COVID-19 July 03, 2020

Independence Day or Groundhog Day? You decide!  I won't belabor the numbers from yesterday.  Unless you've been living in a cave, you all know the situation.

Today, I took data for the number of new cases since June 9 (shown in red on the daily new cases graph), shifted the values, and overlayed that data on the historical data from the beginning of the pandemic.  That shifted data is labeled as "June 9 shifted" in the graph and is shown in green. Surprise!!! The rise in cases since June 9 looks almost exactly like the rise at the start of the pandemic!

We really are right back where we started.  Or are we?  At the end of March, lockdowns were already well underway and the result was that the curve did start to flatten by early April.  For example, our office encouraged remote working on March 12 and less than a week later we shut the office down except for essential personnel.  That's not what's going on now.  Where we are now is roughly equivalent to where we were at the end of March, but rather than being in a process of increasing restrictions we are in an environment of loosening restrictions.  At least some of the state Governors got the memo and are requiring masks.  It seems unlikely that the curve is going to flatten anytime soon under these conditions.  I think we are in a far worse place:  Groundhog day on, I don't know, PCP?

I've had a few questions about why deaths are not rising in concert with the new cases.  I've addressed that here and there, but I'll drop my hypotheses in the form of an abbreviated list for clarity:

1)  Demographic shift in infections.  The morbidity probabilities are heavily skewed to the elderly. The initial pandemic hit the elderly populations hard, especially in nursing homes.  The current rise in infections is heavily skewed to the younger population who are less likely to die. (Note: that doesn't mean the young are immune.  Plenty of healthy, young people with no obvious risk factors are finding themselves dead or injured for life).  I expect this to start to change as young people bring their new viral friends home.

2)  Better treatments.  Through trial and error, treatments have improved at least a little.  This can't account for all the difference, but I'm going to guess it has some impact.

3)  Increased testing.  Part of the rise in infection rates is due to increased testing. If we had the same pathetic level of testing now that we had early on, the rate of increase would likely be different.  (Note: this has implications for whether we really are in a Groundhog Day scenario).  The increased testing cannot however explain all of the increase in new cases.  Not by a long shot.

4)  Virus evolution.  There is evidence that the virus has mutated, particularly in the spike protein that binds to the ACE-2 receptor of the human cell.  There is no evidence that this makes it any more or less lethal, but it does, apparently, make it more infectious.  There have been other mutations, but based on my limited understanding (and I want to emphasize LIMITED.  I AM NOT AN EXPERT and will defer to those that are if I say something that is inaccurate.  Experts: let me know!), there is no strong evidence that the virulence has been altered.

5) Other reasons.  There are almost certainly other influences that I've left off.

COVID-19 July 02, 2020

I barely have time to post, but I wanted to distribute today's graphs.  Text the word "UP" to 321321 for more information.

COVID-19 July 01, 2020

And July starts off with a BANG!  Things are back on the rise as they almost always do on a Tuesday.  Numbers should continue to go up the rest of the workweek.  It's really just a matter of how much.  By the end of the week, we'll have over three weeks of data on the recent rise in infections.  That's enough data that the curve fits should start to become more meaningful.  Right now, the exponential fit is still showing infection rates doubling about every 15 days.  The average is now over 41,000 per day.  So, if the exponential rate doesn't change, we'll be at over 80,000 per day by mid-July.  

The death rate still hasn't gone exponential, but it isn't dropping either.  With each day we diverge more and more from what could have and should have been.  Oh well.  Water on the bridge.  

COVID-19 June 30, 2020

We had a nice little down day yesterday, as expected.  The rest of the workweek should be up, up, up, at least if history repeats itself.

Even though the daily numbers went down yesterday, the 7-day average for daily deaths actually ticked up for the first time in a while.  Is this the start of the anticipated upward trend?  The total number of dead and the daily death rate certainly are showing the divergence from the Gompertz model.  Not unexpected at all, but it's interesting to watch as it develops.  The difference between the data and model tells what could have been if we had done things correctly whereas the actual data tells us how badly we f*cked up.

There was some discussion on my Facebook wall about why it's taken so long for the death rate to respond to the increase in cases.  There's a couple of reasonable hypotheses.  First, the current increase in infections primarily has hit the younger demographics that are less susceptible to complications and death.  The initial peak in deaths had a very strong contribution from the elderly, especially in care homes.  The younger people dominating the cases now are still infected and contagious and they will eventually transmit that virus to the general population.  They just need to take it over to grandma's home.  That process introduces an additional lag.

The second possibility is that the virus is mutating to a less lethal form.  Evolutionary pressure tends to favor viruses that don't quickly kill their host, and our actions over the last few months may be adding to that evolutionary pressure.  The most severe and lethal cases often end up in hospitals where transmission is effectively terminated.  That's the end of the line for whatever virus is the hospitalized person.  Those that are not severe enough to require hospitalization are still out in the wild and can continue to spread their particular strain.

I note that neither of the two hypotheses are mutually exclusive.  There could be an increased lag in transmission and viral mutation.  And, of course, there could be entirely different explanations.  All this is likely to get sorted sometime in the future, but it's extremely hard to tease out the cause in real time.

COVID-19 June 29, 2020

Today and possibly tomorrow should be the last of the down days this week.  It will be interesting to see how much the numbers fall, if any.  Yesterday's numbers came in lower than the day before.  This was expected.  The numbers that come in today and which I will report tomorrow should also be lower due to the well established weekly oscillation.  After that, we'll see how high things go.  The exponential fit to the data since June 9 suggests we will be at over 50,000 new cases per day, on average, by the end of the week.  We are currently at a little over 40,000 per day based on the last 7 days.

COVID-19 June 28, 2020

Another banner day!  For the virus.  Yesterday should have been a down day for new infections based on the historical weekly cycle, but instead, numbers held steady.  It seems likely we will see shocking numbers by midweek as the upward branch of the cycle kicks in.  The doubling time for new infections decreased to 14.5 days based on all data since June 9.  The exponential fit to that is very good (84% correlation coefficient) considering the oscillations that are superimposed on the data.  The best fit for the number of daily cases is now just under 38,000 per day. 

Masks on. Buckle up. 

COVID-19 June 27, 2020

Today I had to do something I hadn't done for quite a while: I had to change the ordinate on the graph of new daily cases.  Yesterday's 44,726 is the highest value ever, and my graph only went to 40,000.  I've now bumped it up to 50,000.  With an average doubling rate that is now under 15 days, I suspect I'll have to continue to adjust the graph.

This weekend should be interesting because the daily numbers in this part of the week tend to be lower.  Will they be?  If not, that indicates a growth rate that is so large that it is actually starting to swamp out the very large weekly oscillation. As indicated above, the doubling rate of new infections based on the data beginning June 9 is just under 15 days.  The doubling time has decreased every day this week.  If you look at the new daily cases since June 9 (it's the red data on the graph), you'll see a dotted black curve that represents the best exponential fit to that data.  If the numbers over the next few days don't start to moderate as they should as part of the weekly cycle, then that best-fit black curve is going to get even steeper, and the doubling time is going to continue to decrease.

I've dropped the semi-log axis plot of cumulative infections back in today.  The last seven days are shown in orange and the best fit linear line to that data is also shown.  The slope is 34,753.  That slope represents the number of new infections per day that best fits the data.  Taken with the data above, we can then expect the number of daily new infections to be about twice that in about two weeks, and that's assuming everything just stays as it is now.  So, about 70,000 new cases each day two weeks from now.  Every indication is that this is a conservative number. 

Now onto the death statistics.  The average number of daily deaths continues to decrease, but it's becoming increasingly clear that they are not dropping as fast as would be expected by Gompertz model (shown as red crosses in the daily death chart).  Daily deaths will lag by at least two if not three weeks the number of new cases, so the numbers from yesterday represent the number of people that died from infections acquired in the early part of June.  This relatively slow decrease in deaths mirrors the relatively slow decrease in infections at that time.  It's going to take a couple of weeks before we start to see the average death numbers start to reflect the dramatic rise in new cases.  Even if we locked everyone in their houses today, a spike in deaths is almost inevitable.  This is why politicians that are making policy decisions based on current numbers are three or so weeks behind the curve.  Today's numbers tell you about the mess you are going to be in two or three or four weeks from now and there's nothing that can be done about it.  Policy changes today only show effect three or more weeks after that fact.