Naked Truth About COVID-19 Spread In The USA: Running Out Of Time In Two Weeks
What the next 2 weeks/200 days can look like, if the Americans do everything we can in US. Or if they do nothing.
A recent article Coronavirus: The Hammer and the Dance by Tomas Pueyo was a pivoting point for many to understand and a wake up call to see a new reality of the Coronavirus pandemic for US and hopefully many other countries around the globe. I support signing the WhiteHouse petition referenced by the article, but unfortunately, it still did not reach the goal. The fact shows that even this thunderbolt-like article did not wake up enough people, although it says 10 million people read it Everyone should have signed the petition. It’s about life and death of tens of millions people in US.
The summary. Strong coronavirus measures should be done TODAY, since there are only is not enough efforts so far to contain and suppress the virus, which is still possible. But we are running out of time, since after 2–3 weeks no measures may help the situation. Full lockdown of the major centers of outbreaks may help a lot to reduce the death toll by hundreds, if not thousands times.
If you have not read the above mentioned Tomas Pueyo’s article, I strongly suggest read it first to better understand the problem and suggested solution.
US, along with UK, Switzerland, and some other countries, waited before fully responding to the pandemic, and as you will see here, US is literally running out of time, unless very strong measures are ordered.
I will heavily rely on an epidemic model SEIR and the epidemic calculator based on that model, although I must mention that this model has an inherent strong hypothesis that the people carry lifelong immunity to a disease upon recovery, but for many diseases the immunity after infection wanes over time. Therefore, the results may not be as predicted, and the outcomes depend heavily on actual immunity developed by people, and whether it is sustainable enough.
Also, my discussion is limited in the scope compared to the article by Tomas Pueyo, as I am running out of time myself, and see it necessary to share the results ASAP. I will only discuss the main outcomes from the model as a result of my recent approximation of the Johns Hopkins data for US COVID-19 ourbreak. The approximation is illustrated in Chart 1 at the beginning of this article.
The chart is using a logarithmic scale, and it can be seen that starting from March 2, the logarithm of the total cases (blue dots) is represented pretty well with a linear approximation (the red line), in accordance to the non-restricted model of the epidemic outbreaks, which is of exponential nature. For the mortality (grey dots) linear approximation (yellow dashed line) is working fairly well also.
Let me note that, according to SEIR model, or simpler exponential models, the number of deaths should follow the total cases, as the death rate is usually considered a constant for modeling purposes. It can change if there is some intervention, and it can be seen that the death count after 3/19/2020 is following slightly different direction. I ran another linear approximation, and the result is shown in blue dashed line on Chart 1.
The total cases data is on the straight line (red) in logarithmic scale after Mar 1, and shows no signs of slowing. It’s exponential and doubles every 2.39 days.
On April 23, total cases from the registered data only may reach 293 MM cases with deaths up to 2 MM. 🔥🔥🔥
On the next day, the total cases extrapolation will exceed the US population.
The picture may be skewed soon, as there are not enough tests to reach 293 MM.
It’s hard to analyze the entire situation since there’s no data published on total number of tests run, which could help get better estimates. And it may be just the lowest number, since there is unknown number of not tested cases or asymptomatic ones.
Using the number of deaths may give a better extrapolation if the death rate actually does not change during this period. So, I have done an approximation of the death count from 3/2/20 and extrapolated to April 23, resulting in predicted 408,286 deaths (yellow dashed line).
Using the latest total death rate of 1.53%, I calculated the approximated total cases (yellow line on top). This gives the extrapolated total number of cases on April 23 equal to roughly 26 MM.
This seems to be a lower estimate, as the linear extrapolation shows deviation down from the latest data points. So, I did another approximation of the death counts from 3/19/20 to 3/27/20 (blue dashed line) with extrapolation resulting in 1.9 MM deaths on April 23.
Dividing the extrapolated deaths data by the same 1.53% death rate I got the extrapolation shown as a blue solid line, and got expected number of cases on April 23 to be 126 MM. The coefficient of the slope in logarithmic scale for this approximation was b=0,264, which corresponds to the doubling time of 2.62 days.
The vast range of the predicted values is due to the nature of the exponential curve, and even the smallest changes in the estimated parameters result in the great variations of the extrapolated data.
Now, enough about this naive exponential model, as I discussed it only in order to be able to get an estimate of the slope of the linear approximation in order to use it as an input for the SEIR model.
To adjust the SEIR model to the discussed data on US total cases and deaths, I used the epidemic calculator and my estimates discussed above.
In the SEIR model, many parameters show strong correlation, as can be seen by changing them in the calculator. In particular, the slope of the logarithmic scale is affected both by the basic reproduction number R0 and incubation period Tinf.
In order to restrict myself to the most reliable set of parameters, I chose the parameters provided in the Coronavirus: The Hammer and the Dance article.
To adjust the slope of the total cases, I used the R0, and received fair approximation with the value R0=3.72, which gives doubling time equal to 2.6 days. The numbers are significantly different from the numbers in Coronavirus: The Hammer and the Dance, (R0=2.4 ) providing much higher growth rate.
The higher growth rate is very significant, since the date range for possible interventions is much more narrow, and urges us to make these interventions faster.
In order to adjust the mortality, I used the case fatality rate, and ended up with the value of 6.19%. This is significantly higher than the direct calculation obtained by dividing the number of deaths by the number of total cases for the same date. It is understandable, as the mortality is connected to the cases that were detected earlier in time, and accumulated from a range of dates because of possible variations in incubation period.
Now, let us take a look at the resulting model representation. The first chart is using logarithmic scale, as it is very useful for illustrating the linearity of the slope and allows for adjusting the slope value to the approximation from the Chart 1.
As we can see, the infections, which we consider equal to total number of cases reported by US, and fatalities are well approximated by linear slopes in the left part of the chart, and are parallel to each other. This justifies our previous assumption made when approximating the data in the Chart 1.
The black triangle points to the position of Day 50 in the chart. We found that the slope of the curve b=0.2553 is close to the value b=0.262 from Chart 1. The values of Infections and Fatalities for Day 50 are close to the actual values for March 29. Thus, we can assume that Day 50 in the model is consistent with March 29 data.
Do Nothing Scenario
To better understand the scale of the events in this model, let us take a look at the same data in linear scale (Chart 3)
As you can see, there is a peak in the the number of Exposed people on Day78, which is 28 days from now (March 29). Also, the black triangle indicates the tiny bar representing Exposed on current date compared to the peak value, which is 87,722,653. You can see also that the Exposed value on March 29 is estimated to be 495,804. Thus, in 28 days, we can expect the number of exposed people to be 177 time higher that today. This is a very alarming, although it is not as big as the simple exponential gives us.
Now, the number of infectious people on March 29 is 139,012. And this is nothing compared to the number of infectious people = 36,342,483 on Day 78 (April 27). And, expected number of fatalities on April 27 is 1,866,471.
As much as these results look chilling, they are not the end of the story. We can see that the number of fatalities is accumulating further down the road, and on day 200, we expect to see the number of fatalities grown up to 20,481,941. That’s 20 million people. It’s roughly twice as much as predicted by calculations in the Coronavirus: The Hammer and the Dance.
So, we have taken a look at the possible future where we do nothing to prevent this.
The model also gives us an opportunity to take a look at the results of applying certain measures, which Tomas Pueyo imaginatively called The Hammer.
The measures are interventions to decrease the transmission. They include social distancing, lockdowns, quarantines etc.
Full Scale Intervention Scenario
Let us see what happens if we do total, 100% restriction by assigning Rt=0 in the model, and start it on Day 50 (Chart 4).
Note that the black triangle is at the far right and showing the numbers for day 198. Drastic change in the scale of the events, as we can see there are 400 time less fatalities than in the Do Nothing scenario. Also, it is very important that the length of the intervention might be just about 3–4 weeks, and then, gradually, the interventions can be decreased.
Wait For 10 Days Scenario
If the government decides to wait for 10 days, here’s what happens in the model.
As you can see, fatalities are increased 12 times.
The government decides to delay actions for 20 days
This is even more disastrous.
As you can see, in this case, the effect of the intervention is very small, since the government waited too long, and passed the peak of the Exposed chart in Do Nothing scenario, and they are left with only the tail of the virus spread distribution, and it’s already too late to do anything. Resulting casualties are 17.2 million people dead.
What if they apply only 75% of the efforts on Day 50?
Here’s what would happen. The chart is in logarithmic scale to better understand the behavior.
In this case, fatalities will be much higher than in full scale scenario, and importantly, the infections will not go down as dramatically. Therefore, the entire period until day 200 there will be a need for intervention.
Cluster Spread And Localized Focused Interventions
Current data show that New York state and NYC have about half the total US cases. This means that focused efforts in interventions may give much better results than spreading the interventions and medical efforts all over the country. This is exactly what China did.
Let us use the model for New York State.
Intervention on Day 50 in New York State would result in the fatalities around 54,000. This is 20 times less than Do Nothing scenario.
The Do Nothing scenario for New York State:
With 1.25 million fatalities, the state will be totally overwhelmed and devastated. There are only about 100,000 ICU beds in US, and even that would be easily overwhelmed by only NY State needs if we do nothing.
Conclusion: There is not time to wait. We can loose the battle
Based on the analysis provided, the first and urgent outcome is this: if the US government waits for another 3–4 weeks, the battle may be lost. And the results provided do not take into account the shortage of all supplies for medical personnel and hospitals, hospital beds total and ICU in particular.
Given all the discussion in the Coronavirus: The Hammer and the Dance still applies now, but only in a compressed time scale. I will follow up with other details, but I hope this report will give bigger impact on the immediate measures in any country. US is in the most difficult situation now, but other countries can learn a lesson from this work: act immediately, or lose precious time and lose the battle. You only may need a few weeks to win.
Note that mitigation is not the option, according to Coronavirus: The Hammer and the Dance, as it only:
create a massive epidemic, overwhelm the healthcare system, drive the death of millions of people, and release new mutations of this virus in the wild.
Intervention might still be an option, but requires a brave heart, decisiveness and smart strategy to focus on the smaller epicenters, and squash them one by one.
Let’s fight together: share the Word
Please, if you agree with this report, share it and help fight the coronavirus.
Millions of lives are at stake. Leaders should hear from us and act now.
