STUMP » Articles » Top Causes of Death for 2021: Heart Disease, Cancer, and COVID » 13 May 2022, 05:30

Where Stu & MP spout off about everything.

Top Causes of Death for 2021: Heart Disease, Cancer, and COVID  


13 May 2022, 05:30

Last week, the CDC did the monthly update of their mortality database. While it’s not finalized data for 2021, it looks like we’re pretty close to settled stats for 2021 mortality data.

Here is a ranking by cause of death, for count of deaths:

So you can see, the provisional ranking is essentially the same as 2020, except when you get down to rank #9. For 2021, liver disease displaces flu/pneumonia in place number 9.

Doesn’t seem to different at that level, does it?

Age grouping ranking

But that is misleading, if we look at death ranking at that level. As I have shown in earlier posts, there were some meaningful differences in death patterns in 2021 compared to 2020, and there was about a 2% – 3% increase in deaths in 2021 compared to 2020.

So let’s grab that table:

Let me point out parts from the ranking, and I’ll push the persnickety explanations to the very bottom of the blog post for those who are already familiar with those details.

  • COVID is the number one cause of death for the age 45-54 group — however, note how close in number heart disease and cancer are for that group. These numbers are provisional, and it is possible that COVID will slip out of the top spot by the time finalized numbers come in.
  • I will be putting the 2020 ranking table below so you can compare numbers, but what’s interesting is how COVID dropped down for the oldest ages. I will address this in upcoming blog posts.
  • The homicide, suicide, and accident counts are too low for each group — because of censoring of “external cause” deaths for 6 months, these aren’t differentiated for November and December 2021. Those are the numbers at the bottom of the table. It’s likely they’ll keep their rank ordering, but the numbers will be higher by the final table, and it may make a big difference for ages 5-34.

How Things Change: first half 2021, 2020, and 2019

Back in January, I did a post where I looked at the first half of 2021, and saw the ranking of causes of death by age. Here it is: 2021 U.S. January-June Cause-of-Death Ranking Table.

Remember, this is before the “Delta wave” (and then omicron wave) of COVID hit.

The ranking table looked like this:

Here is the similar ranking table for 2020:

And finally, so you can compare to a “normal” year, a ranking table for 2019:

These are counts, not rates. But let’s start out simple, with the numbers, so you can see how these numbers have changed over the last couple years.

Coming attractions: COVID, non-COVID, and more!

On Tuesday of this week, I presented at the Actuaries Club of Hartford and Springfield.

I had an hour to talk… and 75 slides. (No, I did not speak to every slide. I’m not crazy. Well, not crazy in that way.)

All my slides and the underlying spreadsheets can be found here. The slides are there both in Powerpoint and PDF. The data are all from CDC Wonder. See below for persnickety details.

I will be walking through some of these slides in the weeks to come. I’m waiting for more data on some of the non-COVID causes (namely – drug overdoses), but I’ve got more than enough data to be pointing out some of the disturbing stuff we’ve seen thus far. There was some worse mortality in 2021 than 2020, and it wasn’t all COVID. And it wasn’t all “external” causes, either.

We’re concerned that there will be a “mortality hangover” extending… how long? No clue.

I’ve mentioned dying fast/slow before, and I’m trying to detect the “slow” causes coming to the fore. Those will obviously be more difficult to see shifting in trend compared to the “fast” ways of dying like motor vehicle accidents.

Blog notes

For those new to the substack, if you want to only see the mortality posts, just click on Mortality with Meep for the death-only stuff.

The STUMP – Death and Taxes portion of the substack is the podcast, and you never know what you will get there. My last episode was on insurance fraud. Before that, it was sumo. And before that, baby booms/busts and entitlements. It’s quite a mix.

The overall substack site is a mix of all this and more – I like writing about public finance and pensions as well.

Methodology and Persnickety Notes

These are the details that will be familiar to those who have been long-time readers of my blog.

You may want to read these if you want to get a better grip on what you are looking at in these ranking tables.

Data come from the CDC Wonder provisional death database, with data drawn on May 6, 2022 or later for 2021 data. For 2020 and prior years, those are from the finalized data databases. These databases can be found at and are based on actual death certificate information. These are compiled, usually at the county level, aggregated at the state level, and then sent to the CDC.

Currently, the CDC updates its provisional database monthly, and the May 6, 2022 and later data draw includes deaths that were reported as of May 1, 2022, and include deaths up to April 16, 2022.

For the causes of death, I’m looking at the Underlying Cause of Death, of which there is exactly one per death certficate. So if it says “COVID”, that means COVID is the ultimate cause of death — not any of this “with COVID” razz-ma-tazz. There is no double-counting of deaths.

For determining rankings, I’m using the CDC’s “rankable” causes of death, which are groupings of ICD-10 codes. I am not using their technical terms for the grouping, but my own simplified forms. For example, for the grouping “#Malignant neoplasms (C00-C97)”, I simply label that as “Cancer”.

The one cause grouping that I see as overly broad — “#Accidents (unintentional injuries) (V01-X59,Y85-Y86)” — which I label as “Accidents” — includes such disparate causes of death as unintentional drug overdoses, motor vehicle accidents, and falls. These causes have very different trends, especially by age group and especially through the pandemic. I will be addressing this (again) in a future post.

In the provisional database, “External causes of death” — accidents, homicides, and suicides — are essentially given a “we don’t know” code for 6 months before their ICD-10 code is finally uncensored for database queries to outside researchers such as me. This is not a new practice to the pandemic. Because intent is a big part of these causes of death (intentional or not?), they would rather not put in a cause of death and then have to change it later. Sometimes if I’m doing an analysis of natural/unnatural causes of death, I will use this information.