STUMP » Articles » Mortality with Meep: Ranking the States (and NYC and DC) by Excess Mortality » 1 April 2021, 07:46

Where Stu & MP spout off about everything.

Mortality with Meep: Ranking the States (and NYC and DC) by Excess Mortality  

by

1 April 2021, 07:46

Death data for 2020 are mostly there for all states, so let us look at how hard each was hit for the full year. The CDC’s update of 31 March 2021 in their excess mortality dashboard has total numbers available for almost all geographies through the end of 2020.

I have to exclude North Carolina from the following analysis, as their data only goes through November 2020, and I don’t even trust the November 2020 data to be all there. . (I explain why NC is a laggard in this post. Basically, they’re three months behind everybody else because they still report their deaths to the CDC on paper.)

For the following, I am measuring excess mortality by comparing the total number of deaths in the given geography against the average number of deaths 2017-2019. I average over three years, because that reduces some of the variability, especially given seasonal variability from flu seasons.

For a proper analysis of death increases, we should really be comparing rates, and adjusting by age. However, the total population and their age distributions will not have changed much from 2017 to 2020, except for the states/locales with very small populations.

I will not only do rankings, but also provide some graphical comparisons, so you can see how much these states really differ in the mortality impact of 2020.

Station break: top causes of death in 2020

I will be doing a post on this later, but I just saw this JAMA article on a preliminary ranking table for top causes of death in the U.S. in 2020 this morning, and will be addressing it in a later post.

In the meantime, a video reaction to what I’m seeing:

The rankings by percentage excess deaths for total population

Before I do this table, just consider for the entire United States, the excess mortality for 2020 was 18% over the average from 2017-2019.

Here’s the whole ranking table:

Geography 2020 Excess Mortality Geography Excess Mortality
1 New York City 49% 27 Missouri 17%
2 New Jersey 28% 28 Montana 17%
3 Arizona 28% 29 Kansas 17%
4 Texas 23% 30 Iowa 17%
5 Mississippi 23% 31 Wisconsin 16%
6 South Dakota 23% 32 Florida 16%
7 New Mexico 23% 33 Oklahoma 16%
8 North Dakota 23% 34 Utah 16%
9 Louisiana 22% 35 Minnesota 16%
10 Wyoming 21% 36 Idaho 16%
11 Nevada 21% 37 Virginia 16%
12 Georgia 20% 38 Ohio 15%
13 Colorado 20% 39 Rhode Island 15%
14 Illinois 20% 40 Massachusetts 15%
15 District of Columbia 20% 41 Pennsylvania 15%
16 South Carolina 19% 42 Kentucky 14%
17 Alabama 19% 43 Nebraska 14%
18 Connecticut 19% 44 Alaska 14%
19 New York 19% 45 West Virginia 11%
20 Maryland 18% 46 Washington 9%
21 California 18% 47 Oregon 9%
22 Michigan 18% 48 New Hampshire 9%
23 Indiana 18% 49 Vermont 7%
24 Tennessee 18% 50 Maine 6%
25 Arkansas 18% 51 Hawaii 3%
26 Delaware 17%

I kept it to whole percentages, because I don’t think the data are good enough for the false precision of additional digits.

The worst location: New York City

New York City’s excess mortality was the worst in April 2020:

That peak was in 11 April 2020, and it was 600% normal mortality.

For the entire year, that made it to about 50% excess. Yes, there was a winter wave, but that did not contribute much excess mortality. Almost all the excess mortality was packed into a few weeks in the spring of 2020.

We saw this pattern months ago, and that New York City sits atop the ranking list is not surprising even if there had been no further excess mortality in the rest of the year.

The lowest excess mortality: Hawaii

Unlike North Carolina, this is a real effect. It shows the benefit of being an island:

Other island nations, such as New Zealand and Taiwan, have had similar good results, because, when you’re an island, it’s relatively easy to control who is coming into the area.

I’m going into Connecticut all the time, for example — there are several crossover points between NY & CT, and there are two different roads I often use to go into CT.

But if you want to get to Hawaii, out in the middle of the Pacific Ocean, you’re pretty much flying there. So to control people coming in and spreading disease, you only have to control the airports.

So I will note they had bad mortality back in that nasty 2017-2018 flu season. Maybe they will think about screening passengers during flu season in the future.

Tile grid map of the excess mortality

The ranking table, though is not really all that meaningful. There are entire swathes of the ranking list where the states are essentially the same in terms of total mortality increase in 2020.

The three worst areas just pop out: Arizona, New Jersey, and New York City.

(Also popping out: the hole where North Carolina should be. I’m going to guess it has landed in the midrange of states, like its neighbors South Carolina and Tennessee.)

New York state excluding NYC (which is what the NY square represents) is middle-of-the-pack for excess mortality, which is hardly surprising – as most of New Jersey is crowded near NYC, New York state is geographically huge and has significant populations on the Great Lakes, far away from NYC.

The lowest excess mortality also really pops out: some states bordering on the edges of Canada, plus Hawaii. It is a little surprising that Washington State, in particular, has not been hit hard by COVID mortality.

I will look at other ways to slice this 2020 full year data in future posts, but for now, one can see that, in general, there seems to be geographic patterns to how hard particular geographies were hit, and that most states/geographies were bunched around the national average level, in the 15 – 20% range.

Endnote: Why not compare COVID deaths?

For those who are new here, the main reason I do total deaths is that these data are less disputable than COVID death data. I’m really not interested in arguing with people about whether official death counts for COVID are legit unless you pay me a lot of money (at about $1000/hour level) or you’re Stu. And then I’m mainly humoring him.

It is really tough to fake death stats in the United States, in terms of just plain counting dead bodies. We even have age and sex for over 99% of the mortality stats.

Some people argue “but they’re slow-walking the death certificates!” … well, slow-walking is what North Carolina is doing. They’re lagging other states by 3 months, because they do paper-based reporting in getting their info to the CDC. Everybody else is digital, now.

The other states will file preliminary death certificates digitally, and they may amend the death certificates later for cause of death, or, more often, contributing causes (as opposed to the underlying cause). You may have seen some data dumps where a bunch of “new” COVID deaths (namely, death certificates filed over a number of months) get updated — by going by “newly reported” you can get absurdities like “new” negative number of COVID deaths. Rarely is a death certificate is rescinded because the person wasn’t dead — not much in the way of negative deaths there, even if the cause of death gets amended on the certificate.

But here’s my main reason: dead is dead.

I’m in life insurance and annuities — in most cases, we don’t really care what you died from, except to the extent that mortality underwriting is off. That’s a practical situation. Only in a few cases does the cause of death invalidate getting death benefits (and no, one of those cases isn’t necessarily suicide.)

But just from a human point of view: dead is dead. If COVID-related policies increased deaths due to drug overdoses and car accidents, that is an important piece of information. Policy choices aren’t made in a vacuum. There are trade-offs.

Thanks to Tables Generator for helping me build that ranking table.

Underlying spreadsheet.


Related Posts
Mortality with Meep: Retrospective on Mortality Trends (and a wee bit about COVID)
Mortality with Meep: Total U.S. Excess Mortality in 2020, by Race and Ethnicity
Mortality Monday: Supreme Court Probabilities