It's 2:14am and your chest has that tightness that isn't quite panic but is definitely panic-adjacent, and you're scrolling your contacts. Somewhere on your phone, saved during a public health campaign or a bad week in 2023, is the number 988: the national crisis line, built and advertised for exactly this moment. You scroll right past it and text a friend. "you up?"
A new study just confirmed that's not a you problem. It's what almost everyone does. Researchers publishing in Health Affairs Scholar produced the first nationally representative estimate of how many American adults actually hit a mental health crisis in a given year. The answer is nearly 1 in 10. And of the people who hit one, 72.6% sought help, but relatively few of them touched a formal crisis service like 988, a mobile crisis team, or the police. Most went to a doctor they already knew, or to family, or to friends. When we're breaking, we don't dial the number on the poster. We reach for someone who already feels safe.
The country spent three years and a lot of money making "988" as memorizable as 911. This study says the front line of American crisis care is actually a group chat. I think that's worth sitting with.
What the study actually found
The research team used a national survey fielded across 2024 and 2025 and asked adults directly whether they'd experienced a mental health crisis in the past year. Not a bad week. A crisis. Close to one in ten said yes.
But the number that stopped me wasn't the headline figure. It was how unevenly the crises fell. Among adults with stable housing, 8.1% reported a past-year crisis. Among adults with housing instability, it was 37.9%. People with a prior depression diagnosis: 27.2%, versus 4.2% without. Younger adults were overrepresented across the board.
Read that pattern for a second. A crisis almost never arrives as a lightning strike out of a clear sky. It's the end of a ramp: money stress, housing stress, a depression that's been circling for months. Which means the useful question isn't only "what number do you call at the bottom of the cliff." It's "what was available on the way down."
This is not a story about other people
If you're in your twenties, you're the demographic center of this data, not a bystander to it. Federal survey figures reported by AJMC put past-year mental illness among 18-to-25-year-olds at 36.2%, higher than any older age group, with 11.6% reporting a serious mental illness. A Blue Shield of California poll of people aged 14 to 25 found 94% saying they deal with mental health challenges in an average month. Ninety-four percent is not a subgroup. That's the weather.
And the help-seeking pattern the crisis study found in adults? Younger people run an even more extreme version of it. RAND researchers reported in November that about 1 in 8 adolescents and young adults now use AI chatbots for mental health advice, rising to 22% among 18-to-21-year-olds, while 40% of teens who had a major depressive episode got no mental health care at all. I've written before about what those chatbot confessions are actually reaching for, and it rhymes with this new data: when the official door feels heavy, people go through whatever door is closest and warmest.
Meanwhile, the Youth Mental Health Tracker found that roughly a quarter of people aged 10 to 24 don't feel supported by their own family. So "just talk to someone at home" isn't a given either. A lot of us have gotten very good at seeming fine to the people we live with, which shrinks the safe list even further. For plenty of twenty-somethings, the entire informal crisis infrastructure is two friends and a sibling.
The hotline is the floor, not the front line
Here's my honest read, and I want to be careful with it: this study is not evidence that 988 failed. 988 is real and staffed, and if you're in danger you should call it. A floor is not a criticism. Every building needs one.
But a floor is not where you live. What the data describes is something researchers have understood about humans forever and policy keeps forgetting: in a crisis, what people want first isn't triage. It's recognition. They want to say the unsayable thing to someone whose voice they already know, someone who won't open with an intake script. There's good evidence that feeling heard does something advice can't, and the crisis study is what that looks like at national scale: given a professional line that never sleeps and a half-asleep friend, most people pick the friend.
Here's the part I'd put on a billboard: this means the heaviest load in American crisis care is being carried by people with no training and no idea they're on duty. The friend who answers the 2am text is the actual first responder. Nobody funded her. Nobody debriefs her. She has a shift at nine.
And the ramp before the crisis? We've built almost nothing there. We have a well-advertised number for the worst fifteen minutes of your year and essentially no shared habit for the ten thousand ordinary minutes that led to it. The mental health burden among young people has been rising for decades, and our answer keeps being to reinforce the bottom of the cliff.
What you can actually do with this
Three things, none of which require becoming a different person.
First, name your people. Literally. The study says that in a crisis you'll reach for someone informal, so decide now, while you're calm, who your two are. Then do the slightly awkward thing and tell them: "If I ever text you 'you up' at 2am, it might not be casual." That one sentence turns an accidental first responder into a chosen one. And accept the reverse assignment, because statistically, you're on somebody's list too.
Second, keep 988 saved and unshamed. Not as the front line. As the floor. The friend and the hotline aren't competitors; one is for being known, the other is for being kept safe, and a bad enough night can need both.
Third, and this is the one I actually believe changes the math: build the talking habit before you need it. A crisis is often just months of unsaid things reaching critical mass. The people in that housing-instability statistic didn't fall apart because they were weak; they fell apart because the ramp got steep and there was nowhere routine to set anything down. Lowering the bar for everyday saying-things-out-loud is genuinely why we built Joice — not a crisis tool, just a place to talk through a Tuesday so that Tuesdays stop compounding. It doesn't much matter what form the habit takes. A rant on a walk. A messy page. As long as the unsaid pile stays small.
Because the study's real finding, underneath the percentages, is that when things get bad, humans reach for humans. That instinct is correct. The only fix it needs is to start reaching a little earlier — back when the text could still say "weird day, can I tell you about it" instead of "you up?"
Send that first one this week. To someone. About nothing much. That's the whole assignment.