Up a little later than normal tonight because I decided to dive into the COVID19 stats a little. It's been awhile since I've had much of an opportunity to look. I'll do my best not to let my own bias get in the way, because the more I look, the more I'm sliding into the camp of "the media is feeding us a big old dose of scary quantitative data, without context of correlations or comparisons."
Hopefully, over the course of this weekend, I can dive in here and really build a strong database of information. If I can do that, I may be able to put something pretty simple up on google docs or one drive, where you can use a drop down menu to see some different stuff. No promises though, I'm basically handling 2 jobs in the healthcare industry right now and even my weekends change with 1 call/email.
Here's some quick hitters from comparisons between the following dates, 6/1, 7/1 and 7/17. I may jump around a little, but feel free to ask questions and I'll answer them tomorrow after my girls' softball games.
(When I discuss vs other nations, this only includes the top 50 most populated countries, as not to let a 1 off like Qatar mess up data due to small sample size, etc)
- As of 7/17, the US is the 3rd most aggressive country in terms of test per 1m. (UK #1, Russia #2)
- As of 6/1, we were behind Italy and Spain as well and fairly close to Germany and Canada. We pulled well ahead of all 4 since.
- From 6/1 to 7/17, # of tests increased by 164.8%
- From 6/1 to 7/17, # of cases increased by 105.2%
- From 6/1 to 7/17, # of deaths increased by 33.8%
- From 6/1 to 7/17, # of active cases increased by 58.6% (due to 221% more recoveries vs only 105.2% more cases)
- From 6/1 to 7/17, # of Serious & Critical cases decreased by -2.6%
- From 7/1 to 7/17, # of tests increased by 36.4%
- From 7/1 to 7/17, # of cases increased by 38.3%
- From 7/1 to 7/17, # of deaths increased by 9.2%
- From 7/1 to 7/17, # of active cases increased by 29.2% (due to recoveries continuing to out-pace new cases on a % basis)
- From 7/1 to 7/17, # of Serious & Critical cases increased by 4.6%
- You can't subtract the % change in the 7/1 to 7/17 from the other period, so don't try. That's just not how it works.
The above is pretty raw information, here's what I'd say with about an hour gathering it and 30 minutes combined thinking about it and typing this here...
- The death variances relative to cases/tests are great... or NY and surrounding states were just that much of a shit show (this is really the truth when you see they hit over 1.2k deaths per 1m in NY, NJ, MA, CT and RI).
- If I gather all of this daily data, then I also need to find a way to pull out those states.
- The data, when looked at it like this, supports the argument that "we're testing more, therefore we've got more cases."
- But... there's cause for some concern in here. For the larger period, that argument holds water. When you look at 7/1 to 7/17, new cases slightly exceed tests. Now, it's not a major alarm because it's close enough (1.9% disconnect) that it should be watched.
- The decrease in serious & critical over the longer period was a nice surprise. This could be driven by the north east states running into early June though. If you consider the 4.6% increase is in July is included in that -2.6% overall decrease from June, then it seems to indicate that's the driver. I tried to check state level data, but they don't have it posted for these cases, so i just have to assume here.
- Many states opened up either May 1st or roughly May 15th. We saw the death curve take off right at 2 weeks after cases took off and we've held a pretty consistent trend of a 2 week gap in relative movement.
- With that said, if you include a full 2 weeks for incubation/increased spread, add the 2 week gap for the death curve, if we were going to see a spike from reopening, it would have hit no later than mid-June.
- If you want to consider the protests as a pivot point, those ramped up around the 2nd week of June, which would lead to expected spikes
around Independence Day.
- Deaths relative to that weekend we start around Aug 1.
The plan, time permitting, would be to pull data for every day, at the same time, going back as far as the the inputs are consistent. Then look at testing, cases, deaths and status through % change and # change, for consistent periods (weekly total, day avg weekly, day avg monthly, etc). Depending on how the data is setup, I could create a pretty simple rolling 4 week or 5 week system that would vet out the consistencies in growths, as well as correlation between cases/deaths. All of that would be up to how clean the data formats are. When doing this stuff, that can be the most time consuming thing and time is the 1 factor here that I don't have much of right now. With that said, I'm glad I've taken the time to do this tonight, because it does mostly confirm my hunch so far. More to see though.
TLDR: If you care, you'll go read it or at least skim the bullets. If you want to be a douche, you'll find a way to take something completely out of context. For others... we're testing a shitload and high level, tests go up, cases go up, but deaths and serious/critical haven't went up nearly as much.