COVID-19 September 17, 2020

It's been almost two weeks since I last posted.  The Labor Day holiday is primarily responsible for the silence.  First, the numbers coming in were rather skewed.  It was clear that testing, labs, and reporting were also respecting the holiday.  Second, I wanted to wait a week so that the holiday was in the rear view mirror and it became possible to better compare week over week numbers.  

 

Daily cases did indeed drop most of last week and have since recovered this week.  Not only have they recovered, but they are starting to rise above the numbers from two weeks ago.  I suspect that new cases are going to at least level off in the next week, and I would not be surprised if they start to rise again.  People are getting weary of quarantine.  Cell phone tracking indicates that people are going out and interacting more and more.  Couple that to the holiday and the return to school and the ingredients are in place for another rise. The death rate is also starting to tick up again and is now on par with the numbers prior to the holiday.  It would not be surprising to see that trend also continue.  

Next week will be key to seeing how holiday interactions and the return to school are impacting the numbers.  I'm not optimistic.

 

  

COVID-19 September 04, 2020

It's been about a week since I last posted.  In that period, the number of daily new cases has nearly leveled off at about 40,000 per day.  It would be interesting to see how much of that is due to the return to school.  There have certainly been a large number of reports of outbreaks at schools and universities.  It's curious, though, that some of these don't seem to be reflected in the individual state numbers.  For example, over a thousand students were supposedly diagnosed at the University of Alabama within the last couple of weeks.  But there is no significant change in that state's trends over the last several weeks.  In fact, Alabama posted a paltry 86 on September 2, which is the lowest since March 31.  Something is amiss.

The death rate continues downward, but the rate of decrease is slowing.  We are now down to about 850 deaths per day.  I expect the deaths to level off in the next week or so given the leveling of new cases.

 

COVID-19 August 26, 2020

We are now solidly past the peak of the second wave based on death rate, which has now been slowly decreasing (7-day average).  New cases have also been on the decline.  And now we wait to see if wave #3 emerges as students head back to school.

I read yesterday that hospitals will, once again, be sending their data directly to the CDC instead of HHS.  At this point I'm not sure it matters any more.  The data collection and distribution is so completely screwed up that it may not be possible to conduct meaningful analysis in real time.  It's going to take researchers months or years to dig through the records we do have and sort it all out.  I continue to get my data directly from the states, but just to illustrate how crazy those data are, I refer you to my state of Colorado.  On August 21, Colorado dumped 3,959 new cases into the database in one day.  Clearly, that's a backlog from many, many days (or weeks) prior.  What am I supposed to do with that?  Colorado is not unique.  

I'll be back again in a week or so, or sooner if something changes.  Stay safe out there!

COVID-19 August 17, 2020

 I'm struggling to get daily or near-daily reports out due to other obligations (a.k.a. my day job!).  Also, considering that there are generally no big changes from day-to-day, and that individual daily numbers shouldn't really be examined I'm going to post less frequently.  

As of today, cases and deaths have leveled off.  The daily death data still looks rather odd to me.  That doesn't mean it isn't real, but it's odd. All through July the death rate was increasing.  Then, at the end of July there was a small bump downward.  Following that the death rate has been near constant.  Why would the upward trend just terminate with a drop and then go constant?  Remember that this coincided with the hurricane in Florida and up the East Coast as well as the system changeover in Texas. 

COVID-19 August 13, 2020

 Rather than posting a rather long narrative, I'm going to drop this article about testing and reporting issues in various states:  https://www.cnbc.com/2020/08/12/accuracy-of-us-coronavirus-data-thrown-into-question-as-decline-in-testing-skews-drop-in-new-cases.html
 

Things are a real mess.  Actually, things are FUBAR.  While everyone is/was worried about the federal government fudging numbers after bypassing the CDC, the real story, I think, is at the state level.  The daily case numbers and total cases are becoming increasingly meaningless with time.  The death rate is still more robust, but that's a relative assessment.  How many different ways and our local, state and federal governments fail us?  

 

COVID-19 August 11, 2020

Wave number two is leveling off.  Sort of.  New daily cases have been dropping since about July 19.  Within the last week that trend has stopped and we've reached a plateau.  We'll have to wait and see if that's a temporary pause, whether we'll continue down, or whether we'll start on a wild ride up again.  Interestingly, California came in with big numbers yesterday: 12,500.  That's the highest number since July 22.  It's just one day, so it may not mean anything.  It's something to watch, however. I'll also be watching what happens as kids return to school.

The death rate dropped last week around the time the hurricane threatened Florida and when Texas switched to a new reporting system.  You can see the drop in numbers around day 212.  After that it plateaued.  This is an odd behavior, but it could be related to Texas and Florida numbers catching up.  We will have to see if this is a temporary pause or whether we're about to go up or continue down. I expect deaths to mirror the new cases with a lag of about three weeks.  To see deaths plateau around the same time as the new cases, with effectively no lag, is odd.

 

COVID-19 August 07, 2020

It looks like the death rate has peaked, but there's a caveat.  The peak coincided with the data anomaly that was likely caused by hurricane in Florida and the reporting system upgrade in Texas.  To my eye, there has yet to be compensation for the data that went missing earlier this week.  You can see the effect in the daily new cases (on the linear scale).  There's a big dip noticeable just a few days ago, but there has not been a reciprocal rebound.  There's a similar signal in the daily death rate, and it has caused a downward offset in the 7-day average. New cases peaked around July 19, so the timing of the peak in deaths is a bit early based on prior lags of about 3 weeks.

Reporting issues aside, it does appear that when masks and social distancing programs are in place it can have a noticeable impact. And the opposite is true.  When all caution is thrown to the wind the numbers can explode.  It's going to get interesting as schools open across the country with varying amounts of precautions in place.  We are suddenly going to have a large segment of the population mixing it up (for those places where school is physically going back into session) like we've never had before.  Other countries have run this experiment to varying degrees with not-so-good results. 

COVID-19 August 05, 2020

Several followers provided additional info related to the numbers from yesterday.  I extend my gratitude to everyone that chimed in.   First, the Florida numbers are likely explained by the suspension of activities due to the hurricane.  That's so obvious an explanation I can't believe it didn't occur to me.  And, apparently Texas was in the middle of upgrading their reporting system, so that explains their missing numbers.  Presumably, we will see things return to normal by the end of the week, and we'll see if the cases and deaths that weren't reported start to show up.

There has been a pretty good rebound in the number of cases and deaths.  This is partly due to the usual upward bounce in the weekly oscillation, but it may also be a result of Texas and Florida adding in previously unreported numbers from the last day or two.  The anomalous numbers have caused a decrease in the death rate. 

It is interesting to note that the hurricane and system upgrade issues were revealed in the data even though I didn't know the source of the anomalies at the time.  This supports my earlier proposition that if there is any funny business with accounting it will likely show up in the statistics. 

COVID-19 August 04, 2020

Apologies again for my inability to post daily.  I do have a day job.  This is all mostly morbid and self-preserving curiosity, but I don't get paid for it.

Back when the Feds ordered a change in data reporting I noted that the data itself would likely show if there was any funny business.  Well, I'm not saying there's funny business, but things don't look quite right.  It could be real. It could be data problems not related to the CDC circumvention, or it could be monkeying around.  I can't tell, and either can you.  At least not yet.  I'm loathe to initiate or validate conspiracies, but I think some skepticism is warranted. 

New cases reported over the last two days were down.  This is consistent with the general downward trend that's been established for well over a week, and it's also consistent with the weekly oscillation, which is usually near the bottom on Sunday and Monday.  What's shocking is just how down it is. It's a very large drop that appears to my eye to be out of family with pretty much all the data since this mess began. Yesterday's total came in at 41,963.  The day before it was 53,561. It's been a month since numbers have been that low, and the drop from the max is far faster than anything we've seen.

Digging into the numbers and individual states a bit more, I went straight to Florida, Texas, and California since these "big three" have been driving the numbers of late.  Texas is on the decrease, but the numbers don't appear to be too out of place compared to the previous week.  For example, last week they bottomed out with 5,810 and 4,267 compared to this week's 6,226 and 5,303.  Both this week and last week are down compared to the prior weeks, but this week doesn't stand out as being anomalous compared to last week.  The same goes for California.  It's down, but not in any obviously odd way compared to the prior week.

And then there's Florida.  Yesterday's number came in at 4,752.  I've cautioned before about looking at individual daily numbers by themselves.  That missive still holds.  But, let's compare that number in the context of the prior number from a week ago: 8,892.  It's down by almost a factor of two compared to last week.  In fact, Florida hasn't reported a number this low since June 23. 

If I then move to some of the other smaller population states, I start to see at least some things that jump out at me.  Tennessee is somehow down by something over 50% in a week.  Arizona, down by something like 50%.  Missouri, down by 40-50% depending on how you do the accounting.  And so on, in several other states.  It just looks odd.  Could be real. Or not.  I think we need to wait and see how things continue to trend and then wait for an explanation from those that do this sort of thing for a living.

Deaths have shown a similar, large drop over the last couple days.  It's a bit unexpected, but compared the amplitude of prior oscillations it's large but not necessarily unrealistic.  It has caused the 7-day average to drop noticeably over the last two days. Death numbers are harder to fudge than infections, but it's not impossible.  All it takes is a reclassification of a COVID death to be something else.  The dead are almost always counted, but not necessarily properly binned.

What's driving the drop in deaths? To start with Texas failed to report at the end of the day on August 2.  From a statistical standpoint, that shows up as a zero.  I might have expected yesterday's number to come in higher to account for that non-report, but it didn't.  Where did those deaths go?  Will they show up in the next few days?  That's a lot of dead people gone missing. If I were to add back in a couple hundred deaths to Texas, things start to look more consistent with the prior trends.  The numbers for Florida are down for the last two days (as expected due to the oscillation) and are comparable to where they were a week ago.  Same with California.  So, this could all be a statistical aberration.  But, at some point, we should be asking to where the Texas dead disappeared.

COVID-19 August 02, 2020

It's been a busy week.  Apologies for the spotty posts.  Here are the data based on yesterday's numbers.  New infections are continuing to slowly drop, but remember we are still nearly double where we were when wave #2 began.  Deaths continue to increase.