I can’t believe another month has gone by, but here we are! I am back to update state level excess mortality data from the CDC website, pulled on 06OCT21. See previous posts for more details about this data.

First up though, here’s an interesting gif someone made that shows the spread of COVID cases over time by region. Definitely shows some interesting seasonality, and also some interesting data anomalies.

Excess Mortality – How bad has it been?

As I’ve talked to a few people about state level data over the past few months, one of the things I’ve noticed is that some people’s perceptions of the pandemic do not match their individual state. I started wondering if this has anything to do with when the peak excess mortality is, and how long the states spend at high levels of excess mortality. Using the same CDC data I’ve been using, I decided to pull the number of weeks each state has a mortality rate >50% above their average. The data goes back to 2017, so we can see that this phenomena only happened three times between January of 2017 and March 28th, 2020: once to Puerto Rico in September 2017 (Hurricane Maria), and twice in Wyoming (October 2018 and January 2020). I’m not totally clear what happened those weeks.

So this happened 3 times in a little over 3 years. How often has it occurred since the end of March 2020? A total of 363 times in 45 states. The only 5 states that haven’t reached that level since the pandemic began are Alaska, Hawaii, Maine, New Hampshire and Oregon. The US as a whole spent 6 weeks in that range, with 25 states exceeding the national average. Here are those states, and how many weeks they spent at that level (so far):

State	# of weeks at >50% excess mortality
Texas	19
Mississippi	17
DC	16
Alabama, Arizona	14
Nevada, North Dakota	13
Oklahoma	12
Georgia, Louisiana, Montana, South Dakota, Tennessee	10
Arkansas, California, Florida,	9
Indiana, New Mexico, New York City (city only)	8
Iowa, Michigan, New Jersey, Pennsylvania, New York (excluding city	7

Just a note on NYC vs NY: only one of those weeks wasn’t overlapping. If we raise the bar and look at only states that have at least one week where they had DOUBLE the number of deaths they usually do, we find only 9 states have hit that bar:

State	# of weeks at >100% excess mortality
New York City (city only)	7
New Jersey, South Dakota	5
California, Connecticut, Massachusetts	3
Florida, New York (excluding city), North Dakota	2
DC	1

Another note on NYC vs NY: the 2 weeks for NY are also in the 7 week stretch for NYC. Not clear why the CDC reports these separately.

Excess Mortality Over Average Updates

First up, here’s the whole US. It’s worth noting that when I did this graph a month ago, the lowest value was 554 excess deaths/million. Now it’s 739 excess deaths/million. The brightest red a month ago was 4107/million, now it’s 4624/million. The greens and the reds mean more than before:

So who were the top movers this month? Let’s see:

State	Excess Deaths Above Average/Million 2/1/20-10/6/21 (change from 9/8)	Change from 9/8 rank
Mississippi	4624 (+516)	No change
Alabama	4000 (+559)	+1
Louisiana	3801 (+534)	+3
Arkansas	3404 (+379)	No change
DC	3749 (+97)	-3
Arizona	3597 (+251)	-1
South Carolina	3453 (+326)	No change
Tennessee	3381 (+326)	+3
Florida	3365 (+538)	+3
New York	3177 (+91)	-2

Note: the NY data here is all of NY, state and city combined. Seems incredible that New York may actually fall out of the top 10 for excess mortality since the pandemic started. To note: there were 4 states that saw substantial gains but are not yet at top 10 level. These were: Georgia (+471, 14th place), Oklahoma (+442, 13th place), Peurto Rico (+407, 37th place) and Kentucky (+390, 17th place).

Excess Mortality Over Upper Bound by State

Okay, here are the states that most exceeded 2 standard deviations from the mean mortality:

State	Excess Deaths Over Upper Bound (change from 9/8)	Change from 9/8 rank
Mississippi	3302 (+443)	No change
Alabama	3004 (+502)	+1
Arizona	2659 (+210)	+1
Florida	2647 (+499)	+5
New York	2646 (+56)	-3
Arkansas	2582 (+324)	No change
Louisiana	2574 (+449)	+3
Texas	2549 (+345)	-1
South Carolina	2471 (+280)	-1
New Jersey	2452 (+52)	-5

To note, there are again 4 states who had a top 10 gain in excess mortality, but didn’t make the overall top 10. These are: Tennessee (+483, 11th place), Georgia (+420, 12th place), Oklahoma (+380, 14th place), Kentucky (+327, 21st place).

As always, let me know if there are any questions and I’ll be back in a few weeks! Given seasonality, I’m going to try to keep this up monthly. I’d also ideally like to see if some states start to regress at all. There is a lot of commentary that COVID mostly killed people who were going to die anyway, but so far that is not what we are seeing. If that’s true, at some point some states excess mortality should start to decrease below the norm. So far I’m only seeing slight decreases for Connecticut, Rhode Island and Minnesota, but those are small and could be adjustments.

More Explanation and Some Links

Well hello again folks! When last we left off about 4 weeks ago, I had updated the state level mortality data provided by the CDC for 2/1/2020 – 8/11/2021. Today I’m updating through 9/8/2021, about 4 extra weeks. All the caveats from my prior post still apply, so go there for any more explanation.

First though, I wanted to clarify some things from my prior post. I find excess mortality data interesting because every state counts COVID deaths differently. There are varying reasons for this, some more valid than others. There are also lots of theories about what the non-COVID excess deaths are. I like looking at state level data because it forces us to think more critically about what those deaths might be and to avoid making sweeping generalizations. In the national press, only the biggest 4 states (California, Texas, Florida and New York) seem to get any air time. Other states may occasionally be cherry picked if something interesting is going on, but otherwise they are mostly ignored.

There is some good work going on with excess mortality, both in trying to estimate it and trying to track it. First up, some good analysis of the 2020 death data, including racial breakdown. While the early phase of the pandemic (when it hit NYC hard) was very skewed towards black and Hispanic deaths, it appears things got far more even as we got towards the winter. For example, here’s the excess death incidence rate for those > 65 years old. Bars are quarters of the year 2020:

Next up is an interesting link (explanation here, site here) to someone trying to catalogue excess mortality in real time, with the concerning hypothesis that we may be seeing an uptick in other kinds of deaths too. Now there are two competing hypotheses here: people could have put off getting treated for other medical conditions due to the pandemic, or people could be more susceptible to other medical conditions after having COVID. Actually, those aren’t competing. It could be both. We know that with the flu there is a well established link between getting the flu and subsequently having a heart attack, and there’s no reason COVID-19 couldn’t act similarly. We also know that in many places hospitals are full and it makes sense people may put off care. We will know more as the data comes in I’m sure, but it’s unfortunate. On that happy note, on to the next updates!

Excess Mortality Over Average by State

I made a more multi colored graph this time:

State	Excess deaths above average/million 2/1/20 – 9/8/21 (change from 8/11)	Change from 8/11 rank
Mississippi	4108 (+473)	No change
District of Columbia	3652 (+111)	No change
Alabama	3441 (+320)	+1 spots
Arkansas	3404 (+392)	+2 spots
Arizona	3346 (+208)	-2 spots
Louisiana	3267 (+166)	-1 spots
South Carolina	3127 (+238)	+1 spot
New York	3086 (+76)	-1 spot
New Jersey	2894 (+33)	No change
Nevada	2842 (+281)	+3 spots

Mississippi’s struggling here guys.

Excess Mortality Over Upper Bound by State

Okay, here’s the updated numbers for deaths only falling outside the upper bound:

State	Excess deaths over upper bound 2/1/20-9/8/21 (change from 8/11)	Change from 8/11 rank
Mississippi	2859 (+379)	+1 spot
New York	2590 (+39)	-1 spot
Alabama	2502 (+229)	+2 spots
Arizona	2449 (+155)	no change
New Jersey	2400 (+5)	-2 spots
Arkansas	2258 (+333)	+4 spots
Texas	2204 (+181)	-1 spot
South Carolina	2191 (+190)	no change
Florida	2148 (+454)	+8 spots
Louisiana	2125 (+107)	-3 spots

There was more motion on this ranking than I expected to see, which is sad because it means there are multiple places where we are seeing truly unusual death tolls.

States of Interest

Since everyone’s always interested in the Big 4, here they are. Change from 8/11 in parentheses:

	Excess Deaths Over Upper Bound/Million	Excess Deaths Over Average/Million	Rank in Excess Deaths Over Average	Rank in Excess Deaths Over Upper Bound
New York	2590 (+39)	3860 (+76)	8	2
Florida	2148 (+454)	2827 (+488)	12	9
Texas	2204 (+181)	2723 (+216)	15	7
California	1761 (+50)	2285 (+92)	30	20

And because I’m always interested in my state and those of similar size, here they are:

	Excess Deaths Over Upper Bound/Million	Excess Deaths Over Average/Million	Rank in Excess Deaths Over Average	Rank in Excess Deaths Over Upper Bound
Arizona	2449 (+155)	3346 (+208)	5	4
Massachusetts	1240 (-9)	1642 (+25)	40	34
Tennessee	1904 (+117)	2660 (+193)	13	11

Age Adjustments?

So on my last post Kyle Watson made an interesting point that there should be some sort of age adjustment if we were going to compare things on the state level. While some of this is sort of inherent to the entire concept of excess mortality (states with older populations likely have more expected deaths in a given year), we would expect a disease like COVID to hit states with older populations harder even if everything else was equal.

Interestingly there was some work done on this by a group using the raw COVID numbers, which also looked at international data. They found that states like Texas actually had a worse pandemic than previously reported due to their young population:

I had to ponder a bit what the fairest way of doing this was though. It turns out the CDC also publishes the numbers by week by age group, so I took a look at the US as a whole from 2015-2021:

So every age group from 45 years on showed a fairly noticeable bump last year. Actually every age group showed an increase in mortality when compared to the previous years, and it wasn’t entirely the groups I expected. It’s hard to see on the graph, but here’s the increase for each age group over the average from 2015-2019:

Age Group	% Increase Over 2015-2019 Average
Under 25 years	5%
25-44 years	33%
45-64 years	20%
65-74 years	30%
75-84 years	27%
85 years and older	19%

I was surprised so many of these increases were so close together, it was just the starting numbers that were different. Please note the bin sizes are different however. There are twice as many ages contained in the 45-64 year old group as the 65-74 group, which is how you get a similar number of deaths in the younger age category.

It’s also interesting to note that while the data for this year is obviously still highly incomplete and anything could happen, there’s a chance the 85+ group may not show a large jump for 2021. Almost certainly not as large as last year.

Back to age adjustments though: I couldn’t find a great source to give me state by state age breakdowns matching the ones above, but I did find a breakdown of how many people in each state are over 65. I assumed that excess mortality followed roughly the same pattern as the overall mortality numbers, and adjusted from there. Here are the new leaders for excess deaths above average:

State	Age-Adjusted (albeit crudely) Excess Mortality above Average 2/1/20-9/8/21
DC	4330
Mississippi	4100
Louisiana	3328
Alabama	3303
Arkansas	3253
Texas	3183
Arizona	3136
New York	3009
South Carolina	2915
New Jersey	2864

The map overall shows there’s a pretty substantial dropoff between Mississippi, DC and everywhere else:

Now here’s the big 4:

State	Adjusted Rank for Excess Mortality Over Average (previous rank)	Adjusted Rank for Excess Mortality Over Upper Bound (previous rank)
New York	8 (8)	3 (2)
Florida	24 (12)	15 (9)
Texas	6 (15)	2 (7)
California	22 (30)	13 (20)

Interestingly, the states most helped/hurt by this adjustment aren’t necessarily the ones you’d think of. For deaths above the upper bound, 4 states added on more than 150 deaths/million and 4 states lost more than that after the adjustment. The ones that gained deaths were: Texas, DC, Georgia and Utah. The ones that lost the most post-adjustment were Arizona, Florida, Pennsylvania and West Virginia. As mentioned, only Texas and Florida receive much air time nationally, and since this worked out differently for both of them I wouldn’t expect to see much on this any time soon.

As always, let me know if there’s more you want to see! I have a lot left on spreadsheets for individual states. Stay safe out there.