State of the States *

Zorgi Scores for July 8, 2020

Today I thought I’d walk through how I use the Zorgi Score calculation table (SDT4) to analyze how a state is doing. Remember that if you really want to get into the guts of the calculations, there’s a complete explanation of the metrics here and the calculations here.

Since I live in California, let’s start there. I’m going to write this pretty much as I look at it for the first time on any given day. Hopefully, you’ll see why the Zorgi Score is a useful analysis tool for me, and even better, it might become that for you!

First, the chart. I notice that CA is not the lowest, but one of the lowest. That said, 12.4 is not really that low a score. It just looks good when you compare it to AZ at 24.2. However, when I complete the Zorgi Score calculations for countries, I suspect that the score for Denmark will be somewhere around 1.0, while the score for Brazil will be around 25, pretty close to where AZ is now. Yes, that means I think the situation there is comparable.

Is GA’s lower score justifiable? After all, aren’t they one of the states that opened up early? Shouldn’t they be doing terribly? Here’s where the Zorgi Score calculations help me avoid motivated reasoning. GA got fewer points than CA for change in daily cases – 98% increase vs. 120%, so that seems reasonable. GA got more points (3.3) for their decrease in doubling days; CA only got 2.4 points. They both got a lot of points because a doubling days should be increasing quite a bit, not decreasing.

In daily fatalities, CA went up by 17%, but GA went down by 29%, so CA got 1.2 points, while GA got 0 points. For fatality doubling days, CA got 1.0 points for a mediocre increase of 53%, while GA got more points, 1.5, for an even more mediocre increase of 43%.

Both states got the same score for testing increases – 0.1. They also got the same score, 2.0, for tests per case. CA was a bit better at 19 than GA was at 10 but the range for that score is 10 to 19. Is that a good number? No. China, for example, is around 850. Germany is around 50. So 19 is not good. The flip side of tests per case is the positivity rate, which was 8% for CA and 14% for GA. If either state were doing enough testing, that rate would be well below 5%. And without a proper rate of testing, we know that case tracking is not happening as much as it should.

Next we come to what I’m sure to some of you is a controversial “adjustment.” CA gets 3 points deducted because they make their hospital data public and easily available; GA only gets 1 point deducted because while they do publish hospital data, they do not publish ICU data. I have found over the past 100+ days that localities that refuse to publish hospital data (assuming they’re at the county level or higher) are obfuscating the real situation, which is typically much worse than the data would indicate.

The same kind of adjustment is assessed for publication of ICU data. Here CA gets a 2 point deduction, while GA gets 1 point added. ICU data is important. Once ICU’s reach their maximum capacity, it’s quite probable that people will start dying because they can’t get care for routine emergencies – heart attacks, strokes, etc.

Also, these penalties offset some of the points that might be added if a locality publishes their data, but gets dinged because the results aren’t very good.

In terms of patients, GA does get hit much harder than CA. GA had an increase of 170%, earning them 3.1 points, while CA had an increase of 71%, giving them 1.3%. In ICU patients, CA got an additional 1.4 points for their 48% increase, while GA, because it didn’t publish any data, got zero points. That’s where the adjustment evens things out.

The HUR for both states was pretty good – 10% and 8%. The IUR for CA is also fairly good at 28%.

So, after reviewing all the metrics, it seems fair to me that CA has a slightly higher score than GA.

Going through the numbers like this, you can look at any state’s score and see for yourself if you think it’s justified. Remember, the Zorgi Score is an analytical tool, not an epidemiological tool. Think of it as a catalyst rather than a predictor.

The data in SDT2 is also useful in examining the ubiquitous question, “Are all these cases just from testing?”

Start with the row, “Max. possible cases from Testing”. For AZ, you see the number 612. This number is computed as follows. Start with the row in SDT1, “Daily Cases @ Start” – in the case of AZ, 1,079. That’s the 7 day rolling average of cases per day on June 8. Then go to the row, “Percentage Change” just under “Daily Tests @ End”. The value here is 52%. That means testing increased by 52% over the course of the month. So even if every single test resulted in a case, the maximum number of cases would be the cases at the start of the period – 1,079 – plus 52% of that number:

Max possible cases from testing = 1,079 + (1,079 * 52%) = 1,645 cases.

How many cases did AZ have at the end of the period? They had 3,503. The row, “Difference in cases” shows that number, less the maximum possible cases that could have come from testing alone:

3,503 – 1,645 = 1,858 cases that could not have come from testing. The row “Pctg. Cases from Testing Change” tells us that only 47% of the increase in daily cases could have come from testing, even if every single test in the increase of testing resulted in a new case, an extremely unlikely prospect.

Another way of looking at this is through the positivity rate. If cases per day stayed exactly the same, i.e., at 1,079, but testing increased the way it did, what effect would that have on the positivity rate? The next row, “Positivity Rate if Cases Stayed at Start” shows that: it would have been 8%, However, the final positivity rate was not 8%, but a whopping 27%. The row, “Difference in Positivity Rate” shows the difference – 19%. Actually, if I put that in pure percentage terms, it’s a 225% difference, but I thought that would be confusing.

Note that in the case of every single state, testing does not explain the increase in cases. The closest it comes is in OK, where the difference was only 387 cases. Even in TX, where testing went up 124%, that only accounted for a maximum of 46% of the increase in cases.


Now that county and state sections are humming along, I’m working on city sections for San Diego. LA readers: do you know of a site that has daily and historical data for LA county cities or neighborhoods that is downloadable, preferably though a json API? If you do, let me know, and I can add that to my daily update.

The San Diego city update will include the following groups and cities:

  • North Coast: Carlsbad, Encinitas, and Oceanside
  • North Inland: Escondido, Fallbrook, Poway, San Marcos, and Vista
  • East County: El Cajon, La Mesa, Lakeside, Lemon Grove, Ramona, Santee, and Spring Valley
  • South Bay: Chula Vista, Imperial Beach, National City
  • Central: San Diego city

There’s no fatality or hospital data for these, so I’ll be reporting on cumulative cases, daily cases, prevalence index, and doubling days. Hopefully, I can construct a Zorgi Score using just these metrics that will make sense, but we’ll see. I’m hoping to have all this done within a week.

Thanks to all of you for your great comments on yesterday’s post about attacks on public health officials. Tomorrow, I’m planning to do a similar post about attacks on “essential workers.”

Stay healthy and safe, everyone!

Comments from Readers & My Responses

The following are some of the comments from readers on Reddit and other social media platforms where I regularly post. Reader comments are in italics and color. My responses are in plain text. If there is more than one commenter without a response, they are separated by different colors.

I’ve been lurking on this sub without signing in for months now. I finally downloaded the Reddit app just so I could comment on your posts 👏🏽 You’ve been doing such a thorough job in presenting this data, I think it’s beautiful! Thank you for all you do.

Thank you as always Zorgi. I see that the case Rate per 100,000 for San Diego County Residents (ages 20-29) is nearly 1000, with over 24% of cases being in that age bracket. Statistically (I never took statistics) but what can we extrapolate from this and do you have a guess as to why its so high among that age bracket?

There is an increasingly higher number of young people ending up in the hospital all over the country. I have a feeling that it’s because older people know it’s deadly, but a lot of young people think they’re immune. Also, young people are far more likely to get together in bars and at inside parties, where a single superspreader can get everyone sick.

Your posts are all I’m really paying attention to regarding Covid-19. Too much hype and fear mongering from regular news sources. The data doesn’t lie. Very calming analysis. Thank you!

Strangely enough, it calms me down too. As you might tell from my political commentary, I’d probably be a raging maniac if all I did was stew about how insane our response to the pandemic has been compared to other countries. If I had school age kids, or if my wife was returning to her teaching job, I’d be very nervous now.

2 comments on “State of the States *

  1. They dwarf the flu. The death toll from COVID is twice that of the highest fatality flu years, and that’s before you factor in “excess deaths.” The seasonal flu doesn’t bring emergency rooms to their knees either. For more on similarities and differences, see https://www.hopkinsmedicine.org/health/conditions-and-diseases/coronavirus/coronavirus-disease-2019-vs-the-flu
    Also, we’re not nearly done with COVID. We’ve still got at least 4 or 5 months to go, and very possibly much longer than that.

Leave a Reply

Your email address will not be published. Required fields are marked *

 

This site uses Akismet to reduce spam. Learn how your comment data is processed.