Top Causes of Death by Age Group, 2020: Death Rates
Comparing 2019 and 2020, non-COVID causes need investigation
And now here comes the numbers where you can start making comparisons between years and between age groups.
As a reminder, I’m doing this for full-year 2020, as we have finalized cause of death database numbers for that year. The 2021 numbers are provisional, and when I do future posts focusing on specific causes of death, you will see that the data peter off at different times for different causes. There are a lot of reasons it takes almost a full year before we get finalized data for cause of death.
If rates are not really comfortable for you, please take a look at two earlier posts: Top Causes of Death by Age Group, 2020: Raw Numbers and Mortality Nuggets: Videos on Death Numbers, Ranking Table for States’ Mortality for 2020. Yes, I will likely be doing a video on the tables and graphs you see below.
If you are reading this on my substack, there will be the spreadsheets with all the data, tables, and graphs at the bottom of the post.
Top Causes of Death in 2020, Death Rates by Age Group
Total death rates by age group are at the top of the table:
A few comments:
I present the rates in percentages, as opposed to the more traditional number (which is per 100,000 people per year), because I do not want people to get this confused with the raw counts of people who died. Yes, that does mean there are a lot of small numbers. For children, I even had to extend some out to 4 decimal places to get a significant figure.
In adulthood, natural causes of death tend to increase in rate with increasing age. More below.
External causes (accidents, homicides, and suicide) will have the similar rates over broad ages but drop dramatically in ranking with increasing age — as the natural causes become more likely to occur.
COVID has a similar pattern in mortality as heart disease — indeed, the heart disease death rate is approximately twice that of the COVID death rate for the entire age range from 15 to 85+ on the table.
Don’t get too excited by that final point, by the way. I will address why this may be in the next section.
Top Causes of Death in 2019, Death Rates by Age Group
For comparison purposes, here is the same table for 2019:
Obviously, with COVID not there as a top cause of death, other causes come to the fore.
In future posts, we will also see some reasons for changes in rank and/or magnitude of certain causes, because there were some prominent changes for certain non-COVID causes of death.
It is obvious (and I’m not going to link to the various surveys, in asking people to estimate death rates) that most people have no clue what a high and a low death rate are. This is why I’m providing the 2019 table for comparison. I don’t expect people to know what normal patterns of mortality look like. 2019 was a pretty normal year.
Age is the predominant factor in death rates, and you can see that for many of the natural causes of death (as opposed to “external”, like homicide, suicide, and accidents), there is an increasing death rate with increasing age. Indeed, not only does that rate increase, but it accelerates. You can see one of my posts from a year ago where I demonstrate a common frailty model structure actuaries use that has this feature.
Once you get old, the force of mortality often climbs rapidly, and that was true in pre-pandemic times.
So this frailty structure is probably related to the 2-to-1 heart disease-to-COVID death rate relationship I observed in the prior section. It’s just not that unusual for natural causes of death to follow a similar pattern described in the model, and thus death rates following such models would be in fairly stable ratios by age groups.
Death rates by age group, 2019 vs 2020
Overall death rates:
Non-COVID death rates by age group, 2019 vs 2020
Let’s remove the COVID deaths to see if there’s still some excess mortality left over:
It’s difficult to see, but there is some there.
Heart disease death rates by age group, 2019 v 2020
As with my prior posts, let’s look at the top two causes of death (going back decades) — heart disease:
It’s difficult to see on this graph, but heart disease is a little higher for all the adult age groups.
Cancer death rates by age group, 2019 v 2020
We can see some of the difference — cancer death rates are lower for some of the groups.
Change in death rates from 2019 to 2020
Let’s put it all together to look at the change in death rates from 2019 to 2020, going by the categories we see above:
Yes, I see that orange bar sticking up in the 1-4 years category. I will get back to it.
I have been looking at 2021 data, and the age effects are somewhat similar, if we compare 2021 against 2019. I don’t have solid estimates for December 2021 deaths, but for the other months, I’m pretty close. There is an out-sized effect on younger adult groups, as we see with those total deaths bars.
We see that from those age 25-44, the death rates increased more than 20% from 2019 to 2020, and when we remove COVID deaths from the calculation, we still see a lot of excess mortality yet to be explained.
Cancer doesn’t seem to be involved — looking at the percentage changes in rates, that looks like extending the general good trend we’ve seen in cancer death rates (very slow decrease in cancer death rates) over decades.
Heart disease might be involved. We do see some substantial increases on that cause of death. As I have remarked multiple times, this may be related to undiagnosed COVID deaths, because when one digs into increased heart disease deaths in 2020, much of it was in spring 2020, and it tended to be heart attacks (COVID-caused blood clots?)
That said, even once heart disease deaths are removed, there is still some excess mortality left over. (Mind you, you can’t just subtract the percentages — these are looking at the changes in rates… yeah, I know it gets confusing). After all, I’m only showing that the heart disease death rate increased a certain percentage… and not many people in those age groups die of heart disease to begin with.
Which takes me to my next bit.
Beware of large changes in small numbers
Yes, yes, I see that large orange finger in cancer deaths pointing up, showing an almost 10% increase in the cancer death rate for those aged 1-4.
You can look to my earlier post on the raw numbers, and you will see that 307 children aged 1-4 died of cancer in 2020. In 2019, 285 children aged 1-4 died of cancer.
So a difference of 22 children between the two years (not to mention the difference in population size, but that has less impact). When that few children die (out of a population of approximately 15.6 million), it doesn’t take that many deaths to move the needle.
This is generally why I stay away from analyzing child mortality.
Not only because it is really depressing (it is), but because death counts are so low, one gets an awful lot of noise just from sheer randomness.
Heck, even the ranking of deaths is fairly unstable for some of the causes due to so few children dying — flu/pneumonia bubbles up and down, as well as others.
Once we get to cause of death #7 in our ranking table, for 2019 and 2020, we’re looking at fewer than 100 deaths for the entire country for the entire year for the category of kids age 1-4. Some of the causes are within one death of each other — I’m glad I didn’t need to use any tie-breakers to make the tables.
I can’t do any reliable statistical analysis when the numbers are so low. In general, I exclude children from my mortality analyses, or make a very large group, like under age 15 (though most of the deaths for that grouping are under age 1. Very sad, but true.)
In posts to come, I’ll be looking at non-COVID causes of death such as accidents, homicides, and suicide. All three have had different trends by age and other factors, such as sex, ethnicity, and geography.
Data were drawn from the CDC Wonder Database as of January 2022, using finalized numbers for 2020 and 2019, using the underlying cause of death database.
Cause of death categories for top leading causes were defined using the 113 cause of death list, using only the “rankable” causes (of which there are 51). I did not use the infant mortality cause list for those under age 1, because (sorry, not sorry), I am not interested in investigating infant mortality trends at this time. Those have their own complexities, even without pandemic influences.
Ten-year age groups were used for the groupings, as seen, for the entire U.S.
Death rates are presented as percentages — calculated as the number of deaths for the category divided by the estimated population for that age grouping.
If you are on the substack post, the spreadsheet with the table and graphs, and all the underlying numbers, are found below in the Excel spreadsheet. You can contact me if you have any questions about the numbers: email@example.com.