Trends in Canadian youth suicide by gender, 2000–2023
In the summer I looked at trends in the US, but what about its neighbor to the north?
During the pandemic I learned of adolescent-psychiatrist Tyler Black from his rebuttals of grim claims of suicide spikes, especially claims of suicide spikes among locked-down or schooled-at-home children. In April 2023 he started a Substack blogletter, posting, inter alia, interesting and informative analyses of pediatric suicides in the US and in Canada. Unfortunately, his posts there stopped in September 2023,1 and I’m curious as to developments since then.
Fortunately, earlier this month, Statistics Canada updated its public dataset of leading causes of death in Canada, enabling me to see whether the pandemic-era declines in Canadian pediatric suicides persisted into 2023. The numbers also offer a chance to check, as I did for the US in the summer, whether they’re consistent with the Haidt-Twenge hypothesis that smartphones and social media sent teen suicides skyward from 2012 (or 2010, depending on which day you ask Haidt2), especially in girls.
Suicide rates in Canadian teenagers by gender, 2000–2023
To start, age-standardized suicide rates per 100,000 among 10–14-year-old girls (in red) and boys (in blue).
Now, even aside from the Twenge/Haidt stuff, or the matter of what’s changed during the pandemic, there’s an instantly notable finding: the girls have the higher suicide rates, not the boys.
Normally the male suicide rate stands above the female suicide rate (men and boys are more likely to be “completers”, in the jargon of suicidologists, even when women and girls are more likely to harm themselves non-fatally and to attempt suicide). But 10–14-year-old Canadians reverse the pattern and, curiously, this reversal is somewhat new. As recently as 2010, the boys’ suicide rate was higher.
Is this finding an error of mine? No, a 2023 Canadian Journal of Psychiatry paper finds the same reversal, in opposition to every “other industrialized country”. Still, neither I nor the paper have special insight to what makes post-2010 Canada special in this way, so I move on.
As to what changed in the pandemic and in 2023 specifically, I confirm Tyler Black’s earlier finding that, although the data are noisy, there was no sign of a dramatic spike in suicides in 2020 or 2021. 2020 saw slight increases among both the boys and the girls, within the ranges of normal variation. 2021 saw a slight decrease among the girls and a halving of suicides among the boys. The new dataset shows the 2022 suicide rates basically holding level for the boys and dipping for the girls, and the 2023 suicide rates returning to pre-pandemic norms.
I don’t want to over-interpret these changes, because they’re built on (mercifully) few suicides. The girls’ peak in 2016 represents 31 suicides; the boys’ halving of suicides from 2020 to 2021 represents a decline of just 10 suicides, from 19 to 9. At the same time, it’s not as if there’s any suggestion of an abrupt lockdown-era jump. The boys’ rate bottomed out in 2021/2022, and its average over the 4 pandemic years was essentially normal; the girls’ rate hit a decadal low in 2022 and its pandemic-era average was less than each of the 7 years before it.
Finally, on the idea that smartphones and social media sent youth suicide rates skyrocketing, there’s no sign of any rise in boys, where the trend is if anything slightly downward. To give the idea its due, there is a clear upward trend among girls for a few years after 2010/2012, but the timing is a little off; the trend really starts from 2008/2009, and it ended at least 5 years ago. It’s surely a little surprising that these girls’ suicides recently declined by something like a third, including the pandemic period when they presumably spent more time on their phones and social apps.
However, again: small numbers. I turn now to 15–19-year-old Canadians, where the numbers involved are larger and the relative scatter of the data can be expected to be less.
These data do look less frothy, and we’re back to the familiar pattern of the male suicide rate being higher. On the social-media-suicide notion, the timing fits worse than it did for the younger kids; any upward trends I draw on this chart would surely start not in 2010 or 2012 but no later than 2006, when suicide rates plainly dipped for both boys and girls.
Moreover, fitting trendlines for 2006 onward, they’re not statistically distinguishable from flat lines. The girls’ trend is upward (1.4% per year), but statistically insignificant (the standard error of the trend’s slope is 0.9 percentage points per year). Nor is it significantly different from the boys’ trend (−0.5% per year, with a 0.6-point-per-year error). So the gender aspect of the hypothesis doesn’t fit. And we again have a pandemic-era decline in suicides, when teens, again, doubtless spent more time on their phones, not less.
This pandemic-era decline is particularly interesting because the new data show it deepening in 2022, even compared to the relatively low rates of 2020, and remaining low in 2023. Indeed the boys and men’s 2023 suicide rate is their lowest on record. In absolute terms, I reckon the total reduction in suicides from 2020 through 2023 is about 120 for the boys and men, and about 30 for the girls and women — and that’s when including those low-suicide years when fitting the trendline.
Comparisons to adults aged 20–34
Statistics Canada supply death data for other adults, too, and I take the chance to repeat an exercise I did for the US: comparing teenagers’ suicide rates to those of adults in the closest age groups. First, boys and men, including men aged 20–24, 25–29, or 30–34. (Watch out for the logarithmic scale, which I use to prevent the lowest line from being crushed next to zero.)
First off, the stereotype of teenagers being more suicide-prone than adults is bunk. (It was bunk in the US data too.) The age groups of 20–24, 25–29, and 30–34 all have higher suicide rates than either teenage group.
More relevantly, the lines move very approximately in parallel. Whatever the precise age, these male suicide rates all have near-flat, maybe downward, trends. The boys aren’t showing worse trends than the men. And in all 4 of the age groups older than the (noisy) 10–14-year-olds, suicides clearly trended down from 2020 onward. Men in their twenties showed upticks in suicide in 2023, but from record lows in 2022, so even in 2023 their suicide rates were relatively low.
Now for girls and women. As with the boys and men, mind the log scale.
Above age 14, the suicide rates for girls and women are lower than those of men and boys, but otherwise the picture is less cheerful here. There are declines in suicide rates from 2020 through 2023, but more upticks in 2023 than for the male 15–34-year-olds. And the female suicide rates seem to follow similar, gently upward, pre-pandemic trends to my eye (though I do mention above that the trend isn’t statistically significant for 15–19-year-olds). One way to interpret that is as evidence against the Haidt-Twenge hypothesis. If teenage girls are being newly driven to suicide by social media, why do their suicide rates move in tandem with the suicide rates of women aged 20–24, 24–29, 30–34?
What about MAID?
Some readers must be wondering about MAID, or “medical assistance in dying”, the Canadian healthcare system’s provision for assisted suicide/death. Under federal legislation passed in 2016, Canadian adults satisfying certain criteria can have medical personnel help them die. Official provision of this MAID service has since expanded a lot, triggering great controversy in the wider English-speaking media, so MAID now springs readily to mind for some of us whenever we hear about Canada and suicide. And so, I fear, I have to put in a whoooole long tangent about MAID.
Alright: does MAID’s growth influence my analyses?
Not to any meaningful extent.
Children are banned from MAID
First, I’m focusing on teenage suicide, and children are forbidden to receive MAID. Informed consent, freely given, is not sufficient to allow a Canadian doctor to shoot you up with a Michael Jacksonian dose of benzos and propofol; you must “be at least 18 years old and mentally competent”.
An obvious riposte is that the law is just words on a page, and doctors could simply be flouting it by blithely shooting up 13-year-olds with fatal drug cocktails. They could. Not only that, but I’d guess that they did occasionally give a few children deadly drugs before MAID was instituted. Doctors informally providing deadly drugs to terminally ill patients is widely recognized as a known, if irregular, practice, and I expect a few terminally ill patients getting those drugs have been minors. I also expect that Canadian doctors occasionally helped sick children die3 before MAID, do it now, and would continue to do it if MAID were abolished tomorrow.
Really, whether doctors assist children in dying is not germane. The pertinent question is whether they assist children in dying in the guise of MAID. My answer to that is “seldom, perhaps never”; MAID requires paperwork, so a doctor killing a child under MAID is either
not completing and filing the MAID forms (in which case, what’s the basis for saying they’re killing the child under the aegis of MAID?) or
completing the MAID forms with an easily detected falsehood (misrepresenting the child as an adult, at least).
As such, any medical professional who uses MAID as a pretext to help a child die is setting a very big, and very sensitive, bomb under their career. This is a strong incentive to not use MAID as a pretext to kill children!
There is another strong incentive: the loud group of people who find MAID highly objectionable, and are keen to publicize the story of anyone who might be characterized as a victim of MAID. A significant fraction of media coverage and online commentary about MAID discusses potential victims of MAID, whether they’re people who might’ve been coerced into being medically killed, vulnerable people (usually the disabled or simply poor) who resorted to MAID because they couldn’t get other help for medical problems, or younger adults free of terminal illness who request MAID despite having most of their probable life left to live.
Yet for all that coverage, I have located no reliable reports of a child receiving MAID, despite how juicy it would be as a news story.4 Instead I find
National Catholic Register describing a judge allowing an autistic 27-year-old to get MAID over her father’s objections;
America (“a monthly Catholic magazine published by the Jesuits of the United States”) reporting on a seminary professor who was “stunned” by her 89-year-old grandmother with colon cancer receiving MAID two days after requesting it, and quoting her (the professor, not the grandmother) as saying that her “19-year-old brother said he was going to apply for MAID when he gets old”;
Christian Action, Research and Education quoting a 19-year-old veteran who declares it “unacceptable” and “infuriating” that other veterans “have been offered euthanasia by authorities”;
the New York Post reporting on another veteran, of unspecified age, who was seeking support for PTSD from a Veterans Affairs Canada employee, an employee who responded by mentioning MAID;
a Journal of Pain and Symptom Management paper about 3 octogenarians who were among the first people denied MAID who subsequently attempted suicide;
an Associated Press report of disabled people, without terminal illness, receiving or being offered MAID, with the specific people named being aged 61 (Alan Nichols), 41 (Sean Tagert), and 25 (Candice Lewis);
CTV News reporting on generic objections to “Track 2” MAID, the provision of MAID to the 4% of MAID recipients who had no terminal illness;
Agence France-Presse rebutting false online claims that a 13-year-old cancer patient signed MAID forms when her family left her in a hospital for an hour, and was killed, over her parents’ objections, two days later; and
a mom campaigning, after her 17-year-old son died of bone cancer, for “mature minors” to be allowed to get MAID;
but not one verifiable report of a Canadian child actually getting MAID.
Few young adults get MAID
I’m satisfied that minors don’t get MAID. But, a skeptic might ask, what about 18- and 19-year-olds, who are in principle old enough to get MAID? Well, it’s not clear any of them get MAID either! The youngest MAID recipient I’ve located is 20-year-old Eric Coulam, who opted for MAID rather than undergo a record-breaking 6 organ transplants. (A lot of media reports about Coulam from mid-2022 don’t make clear whether he ultimately died; his Facebook support/fan group confirms he did.)
The Government of Canada’s new annual MAID report also provides evidence that very few teenage adults get MAID.
In 2023, 15,343 people got MAID, 14,721 via Track 1 and the remaining 622 via Track 2. Only 1.1% of the Track 1 MAID recipients, and 4.0% of the Track 2 MAID recipients, were younger than 45. So 25 of the Track 2 MAID deaths, and 155–169 (allowing for the rounding of the 1.1%) of the Track 1 MAID deaths were of adults under 45. The 18–44 age range is wide; supposing that MAID deaths were spread evenly through the age group (i.e. that there were equally many MAID deaths of 18-year-olds, of 19-year-olds, of 20-year-olds, etc., all the way up to the 44-year-olds), then 11–13 teenagers died by Track 1 MAID and two teenagers died by Track 2 MAID, or 13–15 teenagers in total. And these last numbers must be overestimates; in either Track, every older age group contributes more MAID deaths, and a similar positive correlation between age and MAID deaths presumably holds among the 18–44-year-olds, in which case 18- and 19-year-olds contribute relatively few MAID deaths, and the 13–15 total is (far) too high.
I’m confident that the real number of teenage adults’ MAID deaths in 2023 is in the single digits. It could be nil. By contrast, Statistics Canada counts 196 teenagers who killed themselves in 2023, dwarfing the number of MAID deaths.
Finally, since I also invoke suicide statistics for adults aged 20–34, I note that 903 of them killed themselves in 2023, outnumbering MAID deaths at least 4:1, and almost certainly more than 8:1 (the ratio I get by again assuming equal MAID deaths throughout the 18–44 age group). There aren’t very many MAID deaths among twentysomethings.
Conclusion
I have happier news about teenage suicide trends in Canada than I had for the US.
The rates are lower in Canada, except in 15–19-year-old girls, for whom they recently converged to the US rate of 5 per 100,000 per year.
The rates declined in the 2020s, and were below trend for every age-sex group in 2022 in particular.
Longer-term trends are slight, possibly downward for boys and upward for girls, but statistically tough to distinguish from flat.
Trends in teenagers and 20–34-year-olds are broadly similar: notably lower in the 2020s, slightly upward in women before then, and near flat (perhaps downward) in men.
MAID is statistically irrelevant to teenagers. Minors are de jure and de facto barred from MAID, and I estimate that fewer than 10 (and quite possibly zero) teenage adults got MAID in 2023.
The potentially upward trends in girls might be read as support for the Haidt-Twenge idea that smartphones and social media hurt teenagers’ mental health, but the pandemic-era declines in suicides, lack of an upward trend in boys, and similarities to trends in 20–34-year-olds are all evidence against it.
This is not to say that everything’s fine. For instance, a national survey found increases in suicidal ideation in young Canadian adults in 2021. But given the readiness of the news to report on “dramatic” increases in suicide in Alberta, or to announce that suicide “among young women is on the rise”, or to reveal “an unprecedented spike in suicide deaths among young girls”, recognition of lower and falling teenage suicides is surely apposite too.
He continues to post elsewhere, including on Bluesky, where he highlighted a month ago that Canada’s overall pandemic decline in suicide was sustained in 2023.
And assuming he’s not too busy emailing the Premier of South Australia to privately bash his critics.
By the way, I count withdrawal of medical treatment necessary to maintain life as helping someone die. I see no deep, morally relevant difference between sliding a life-ending needle into someone’s arm and sliding a life-sustaining needle out of someone’s arm.
Juicy enough that bullshit artists sometimes just make shit up about “Canada […] encouraging minors to commit suicide” or how children “can now kill themselves with Canadian government assistance”, citing an advocacy group’s aspirational webpage instead of any actual cases.