On Sunday, March 29, the federal government’s task force was predicting 100,000 – 200,000 deaths from the Wuhan virus with a peak in about 2 weeks.  Since March 19, 2020 I’ve been gathering data on the virus, at least 3 times a day. Here is a plot of cumulative U.S. Deaths through March 30:

Fig. 1

The x-axis is days since the first reported death in the U.S., Feb. 29. I plotted based on the exact time I grabbed the data and you can see it is a very smooth curve fit very precisely using a power equation with a correlation coefficient of 0.99647. If the current trend holds, 7,000 deaths would be expected by day 35 (end of day Friday, April 3).

If you look closely you can see that that rate appears to be slowing down slightly. For example, the death rate doubled from 1,000 to 2,000 in about 3 days, whereas it is projected to slow to doubling every 3½ days from 2,500 to 5,000. This can be seen more clearly when using a log y-axis where you can clearly see the curve folding over:

Fig. 2

On Sunday, March 29, health officials from the Trump administration indicated that they expect the peak of the epidemic to be in about 2 weeks and a best case scenario of 100,000 - 200,000 total deaths if everyone practices social distancing, religious hand washing, sneezing into your sleeve, staying home when you’re sick, etc., using their current modeling, whatever that is. Based on those vague indications I calculated a standard cumulative normal distribution using a maximum number of deaths equal to 100,000, centered about 2 weeks from Sunday and a standard deviation that allowed the curve to go through the exiting data. The actual midpoint is at 49 days (30 + 19) with a standard deviation of 9.5 days:

Fig. 3

When plotted as a standard bell curve you can see that the maximum daily death rate would be about 4200. Note that the maximum number of deaths to date has been < 600.

Fig. 4

A standard normal distribution seems reasonable given the history of the three outbreaks of the Spanish Flu:

Fig. 5

How reasonable is this scenario? Here’s the epi curve from the CDC through March 29:

Fig. 6

It looks like there might be a peak at March 16. The typical incubation period for the virus appears to be about 5 days so people who reported onset of symptoms on March 16 were probably infected around March 11. The curve for Cook County, Illinois is very similar, peaking on March 18. It shows an even clearer peak than the national data. (This graph does not include Chicago because Chicago has it’s own public health department and so Cook County does not collect that data.)

Fig. 7

Five days earlier, on March 13, president Trump stopped the flights from Europe and the Chicago Archdiocese closed all churches in Cook (where I live) and Lake counties, which includes Chicago. This was the beginning of the panic buying of toilet paper and other grocery items. On March 16, the CDC announced the “15 Days to Slow the Spread” campaign that recommended the extreme social distancing and other measures alluded to above. On March 21, Illinois governor Pritzker shut down all non-essential Illinois businesses. The mayor of Chicago has been even stricter, limiting people’s walks and bicycle rides. Perhaps anti-social distancing would be a better term.

It appears very clear that the mitigation measures are working as the peak coincides with a 5 day incubation period starting March 13. The Cook county data is also broken down by travel, close contact and community spread which further supports the notion that these actions have been effective as they certainly should be based on what we know about how the virus spreads (direct exposure to droplets from sneezing/coughing and through contact of surfaces containing the virus, typically through prolonged close contact of infected persons). If we didn’t see this peak in the data than we would conclude that people are not following the protocol, our understanding of the spread of the virus is faulty or our health care system is overwhelmed.

Given the above, is it reasonable that we can expect the maximum daily death rate to rise to be 7 times what it is currently? Current reports seem to indicate that time to death may be typically 2-3 weeks from onset of symptoms so for those that were infected at the peak of March 18 that go on to die, we expect most will have died by April 8. That would imply that the maximum death rate would be achieved by that date. Here I’ve plotted two more distributions showing maximum death counts of 50,000 (blue curve, peak at day 44) and 25,000 (green curve, peak at day 41, which is April 9).

Fig. 8

I extended the power curve fit (black line) through day 41 and it perfectly matches the red curve. But if this turns out to be the peak date we can see that the red curve is highly exaggerated. The green curve’s peak rate is day 41 with a much more optimistic death count of 25,000. The blue curve’s death count is about what the high end of the estimated death count from the season flu might be this year. If the green curve is more correct, we should begin to see the data start to deviate from the current projection by the end of this week (April 3). If the blue curve is more representative then we should start to see the data deviate at around day 41 (April 8). Note how big of a guess all of this is given the very small amount of data we have so far. Any number of curves could be made to fit the existing data based on guesses as to total deaths and when the peak will be. Place your bets now.

What about the current therapies of hydroxychloroquine and an the anti-viral therapies that are showing significant clinical success? These should have a dramatic impact on the curves and so I suspect that the green curve is a very real possibility.

How does it compare to the flu? I’ve added in the weekly Flu and Pneumonia deaths from the CDC since August 2019, which is when the season Flu rates began to rise.

Fig. 9

These are based on analysis of death certificates; however, the CDC estimates as many as 50,000 deaths this flu season based on their models. That’s the number you typically hear reported, which is 5-10 times the number reported on death certificates. You can see that we have already reached 110,000 flu and pneumonia deaths over the past 7 months. Note, however, the slope of these curves vs. the projected slope of deaths due to the Wuhan virus which gives a feel for the potential impact on our health care system.

©Richard Wright, March 30, 2020