A path out of the pandemic

Apr 17, 2020
How are we going to get out of this coronavirus pandemic? One thing is for certain, we're not going to social distance this virus out of existence. I don't know if the public at large thought that if we social distanced ourselves the virus would eventually go away. No, that is not the point of social distancing. Social distancing was meant as a measure to help prevent our healthcare system from being overwhelmed. The actual act of of social distancing is preventing us from really moving forward with battling the virus. That's not to say that all social distancing measures should be removed right now, it's just saying that the longer a population uses social distance as their primary means to battle this virus... the longer you're going to be practicing social distancing.

A few things to look at before going in-depth with all of this:

Why are 80% of the people that are being tested coming back as negative (don't have the virus) in all of the molecular (active) tests? I don't have anything to compare this to and I'd love to know how this stacks up against other viruses and diseases testing. This may in fact be the expected behavior. But just the number... seems high to me. Are we testing too many people that don't need to be tested? Is the virus not that contagious? Is this molecular test really that accurate? Are we double dipping too much in the number of tests administered? Are unique individuals being tested twice, thrice, or even more often? Keep in mind... just because an individual tests negative for the virus, that doesn't mean they won't contract it later. It just seems to me that a lot of people are being tested and coming back as negative.

The governments (United States) of the country have been ridiculed for their lack of response to this. Some of that is warranted. Although, from best I can tell, the ridicule comes from the lack of these molecular tests. But I would argue that the ridicule should be aimed at the lack of planning in response to this virus. A widespread molecular test would only be feasible if you could test EVERYONE in the country at the same time. That would be 320 million+ tests all done at largely the same time. This was never going to be possible - the governments (federal, state, local) should have realized that this was never going to be feasible. Instead the goal of the government should have been focusing on reliable antibodies testing. Money should have been thrown at developing and testing a reliable antibodies test so that you could push that testing out widespread, 6 to 8 weeks after this all started. Antibodies testing may or may not mean immunity... but it would at the very least tell you how widespread the virus has reached. If a percent of your population has the antibodies but never experienced any symptoms or had very mild symptoms, then that would alleviate some of the fear that goes along with this virus. And just to note, the fact that 80% of people that are being tested come back negative for this virus would seem to contradict some of the antibodies studies that suggest that this virus is more widespread, that is something that needs to be studied.

Further the governments should have recognized that basing this test at a state level was much to broad. A state level test might work in the smaller states of New England, but as you get more and more out towards the Midwest, the testing needs to be a bit more finer. Doing it by county is much too small (there are over 3,000 counties in the United States). I would suggest doing it by US congressional districts - mainly because these are already defined. It's not exactly fair to take a state like Texas and treat it all as one. Is it fair to the residents of Pecos, TX to have to suffer through restrictions because the number of cases are high in Houston? This is another miss from how the governments should have been planning for this.

Just to be clear, this is not a political issue. I'm not saying either party is to blame for this. There are states with governors from both parties and none of them would appear to have done any planning like this. I'm saying the governments as a whole failed in this planning. But also to be fair... hindsight is 20/20.

The problem we are facing right now is an extreme lack of data. We've social distanced ourselves to the point that no new data is coming in. Are people immune to it? Is the virus as contagious as we once believed? Are the symptoms that bad? We're not going to get any new information until we change things up a bit.

Now, seeing as how a widespread reliable antibodies test is probably not viable - at least in the short term, how can this situation be remedied? I would focus my attention to the percentage of hospitalizations of confirmed cases. Or if you really wanted to take it a step further, the number of people in the ICU of confirmed cases. The number of confirmed cases isn't telling us a whole lot right now. You can have a confirmed cases, but not affected by it very much. You could have a case of the virus, but not be confirmed (because you didn't get tested). You could have a case of the virus and not be showing any symptoms. You could have had the virus weeks ago and be recovered. But if you're in the hospital - or specifically in the ICU - and you have the virus, then you have a really bad case of the virus. That's not to say that it's peachy having it while you self-isolate at home. But if you've got it, and you're able to stay at home, then at worse it's probably like a really bad case of the flu.

Then to add on to this, I would again look at this data in terms of US congressional districts. Again, this would have ramifications in how to potentially treat a town like Pecos, TX vs. Houston, TX. If you've got a better way of dividing this up - I'm all ears.

I would also weigh in hospital capacity (or probably more specifically ICU capacity) in those divided areas and how much excess capacity those areas have.

All of this together would tell you which areas could potentially be opened back up. If an area has a relatively low percentage of hospitalizations - then it stands to reason that that area is not experiencing a widespread outbreak. If that area also has excess hospital and ICU capacity, then this would seem to be a prime area where things could be cautiously opened back up. That doesn't mean that an outbreak won't happen in that area. But that area would appear to have hospital capacity to treat people if the number of cases start to rise. And if a large outbreak then happens, it may require the area to be closed down again. But either way, it tells you something. It gives you more data that you can feed into the model about this virus.

To attempt to model some of this, I took data from the John Hopkins website - from April 21st - and tried my best to fit this into a useful display of information. There's a few deficiencies with the JHU data. Mainly, it doesn't give data for active cases in each state. It only gives information about confirmed cases in each state. It also does not give information about recovers in each state. Calculating active cases should be a simple matter of (confirmed cases - recovered cases - deaths). Also the JHU website lists deaths by counties or city and not so much by state. Since I don't have recovered cases and deaths for each state (and this could easily be a reporting issue and not necessarily with how JHU presents the information), what I did was took a percentage of active cases to confirmed cases from the total US population and applied that to each state's confirmed case number. (816,240 total US confirmed cases at the time I took this information, 44,228 deaths, 75,177 recovers ... subtract all of that out and you get 696,835 ... 696,835 divided by 816,240 equals about an 85% active case rate. Multiply each state's confirmed case by 0.85 to get a "guess" at how many active cases exist in each state. Kentucky had 3050 confirmed cases, multiplied by 0.85 equals 2603.83 "active cases guess"). This is fudging the math I know - but it's all I've got to work with. If there's a website that gives me the an accurate number of active cases and the number of active cases in the hospital for each state, that would definitely be preferable. I also took a snapshot of the hospitalizations from JHU for each state one day (April 16) and then again on a later date (April 21) and in some states the hospitalizations decreased... this told me that the hospitalizations figures JHU is reporting is a function of active cases and is not a growing number of total hospitalizations ever within the state - although I suppose that information could be a reporting issue explaining the reduction. The bottom line of it, I really need more accurate and better defined data to build this... but this is the best I've got. Also, some states did not report hospitalizations on the JHU website, so I had to remove them - I think Indiana was one such state - there were others but that's one I remember.

The information looks like this:

StateActive (Guess)HospitalPercentStateActive (Guess)HospitalPercent
Kentucky2603.83105940.67District of Columbia2644.8040215.20
New York220566.245710325.89Connecticut16916.33191911.34
Colorado8631.90188021.78West Virginia775.17779.93
South Carolina3789.6377620.48North Dakota549.79539.64
Virginia8221.26158119.23New Jersey76748.8369869.10
New Mexico1682.6729117.29Texas17302.2114118.16
Arkansas1698.8929117.13Rhode Island4695.423317.05
Illinois28222.91459916.30North Carolina6077.593736.14
New Hampshire1235.3220116.27South Dakota1498.27875.81

Now, looking at this table, this shows that Kentucky is actually faring pretty poorly with 40% of their cases being hospitalized. But also likewise, Washington state only has 4.71% of their cases in the hospital, but that's still 503 people. Does Washington state have the hospital capacity to handle 503 (presumably ICU) patients with excess capacity? (I really don't know). Again, this is probably why a more granular data set would be preferable... like US congressional districts, but this is the information I have to work with. I would arbitrarily consider any state that has less than 20% of their cases in the hospital and less than 100 total hospitalized would be areas that could potentially handle being opened up a bit. That would be the states of: Alaska, Montana, Wyoming, Hawaii, West Virginia, North Dakota, Vermont, South Dakota, and Guam (not a state, but the information was there). Maybe the "under 100 total hospitalized" is way too low... I don't know, I'd leave that up to people with a bit more knowledge of this information. And again, if this was divided up in congressional districts then this might yield different results. While the hospitalization rate in Kentucky is at 40% for the whole state, that might not be true for the far western end of the state where there's not a lot of densely populated areas. Again, I'm kind of assuming that all hospitalizations are ICU cases - that may not be the case and having a true number of ICU cases would probably be better.

As those states or areas open up, see what happens. Monitor those areas for 2 weeks or 4 weeks, what happens in those areas? No doubt the number of confirmed cases will likely go up, but to what degree? And is it manageable? And then you start to roll this out to other areas, which hopefully have seen their hospitalization numbers decrease while this monitoring is going on. Or hopefully you get reliable antibodies testing available by then and you could start to base this opening up on which areas have seen a widespread number of cases.

This would be what my plan for opening things up would look like. And this is where the lack of planning by the governments seem to be hurting. But I just don't see where the governments really had any plans to combat this. And again I want to point out that the actual numbers I used may not be the best numbers to use - but it's all I had. The logic behind using those numbers (whatever they might actually be) still remains the same. More accurate numbers is going to lead to a more accurate model.
Mar 4, 2020
This episode ain't over, and it's going to be hell to analyze. We traded our normal society life temporarily, to lower the death count of our most vulnerable, until we could come up with viable treatments. Modern medicine has allowed formally fatal conditions to become chronic conditions now. This increases the number of vulnerable people.

Just the vulnerable can overwhelm our present lean system. And a bunch is added every year.

This delayed the normal population progression of the event. This alone will complicate the comparison to other outbreaks. And judging the strategy.

The data and actions will be debated for decades. But we will experience the effects.

The event is truly historic. And worrisome.
Apr 17, 2020
Just to add to this. I think my data is either outdated, inaccurate, or I misinterpreted it.

Saw a report out of Kentucky today. There are 3,481 total cases there. 1,335 have recovered. 191 have died from this virus. Subtract all of that out and you get 1,955 active cases in Kentucky.

That same press release stated that there were 302 in the hospital. And 163 of those are in ICU.

So that puts Kentucky's hospitalization rate at 15%, that's well off of the 40% I stated in my post. If you looked at ICU cases, that's 8%.

So I'm really starting to think that the data I used to make that table is wrong or needed to be updated. But the procedure is still the same, just feed it in with the correct data. I don't suppose anybody knows where I can get a hold of this type of data for every state in a single location, with a pretty ribbon on it?
Mar 4, 2020
Isn't that data suppose to be on the CDC site? I have tried to use that site before, but I have problems navigating it.

But it will be just a snapshot. We'll have to wait for awhile, to make true sense of it.
Apr 17, 2020
It may be on the CDC website... but I can't find it. Maybe I'm not looking hard enough. All I really see on the CDC website is a bunch of graphs where they probably used data like this to create. But they don't actually provide the raw data.

But... um... yea... it's always going to be a snapshot until we're talking about this event using past tense. But that's the point of building models. If you're waiting for this to be over before analyzing the data, you're going to be waiting a while... a long while. The point of building models is watch trends and see what is happening now.
Apr 17, 2020
OK, I found better data (I think) at:

Although, it shows Vermont with 825 confirmed cases and 1,710 recovered. Recovered cases can't be more than confirmed cases. One thing that whole thing has illustrated... there's a severe lack of funneling data into one spot.

But this data also shows the correct Kentucky information, so I'm going to assume that it's a bit more spot-on for currently active cases.

StateActiveHospitalHospital %StateActiveHospitalHospital %
West Virginia57010818.9474New Hampshire1068928.6142
Washington DC257440215.6177New Jersey9462172407.6516
Kentucky195530215.4476New Mexico15921217.6005
Illinois35246487713.837New York223833150216.7108
California35900492913.7298North Carolina73554866.6077
Michigan29077361112.4188South Dakota883586.5685
Iowa233628212.0719Rhode Island57252674.6638
Delaware257329011.2709North Dakota425184.2353

Again, not all of this information is available for every state.

Looking at how the data is distributed and without knowing the capacity and remaining capacity in each state's hospitals system. I'd say any state that has less than an 11% hospital rate and less than 500 people in the hospital is probably safe to consider opening things up a bit.

This would be the states: Montana, Maine, New Hampshire, New Mexico, Wisconsin, North Carolina, South Dakota, Arkansas, Rhode Island, and North Dakota

This website also reports ICU numbers - although not a lot of states appear to report that. But the information for that would be:

StateActiveICUICU %
West Virginia570447.7193
Washington DC25741204.6620
New York22383350162.2410
New Jersey9462119902.1031
Rhode Island5725721.2576

Although, there's just really not a lot of data to go on here. But of the states reporting, and of the states that would potentially be able to open up a bit, Maine reports the highest ICU rate with 4.4% of active cases for a total of 18 ICU. Wisconsin has the most in ICU at 146 or 3% of their active cases. Does Wisconsin have excess ICU capacity? That might give them pause on opening things up.

But if your state is listed in one of these tables and you're wondering why your state isn't opening up, I'd consider looking at where you fall in these tables and that might help you to understand why it's not opening up and what needs to happen before it can consider opening up.
Mar 4, 2020
I'm guessing analyzing this data is a lot like the art of troubleshooting. Troubleshooters are called in when previous attempts and mitigations have been tried. This colors the current symptoms of the problem. A good troubleshooter has to filter the previous attempts, to track down the actual problem.

It will be a cluster-buster.