It’s 2 AM. You have to be up in five hours. Your body is dead, you can barely keep your eyes open, and your brain has decided right now is the perfect time to replay every awkward thing you’ve ever said, figure out how short you’ll be on rent, rehearse a conversation with your boss you may never have, and rotate through worst-case scenarios about people currently asleep and fine. You check the clock, do the math on how much sleep you can still get, frustration kicks in, frustration makes you more awake, and the loop just runs until the alarm goes off.
If you’ve tried the chamomile tea, the blue-light glasses, the melatonin gummies, and the bedtime meditation app, and none of it worked, you’re not bad at sleep hygiene. What’s actually happening is that your nervous system has forgotten how to downshift, and there’s a real clinical reason for that, which the wellness industry has not done a good job of explaining because explaining it doesn’t sell anything.
The hyperarousal problem
Insomnia in psychiatry isn’t really a sleep disorder, it’s a wakefulness disorder. Your sleep drive is probably fine. What’s broken is the part that’s supposed to let go of the day. We call this hyperarousal, and it shows up in the body before it shows up in your head. Elevated resting heart rate. Higher core body temperature at night. Cortisol (your stress hormone, the one that’s supposed to spike in the morning and crash by bedtime) still rolling at 11 PM when it should have crashed two hours ago. Your sympathetic nervous system (the gas pedal half of your nervous system, the one that runs fight-or-flight) is working a shift it was never supposed to.
Insomnia in psychiatry isn’t really a sleep disorder, it’s a wakefulness disorder.
The other piece is what neuroscientists call the default mode network, which is a fancy way of saying the part of your brain that lights up when you’re not doing anything specific… daydreaming, autobiographical replay, future-planning, social rumination. It’s supposed to quiet down as you fall asleep. In people with racing-thoughts insomnia, it doesn’t. It stays online, churning through your day, your relationships, the email you didn’t answer, the thing your kid said at dinner. You’re not thinking on purpose. The thinking is happening at you, which is honestly the more accurate way to describe it.
You’re not thinking on purpose. The thinking is happening at you, which is honestly the more accurate way to describe it.
This pattern shows up across anxiety, depression, PTSD, untreated ADHD, perimenopause, alcohol withdrawal, and the version most patients have, which is chronic overstimulation from modern life. Phone at dinner, email after 9 PM, six tabs open in your head at all times. The brain doesn’t get a runway to slow down, then you turn the lights off and expect it to land like a jet without flaps.
The stimulant problem nobody talks about
Caffeine has a half-life of around five to six hours. The coffee you drank at 2 PM is still in your bloodstream at 8 PM at half-strength and at midnight at quarter-strength. People tell me they’re fine, they can fall asleep no problem after coffee, and I believe them, the issue isn’t usually sleep onset. Caffeine fragments deep sleep architecture even when you don’t notice it, so you fall asleep fine and wake up at 4 AM with your brain switching back on like a light.
The kind of guy who shows up convinced he has treatment-resistant insomnia, on three different sleep meds, is almost always also drinking six cups of coffee a day with the last one around 4 PM before a workout, taking 30mg of Vyvanse for ADHD, and vaping nicotine because he quit smoking but the receptors are still there asking. We don’t change his sleep meds first, we pull the late coffees, move the Vyvanse earlier in the day, and taper the nicotine, and three weeks later he’s sleeping seven hours and we’re talking about coming off the trazodone he never needed in the first place.
Stimulant withdrawal at night is one of the most underdiagnosed drivers of insomnia anywhere. Caffeine, nicotine, prescription stimulants taken too late, even the rebound dip off an afternoon energy drink your buddy at the gym told you was clean. The brain’s been propped up all day, the chemicals leave, and it has nowhere to land except wired and tired, which is a real phenomenon and not just a phrase.
If you’ve been lying in bed trying to fall asleep for an hour every night for three months, your bed has started meaning “the place where you fail to sleep,” which is a problem you don’t fix by buying a better pillow.
Why CBT-I is the actual gold standard
CBT-I (cognitive behavioral therapy for insomnia, the structured sleep-restriction-and-stimulus-control kind, six to eight weeks of work that almost feels like punishment for the first two weeks before it starts paying off) is what every sleep-medicine guideline currently lists as first-line treatment, not Ambien, not melatonin, the structured behavioral protocol. It outperforms sleep medications at six and twelve months in head-to-head trials, and the effect holds after treatment stops, which medications don’t do.
The version that works has four real components and almost nobody does all of them on their own. Sleep restriction, where you spend less time in bed than you want to so the time you spend in bed gets denser. Stimulus control, where the bed is for sleep and sex and nothing else, no reading, no phone, and if you can’t sleep in 20 minutes you get up. Cognitive work on the catastrophic thoughts about not sleeping, because the fear of not sleeping is what’s keeping you not sleeping. Relaxation training to give your parasympathetic nervous system (the brake half of your nervous system, the one that’s supposed to take over at night) a runway.
Six to eight weeks of that, done seriously, fixes most cases of chronic insomnia. The catch is that the first two weeks feel worse. Sleep restriction is genuinely brutal at the start, people bail, and doing it with a therapist or a structured app like Somryst makes the difference between finishing and giving up at day nine. If we’re being honest, this is the protocol that works and nobody wants to do it, which is roughly the story of every effective intervention in psychiatry.
CBT-I
Six to eight weeks of structured work. Outperforms sleep meds at one year. Effect holds after you stop. The version with sleep restriction is the version that works.
Trazodone 25-100mg
Sedating antidepressant used off-label for sleep. Not addictive. Doesn’t fragment architecture. Mild morning grogginess in some people. Good while you’re sorting the underlying cause.
Trazodone, Ambien, and the rest
Trazodone is what most psychiatrists prescribe first. Ambien is what most primary care doctors prescribe first. The psychiatrists are right on this one. Trazodone at 25 to 100mg is a sedating antidepressant used off-label for sleep, it’s not addictive, it doesn’t build tolerance the way Ambien does, and it preserves the sleep architecture (the actual shape of your sleep cycles, deep sleep and REM and so on). Downsides are some morning grogginess and occasionally vivid dreams, which patients tend to either tolerate or hate.
Ambien works, that’s the problem. It works so well the first night people fall in love with it, three months later they can’t sleep without it, and the dose that used to do the job doesn’t anymore. The data shows tolerance building within weeks and rebound insomnia when you stop, plus the parasomnia stuff (sleepwalking, sleep-eating, sleep-driving) that looks hilarious in news stories and is not hilarious when it’s your patient and the cereal box is on the kitchen counter and they don’t remember putting it there.
Hydroxyzine, an antihistamine, is what I reach for when insomnia is clearly anxiety-driven and the patient wants something as-needed without dependency risk. Mirtazapine at 7.5 or 15mg is what I use when depression and insomnia are tangled together. Melatonin is useful for circadian shifts (jet lag, shift work) and useless for the kind of insomnia where your brain won’t shut up. Doxepin at 3 to 6mg has good data for middle-of-the-night awakenings and gets underused because nobody marketed it.
What this means if you’ve been not sleeping for a year
If you’ve been running on five hours for months, a lot of what feels like your personality right now is sleep deprivation. The short fuse, the forgetfulness, every small problem feeling like a crisis. Chronic insomnia wrecks the part of your nervous system that’s supposed to keep you from blowing up at people, tanks your immune function and your blood sugar control, and makes every psychiatric condition we treat worse. Depression is harder to budge in a sleep-deprived patient, anxiety is louder, ADHD looks worse than it actually is. People come in for what looks like a mood disorder and what they actually have is two years of broken sleep wearing a mood-disorder costume.
Order of operations. Pull the obvious culprits first, caffeine after noon, alcohol in the evening (which sedates you for two hours and then fragments the back half of the night), phone in bed. Then figure out what’s actually driving the hyperarousal, because something is. Untreated anxiety, untreated ADHD, sleep apnea (which a surprising number of insomnia patients actually have without knowing it), perimenopausal hormone shifts that don’t get screened often enough. Once you know the driver, treat the driver. Use trazodone as a bridge while you do the work.
The thing about meds in general for sleep, and this applies to anything you walk in asking for. If you want a sleep medication and we’ve talked about what it does and doesn’t do, you get one. I’m a provider, not a parent. My job is the honest take on what’s likely to work and what the trade-offs are, your job is the choice. What I’ll push back on is the autopilot script… where somebody’s been on Ambien for three years and nobody’s revisited it, where the original prescription was for a bad week and turned into a habit nobody named. That’s a different conversation, and it’s the one that costs people the most time.
What’s actually nice to hear, if you’ve been in this loop for a while: most chronic insomnia is fixable. People will spend a decade not sleeping, treating the daytime symptoms one at a time, going from psychiatrist to therapist to functional medicine clinic to supplement stack, never fixing the sleep itself. Fix the sleep and a surprising amount of what felt like separate problems walks itself back to something manageable, sometimes within a couple of months. The work isn’t dramatic. It’s just specific.
Sources
- Sateia MJ, Buysse DJ, Krystal AD, et al. Clinical Practice Guideline for the Pharmacologic Treatment of Chronic Insomnia in Adults: An American Academy of Sleep Medicine Clinical Practice Guideline. J Clin Sleep Med. 2017;13(2):307-349. PMID 27998379.
- Edinger JD, Arnedt JT, Bertisch SM, et al. Behavioral and psychological treatments for chronic insomnia disorder in adults: an American Academy of Sleep Medicine clinical practice guideline. J Clin Sleep Med. 2021;17(2):255-262. PMID 33164742. (CBT-I as first-line)
- Walker MP. Sleep essentialism. Brain. 2021;144(3):697-699. PMID 33787879. (Sleep necessity and cognition)
Six to eight weeks of structured work. Outperforms sleep meds at one year. Effect holds after you stop. The version with sleep restriction is the version that works.
Sedating antidepressant used off-label for sleep. Not addictive. Doesn't fragment architecture. Mild morning grogginess in some people. Good while you're sorting the underlying cause.
Ambien, Lunesta, Restoril. They work fast and they build tolerance fast. Dependence is real, even at prescribed doses. Useful in short courses. Bad as a long-term plan.
| Common comorbidity | Clinical note | |
|---|---|---|
| Anxiety disorders | Hyperarousal and default mode network over-activation are shared mechanisms | Treating anxiety often improves sleep; CBT-I and anxiety CBT have overlapping components |
| Major depression | Depression breaks sleep architecture; broken sleep deepens depression; bidirectional loop | Can't fix depression on top of unfixed sleep; fix sleep first and reassess mood diagnosis |
| ADHD | Stimulants taken too late fragment sleep; ADHD itself disrupts circadian anchoring | Move Vyvanse or Adderall earlier in the day before changing sleep meds |
| PTSD | Trauma nightmares and hyperarousal produce classic racing-thoughts insomnia picture | Prazosin 1-10mg at bedtime for nightmares; treating PTSD often improves sleep more than sleep meds |
| Alcohol use | Alcohol sedates for 2 hours then fragments back half of night; REM rebound wakefulness | Stop alcohol for 2 weeks; baseline insomnia often substantially improves |
| Sleep apnea | Presents as insomnia (middle-of-night awakenings) or hypersomnia; frequently missed | Screen before adding sleep meds; CPAP can resolve what looked like psychiatric insomnia |