Geeking Out: Finalized 1999-2022 U.S. Cause of Death Spreadsheet (a gift from me!)

(If you have Microsoft Excel)

Rather than wait until I have pretty graphs for other people to look at, I thought I’d share a downloadable spreadsheet for other people to use.

Spreadsheet

1999 2022 U

3.21MB ∙ XLSX file

Download

Download

Video demo

There is no audio on this — just showing me navigating the spreadsheet:

Basically, I’m taking the finalized data from CDC WONDER from 1999-2022, doing look-ups using the 10-year age groups and their standard major causes of death.

On top of that, I put in some default sparklines and year-over-year percentage changes. I will put in a big caveat below (and I’m about to lay in some actuarial stuff at the very end, because there are very good reasons for all of this.)

I know many people have issues using CDC WONDER. I did do some videos on how to use it (here’s the playlist: Working with CDC WONDER). WONDER hasn’t changed in how it’s operated in the last couple of years, and I find it as much of a delight as ever. But I understand that others do not like it.

So I am gifting this relatively simple-to-use spreadsheet to do WHATEVER YOU WANT TO DO WITH IT. Okay? Just take it.

But if you bug me about it… be prepared to have some actuarial standards dumped on your head.

With Actuarial Standards Come Great Responsibility…. Or Rather… Get Ready To Handle the TRUTH

Okay, it’s not that bad.

Yesterday, I was at the Spring Meeting of the Actuarial Club of Hartford and Springfield (Massachusetts, no relation to the Simpsons… probably).

I had a great time — in general, actuaries who go to actuarial conferences are the type of actuaries who love being at conferences and we all have a fabulous time.

Okay, I may be digressing from my theme. Or not. I saw a lot of my actuarial friends…. (and my enemies? We didn’t even look at each other’s shoes. That’s cold.)

We had several great sessions yesterday — two of which I will bring to people’s attention.

One began the day: Dave Dillon: Professionalism in the Everyday life of an Actuary.

The other ended the day: All Things AI: Principles, Standards, and Best Practices by Mitch Stephenson

Both drew heavily on Actuarial Standards of Practice,

Actuarial Standards Involved: Data Quality and Modeling

The main actuarial standard being engaged here is ASOP 23, Data Quality, pretty much my favorite ASOP.

There is one major issue with the spreadsheet above, and that’s the 2021 population estimates, which are the denominators for the death rates.

The 2022 population estimates in the finalized database are good, but the 2021 numbers are still not great.

More here:

For ages 85+, the death rates are overstated in 2021, due to underestimated population for this group. For age 75-84, and 65-74, there are still some underestimates, but not as extreme. This has to do with undercounts for the nursing home population in particular.

There were some effects on the younger adult population as well, to the extent they may have been in group living arrangements in 2021.

However… who was in a group living arrangement in 2021?

Many or most colleges just completely shut down their dorm arrangements - and this was for July 1 estimates, so not many of college students would be affected. However, some disabled adults would have been affected in the estimates if they were in group homes.

So keep in mind that the CDC is aggregating death numbers, categorizing by Underlying Cause of Death (which there can be one and only one per death certificate), using ICD-10 codes.

I am using the CDC’s aggregation of 113 causes list (which isn’t 113, but we’ll ignore that for now).

But the most significant disparity is that the population estimates in 2021 for older ages is not consistent with 2022 and 2020, so that the year-over-year changes are not the best comparisons for those groups. Don’t over-interpret that.