

The purple line of predicted deaths fits nicely with the actual number of deaths. The errors were calculated using the sum of absolute errors (not used here) and the sum of squares (used for adjusting the scale of the probabilities and for optimizing the lognormal approximations). Using these two algorithms, sum of errors were calculated between January 20 and April 30 for the 7-day moving means (gray curve above), for the sum of the arctangent and lognormal curves (green curves in a previous chart) and for the final result (purple curve above). Here are the results in a table:

Table 1: Error values

The sum of absolute errors is not very meaningful with figures which vary wildly from one day to the next. The sum of squares is more adequate in this case and shows very well that the whole processing through assuming a curve of daily new infections, calculating the daily probability of death per case and calculating the number of deaths from these two data sets increases the error from 163.3 to 187.2, as compared to a simple 7-day moving mean. We can therefore assume that the method outlined above to calculate the daily probability of death provides a good approximation. Obviously, these results depend strongly on the assumptions made for the distribution of daily new infections. A closer collaboration with South Korean officials or researchers could probably provide a more robust estimation of the time distribution of new infections than what can be assumed based on media reports. Among others, it might be that case #31 was infected by other members in the sect before she was hospitalized due to a minor traffic accident, and that the virus spread in that community well before Feb. 10. It might also be that patient #31 infected members of the sect even before she had a fever on Feb. 10 (presymptomatic transmission). This paper should be seen as a first attempt to calculate the daily probability of death. Hopefully, the algorithm presented here will be refined and applied to other cases. Here is the table of the final result for the daily probability of death:

Table 2: Values of daily probability of death

Now we have got all the data we need to calculate the total IFR by simply adding the daily probabilities of deaths, and we get 2.58%. If we put all this together: naïve IFR (dividing the number of deaths through the confirmed number of cases) on April 9 0:00 is 1.96%. If we complete the still missing data after April 9 with data from China (?#1), we get an IFR of 2.11%. If then we calculate the daily probability of death in order to exclude the underestimation due to the deaths from the 100 cases per day on average which continue to April 9 after the initial sharp peak in cases (?#2), IFR increases to 2.58%. How is it that most experts consider that IFR in Western countries is around or below 1%, with only a few experts considering a value between 1% and 2% as being plausible? We will try to answer this question by comparing the South Korean data to the Swiss date. In a first step, we adapt the IFR from South Korea to the Swiss age pyramid.

Adapting South Korean CFR/IFR to another age pyramid and age distribution of cases

The fatality rate of COVID-19 varies massively according to age. It is therefore of utmost importance to adapt the various fatality rates obtained for one country at a certain time to the age pyramid and case age distribution of other countries or regions and/or of other times before drawing any conclusion.

Let us start with calculating what IFR would have been like in the recent outbreak in Switzerland if we assume that it had been caused by the same virus strain as the one in South Korea and if Swiss bodies reacted in the same way to each virus strains as South Korean bodies. We will see that at least one of these assumptions must be rejected.

The table below shows that if we adapt the South Korean IFR to Swiss conditions (age pyramid and age distribution of the cases), we get a CFR of 5.36%. This would lead to a mortality of 2.68% of the population if we assume an infection rate of 50% (still below the level required for herd immunity)

Table 3: Adapting South Korean fatality rates per age group to Swiss age pyramid and age group distribution

Fortunately, this is not what actually happened. We must take into consideration that testing in Switzerland was very far from being as massive and efficient as in South Korea. A random sample PCR test study in Austria shows that in a situation very similar to Switzerland, a reasonable under-ascertainment rate would be at least 2 or (maybe much) more. Therefore, if in Switzerland we had the same virus strain as in South Korea, and if this strain had the same impact on the Swiss population, we would have an nIFR which would be well above 10%, but this is not the case, as the chart below shows; nIFR will flatten out somewhere around 5–6%, with plausible estimates for IFR around 1–2%:

Figure 8: Evolution of nCFR in Switzerland

It is not only a question of fatality rate, but also of the delay between infection and death and, more importantly, its distribution. We have seen above that in the case of South Korea, the first deaths can occur very fast after infection, but that the peak is reached 34 day after infection, and the daily probability of death becomes insignificant roughly 60 day after infection. This pattern is coherent with what we saw in Wuhan, except for the fact that testing capacity was insufficient at the beginning of the outbreak, even for testing all the deaths possibly due to the virus. So how is it that in European countries, the number of deaths decreased relatively fast after lockdowns brought the number of new infections down?

To investigate this question, we can apply the daily probability of death calculated from South Korean data to the Swiss data. It is obvious that they don’t fit at all. The chart below shows the daily number of confirmed cases (blue area) and the actual number of daily deaths (orange curve), both for Switzerland, and the hypothetical number of deaths calculated from Swiss case numbers using the South Korean figures for daily probability of death (grey curve):

Figure 9: South Korean daily probability of death does not fit Swiss data

The probability of daily death calculated for South Korea was originally calculated from the date of infection. For the chart above, the calculated number of deaths is calculated not from the date of infection, but from the date when cases were confirmed. A delay of 7 days between infection and announcement of the confirmed case was assumed. Research from Germany confirms that this assumption is reasonable (9). The amplitude of the curve was adapted to the actual figures of daily deaths from Switzerland. During the first weeks of the outbreak, the grey line follows the orange curve of actual deaths quite closely. However, the peak of actual deaths in Switzerland is reached much sooner than the calculated deaths based on South Korean figures. It is impossible to get a better fit by changing the assumed delay between infection and confirmation, because the width of the peak of the grey curve is much broader than the width of the peak of actual deaths (orange curve). In other words: whereas in South Korea, deaths are spread out over a much longer duration following infection, in Switzerland, death intervenes earlier and with a much smaller standard deviation. It is difficult to imagine that such a massive difference could be due to differences in treatment in South Korean versus Swiss hospitals.

Another possible explanation could be the levels of air pollution. According to a Harvard study (10), an increase in 1𝜇g/m3 in average PM2.5 exposure increases COVID-19 mortality (deaths per million people) by 8%. This can obviously not be true in general throughout the world, because otherwise mortality in Asian countries would be sky high, even if the 8% are considered to increase in a linear way with PM2.5 air pollution and not exponentially. The following table shows PM2.5 pollution levels (average exposure) for a few countries with well-documented COVID-19 outbreaks, according to World Bank data (11):

Table 4: PM2.5 mean exposure values for selected countries

The mortality measured in deaths per millions people, even if compensated according to the beginning of the outbreak, depends much more on the speed of spreading of the virus than on the fatality rate. Among China, Italy and Germany, the spreading rate of the virus in the initial phases of the outbreak is clearly negatively correlated to PM2.5 pollution levels: slowest in Wuhan (11–18% per day, according to estimates), somewhat faster in Italy (17% per day) and fastest in Germany (23% per day). On the other hand, if we believe the Harvard study mentioned above, the spreading rate should be much higher in Wuhan than in Germany, leading after several weeks to a mortality more than 300% higher than in Germany. This shows the problem of building models which rely on one single data set. An artifact from a hidden variable can never be excluded.

There are indications that the spreading rate depends significantly on the virus strain (see below). In the US, the currently fastest spreading strain (from Northern France) arrived first in New York and spread from there throughout the US. The slowest spreading strain, on the other hand, arrived on the West Coast much earlier and slowly spread eastwards. Since the East Coast is much more heavily polluted than the West Coast, the effect of the two virus strains prevalent in the US can perfectly well explain the apparent impact of PM2.5 pollution levels.

In the absence of any other explanation, the results shown above clearly indicate either that in South Korea, they had another virus strain than the one in Switzerland and that these strains have got significant functional differences, or that Swiss bodies react differently to the same virus strain, or both. Actually, there are indications from research that both are the case: the virus strain which caused the outbreak in Wuhan and spread through many Asian countries seems to spread more slowly in general than the one we have got in Europe now and replicates also more slowly, leading to a longer delay between infection and possible death, and it seems to be less well adapted to Western bodies. Both factors are supported by many research results, as the next section shows.

Putting the South Korean-Swiss contrast in context

Data from places with a relatively high infection rate like Nembro (Lombardy/Italy) (12) (13) (14), Spain and Madrid (15) (16), New York and many other places puts IFR consistently between 1% and 2% if we take the infection rate determined by antibody studies and the total excess mortality as data sources. On the other hand, IFR rate calculated for Gangelt in Heinsberg district is way below 0.5% (17). One difference between Gangelt and the other places is that in Gangelt, they probably had the original Wuhan strain of the virus, whereas everywhere else, they had the Italian strain which derived from it through two mutations, or (in New York) a mix of Italian and Northern France strain, which in turn derived from the Italian strain.

Several research papers conclude that there are significant functional differences between SARS-CoV-2 strains, using epidemiologic (18), phylogenetic (19) (20) and experimental (21) data. Most of this research comes from Asia, but it is generally ignored by Western experts or considered to be “controversial”. There is also increasing evidence that at least some people of Asian origin are more vulnerable to the virus (and coronaviruses in general) than people of European origin (22).

The importance of this research goes way beyond abstract academic significance. If we get this topic wrong, we cannot put together a coherent picture of how this virus spreads across human society, and the massive inconsistencies end up destroying the coherence of the Western academic and media discourse.

The danger from white supremacists and similar groups

For many decades now, there has been a dangerous trend among certain political orientations to dismantle the right to access to healthcare, which is part of the UN human rights concept (23), but this fact is less and less mentioned even in Western mainstream media. At the same time, we see a strengthening of various white supremacy movements. COVID-19 is apparently seen by members of these movements as a welcome tool to increase mortality among non-whites, and unfortunately, certain characteristics of this virus make it appear adequate for this purpose. Beyond the fact that non-white bodies seem more vulnerable to it, there has probably never been another virus in history whose mortality depends so much on costly hospital treatment. For most illnesses, cheap cures are available by now. For COVID-19, no such cure is known yet, and the only way of reducing mortality among severe cases is a lengthy stay in hospital with at least oxygen supply or in some cases heavy machinery like respirators, ECMO machines, dialysis etc.

These factors are well known among white supremacy groups; exposing them here will therefore not fuel these ideologies. On the other hand, few people who care about human rights seem aware of this aspect of the pandemic. There is no international movement to expose and protest against the disastrous COVID-19 policies in many countries in the Americas. Letting this virus spread when many countries across the world show that it can be contained is highly problematic in itself. If we add to this the fact that high income families can shield themselves by staying at home, and that low income families are always most exposed, that in many countries healthcare is unevenly accessible, in combination with the differential in fatality rate explained above, it becomes a form of ethnic cleansing. This is happening right in front of our eyes, but nobody seems to care.

Bibliography

1. COVID2019app.live [Available from: https://docs.google.com/spreadsheets/d/1Z7VQ5xlf3BaTx_LBBblsW4hLoGYWnZyog3jqsS9Dbgc/htmlview#.
2. South Korea CDC COVID-19 updates [Available from: https://www.cdc.go.kr/board/board.es?mid=&bid=0030.
3. National Health Commission of the PRC outbreak reports 疫情通报 [Available from: http://www.nhc.gov.cn/xcs/yqtb/list_gzbd.shtml.
4. Bendix A. South Korea has tested 140,000 people for the coronavirus. That could explain why its death rate is just 0.6% — far lower than in China or the US. Business Insider. 2020 06.03.2020.
5. Terhune C, Levine D, Jin H, Lee JL. Special Report: How Korea trounced U.S. in race to test people for coronavirus. Reuters. 2020 18.03.2020.
6. Salmon A. Why are Korea’s Covid-19 death rates so low? Asia Times. 2020 11.03.2020.
7. South Korean city on high alert as coronavirus cases soar at ‘cult’ church. The Guardian. 2020 Feb. 20 2020.
8. Dick S. Cult-like doomsday church linked to South Korea’s COVID-19 outbreak. The New Daily. 2020 23.02.2020.
9. Dehning J, Zierenberg J, Spitzner P, Wibral M, Pinheiro Neto J, Wilczek M, et al. Research Article Summary: Inferring COVID-19 spreading rates and potential change points for case number forecasts. 2020.
10. Wu X, Nethery RC, Sabath BM, Braun D, Dominici F. Exposure to air pollution and COVID-19 mortality in the United States: A nationwide cross-sectional study. medRxiv. 2020:2020.04.05.20054502.
11. World Bank. World Development Indicators 2015 [Available from: http://data.worldbank.org/products/wdi.
12. Coronavirus, test sierologico ad Alzano e Nembro: 61% positivo. Rai News. 2020 30.04.2020.
13. Viganò F, Samotti L. Coronavirus, a marzo 5.900 i morti in Bergamasca: i dati comune per comune. Bergamo News. 2020 05.05.2020.
14. Viganò F. Nembro torna a respirare, ma mentre conta le vittime preoccupa la crisi socio-economica. Bergamo News. 2020 01.05.2020.
15. Carreño B. Spanish antibody study suggests 5% of population affected by coronavirus. Reuters. 2020 13.05.2020.
16. Llanera K, Andrino B, Grasso D. Spain’s excess deaths during coronavirus crisis reach 43,000. El Pais. 2020 28.05.2020.
17. Streeck H, Schulte B, Kuemmerer B, Richter E, Hoeller T, Fuhrmann C, et al. Infection fatality rate of SARS-CoV-2 infection in a German community with a super-spreading event. medRxiv. 2020:2020.05.04.20090076.
18. Korber B, Fischer WM, Gnanakaran S, Yoon H, Theiler J, Abfalterer W, et al. Spike mutation pipeline reveals the emergence of a more transmissible form of SARS-CoV-2. bioRxiv. 2020:2020.04.29.069054.
19. Tang X, Wu C, Li X, Song Y, Yao X, Wu X, et al. On the origin and continuing evolution of SARS-CoV-2. National Science Review. 2020.
20. Yeh T-Y, Contreras GP. Faster de novo mutation of SARS-CoV-2 in shipboard quarantine. Bulletin of the World Health Organization. 2020.
21. Yao H, Lu X, Chen Q, Xu K, Chen Y, Cheng L, et al. Patient-derived mutations impact pathogenicity of SARS-CoV-2. medRxiv. 2020:2020.04.14.20060160.
22. England PH. Disparities in the risk and outcomes of COVID-19 2020 [updated 02.06.2020. Available from: https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/890258/disparities_review.pdf.
23. United Nations. International Covenant on Economic, Social and Cultural Rights 1966 [Available from: https://www.ohchr.org/EN/ProfessionalInterest/Pages/CESCR.aspx.