STUMP » Articles » Mortality Monday: Suicide -- the Absolute Numbers » 24 July 2017, 04:35

Where Stu & MP spout off about everything.

Mortality Monday: Suicide -- the Absolute Numbers  


24 July 2017, 04:35

Looking at the create date for this post, it was 14 August 2014. I started it, and built up a lot. But didn’t post it.

If you’re here because of recent events: Statement by the American Foundation for Suicide Prevention on Chester Bennington

The reason I started the post back in 2014 was Robin William’s death, and the reason I didn’t post it… it’s tough. I have way too much experience with suicide, on both a personal and professional level. When I worked at a reinsurer, I had to review death claims files, and we’d focus specifically on large covered amounts. I will talk about that at the end of the post.

For this post is mainly just going to be about the absolute numbers – who is committing suicide in terms of race, age, and sex. I will talk about rates and trends next week.

I will not be talking about any specific individuals, causes, or even methods of suicide in this post. I will likely not write about those this year. It took me this long just to post the dry numbers.

All of the numbers/graphs in this post are based on the 2014 U.S. Deaths Database I’ve used in earlier Mortality Monday posts.


I am going to start out at the highest level, and then get more detailed in terms of demographic details.

In my database, I have a category for “Manner of Death”, which breaks out in this way:

Now, I don’t want you to get het up about the “Not specified” category. It could be that it’s hiding suicide stats, but you’ll see where that category gets reported a little later on.

Even so, note that suicide numbers about 150% higher than homicide, and is about a third of accidental deaths.


My biggest theme is going to be the gender gap in suicide. Let me give you the high level numbers:

Notes: numbers by violent death — especially homicide, and suicide, but also accident, are much higher for men.


Now, I’m going to start digging in, but here is where the numbers may be a bit misleading. This is why I will be revisiting the numbers next week, but looking at rates instead of the base numbers. We’ll also see this issue when I break out suicide numbers by race/ethnicity.

Another note: the total deaths do not always match up for each of these split-outs, because there are always some deaths for which a particular characteristic is not recorded. I exclude the “unknown” category when I graph.

Let me give you the straight pattern of deaths by age ranges:


  • Yes, I made the suicide data series black.
  • You can barely see the suicide bars at younger ages – yes, the numbers are low.
  • While the numbers grow into middle age, the suicide bars fill up larger amounts at younger ages… because there are so few deaths.

Let me do this as percentages:


  • Suicide is a substantial manner of death for ages 15 – 34
  • Accidental deaths as well as homicides is also elevated for these ages
  • Unsurprisingly, old folks tend to die of natural causes… and yes, so many of them die

It can be tough to see how the suicides actually fall out by age, so let me do this with suicide by itself:

The peak number of suicides falls in the age range of 55-59.

You may not have been expecting that.

Again, this is partly misleading — these people were born in the late 1950s — aka the late Baby Boom… which was actually the peak of the baby boom in the numbers of births. We will see this next week when we look at suicide rates.

So there were lots more people around to commit suicides to begin with.


So let’s pull this out by both age and sex, and not have the ranges anymore.

Yes, the youngest suicides recorded were at age 9.

Notice the difference in the pattern of male and female suicide numbers: a rapid climb through the teens for males, a drop-off in early middle age, and then climbing again to peak in later middle age.

Female suicide numbers steadily climb throughough teens into middle age, with no extreme spikes like you see with males.

Again, keep in mind that we will later need to look at rates to get a better handle on what’s going on. But often the focus for suicide in media has been for younger people; part of this is because so few people in those age ranges die from any reason. So suicide stands out.

In middle age, so many more natural causes are killing people, one generally doesn’t notice that more middle-aged people, especially middle-aged men, are committing suicide than are teens.


The data I have here are a bit less reliable, because sometimes race and ethnicity are imputed, and I have more missing data here.

Here’s the crudest cut:

This crude cut shows:

  • Over-representation in suicides among Non-Hispanic whites, Hispanics, and Non-Hispanic other (which is primarily east Asian, from census records because they didn’t give me that breakout in the death records)
  • Interestingly, there is a low amount of suicides among non-Hispanic blacks.

Let’s do these tables in a couple ways.

First, let’s look at percentages so that the row percentages add up to 100% – basically, for each manner of death, what % were of which race/ethnicity?

Yeah, non-Hispanic black is way over-represented among homicides. I will deal with homicides at another time.

Here is where I look for a given race/ethnicity, how much of a percentage the manner of death makes up.

And here is where we come back to earth a bit — while non-Hispanic blacks have a disproportionate number of homicides, it’s still a very little cause of death for the overall population.

Still, interesting how high accidental deaths rate for certain groups.

Because we’re looking at raw numbers, we may be missing the underlying age distributions — the disproportionate number of accidental deaths may be linked to a disproportionately younger population for a specific race/ethnicity.


Here’s our next slice – race/ethnicity and age:

So… there’s a lot of white people. Let’s try something different – let’s look at percentage of deaths that are suicides for a specific age/race bucket:

That’s quite a different pattern, isn’t it?

And you can see why people focus on adolescent suicides — it’s a large component of deaths at those ages. Of course, so are accidents and homicides.


Now let us look at the numbers for all the dimensions at once. I will bucket the ages as above.

And we’ll do it as a table. I will take male/female ratios for suicides for each bucket.

I restricted which age buckets to look at, because there were too few deaths to have reliable ratios. But you can see the real result in the total line:

  • While Non-Hispanic Other has disproportionate suicide numbers… the gender gap is less.
  • Yes, suicides are disproportionately low for Non-Hispanic blacks — but it’s really skewed towards males, more than any other group
  • Hispanics also have a heavy male skew — but it may be that the underlying sex ratio of the Hispanic population.

In fact, I did a quick look — while both Non-Hispanic whites and Non-Hispanic blacks had similar overall male/female death ratios of 102%, the other two groups, which are immigrant-heavy, have heavier skews: 129% and 115%.

With regards to immigration, in general you get more male immigrants. The men come over first, and then either go back, or bring the women & children later.

So the bottom line in all this is that while we mainly pay attention to bulk numbers, it’s really the rates that matter.

And that’s what I’ll look at next week.


As for my professional experience, one thing I used to do at a reinsurer was review death claims, with a particular focus on early death ages and/or large insured amounts. Yes, there were suicides on the list, as well as cancer, primarily, for the younger ages.

No, it wasn’t to deny claims — first off, the reinsurer wouldn’t be making that call in terms of what beneficiaries get. It would be original insurer that issued the policy.

But one thing people often assume is that you can’t get a life insurance death benefit if the insured died via suicide. That’s not true in all cases — at least, not in the U.S. There is something called a suicide clause — yes, the claim will be denied if it was death by suicide… but only within a certain amount of time after the policy was issued (usually 2 years). After that period, the death claim would be paid.

The only unusual time I have seen where the company won’t pay the death benefit to beneficiaries is when the beneficiary is charged for murdering the insured. The company would pay the money to a trust — the point is insurers aren’t interested in paying off murderers. Recent example: Man who killed mom wants to get paid from her life insurance – guy was acquitted, but … well, read the story.

As for why I was reviewing death records — well, some of it was to see if there were underwriting issues in our book of business, but it can also be a way of checking for fraud. If you want to see why that would be, consider the TV movie Billion Dollar Bubble – yes, it’s on YouTube and only an hour long. It’s got James Woods (as the actuary!) and Christopher Guest, and the scene of them smoking pot as they talk about creating fake policies and then fake deaths to get reinsurers to pay claims… well, you just gotta watch it.

Spreadsheet behind these tables and graphs

Related Posts
Mortality Monday: All About Traffic Deaths!
Mortality Monday (with a little Trumpery): Supreme Court Probabilities Take 2
Mortality Monday: Broken Heart Syndrome