Monday, June 14, 2021

Post 7900 - Excess Mortality

Good news and bad news from Lyman Stone again. His big advantage is that his models have proven very close once the data trickles in over the following weeks.  I don't know if his model makes long-term predictions. These are usually what is happening just as it breaks, when only the preliminary death reports are in, showing what the real excess mortality numbers are. A few weeks or a month later, he has been consistently right, though as he himself admits, there have been occasional weeks where results were odd.

The first tweet is the summary: Excess deaths remain significant - they are the same as June of last year - but there is no current indication we are going to have a fourth wave. We are telling ourselves that "this covid thing is over" largely because we are sick of it and we want to to be, not because it really is. Rates of illness among the unvaccinated are as high as they were from the whole population last year, so the rest of us getting vaccinated hasn't protected them any. 

I said early there would be both undercounts and overcounts, just because it was new and the decision-and-tracking methods were still in development and  there would be cross-incentives in reporting. For some reason panicked conservatives have been determined to see only over-reporting and on the basis of incidents here and there declare that the whole thing is unreliable and overcounted.  Liberals have not had a corresponding panic that covid is being radically undercounted, so I have to say this is on us along.  I suppose it is brought along as part of the package that insists that we have overreacted.  We may have overreacted, but it's always best to argue from facts, not feelings and anecdotes on such things. 

So state-by-state data gets interesting when Covid is compared to excess mortality.  I continue to use that metric because no one has convincingly shown me that anything except drug overdoses are up over the Covid period, and those do not come close to the full discrepancy. If you want to look at a place where there may be an overcount, it might be Massachusetts, where the official Covid deaths and excess mortality are very close. A friend from MA emails me these things, annoyed because other states seeming to be "getting away with" underreporting in order to look better. So the focus is on two states with similar populations, TN and AZ, which look on the official lists like they are doing much better than MA, but the deeper look contradicts that. Then there is a look at the three big states in the news about how they are or are not doing:  California, Texas, Florida. You know, the ones that conservatives were pointing to a year ago as the ones who had done things right because they didn't have many covid deaths at all. They were spared the first wave, but have done worse since then. My friend combed these from the CDC site directly, so there isn't a simple link that summarises the work. If it doesn't look right to you, the data is open, but you'd have to find any contradiction yourself.

If you just look at the Worldometers Coronavirus deaths per million, as I usually do, Massachusetts is one of the worst states, California well down the list.

3. Massachusetts 2603

6. Arizona 2441

National average 1858

22. Tennessee 1830

24. Texas 1799

26. Florida 1735

33. California 1599

But if you look at the Excess Mortality of all causes, things look different

Arizona 3011

Tennessee 2461

Texas 2367

California 2145

Florida 2082

Massachusetts 1633

How can that last number be true?  How can the number of Covid deaths exceed the excess mortality? Well, my friend cautions that different states have different ways of counting (Yeah, you got that right!). But if you were going to look for an overcounted covid place, you should probably start with Massachusetts.  Florida's totals look sort of okay, playing fair.  Add some drug overdoses to that covid number and you are right in the ballpark. But the other states have EM totals 25-30% higher than covid, with no convincing sector of deaths to explain that gap.  They are undercounting. Again, different agencies and states county differently, so I may have missed a trick here. But if you remember from the Twitter thread at the top that the more recent numbers are clustering around respiratory deaths, it looks even more suspicious.

Yes, all those hospitals that have been losing money on everything else for the last fifteen months have an incentive to over-report to get some of that federal money to make up some of it.  But governors and health departments have greater incentives to underreport so that their state and their personal competence looks better. They make many of the counting rules. That latter seems to be winning in most places.

BTW, I dug down into the more recent numbers to try and see trends.  Even though deaths are down for everyone, they are not equally down, and Texas, Florida, and California have not been dropping much over the last couple of months, but holding steady.  Texas and Florida are still showing excess deaths as recently as May22.  With the time lag for reporting, they may still be up.  Look for those three states to start passing others in deaths per million. New Hampshire continues to do well in 44th place. Not as well as best-comps Maine and Vermont, but we have much more interaction with Greater Boston area.

Possible grim news from Sweden is up next.


Aggie said...

Why is it that a country that has the power and the willingness to close the economy and confine substantial portions of its population to their homes, doesn't have the will to impose standardized methodologies for identifying cases accurately, or to specify standardized parameters and metrics for testing and certification of cases and causes of death?

Assistant Village Idiot said...

Standardised would mean federal in this case. Because such things as health interventions are one of the few things under the control of the states at this point I would hesitate before insisting they do its Washington's way. Some states required a formal covid test to call it a covid death, others relied on the physician's judgement. Both of those have weaknesses. The CDC starts from the numbers of actual dead people and works outward from there.

Cranberry said...

I would not place a great deal of faith on comparing different states' "excess deaths," as a method to gauge how they handled Covid.

For one, looking at the charts at the CDC's excess death from all causes pages, states vary wildly in how close they come to their expected death rate:

For one thing, look at Vermont. And then look at Florida. Florida comes much closer to its expected deaths than Vermont, even before the pandemic. Their median ages are very close, according to Wikipedia--they're the 4th and 6th oldest states, by median age.

However, they're leagues apart in death taxes. Vermont charges 16% on estates over $2.8 Million.

You know, there are quite a few people who live in two states, especially Florida. They winter in Florida, then return to the northern states in the summer. If your grandma has a large estate, you want her to die in Florida, at least taxwise. I did know of a number of older people who spent the pandemic in Florida.

I assume there are similar pairings in the midwest...I search...ta dah!

Minnesota and Illinois both have estate taxes. (Source:

So, I would say that comparing states is difficult at best, unless you know how to divvy up people who own properties in more than one state. Certainly, for the heirs who do the transporting, there's a heck of an incentive for their ancestors to be residents of states without estate taxes. The age range most in jeopardy from Covid--over 75--is also the age most likely to go south for the winter. Florida and North Carolina are home to many residents of Massachusetts, New York, Connecticut, etc.

The excess deaths statistics for the country as a whole, for all causes, is interesting, but it does suffer from bureaucratic delay. Not all counties send death statistics promptly. The Wall Street Journal covered this some time ago.

Assistant Village Idiot said...

Agree that there is a delay in reporting, but that does catch up. We might not have a good picture of what happened two weeks ago, but we are pretty sure what happened two months ago - again, recognising that different states are going to assign deaths a little differently. But ultimately, you have x number of dead people, and you can compare that to how many were expected. My friend started pointing out AZ and TN underreporting covid months ago, and it has held steady.

It's an intriguing idea to look at whether people have moved to their other residences for the pandemic. I imagine there must be some. But I don't think it goes a long way to explaining the overall covid death rates. The dual-residence Vermont/Florida would have all been in Florida as of February- March last year. Nobody thought we would be without travel for a year, so that would not have been in anyone's plans. The Vermonters were looking forward to heading home, getting away from the heat in April. Also, no one knew which places were going to get hit hard then, so that couldn't have factored in. So taking all the northerneres in Florida, you would have more old people than usual, and Vermont would have had fewer. But I don't see how that comes close to explaining the huge difference in state rates. If mobility was a factor, it would have had to have been immediate, because Vermont was low right out of the gate despite bordering MA and NY.

People weren't moving around much. By fall, the smowbirds I know were not heading for Florida, because Florida's covid rate was getting very high, and Vermont's was still low.