Forecasting:
We are due to see a slight rise in deaths beginning about July 7, if the time lag of cases to deaths of 18 days is correct.
There is an argument that the new cases being observed now are the result of additional testing, and that additional deaths will therefore not occur.
One further indicator is hospitalizations.
These have also risen recently.
The USA aggregate cases began rising again about June 17 or 18 (depending somewhat on the specific data source, since their reporting lags differ). The chart below shows new cases increasing beginning about June 18.
The same data source shows USA hospitalizations increasing starting about June 23.
If the 18-day lag is correct, this puts hospitalizations about 13-14 days behind eventual deaths.
We should be able to see that same gap directly, between deaths and hospitalizations, if the relationship holds.
The hospitalizations in the USA began to drop from their peak about April 24. Deaths began to drop from their peak about May 8. This is 14 days - the relationships are consistent.
Using the 14-day gap from rising hospitalizations, then, the prediction for a rise in USA total deaths should appear about two weeks after hospitalizations began to rise gradually again, which puts the appearance of the rise in deaths at about July 7.
This is calling it with greater apparent precision than is likely really possible. And as before, we hope that improved treatments will cut the number of deaths.
The data relationships say "next week".
We'll see...
We have markers already in the Texas data for all three, because deaths have already started to rise.
In Texas, the hospitalizations began a gradual rise about June 9, and began to take off about June 16.
The Texas deaths began to rise gradually about June 16, and began to rise more swiftly about July 1.
This gives us two possible markers from the Texas data for changes which could be matched for time lag - they are consistent with the USA aggregate data: death counts lag hospitalization counts by about 14 days.
Since we have a hypothesized time lag between hospitalizations and deaths, the Texas data also allows for an estimate of deaths to be expected from the hospitalizations. These range from about 1.2% to 1.7%. Call it conservatively, about 1.4% fatality rate expected, relative to the corresponding number of hospitalizations.
Two weeks before July 7 was June 24. The number of hospitalizations that day in Texas was 4389, meaning that Texas could see 61 deaths on July 7 - that would be a new Texas record.
In the USA, the ratio of deaths to hospitalizations recently is about 2.2% (Texas undercounting its deaths, ya think???).
Two weeks before July 7 the USA had 31083 new hospitalizations. If 2.2% of them result in deaths, then on July 7 we would estimate 684 new USA deaths. Since the US average deaths have been flat around 600-650 recently, this would be the first uptick of a consequent rise.
And if the ratios and lags hold up, we could be up over 1200 daily deaths by mid-July.
All opinions expressed in this blog are solely those of the authors, and do not reflect those of University of Texas Rio Grande Valley, or any organizations of which either is a member.
Commentaires