Crunching U.S. Covid 19 Numbers

NOTE: This post is not about PennEast.

So this week I started looking heavily into the John Hopkins Covid-19 website visualizations. They are extremely useful to help understand what’s happening with the pandemic both in the U.S. as well as world wide.  In particular, I wanted to focus on State-by-State coronavirus daily infection curves, and what influences those curves.

However, I had some questions that the visualizations couldn’t easily show, such as deeper insights into what the curve looked like across the United States, how opening/closing of States has impacted the curve, etc.

Note that this post doesn’t talk about hospitalization rates, mortality, age or ethnic groups, or similar data.  Those topics are important, but for another time.  This post is all about the curve.

Summary

The summary of this research is pretty simple.  States that opened up places like bars, gyms, or pools prior to the Memorial Day weekend are now seeing huge spikes in Coronavirus infections.  And we are likely going to see another one after July 4th.  

Even States that had daily case rates of only 100 new cases a day have been hosed when they opened up bars.  It is clear you need near-zero daily cases or extensive contact tracing before certain kinds of facilities can be safely open (the good news is that beaches and most outdoors stuff seems to be OK).  

And whatever you do, don’t travel to the States whose curves I’ve designated as “Rising Steep”.  They are in serious trouble and in most of them the infection rate is far worse than it was at its height here in NJ in April.  

Finally, this is not about fear mongering.  It’s clear where States have ignored science and math or have given into political pressure and opened too early, the results have been horrible.  And in States who have been carefully doing the math and holding firm based on scientific principles, they are weathering the storm exceptionally well.

This is all about being smart and using information to drive decisions, not politics.

United Staters Covid-19 Trends

And now onto the detail.  I was able to pull a spreadsheet together from a variety of sources to help visualize trends in a bunch of dimensions.  See the “Where This Data Came From” section at the end for details and a link to the spreadsheet.

First of all, let’s look at Curve statistics for the U.S.  This is my categorization of the state of a State’s curves today (the “today” part is important).

  1. Low & Flat: 13.  There are 13 states whose curves peaked quite awhile ago, and are now in a “long tail” of a smattering of cases that are keeping them more or less flat for the past several weeks.  These are mostly in the NE.
  2. Flat: 6.  These are states where they’ve been at a steady state of infection for some time.  It is not running out of control or improving.  Note that many Flat States have had a small trend downward and may be about to bump up again.  I expect many of these will transform into  “Second Curve” states soon since the 4th of July weekend.
  3. Second Curve: 11.  The dreaded “camel double hump” curve is there for 11 States.  This is where infections peaked, started coming under control and then have started rising again because they reopened certain facilities too early.
  4. Rising: 11.  In 11 States they never really got the virus under control, and it’s been steadily rising since March, but at least the rise is moderate.
  5. Rising Steep: 11.  In 11 States there has been alarming gigantic leap in cases in the past couple of weeks.

Here’s what these look like.  All examples are from the John Hopkins covid-19 curve tracker website.

A sample “Low & Flat” Curve for today (Example: NJ).

Screen Shot 2020-07-11 at 4.28.11 PM

A sample “Flat” cuve for today (Example: Indiana):

Screen Shot 2020-07-11 at 4.30.56 PM

A sample “Second Curve” state for today (Example: Ohio):

Screen Shot 2020-07-11 at 4.33.22 PM

A sample “Rising” curve for today (Example: North Carolina):

Screen Shot 2020-07-11 at 4.34.28 PM

A sample “Rising Steep” curve for today (Example: Texas)

Texas-Curve

 

 

The Rising Steep states are terrifying.  Here is the infection data from the top 4 “Rising Steep” states over the past two weeks (not since March, just the last two weeks!).

Screen Shot 2020-07-11 at 4.44.29 PM

In total these four States alone have managed to wrack up 379,205 cases in just two weeks.

For reference, at our worst the State of NJ peaked at 3,972 new cases on April 4th.  Compare that to a 14 day average of 3,528-8,657 cases for the four states above, and imagine where they are going to be in 3 months time.

The State of the United States in July is far worse than it was in April, when we thought this thing was at its worst.

Regional Data

When you look at regional data you can also see obvious trends – where you live in the Country makes a huge impact on Corona virus status.   My regional definitions are a bit arbitrary but I think are useful in this context, this map shows what I used:

Screen Shot 2020-07-11 at 5.19.04 PM

…with the exceptions of WV (which I put into the MidWest) and VA (which I put into South East).

Here’s the breakdown.

Screen Shot 2020-07-11 at 5.24.17 PM

As you can see the North East is in pretty good shape.  Most are Low & Flat, except for Maryland (Flat) and Pennsylvania (Second Curve).  

The Mid West has a whole lot of Second Curve States – States that were doing good but then likely re-opened too soon (or saw crowds violating orders).  Only Minnesota has managed to get their curve under control and is Low & Flat.

The South East is a mess.  Only Virginia is flat.  And Floria, Georgia, and South Carolina have disastrous Rising Steep curves.

The Plains States show the full range of curves.  Relatively low-population states like Nebraska and Wyoming have been mastering their curves, while Texas and Oklahoma both have alarming Rising Steep curves (especially Texas, who-boy).

The Western States are a mess like the South East is.  5 Western States are in Rising Steep mode, including California where the Los Angeles/Orange County area is simply getting swamped with new cases daily.

Curves by cessation of Stay At Home orders

Next up we’ll look at how the curves look as related to when any Stay At Home orders were expired or revoked.  The data is fairly mixed, which seems to indicate that Stay At Home Orders in and of themselves didn’t do much on their own (but in combination with other orders and effects, perhaps they did).

Screen Shot 2020-07-11 at 5.41.13 PM

Only 1 State seems to still have a Stay-At-Home order in effect, Kentucky (but it’s not 100% clear).  That State was doing Flat for long time but is now Rising.  

States that lifted stay-at-home in April (early openers) haven’t faired too well.  Only low-population Wyoming has managed to stay Low & Flat.  The others are Rising and 1 Rising Steep (Georgia).

May is when most States opened up and the data is very mixed.  The Region dominates what happens to your State if you reopened in May.

Same with June and those States that did not have a formal Stay At Home order.

Overall, general Stay-At-Home orders by themselves don’t seem to have made a difference.  Closing indoor places of employment probably did, but otherwise it’s less about staying at home, and more about wearing masks, social distancing, and not letting certain types of facilities to open unless your cases are near zero.

Looking at Gym/Bar/Pool Opening Along with Memorial Day

We hit serious pay dirt when you look at the gym, bar and pool status of states in relationship to Memorial Day weekend.  States who had these facilities open on Memorial Day have been generally creamed by the coronavirus.  States who did not have them open have faired directly to their opening dates of these facilities, and, critically, the status of their curves at that time.  Note that beach openings are not part of this data, as that does not seem to have played a role in case spikes that I can see (although it may have contributed when drinking at beaches was allowed).  States like California opened their gyms & bars on June 11th, which is satisfyingly late, but their curve was still slowly rising at the time (they had over 3,000 new cases on the days they opened their bars and gyms).  The result is disaster.

Screen Shot 2020-07-12 at 5.24.44 PM

In general, nearly every state that opened pools, gyms, or bars prior to Memorial Day weekend has seen a severe spike in their coronavirus cases.

In all, 25 States opened facilities prior to Memorial Day.  Here is where their curves are at today:

Screen Shot 2020-07-12 at 4.47.15 PM

The two “Low & Flat” States are Delaware and Wyoming, which are low population states with very low daily case counts.  This group is dominated by 9 “Rising Steep” states along with 7 rising and 5 Second Curve States.  It seems very clear that opening facilities like Gyms and pools with any kind of significant daily new case count going into Memorial Day has been disastrous.

Opening bars specifically has also been a mistake in every case.  Even in States with very low case counts like West Virginia started seeing a spike after they opened their bars.  So unless you are down to effectively zero new cases per day, or have extensive contact tracing in place, opening bars seems to be a recipe for covid-19 disaster.

Oklahoma is another example.  They were averaging about 100 new cases a day for months.  Then they opened bars & gyms in early May.  It stayed OK for awhile, until Memorial Day hit.  And then boom.  Now their daily case count is 7x what it was.

Screen Shot 2020-07-12 at 4.56.33 PM

This means I am really worried about Pennsylvania now.  They opened gyms and bars in a few counties on May 29th, and have expanded that program throughout June.  This “county by county” approach seems like genius until you look at the numbers.  PA’s cases are now on the rise again. Because, you know, people have these things called cars, and they will go to where the open bars are (especially since you closed liquor stores for months!).

Screen Shot 2020-07-12 at 5.00.08 PM

What Didn’t Seem to make much difference

Stay-at-home orders and closing beaches doesn’t seem to have made much difference either way.  I see no correlation to letting stay at home orders lapse or opening up of beaches, other than closing indoor work probably was effective.  In general outdoor stuff where you can disperse seems to not be impacting the curves significantly here in the U.S.  

Limited indoor seems to be OK as well.  Opening malls seems to have not been a problem.

What it all Means

I think it’s clear that strict stay-at-home orders weren’t really necessary and didn’t have a big impact (hindsight is 20/20!).  What does have an impact is masks, and avoiding activities where you are in prolonged contact with others.  Opening gyms, bars, and public pools when your curve is still rising is an incredibly bad idea.  It can only be safely done when your number of new cases per day is vanishingly small, and contact tracing can be done on nearly every new case you see.  

Looking at Hunterdon County and NJ, we seem to be in a pretty good spot overall.  It is likely that schools can be open safely to some degree in the Fall.   But I really hope parents and school administrators keep a firm eye on the state data, and that they will adjust plans if July and August data starts going south.  So far Murphy seems to have done a good job in that department.

Here is NJ’s data right now.

Screen Shot 2020-07-12 at 5.05.29 PM

Looks good, right?  But look at that long tail on the right side of the graph- we are still averaging around 300 new cases a day for the past month or so.

Hunterdon County numbers look even better.  We have the third lowest case count in the State (1,094) and new daily cases are down to a trickle.  But remember – people travel.  As we saw with PA and the closed liquor stores, all that does is give people to travel to States or Counties nearby that have fewer restrictions.  If people push for “Let’s open just Hunterdon because we’re so low!”, that will just entice people from other counties to flock here.  Bet on it.

As we look towards September, let’s hope the school administrators and Board of Ed are diligent with cleaning, air purification, and mask requirements in schools (and really, really enforce them).  Kids seem to be more immune than adults, but that doesn’t mean they all are, and it only takes a few super-spreaders to bring it home to parents and grand parents, family and friends.

For States like PA that are starting to start a second curve, it’s a tough decision.  In States like CA and FL and TX, the decision (I hope) should be crystal clear – opening schools there would be suicidal.  

Where This Data Came From

For those who haven’t seen it yet, here is the link to the John’s Hopkins sites:

The US Dashboard: https://coronavirus.jhu.edu/us-map

The World Dashboard: https://coronavirus.jhu.edu/map.html

Curves by State: https://coronavirus.jhu.edu/data/new-cases-50-states

As I said, all of that is extremely useful, but also somewhat limited. But fortunately we live in a world of highly transparent science facilitated by technology, and the Johns Hopkins researchers have made their underlying dataset available on an open platform (GitHub for the geeks out there):

John Hopkins Datasets: https://github.com/CSSEGISandData/COVID-19

For my analysis I started with the July 9th daily data dump. This includes total confirmed cases of covid-19 per state, along with total deaths, hospitalization, testing rate, etc.

I then folded in the dataset from HealthData.org (thanks Lorraine for the link!).

HealthData.org Dataset: https://ihmecovid19storage.blob.core.windows.net/latest/ihme-covid19.zip

This dataset is fantastic in that it has a trove of information on when States have taken various actions (closing schools, closing non-essential businesses, stay-at-home orders, etc).

I then annotated all this data with some additional data points:

  • U.S. Region a State is in (to allow for regional analysis)
  • Curve State
  • Governor’s Party
  • Did Memorial Day weekend have an impact
  • Date bars, gyms, or pools were re-opened
  • Cause of recent curve spike (if applicable)

A lot of this extra data I took from this great Washington Post page that shows the reopening status and curves for every State in the Union:

https://www.washingtonpost.com/graphics/2020/national/states-reopening-coronavirus-map/

With that info we can do some really basic spreadsheet manipulations you see above.

The resulting spreadsheet is available below.  It includes the probable reason for each State’s spike in cases where they have in fact spiked.  

U.S. Covid-19 Snapshot July 9, 2020: US-CoronaVirusBreakdown-07-09-2020

Published by

Mike Spille

I'm a thinker, an analyzer, a synthesizer. Maybe not in that order. I live in West Amwell NJ with my wife Kristina, our two kids Day and Z, our two dogs Fern and Cinna, and two cats Ponce de Leon and Xavier.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s