19. Weekly Update: The Good Get Better, The Bad Get Worse

(5/28/20) Interestingly those populations international and domestic who got hit the worst, GET IT! They are recovering fast. Elsewhere, it is a slow drip of agony.

This may be my last update for awhile if I can actually resist. The IHME model is now becoming as good as mine and my voice doesn’t carry very far. But sometimes I get punchy and upset, and when I do, I blog! But I promise no political undertones, well maybe, but definitely no overtones. Although there is still a lot of misleading news that comes out of respectable press it is inadvertent as opposed to misinformation campaigns mounted by some governments (that’s an undertone). So, my mission is now less essential.

International News

I am going to focus on what is critical. So, I am dropping China and S. Korea as solved (at least officially) and dropping Iran because they’ve reached a peak quite a while ago, but can’t get out of their trickle mode plateau, which seems to indicate they are fine with it. We have picked up Belgium and Brazil. Here are some death rate plots:

The symmetric Gaussian dotted line is for visualization to accentuate the varying deviations. The Gaussian fits the rising side, but not falling side well. Notice the Sunday under-reporting effect for most countries.

Belgium has the notoriety as having been the deadliest country in the world with 806 deaths/million (as of 5/27/20), but as I said above now GET IT, but still very deadly as the Tables below show. At its peak Belgium had 183 deaths/ week per million people, by far greater than any other nation (but less than several U.S. states as shown below). Brazil is now in our Hall of Shame for a government that ignored and now is denying the pandemic. Whereas almost all other major countries have peaked in death rate (and prevalence) weeks ago, Brazil continues to climb. But then again how bad is Brazil on a per capita basis? Oddly it is no where near the worst country. Its numbers are big because it has a population of 212 M compared to say Belgium at 11.6 M. So, on a per capita basis Brazil may not be so bad. It’s death rate also appears to be peaking (although that is a big if) and if that is true then we can make the following comparison: It’s forecasted cumulative deaths per million is comparable to the U.S. and its peak death rate is less. One could look at this positively in that given Brazil’s wanton disregard for public health even worse than in the U.S. (that is an undertone, not overtone) it is less deadly and hopefully that is a Southern Hemisphere effect indicating that the virus may indeed be seasonal.

Notice that worldwide there is a strong death rate tail indicating there are new and significant outbreaks around the world of which Brazil is the most noticeable.

The negative value for Spain is due to their readjusting their deaths downward this past week.

Domestic News

Now let’s look at what’s going on in our country. I am going to de-emphasize Washington and Louisiana and focus on the remaining trouble spots, the Northeast and California. Here are some death rate plots:

The symmetric Gaussian dotted line is for visualization to accentuate the varying deviations. The Gaussian fits the rising side, but not falling side well. Notice the Sunday under-reporting effect for most states.

Key observations are as follows:

  • NY is showing the WE GET IT mentality having seen the worst death rates in the world. But there is still a stubborn persistent tail, but at least it is declining.
  • NJ is pretty much showing NY fortitude in its recovery, but either a little behind or a little less effective.
  • CA is simply pathetic; Gov. Gavin Newsom did a good job quickly implementing social distancing mandates and keeping the virus down, but then succumbed to public pressure to ease prematurely and the results are obvious. For that we downgrade CA back to the red zone in our 3-color ranking system. From what I see people in CA don’t care.

So next are Tables that compare to the international ones above. So, although the U.S. might only be the sixth most deadly country based on per capita (hardly an achievement despite Administration claims given 197 countries in the world – undertone), the northeastern states are in fact the deadliest places in the world, led by NY, NJ, CT and MA, all greater than Belgium.

OK, it is now time for the nerd stuff. Below is an update of our regular update Table. Some of this repeats the above but I include it because I am a data junkie and because I spent a lot of time updating it. I reiterate again that this model does not anticipate future changed behavior such as social relaxing, but rather past behavior, so we expect to see some upward movements that could be quite severe for certain populations.  We will be developing a revision to the model to explore various scenarios for relaxation of social restrictions.

The easing date threshold was raised from a prevalence of 200/million to 500/million (i.e., 1 in 2000 people are infected) to provide a little more leniency, though perhaps unadvisedly.

Key observations include:

  • Social easing: This model says no country is safe to ease social restrictions until June earliest, but that isn’t stopping them. So, I don’t push this Easing Date other than to use it as a relative indicator of which countries could ease sooner or later. Not surprisingly, based on the rate of decline in deaths and prevalence (active cases), the U.S. is lagging every major hard-hit country. We are just not recovering as effectively. Disturbingly with the new wave of deaths and prevalence in CA, it is now projected to be the last of the major states that should be easing.
  • Other tidbits: I am always surprised by some of the outcomes of this model and you have to take forecasting models with a grain of salt, this one included. But this one indicates that the U.S. has more than a third of the world’s active cases (prevalence). The news media says it is about a quarter, but that is because of the missing cases in the U.S. due to still inadequate rates of testing.

And now for the finale, we reach our model comparison of forecasted total deaths to the benchmark model from the Institute for Health Metrics and Evaluation (IHME) at the University of Washington (UW) (http://www.healthdata.org/covid/).

We’ve talked about these two models before. We are tracking similarly, but the IHME is now forecasting a little higher than us particularly internationally and may be incorporating social easing, which we have not done yet. Our domestic forecasts are very close, but that reflects that most of the deaths have already happened and we are forecasting the fraction that still remains. Our forecasts, as measured statistically, still remain more stable regarding without any apparent penalty in forecasting accuracy (and precision).