United States · 1999–2024
School days are associated with youth suicide
No matter how you chop the data, the result is always the same: kids struggle when school is in session.
Pediatric Suicide Heatmap
| Sun | Mon | Tue | Wed | Thu | Fri | Sat | All | |
|---|---|---|---|---|---|---|---|---|
| January | +26% | +66% | +46% | +41% | +44% | +20% | +16% | +37% |
| February | +27% | +49% | +53% | +53% | +58% | +29% | +17% | +41% |
| March | +14% | +56% | +35% | +50% | +50% | +38% | +16% | +37% |
| April | +34% | +51% | +58% | +49% | +41% | +32% | +31% | +42% |
| May | +12% | +30% | +40% | +50% | +38% | +21% | +28% | +31% |
| June | −8% | +15% | +8% | +16% | +13% | +18% | −5% | +8% |
| July | −10% | +9% | +3% | +10% | −8% | −5% | −0% | ref |
| August | +20% | +33% | +27% | +21% | +20% | +6% | +10% | +20% |
| September | +19% | +53% | +41% | +47% | +38% | +18% | +8% | +32% |
| October | +39% | +58% | +54% | +53% | +40% | +24% | +13% | +40% |
| November | +32% | +61% | +48% | +43% | +32% | +33% | +1% | +36% |
| December | +6% | +19% | +11% | +28% | +15% | −3% | −0% | +11% |
| All | +17% | +42% | +35% | +38% | +31% | +19% | +11% | +28% |
Visualization by Dr. Tyler R. Black – pediatricsuicides.ca – source: CDC WONDER
A mechanism to test "does school cause youth suicide?"
In the following chart, we subdivide months into Non-school months (July), Partial School months (June, August, and December), and Full School Months (January-May, September-November). By doing so, we can test the "dose response curve" (an important part of testing cause in a correlation). As is clearly shown, the dose-response curve holds both ways. As school days increase, suicide risk increases. As we go from weekends to weekdays, suicide risk increases. As these are still associations, they do not definitively show cause, but they are what we would expect if there was a causative relationship.
| Full school months | Partial school months | No school | |
|---|---|---|---|
| Weekdays | +52%95% CI +39% to +65%4.73 per 100,00016,512 deaths · 349.0M person-years | +23%95% CI +13% to +34%3.84 per 100,0005,091 deaths · 132.6M person-years | +8%95% CI −2% to +19%3.37 per 100,0001,497 deaths · 44.4M person-years |
| Weekends | +28%95% CI +17% to +39%3.98 per 100,0005,555 deaths · 139.4M person-years | +10%95% CI −0% to +20%3.42 per 100,0001,813 deaths · 53.0M person-years | reference3.12 per 100,000564 deaths · 18.1M person-years |
Visualization by Dr. Tyler R. Black – pediatricsuicides.ca – source: CDC WONDER
These six cells hold 31,032 deaths across 736.5 million person-years of American childhood.
A model year, day by day
This model year was created mathematically by summing weekly data from the CDC over the past 7 years (excluding the pandemic 2020 year). This visually demonstrates the drop of suicide risk as school turns to summer, the increase as kids get ready to return (and in the states, many do return in August), and the substantial and obvious drop off for the winter break. Spring break is more difficult to model because it is handled differently in almost every jurisdiction, but for whatever reason, we see no 'spring break effect.'
Visualization by Dr. Tyler R. Black – pediatricsuicides.ca – source: CDC WONDER
School breaks against the school year
| Period | Rate | Compared with | Change | ||
|---|---|---|---|---|---|
| Winter break | vs the school year | 2.33 | 3.21 | −27% | 95% CI −35% to −18% |
| Easter span (2 wk) | vs the school year | 3.24 | 3.21 | +1% | 95% CI −9% to +12% |
| July to mid-August | vs the rest of the year | 2.51 | 3.17 | −21% | 95% CI −26% to −15% |
What happened when schools closed for COVID
The commonly held myth is that suicides increased with lockdowns. In fact, suicides decreased significantly when lockdowns were universal, such that march-may did something for the first time in recorded history, which is approximate July's rate. The data from the US clearly shows that it was returning to school, not lockdowns, that was associated with upward deviations from suicide historical suicide rates. 2021 was abnormal early on, but by April, suicides followed expected trends.
Visualization by Dr. Tyler R. Black – pediatricsuicides.ca – source: CDC WONDER
Age group and sex follow the selector at the top of the page.
Important information on these charts
Associations are just that, they are links. They do not establish cause, and never really can. What I can say is that my hypothesis that school days predict pediatric suicide rates has held every year for 15 years, and held during the pandemic year, in which when schools were universally closed, pediatric suicide rates plummeted. As well, we have a dose-response-relationship between "how full the month is of school days" and "weekends/weekdays" that fits the causative model as well. As randomizing kids to receive school at home or in person is challenging to do, we will likely never get a true causative test.
I've included the 18-24 comparison group because I care about them immensely (we call them "transitional youth", both transitioning from childhood to adulthood as well as moving from the relatively protected youth world to the extremely variable and unprotected adult one. Because we do not have "school status" as a CDC wonder metric, there is no way for me to know what the impacts of schooling is on this group (one could assume the association holds for university). With <40% of 18-24s pursuing significant higher education, these national numbers simply cannot show us.
I have written about "how schools can improve suicide risk for kids" in the article "Children’s Risk of Suicide Increases on School Days".
I can be contacted about any of the above, and am willing to be interviewed in print, audio, and video at any time. Please contact me at dr.tylerblack@gmail.com.