Following up: Where did major increase in deaths come from in 2020-21?

Last week’s post looked at the overall increase in deaths across all age groups in 2020-22 over 2018-19. Generally, large increases in 2020 were followed by additional increases in 2021, and then mild to moderate decreases in 2022 (but still well above 2019 levels). That post showed that for most age groups, the large increases were not due primarily to deaths labeled as COVID. The obvious question to investigate is then, to what are the increases attributed?

To answer this, I went back to the CDC Wonder data and queried for 2018-2021 data that included state, 10 yr age group, and cause of death. I then subtracted the drug and alcohol related deaths from the “Non-transport accident” category as I was very interested in more detail on these topics. Also, given my recurring interest in data around pregnancy during COVID, I also gathered data on deaths related to pregnancy even though this is not a large number overall. In the end, doing some of my own categorization, my final way of looking at causes of deaths uses this framework:

Below I will present by age groups how overall deaths and deaths by major causes changed in 2020 & 2021 vs the average of 2018-19. This data will be presented as rates per 100K population to account for population changes. The presentation will use bullet charts, which essentially is a bar chart with a reference line displayed to show how the bar relates to a fixed value. In our case, the fixed value is the 2018-19 averages. Charts will be presented at a high level (total & COVID deaths) and at a more granular level (major cause of death).

How to read the charts

Totals

---

Causes of Death

---

---

* Note, in graphs that follow, if a specific cause of death is not shown in chart, this indicated there were not enough deaths from that cause in that age group to have confidence in comparisons.

All ages (5-84)

• 2020: Total deaths ⬆ 22% over 18-19 avg.

• 2021: Total deaths ⬆ 26% over 18-19 avg.

• Respiratory diseases and suicide (see below on overdoses… are some suicides/overdoses hard to label?) decreased moderate or mildly. Cancer was basically flat (mild decrease).

• Huge increases in drug & alcohol overdoses, other alcohol-induced causes, liver disease (also alcohol related, presumably), diabetes and homicides.

• Also significant increases in major categories like cardiovascular, Alzheimer, Parkinson, accidents (Transport and Non-transport), and “Other”

Young people (5-24 year old) summary

• 5-14 year olds (left)

  • Compared to other groups, relatively small increase in total deaths. Still, in normal times, we’d be alarmed that ‘20 & ‘21 deaths ⬆ 4.5% and 8.5% above 2018-19 average.

  • Numbers are too small in this group to report accurately on cause of death trends.

• 15-24 year olds (right)

  • ‘20 and ‘21 total deaths were ⬆ 22% and 27% above 18-19 average.

  • The numbers on drug overdoses and homicides are almost too horrific to believe… I’m actually nervous that I have made some error in my work, but I have tried a few ways to verify these numbers and I believe they are accurate (I welcome any peer review from readers).

  • Transport accident deaths also off the charts.

  • In the future I may try to dig more into “Other”… this applies for all ages.

Adults leading up to middle-aged (25-44 year old) summary

• 25-34 year olds (left)

  • ‘20 and ‘21 total deaths were ⬆ 26% and 41% above 18-19 average 🤯.

  • Similar, only worse, to 15-24 year olds in terms of drugs and alcohol, homicide, and transport accident data.

  • In addition, diabetes and liver disease are terrible.

  • Suicides ⬆ a decent amount.

  • Cardiovascular ⬆ 11 & 17% respectively is also not good.

• 35-44 year olds (right)

  • Story is essentially the same as 25-34 year olds, only somehow it’s actually worse basically across the board…. 2021 showing 61% increase in drug overdoses compared to 18-19 😢🤬

Middle-aged (45-64 year old) summary

• 45-54 year olds (left)

  • ‘20 and ‘21 total deaths were ⬆ 23% and 35% above 18-19 average.

  • Again, drugs and alcohol, homicide, diabetes, liver disease and transport accident data are really, really bad.

  • Suicides ⬇ a significant amount🤔

  • Cardiovascular ⬆ >11% above 18-19 levels in both ‘20 & ‘21.

  • Cancer ⬇ moderately

• 55-64 year olds (right)

  • ‘20 and ‘21 total deaths were ⬆ 20% and 26% above 18-19 average.

  • Story is very similar to 45-54 year olds, but moderately better on most accounts

Seniors (65-84 year old) summary

• 65-74 year olds (left)

  • ‘20 and ‘21 total deaths were ⬆ 19% and 21% above 18-19 average.

  • Again, drugs and alcohol are large increases, but smaller numbers in absolute than other ages.

  • Diabetes, Alzheimer, Parkinson, and liver disease are pretty bad.

  • Cancer is down mildly.

  • Respiratory diseases down moderately.

  • Cardiovascular disease ⬆ >6%

• 75-84 year olds (right)

  • ‘20 and ‘21 total deaths were ⬆ 17% and 28% above 18-19 average.

  • Very similar trends as 65-74 year olds, but milder in nature.

Explore the data yourself

Another day, another dashboard 😁. Click on the picture below to explore this data yourself. Click on the map to choose a state(s), select an age group as desired, and charts like the above will display on the right hand side of the dashboard. Below the map you will see some high level health indicators of the chosen geography, with a comparison to national average. Note 1: these indicators are not by age, so age group filter does not affect it. Note 2: If you do not pick a particular state(s), then health indicators will appear to have duplicate rows because it is comparing the US as a whole to the national average (same data).

Here is an an example of dashboard with California selected:

P.S. Again, some of the numbers 👆 are so horrible it makes me question if I made an error somewhere. If you think (know) that something is off, please do let me know.

UPDATE

As a spot check on data accuracy. Above I show an alarming number for drug overdose increases for 35-44 year olds in 2021 vs 2018-19. This comparison will not be exact for a couple technical points, but using a quick high level pull of data from CDC Wonder, the number above looks to be accurate: