Expect a Million Deaths

Let's see where we are over the last 5 days with the total deaths, the only reliable number:

April 4 = 1,331
April 3 = 1,045
April 2 = 974
April 1 = 1,049
March 31 = 912
March 30 = 558

Let's calculate the average Death Growth Factor for the last 5 days (once it drops to 1.0 we have the inflection point):

April 4 = 1.27
April 3 = 1.07
April 2 = 0.93
April 1 = 1.15
March 31 = 1.63

OK, March 31st was an outlier. It was diffused that it had to do with incomplete reporting over the weekend, so it's fair to toss it.

If we count it, we have an average of 1.21, still far from the inflection point.
If we don't count March 31st, we have an average of 1.1, which is lower than it had been in two previous 5-day periods I calculated before, indicating that we may be slowly approaching the peak.

For those not following this discussion: the DGR (death growth rate) is obtained by dividing the number of deaths in a given day by the number the previous day. Given daily up-and-down variation in the context of spurious factors (like, the virus gets to a cluster of nursing homes), we use a 5-day average.

When this 5-day average remains around 1.0 in two consecutive 5-day periods, we likely have reached the peak or inflection point. That's the mid-point of the curve. The total deaths by that day, will be likely to be half of the final tally at the end of the pandemic, so, when you reach the inflection point, you double the number of deaths you had thus far, and you have a rather good estimate of the total, at the end of the pandemic.

Last time I calculated a 5-day average it was around 1.25. If it dropped to 1.1, it's a good sign, but in order for it to get down to 1.0, we may still have a way to go, since it will bounce up again once other major cities peak.

Still, April 21st is 17 days from now, so I continue to estimate that this will peak by April 21st.

The only way (I could explain why, but I don't feel like it, right now) that my calculation is wrong, is if I'm underestimating the CFR. If it is 1.4%, or up to the WHO's estimate of 3.4%, then my model overestimates the total infected cohort, meaning that it is currently smaller than I though, thus prolonging the time for herd immunity that would flatten the curve, so the peak will be later than April 21st.

GreatNews2night said:

Let's see where we are over the last 5 days with the total deaths, the only reliable number:

April 4 = 1,331
April 3 = 1,045
April 2 = 974
April 1 = 1,049
March 31 = 912
March 30 = 558

Let's calculate the average Death Growth Factor for the last 5 days (once it drops to 1.0 we have the inflection point):

April 4 = 1.27
April 3 = 1.07
April 2 = 0.93
April 1 = 1.15
March 31 = 1.63

OK, March 31st was an outlier. It was diffused that it had to do with incomplete reporting over the weekend, so it's fair to toss it.

If we count it, we have an average of 1.21, still far from the inflection point.
If we don't count March 31st, we have an average of 1.1, which is lower than it had been in two previous 5-day periods I calculated before, indicating that we may be slowly approaching the peak.

For those not following this discussion: the DGR (death growth rate) is obtained by dividing the number of deaths in a given day by the number the previous day. Given daily up-and-down variation in the context of spurious factors (like, the virus gets to a cluster of nursing homes), we use a 5-day average.

When this 5-day average remains around 1.0 in two consecutive 5-day periods, we likely have reached the peak or inflection point. That's the mid-point of the curve. The total deaths by that day, will be likely to be half of the final tally at the end of the pandemic, so, when you reach the inflection point, you double the number of deaths you had thus far, and you have a rather good estimate of the total, at the end of the pandemic.

Last time I calculated a 5-day average it was around 1.25. If it dropped to 1.1, it's a good sign, but in order for it to get down to 1.0, we may still have a way to go, since it will bounce up again once other major cities peak.

Still, April 21st is 17 days from now, so I continue to estimate that this will peak by April 21st.

The only way (I could explain why, but I don't feel like it, right now) that my calculation is wrong, is if I'm underestimation the CFR. If it is 1.4%, or up to the WHO's estimate of 3.4%, then my model overestimates the total infected cohort, meaning that it is currently smaller than I though, thus prolonging the time for herd immunity that would flatten the curve, so the peak will be later than April 21st.

Click to expand...

IMO, the daily death rate across the nation will be 10,000 before this thing ebbs.

calamity said:

US reports most coronavirus deaths in one day as 1,224 die on Saturday

Click to expand...

Different sources have different cut-off points so it varies slightly. Worldometers has today's total at 1,331.
They do say that there was some controversy in the total of deaths reported by the State, so they had a larger tally before but lowered it.

Still, that's the record so far.

Sunday this number will be lower, but then, since Health Departments are closed on Sundays, it's never very accurate.

On Monday or even better, Tuesday, we'll know if the GDR is still in the 1.2 range, or if it is coming down to 1.1.

As you've mentioned, this model may be less predictive because of the heterogeneous death rate in different states. If we had ONE compact country with ONE epicenter, this model I'm using would be spot on. Given that we have multiple foci, very far from each other geographically, we'll se the DGR bouncing up and down for quite some time.

Yes, if the virus ravages the 4 million seniors who live in Florida, with CFRs similar to Italy's (10%) in Florida alone we'd have 400,000 deaths which would completely throw off this model.

This said, I don't think it will come to that. Probably seniors in Florida are self-isolating, and their governor has finally declared a stay-home order, so one, it might not get all seniors, and two, the death rate might be lower than in Italy.

Still, regardless of specific state situations, there *is* a ceiling: if the virus gets to 100% of the US population, it won't have any room left to keep spreading, and it will then peter out.

100% is unlikely given social isolation and social distancing efforts.

calamity said:

IMO, the daily death rate across the nation will be 10,000 before this thing ebbs.

Click to expand...

Yes, but if we keep doubling it every 3 days we will reach this daily death total in 9 days. It would be on April 13, so my "by April 21st" deadline for the peak would remain accurate.

calamity said:

IMO, the daily death rate across the nation will be 10,000 before this thing ebbs.

Click to expand...

When I put those numbers in excel and plotted an exponential trend:

4/19/2020 9,832
4/20/2020 11,238

trouble13 said:

If it does not reach anywhere near 1 million can we consider them not experts any longer?

People making these dire predictions should do so on record so we know who not to listen to in the future.

Sent from my SM-G965U using Tapatalk

Click to expand...

So if it doesn't reach a million that means Trump did a great job!

GreatNews2night said:

Yes, but if we keep doubling it every 3 days we will reach this daily death total in 9 days. It would be on April 13, so my "by April 21st" deadline for the peak would remain accurate.

Click to expand...

I'm seeing an increase in deaths not being much over about 10% per day. So, starting at 1000 deaths to get to 10,000, we get ln(10)/ln(1.1) = 24 days.

So, yeah. You might be right, if my 10,000 number is correct.

SLC said:

So if it doesn't reach a million that means Trump did a great job!

Click to expand...

Yeah, run on that.

Sandy Shanks said:

The number of cases does not matter because the CDC is tracking the number of people who have come down with the common cold since the beginning of the year.

The number of deaths is the true measure of the casualties of a virus. That, too, is a running total. 8,434 sick Americans have died in three months.

That number is not alarming.

Click to expand...

Over 8,000 have died from the common cold in three months? Do you have a cite for that? Seems way high to me.

Risky Thicket said:

Marke said as he beat people over the head with Jesus. :roll:

Click to expand...

My blows are not offensive, but defensive. I resist barbarian attacks on good people and people like Trump trying to do the best they can for the most people they can within fair and reasonable boundaries.

GreatNews2night said:

No, there is solid calculations going into it. It all depends on the CFR. If it is about 1% then based on the number of deaths we can extrapolate the number of infected people. So, if we project the exponential curve of death to 100 times, we get the daily total population infected. The way the progression is going, my calculations were that by April 21st if the progression kept going unchanged, we'd reach a number of deaths that multiplied by 100 would mean 100% of the US population being infected which is both not likely and a ceiling, so, necessarily, the curve will have to inflect and start coming down by then.

It seems like the epidemiologists at the University of Washington agree with me.

There is a ceiling. The virus can't go above it because it would run out of new people to infect. That's why at a certain point, like in all pandemics, the curve will flatten, then come down. I estimate that the national inflection point will happen by April 21st, give or take a few days. It doesn't mean deaths won't spike in other places like New Orleans, Houston, Dallas, etc. Still, the total deaths per day will start coming down after April 21st.

Click to expand...

Repeat after me...."We dont even now have proper testing, so therefor and whatnot we have no idea how many people this bug has visited".

Therefor your argument is crap.

SLC said:

So if it doesn't reach a million that means Trump did a great job!

Click to expand...

I never believed the "experts" predictions were accurate so I would not use their predictions as a metric to evalate Trumps performance but it does create an interesting quandry for those who do accept those numbers as gospel.

Sent from my SM-G965U using Tapatalk

Lord of Planar said:

When I put those numbers in excel and plotted an exponential trend:

4/19/2020 9,832
4/20/2020 11,238

Click to expand...

Gentlemen,

No point in uninformed speculation. Todays data confirms that the growth rate is declining. After April 8th, the growth rate will be below 1.0 consistently and should meet the target of April 16th - the top of the peak.

Of course, there is always the chance that we will wake up one morning a find that either the UofW IHME has revised its estimates way upward, in which case the polynomial curve up will continue.

maxparrish said:

Gentlemen,

No point in uninformed speculation. Todays data confirms that the growth rate is declining. After April 8th, the growth rate will be below 1.0 consistently and should meet the target of April 16th - the top of the peak.

Of course, there is always the chance that we will wake up one morning a find that either the UofW IHME has revised its estimates way upward, in which case the polynomial curve up will continue.

View attachment 67277228

View attachment 67277229

Click to expand...

I hope you are right, but I doubt that you are. We’re going to pick up 35,000 new cases today, and at least 1000 of those people will die in about a week.

We are still climbing the death curve, not seeing a plateau.

Keep an eye on India. They're trying social distancing. But among the hundreds of millions of poor there, S/D is impossible. People in that country literally live on top of each other.

Remember, this is a country where 600,000,000 million people lack running water. No water!


There Are At Least 75,000 Coronavirus Infections In India TODAY: 21 Days Not Enough

...on March 11th, in India there were 100 coronavirus infections for each deaths TODAY. So, there were a total of 8600 people infected with the new coronavirus in India on March 11th. If you understand how we calculated this 8600 figure, it's relatively easy to understand the rest of our model.

The issue is that India is a poor country. Three-generation households are very common. On March 24th there were at least 75000 infections in India and these people were locked down in their small homes. Many of these people are asymptomatic and will spread this virus unknowingly.

Half of these 75000 people contracted this virus between March 20th and March 24th. This means most of these 37500 were asymptomatic or pre-symptomatic on March 24th. The average incubation period is 5-6 days, but it could be as much as 14 days in some people. This means it may take them as much as 2 weeks to spread this virus to one of their family members. If their family members don't develop any symptoms before the end of the 21-day lockdown, they are very likely to go on with their lives and start spreading the virus again.

That's why we believe Modi's 21 day lockdown will slow down the virus significantly, but won't be enough to contain it.

Click to expand...

The trouble is, at least in my opinion, India cannot survive a lockdown. People will starve there because they live hand to mouth, day by day. Very few Indians have one month financial reserves.

In India, they might be better off just dealing with the virus. After all, they won't beat it anyway. And, their demographics are younger than the USA, Italy and Germany. Basically said, unless you are rich in India, you don't live much past the age of 60 anyway.

L/E India: 68.56 years

maxparrish said:

Gentlemen,

No point in uninformed speculation. Todays data confirms that the growth rate is declining. After April 8th, the growth rate will be below 1.0 consistently and should meet the target of April 16th - the top of the peak.

Of course, there is always the chance that we will wake up one morning a find that either the UofW IHME has revised its estimates way upward, in which case the polynomial curve up will continue.

View attachment 67277228

View attachment 67277229

Click to expand...

You’re assuming this affects the nation equally and smoothly.

It’s not. The geographic separation in the US is much larger than most other nations, and as one area peaks, another one will be starting.

The peak will be extended a while.

Sorry you wasted so much time on your ‘models’.

JasperL said:

Over 8,000 have died from the common cold in three months? Do you have a cite for that? Seems way high to me.

Click to expand...

Don't pay attention to Sandy Shanks. He doesn't even know the difference between the common cold and the flu. He probably means the flu. He kept talking about how people die of the common cold, which doesn't really happen (except to severely immunocompromised patients who will die of just about any infection, and except for those who get a *bacterial* super infection on top of the common cold and die of the bacteria, not really of the common cold). The case-fatality rate of the common cold is close to zero. He talks about "the common cold season" which doesn't exist; while there are more cases in the winter we don't even call it a season since it happens all year long and the average American gets the common cold two to five times per year, unlike the flu that does have a season. Sandy Shanks said the COVID-19 is not worse than the common cold which takes the cake of the most misguided statement ever made on Debate Politics about this pandemic. He keeps saying that the problem is only in New York and there isn't a "national pandemic" which is not only an oxymoron, but ignores the fact that it's more in New York just because it got there first, but the virus is present in all 50 states and it will peak there too, eventually. So, Sandy Shanks is so incredibly misguided that I stopped reading him and responding to him. It's a waste of time. And it is completely useless, since the virus will defeat his ideas pretty soon. All that I have to say to Sandy Shanks is, just wait and see.

GreatNews2night said:

Don't pay attention to Sandy Shanks. He doesn't even know the difference between the common cold and the flu. He probably means the flu. He kept talking about how people die of the common cold, which doesn't really happen (except to severely immunocompromised patients who will die of just about any infection, and except for those who get a *bacterial* super infection on top of the common cold and die of the bacteria, not really of the common cold). The case-fatality rate of the common cold is close to zero. He talks about "the common cold season" which doesn't exist; while there are more cases in the winter we don't even call it a season since it happens all year long and the average American gets the common cold two to five times per year, unlike the flu that does have a season. Sandy Shanks said the COVID-19 is not worse than the common cold which takes the cake of the most misguided statement ever made on Debate Politics about this pandemic. He keeps saying that the problem is only in New York and there isn't a "national pandemic" which is not only an oxymoron, but ignores the fact that it's more in New York just because it got there first, but the virus is present in all 50 states and it will peak there too, eventually. So, Sandy Shanks is so incredibly misguided that I stopped reading him and responding to him. It's a waste of time. And it is completely useless, since the virus will defeat his ideas pretty soon. All that I have to say to Sandy Shanks is, just wait and see.

Click to expand...

Yes, common cold is sniffles, sore throat and a slight cough for a few days. Flu knocks you on your ass for a week. COVID-19 does it for two.

calamity said:

I'm seeing an increase in deaths not being much over about 10% per day. So, starting at 1000 deaths to get to 10,000, we get ln(10)/ln(1.1) = 24 days.

So, yeah. You might be right, if my 10,000 number is correct.

Click to expand...

Yes, the average of the last 4 days was precisely 1.10, that is, 10% daily increase, if we don't count the 3/31 outlier of 1.63 for a five-day average. If we do count it, the average goes to 1.21 or 21% daily growth. Like you said, the average may be coming down because the peak in NY is expected now to happen in 4 days. The average will soon flatten there, bringing down the national average. But then the next epicenter in Lousiana, followed by Florida and Texas and some other stays (I hear that Nevada will be bad, and so will Tennessee), are likely to push the average up, again.

Hawkeye10 said:

Repeat after me...."We dont even now have proper testing, so therefor and whatnot we have no idea how many people this bug has visited".

Therefor your argument is crap.

Click to expand...

Have you noticed that my "argument" is based on the daily DEATHS which IS an accurate number????????????
Do you have a reading comprehension problem?
Like I said, it depends on the CFR, but you wouldn't know what that is, would you?

I'll try to explain one last time but I don't think you'll get it.

The CFR or case-fatality rate can only be calculated accurately at the end of an epidemic. But it can be estimated. We do have data to estimate it, coming primarily from South Korea, a country that did extensive testing and is likely to have achieved testing that captured half of the total infected cohort. They came up with 0.9%. If we halve that, we get 0.45%. But then we have a cohort that was tested fully, every single one of them, and that's the Diamond Princess cohort. Over there, all passengers and crew were tested and they remained isolated so it's a great opportunity for epidemiologists because no mild/asymptomatic cases were missed. Their CFR was 1.4%. Now, a downwards correction is needed because although the crew being younger it self-corrected this a bit, still, the average age was up due to a predominance of senior citizens among the passengers. So it's likely to be below 1.4%. But it is also likely to be above 0.45% because in Korea, they secluded their elderly and only 3% of the deaths happened in the very elderly group, the opposite of Italy that got 10%. The bottom line is that the REAL CFR is between 0.45% and 1.4%. Add the two and divide by two, you get 0.925. To be on the safe side and avoid OVERestimating the infected cohort, I dropped the rate to 0.8% in my calculation. Several epidemiologists got similar ideas and the 0.8% is a very far estimate given the virological characteristics of the COVID-19 and extensive data collected in several parts of the world. It's becoming the prevailing estimate.

Now, if this 0.8% is fairly accurate (and again, most epidemiologists think that it is), you get the total death so far, and make it project up to 100% based on the 0.8% rate. You get a certain number of estimated infected Americans, REGARDLESS OF TESTING. Then you follow the exponential growth of the number of deaths when LEFT UNCHECKED by containment measures, and see at what point you'd reach 100% of the US population being infected by making of that the 0.8% that gets you to your 100% of infected cohort, obviously a ceiling because once the virus gets everybody it can no longer grow since each infected person is surrounded by other infected persons or those who had it and recovered and are now immune. In other words, the virus runs out of new people to infect. That's the ceiling. If you reach the ceiling you have necessarily reached the peak. In this situation of a virus so infectious that it gets 100% of us, the ceiling and the peak are the same.

In my calculation, this ceiling is likely to happen by April 21st.

Now, ANYTHING that slows it down (stay home orders, social distancing) will make the percentage of infected Americans lower, and will lower the ceiling as the virus won't get to isolated people (like the elderly in South Korea). So by lowering the infected cohort, the inflection point will happen BEFORE the point when it would have reached the bigger ceiling, the one with no containment measures. Therefore what I'm saying remains true: the peak would still happen BY April 21st. If you drop the rate of infection to 70%, the peak happens two days earlier, on April 19th. I didn't calculate it for 40%, but hopefully you'll get my point, now.

No, this is not crap. This is a science called epidemiology. That you have no clue about it, doesn't surprise me.

GreatNews2night said:

Yes, the average of the last 4 days was precisely 1.10, that is, 10% daily increase, if we don't count the 3/31 outlier of 1.63 for a five-day average. If we do count it, the average goes to 1.21 or 21% daily growth. Like you said, the average may be coming down because the peak in NY is expected now to happen in 4 days. The average will soon flatten there, bringing down the national average. But then the next epicenter in Lousiana, followed by Florida and Texas and some other stays (I hear that Nevada will be bad, and so will Tennessee), are likely to push the average up, again.

Click to expand...

Florida has a lot of old people (20% over age-65). So that will skew the curve. IMO, FL is an outlier. Louisiana may be too, due to the Mardi Cras transmission vector. So, it too may be hard to use as a yardstick.

Texas (11% over age-65) is probably a better measure. It's demographic is not too far outside the norm and it's not California in being rapid acting. TX is more like the US, a mix of crazy and sensible, with lots of people who are both young and old.

maxparrish said:

Gentlemen,

No point in uninformed speculation. Todays data confirms that the growth rate is declining. After April 8th, the growth rate will be below 1.0 consistently and should meet the target of April 16th - the top of the peak.

Click to expand...

I'm very far from uninformed speculation, as I'm actually a professional in this area. I used a similar method to your graphs. New cases get influenced by insufficient testing and bouts of testing when a health care system receives a supply of test kits, which is why it is desirable to proceed instead, by solidly estimating the CFR (see post #1896) and base the evaluation of the growth of the infected cohort by extrapolating to 100% from the CFR (best estimate, 0.8%) from the ONE number we are certain of, that is, the death tally (as per your first graph).

Still, it's not terribly important because the two ways to estimate the progression of the contagion do go hand-in-hand so my calculations are not that far from those that go by total diagnosed cases. You have the inflection point on 4/16, and all that I said was that it would occur, using my model, BY 4/21, *if* we got a total infected cohort of 100% of the American population which we will not. If we get 70% then my model anticipates the peak to 4/19. While I stopped calculating for smaller rates, some epidemiologists say 60% so it would probably bring it down another day at 4/18 so our predictions are overlapping almost perfectly.

Funny that your assertion that others here are "uninformed", gets to a prediction almost exactly overlapping with yours, huh? So, if I'm uninformed, you'd have to be, too.

And yes, the DGR has had an average of 1.1 over the last four days (10% daily growth), so, the inflection point is getting closer; the caveat, though, is that other epicenters are developing so it's not a given that the slow-down will continue, as the DGR may jump up again, soon. Still, that wouldn't affect the overall calculation because my model does have some built-in buffer of April 21st, not mid-April like the University of Washington Department of Epidemiology found it to be.

The bottom line is, if any of these models gets thrown off due to different rates of infection in different states, it's still VERY unlikely that the peak will happen past April 30th (9 more days for my model or 14 more for the University of Washington model are a lot of days; the virus will necessarily start bumping into trouble finding new people to infect, by then.

So, the peak will be in April. Then we get to the descending part of the curve, which of course will still kill a bunch of people. Let's say that the curve takes 2 months to dwindle to single digits, May and June. We should be fine in July, and we certainly will be fine in August.

Hawkeye10 said:

Repeat after me...."We dont even now have proper testing, so therefor and whatnot we have no idea how many people this bug has visited".

Therefor your argument is crap.

Click to expand...

I'm pretty sure that myself and others i know have already had it. Just no way to know with certainty.

maxparrish said:

Gentlemen,

No point in uninformed speculation. Todays data confirms that the growth rate is declining. After April 8th, the growth rate will be below 1.0 consistently and should meet the target of April 16th - the top of the peak.

Of course, there is always the chance that we will wake up one morning a find that either the UofW IHME has revised its estimates way upward, in which case the polynomial curve up will continue.

View attachment 67277228

View attachment 67277229

Click to expand...

I have already speculated a week or more ago, a death of 89,100 in the USA, based on what I knew then, and certainly... yes... speculations. I have within a few fays said this estimate was probably high.

I am not saying the 4/19/2020 9,832 and 4/20/2020 11,238 deaths will be accurate. I only stated that based on following the exponential trend of the previous data, that is where the 10,000 point would be. If we fall significantly short of that, I think we can be certain the rate has tapered off.