Daniel sent us this one — he's asking what sleep maintenance insomnia actually is, how it differs from the classic can't-fall-asleep kind, and what treatments actually work for it specifically. And this is one of those topics where the standard advice most people get is exactly wrong for what's actually happening in their brains. Because most people think insomnia means lying there at midnight unable to drift off. But the more common version, and the one that's harder to treat, is waking up at three in the morning with your mind already running and no way back.
The numbers bear that out. Sleep maintenance insomnia — that's the clinical term for waking up and not being able to return to sleep — accounts for roughly sixty percent of chronic insomnia cases. Yet it receives less than twenty percent of insomnia research funding. That's from a twenty twenty-four analysis in the Journal of Clinical Sleep Medicine. So we've got this huge mismatch between what people actually experience and what the research establishment studies.
Sixty percent of cases, twenty percent of funding. That's the sleep equivalent of building flood defenses for the wrong river.
And the clinical definition matters here because it sets the bar for when this becomes a diagnosable condition. Under the DSM five TR criteria, we're talking about difficulty returning to sleep after a middle-of-the-night awakening, with at least thirty minutes of wakefulness, occurring at least three nights per week, for at least three months. So this isn't the occasional bad night. This is a chronic, patterned disruption.
Let's define the thing properly before we get into fixes. What's actually different in the brain between someone who can't fall asleep at eleven PM and someone who wakes up at three AM and can't get back?
This is where it gets interesting, because they're genuinely different conditions with different neurobiological substrates. Sleep onset insomnia involves hyperarousal at what's called the sleep-wake gate — that's in the ventrolateral preoptic nucleus, the VLPO. Think of it as the switch that flips from wake to sleep. In onset insomnia, that switch is sticky. But sleep maintenance insomnia is about what happens in the second half of the night. It's a failure of arousal regulation during the transition from NREM stage N two to N three — that's deep slow-wave sleep. The brain fails to sustain the deeper sleep stages that protect against awakening.
One is failing to enter the building, the other is getting kicked out at two in the morning and the door locks behind you.
That's actually a pretty good way to put it. And to understand why this happens, we need to talk about the two-process model of sleep regulation. This is foundational. Process S is your homeostatic sleep drive — it builds up the longer you're awake, like pressure in a tank. Process C is your circadian timing system — the internal clock that tells your body when to be alert and when to wind down. Sleep maintenance insomnia represents a failure of Process C to sustain sleep drive through what's called the circadian trough — that's the lowest point of alertness in the twenty-four hour cycle, typically around three to five AM.
That's exactly when people wake up.
And here's the mechanism: adenosine is the chemical that builds up in your brain during wakefulness and creates sleep pressure. In a normal sleeper, adenosine clears gradually through the night. But in maintenance insomnia, there's evidence that adenosine clearance accelerates prematurely. So by four or five AM, your sleep pressure has already dropped below the threshold needed to stay asleep. The tank empties too fast.
The brain is basically taking out the trash too early. What's driving that?
This brings us to the hyperarousal hypothesis, and this is where the neuroimaging data from the last few years is really compelling. PET and fMRI studies from twenty twenty-three to twenty twenty-five show that maintenance insomnia patients have elevated glucose metabolism in specific brain regions during NREM sleep compared to good sleepers. We're talking about the prefrontal cortex, the anterior cingulate, and the amygdala. These are regions associated with executive function, error monitoring, and emotional processing. Essentially, the brain remains in a low-grade "on" state that primes it for awakening. It's not fully offline.
The brain is running a background process that should have been killed hours ago. That explains the three AM anxiety spiral — you wake up and suddenly you're mentally reviewing every mistake you've made since high school.
That's the anterior cingulate doing exactly what it does during wakefulness — scanning for errors and threats. Except it should be quiet during deep sleep. And this connects directly to why standard sleep medications so often fail for maintenance insomnia, and can actually make things worse.
Let's go there, because this is where I think a lot of people get tripped up. They go to their doctor, they say they can't sleep, they get a prescription for something like Ambien, and it helps them fall asleep faster but then they're wide awake at two thirty in the morning feeling worse than before.
There's a clear pharmacological reason for this. Let's start with the half-life problem. Zolpidem — that's the generic name for Ambien — has a half-life of two and a half to three hours. So if you take it at eleven PM, it's effectively cleared from your body by about two to four AM. Right when you're entering the vulnerable window for maintenance insomnia. It wears off exactly when you need it most.
You get the falling-asleep benefit and then it abandons you at the worst possible moment.
The alternatives aren't great either. You could use a longer-acting benzodiazepine like temazepam or flurazepam, but those have half-lives of eight to twenty hours plus active metabolites. So you get next-day sedation, cognitive impairment, increased fall risk in older adults — it's a terrible trade-off. But the more fundamental problem is what all GABA-A positive allosteric modulators do to sleep architecture. That's the drug class that includes z-drugs and benzodiazepines. They all suppress slow-wave sleep. N three — that deep, restorative stage. And here's the cruel irony: N three is the very stage that protects against awakenings. So you're taking a drug that reduces the sleep stage you most need to stay asleep.
It's like taking a medicine for a headache that thins your skull.
There's a twenty twenty-four meta-analysis in Sleep Medicine Reviews that quantified this. They found that z-drugs actually increase wake-after-sleep-onset — that's WASO, the standard measure of sleep maintenance — by an average of twelve minutes in maintenance insomnia patients after four weeks of use. So not only are they not helping, they're actively making the problem worse over time.
Twelve minutes more awake time after a month. That's moving in the wrong direction. So what's the alternative pharmacologically? You mentioned orexin earlier.
The orexin system is really the key to understanding the modern approach to maintenance insomnia. Orexin — also called hypocretin — is a neuropeptide produced by neurons in the lateral hypothalamus. These neurons project throughout the brain and they are fundamentally wake-promoting. They stabilize wakefulness. In a healthy sleeper, orexin levels decline during sleep, especially during the second half of the night. But in maintenance insomnia, orexin levels fail to decline appropriately. You get this persistent wake signal that overrides the sleep drive. It's like someone's leaning on the "awake" button while you're trying to stay asleep.
The problem isn't that the sleep signal is too weak — it's that the wake signal is inappropriately strong.
That's exactly the right framing. And this is why dual orexin receptor antagonists, called DORAs, are so promising. They block both orexin one and orexin two receptors, essentially turning down that wake signal. The first DORA, daridorexant, was approved in twenty twenty-two. And then just a couple of months ago, on March seventeenth, the FDA approved suvorexant XR — that's the first extended-release DORA specifically indicated for sleep maintenance insomnia. The extended-release formulation matters because it provides coverage through that vulnerable four to five AM window.
It's timed-release to match the actual problem window. That's pharmacologically elegant.
Critically, DORAs don't suppress slow-wave sleep the way GABAergics do. They preserve sleep architecture. They're not forcing sedation — they're removing the inappropriate wake signal and letting the natural sleep processes do their job.
Let's make this concrete. You've got a case study that illustrates the difference.
Consider a forty-five-year-old woman with an eighteen-month history of waking at three thirty AM, unable to return to sleep for sixty to ninety minutes. She tried zolpidem five milligrams. She fell asleep faster, but started waking at two thirty AM instead, and with more anxiety than before. That's the z-drug wearing off and the sleep architecture disruption kicking in. She switched to daridorexant fifty milligrams. After six weeks, her wake-after-sleep-onset dropped from seventy-eight minutes to twenty-two minutes. That's a clinically meaningful difference — going from over an hour of middle-of-the-night wakefulness to just over twenty minutes.
Seventy-eight to twenty-two. That's the difference between a ruined night and a minor inconvenience.
It illustrates why mechanism matters. She didn't need more sedation. She needed her wake signal turned down during the second half of the night.
That's the neurobiology of why you're awake at three AM. But understanding the mechanism is only half the story — the real question is what actually fixes it. Let's start with the treatment that has the strongest evidence base, even though most people have never heard of it.
Cognitive Behavioral Therapy for Insomnia — CBT-I. And this is the gold standard. The American College of Physicians and the American Academy of Sleep Medicine both recommend CBT-I as first-line treatment for chronic insomnia, before medication. But here's the problem: most people can't access it. There's a massive shortage of trained CBT-I practitioners. Wait times can be six months or more.
The best treatment is also the hardest to get. That's the healthcare system in a nutshell.
Let's talk about what CBT-I actually does for maintenance insomnia specifically, because the components are tailored to the problem. Sleep restriction therapy is the heavy hitter here. You calculate your actual average sleep time using a sleep diary — say you're in bed for eight hours but only sleeping five and a half. You then compress your time in bed to match your actual sleep time, initially. So you're only allowing yourself five and a half hours in bed. This builds up homeostatic sleep pressure. Then you gradually expand the window as your sleep efficiency improves.
You're deliberately sleep-depriving yourself in the short term to rebuild the sleep drive. That sounds brutal.
It is initially difficult, but the data supports it. A twenty twenty-five randomized trial in JAMA Internal Medicine — four hundred and sixty-two participants — found that CBT-I reduced wake-after-sleep-onset by fifty-four percent in maintenance insomnia patients at twelve-week follow-up. Versus eighteen percent for sleep hygiene education alone. So sleep hygiene — you know, keep the room dark, no screens before bed — that helps a bit, but it's not the active ingredient. The active ingredient is sleep restriction plus stimulus control.
Stimulus control being the rule about getting out of bed if you're awake.
If you're lying in bed awake for more than about twenty minutes, you get up, go to another room, do something boring in dim light, and only return to bed when you feel sleepy. This breaks the association between the bed and frustrated wakefulness. Over time, the bed becomes a sleep cue again rather than an anxiety cue. And for maintenance insomnia specifically, there's a cognitive restructuring component that targets clock-watching and catastrophizing. The three AM "I'm going to be exhausted tomorrow and my whole day is ruined" spiral.
That mental loop is so familiar. You wake up, you check the clock, you do the math on how many hours you have left, and then you're doing arithmetic instead of sleeping.
That checking behavior — looking at the clock — it amplifies arousal. The cognitive restructuring in CBT-I specifically addresses the catastrophic predictions. "I'll be useless tomorrow" becomes "I've functioned on bad sleep before and I'll manage." It sounds simple but it measurably reduces the anxiety that keeps you awake.
What about digital versions? Because if the waitlist for in-person CBT-I is six months, most people are going to reach for whatever's available.
There are FDA-authorized digital CBT-I programs — Somryst and Sleepio are the two main ones. They're not as effective as in-person therapy with a trained clinician, but they're significantly better than nothing, and they're accessible immediately. A lot of sleep centers will recommend them as a bridge while you're waiting for an appointment.
You mentioned chronotherapy earlier — timed light exposure and melatonin timing. How does that fit in for maintenance specifically?
This is where the distinction from sleep onset insomnia becomes really important, because the advice is different. For sleep maintenance, you want timed bright light exposure in the early morning — thirty minutes of ten thousand lux within thirty minutes of awakening. This advances your circadian phase, which strengthens the sleep drive later in the night. It's essentially telling your clock to shift earlier so that by three or four AM, the sleep signal is more robust.
Here's where people get it wrong. For sleep onset insomnia, you might take melatonin at bedtime — though even that's debated. But for maintenance insomnia, taking melatonin at bedtime can actually worsen the problem by causing a phase delay. What you want instead is a very low dose — zero point three to zero point five milligrams — taken three to four hours before bedtime. This reinforces the circadian signal that says "sleep is coming" without the phase-shifting effects that can push your sleep window later.
The standard drugstore melatonin gummy at bedtime is potentially counterproductive for the sixty percent of insomniacs who have the maintenance type.
And most people don't know this. They grab whatever's on the shelf, take five or ten milligrams — which is way more than needed — right before bed, and then wonder why they're still waking up at three AM. The dose and timing matter enormously.
That covers the behavioral and chronotherapeutic approaches. But for listeners who've tried CBT-I and still struggle, or who need faster relief, there's a pharmacological hierarchy that makes more sense than what most doctors prescribe. Here's where the orexin story comes back in.
Let's lay out the treatment ladder for maintenance insomnia specifically. First-line pharmacological option: DORAs — daridorexant or suvorexant XR. They target the mechanism, they preserve sleep architecture, and the side effect profile is relatively favorable. The main concern is next-day somnolence at higher doses, but that's manageable with dose adjustment. Second-line: low-dose doxepin, three to six milligrams. Doxepin is a tricyclic antidepressant, but at these very low doses, it's highly selective for histamine H one receptors. It blocks the histamine-mediated wake signal without the anticholinergic effects you get at higher antidepressant doses. And critically, it doesn't suppress REM or slow-wave sleep.
Doxepin at three to six milligrams is a completely different drug than doxepin at antidepressant doses.
At antidepressant doses — seventy-five to one hundred fifty milligrams and up — you're hitting serotonin and norepinephrine reuptake and a bunch of other targets. At three to six milligrams, you're basically just a very clean antihistamine with a long half-life — fifteen to twenty hours — which provides coverage through the whole night. It's actually one of the few medications with an FDA indication specifically for sleep maintenance.
What about the elephant in the room? Because that's what everyone gets prescribed.
We've talked about what works. But there's a trap that catches a lot of people — and it's the most commonly prescribed off-label sleep medication in America. Let's talk about why trazodone is probably not your answer.
It's everywhere. Every primary care doctor reaches for it. It's cheap, it's not a controlled substance, it feels safer than Ambien.
The evidence base is remarkably weak. A twenty twenty-three Cochrane review — and Cochrane reviews are the gold standard for evidence synthesis — found insufficient evidence for trazodone in chronic insomnia. The studies that do exist are small, short-term, and of low quality. Meanwhile, the side effect burden is real. Significant next-day sedation — that's the histamine and alpha-one blockade. Orthostatic hypotension — that's the drop in blood pressure when you stand up, which increases fall risk. And there's a rare but serious risk of priapism — that's a prolonged, painful erection that can require emergency intervention.
That's quite the side effect to roll the dice on for a drug that doesn't have good evidence it works.
Yet it remains the most commonly prescribed off-label sleep medication. Part of that is habit, part of it is that it's not a controlled substance so there's less regulatory friction, and part of it is that doctors perceive it as safer than benzodiazepines. But the evidence just doesn't support its use for maintenance insomnia, and the side effect profile is worse than the DORAs or low-dose doxepin.
If someone's on trazodone for sleep maintenance and it's not working, what's the conversation they should have with their doctor?
They should ask specifically about switching to a DORA or low-dose doxepin. And they should frame it in terms of mechanism — "I understand that trazodone works primarily through histamine and serotonin receptors, but my problem is orexin-mediated wake signaling in the second half of the night. Would a DORA target my actual problem more directly?" Doctors respond better when patients demonstrate understanding rather than just asking for a different drug by name.
That's a good practical tip. What about the combination approach? CBT-I plus medication?
This is where the data gets really interesting. A twenty twenty-five head-to-head trial compared CBT-I alone, DORA monotherapy, and the combination. CBT-I alone produced the most durable improvements at twelve-month follow-up — the effects persisted and even strengthened over time. DORA monotherapy worked faster but the benefits faded after discontinuation. But the combination — DORA plus CBT-I — had the fastest WASO reduction, about two weeks versus six weeks for CBT-I alone, and the durability was comparable to CBT-I alone because the behavioral skills were in place.
The ideal protocol might be: start both, use the medication as a bridge while the behavioral changes take hold, then taper off the medication.
That's exactly the model that a lot of sleep specialists are moving toward. And let me give you a concrete example. Sixty-two-year-old man, five-year history of maintenance insomnia, failed both zolpidem and trazodone. He implemented sleep restriction — six hours time in bed initially — plus morning light therapy. His WASO dropped from sixty-five to twenty-eight minutes over eight weeks. But he added daridorexant for the first four weeks to accelerate the response and make the sleep restriction more tolerable. By week eight, he was off the medication and maintaining the gains with behavioral strategies alone.
The medication bought him a softer landing while the behavioral changes did the heavy lifting long-term.
That's the pattern we see repeatedly in the clinical literature. And it makes intuitive sense — you're fixing the underlying dysregulation while using pharmacology to reduce suffering in the short term.
What about some of the newer approaches? You mentioned cognitive shuffling earlier.
This is fascinating and it's something anyone can try tonight for free. There was a twenty twenty-four study from Oxford that looked at what's called serial diverse imagery — or the cognitive shuffle technique. When you wake up in the middle of the night, instead of trying to "clear your mind" — which is basically impossible when you're anxious — you engage your working memory with a specific type of mental task. You think of a random category — say, things that are red — and then you slowly visualize five to ten items, one by one. A fire truck. A stop sign.
The point is you're occupying the brain's limited working memory with something that's not anxiety-producing.
Working memory has limited capacity — roughly four to seven chunks of information. If you're filling it with random, emotionally neutral imagery, there's less room for the rumination and catastrophizing that drive the hyperarousal. The Oxford study found this reduced sleep onset latency after middle-of-the-night awakening by forty percent compared to distraction or counting sheep. And counting sheep, interestingly, is actually too boring — it doesn't occupy enough cognitive bandwidth to block the anxiety loops.
Counting sheep is the original bad sleep advice. It's the "have you tried turning it off and on again" of insomnia.
It really is. And the cognitive shuffle works because it's just demanding enough. You have to generate novel images, which requires some cognitive effort, but it's not so demanding that it keeps you awake. It hits the sweet spot.
If someone's listening and they've been waking up at three AM for months, what's the actual first step? Not the ideal-world protocol where they have instant access to a sleep specialist and a CBT-I program — the thing they can do starting tonight.
Step one is a sleep diary for two weeks. Pen and paper or an app — it doesn't matter. You're tracking: what time you went to bed, roughly when you fell asleep, when you woke up during the night, how long you were awake, what time you got up in the morning. From that, you calculate your sleep efficiency — total sleep time divided by time in bed. If it's below eighty-five percent, sleep restriction is likely your most powerful tool. And you can start a modified version on your own: if you're averaging five and a half hours of actual sleep, set your time in bed to six hours. Pick a consistent wake time and stick to it seven days a week, no matter what. Count backward from that wake time to determine your bedtime.
If I need to be up at seven, and I'm allowing six hours in bed, I go to bed at one AM. Even if I'm tired earlier.
And that's the hard part. You will be tired. But you're building sleep pressure. After a week or two, your sleep efficiency should start climbing above eighty-five or ninety percent, at which point you gradually expand the window — add fifteen minutes per week. It's not pleasant, but it works.
If you're going to consider medication, ask specifically about DORAs or low-dose doxepin. Not zolpidem, not trazodone. Say "I have sleep maintenance insomnia — I wake up and can't get back to sleep — and I understand that these medications target that specific mechanism.And if your doctor hasn't heard of suvorexant XR, which was only approved in March, that's not surprising — it takes time for new drugs to enter common prescribing practice. But you can ask.
There's also the cognitive shuffle technique tonight — that costs nothing and has no side effects.
And combine it with stimulus control: if you're awake for twenty minutes, get out of bed. Go sit in a chair in another room. No phone scrolling — that's crucial, the blue light and the content both amplify arousal. Read something boring in print. Do the cognitive shuffle. Only return to bed when you feel sleepy. The first few nights, you might be up for an hour. That's okay. You're retraining the association.
What about the new stuff in the pipeline? You mentioned an orexin receptor agonist trial.
This is a fascinating development that actually reinforces how delicate the orexin system is. In April, a phase two trial of the first orexin receptor agonist — that's the opposite of a DORA, it activates rather than blocks orexin receptors — showed promise for narcolepsy. The drug, code-named OXA one oh one, is an OX two R selective agonist. It promotes wakefulness in people who have excessive daytime sleepiness. But here's the twist: in healthy volunteers, it paradoxically worsened maintenance insomnia. It increased middle-of-the-night awakenings.
Turning up the wake signal makes you wake up. That's not surprising.
No, but it's informative. It confirms that the orexin system is the wakefulness rheostat. Too little signal — that's narcolepsy. Too much — that's maintenance insomnia. And we now have drugs that can modulate it in both directions, which is a remarkable pharmacological achievement.
You mentioned something about a melatonin receptor drug in the pipeline?
There's a melatonin receptor MT two selective agonist called neu P eleven in phase three trials. The interesting thing here is that existing melatonin drugs hit both MT one and MT two receptors non-selectively. MT one is involved in the acute sleep-promoting effect, but MT two is involved in circadian phase shifting. By selectively targeting MT two, neu P eleven aims to improve sleep maintenance without the phase-shifting effects that can make melatonin counterproductive for maintenance insomnia. The twenty twenty-six to twenty twenty-seven pipeline is going to be interesting to watch.
We're moving from the era of blunt sedation to the era of targeted circadian and arousal modulation. That's a genuine paradigm shift.
It really is. And it's worth stepping back and appreciating how far sleep medicine has come. For decades, the approach was essentially chemical cosh — knock the person unconscious and hope for the best. Now we understand the specific neural circuits, the receptor subtypes, the timing dynamics. We're treating the actual pathophysiology rather than just papering over the symptom.
The open question, I think, is whether we'll eventually move beyond daily pills entirely. If the orexin system is druggable from both directions, and we're developing more selective circadian modulators, do we eventually get to gene therapies or long-acting implants that reset the circadian system at its root?
There are labs working on exactly that. The idea would be to deliver a gene construct that upregulates or downregulates orexin receptor expression in specific brain regions. You'd essentially recalibrate the set point of the arousal system. But we're years away from that being clinically viable, and the safety questions are enormous. You don't want to permanently alter someone's arousal system and then discover a problem five years later.
It's not like adjusting the thermostat — there's no undo button on gene therapy.
But the fact that we're even having that conversation is remarkable. Twenty years ago, sleep maintenance insomnia was barely recognized as a distinct condition. Now we have targeted drugs, structured behavioral protocols, and a mechanistic understanding that spans from adenosine kinetics to orexin receptor pharmacology.
Let's hit the key takeaways for someone listening who's been dealing with this. What should they actually do?
First, if you wake up at three or four AM and can't get back to sleep, the first step is not a sleeping pill. It's a two-week sleep diary to calculate your sleep efficiency. If you're below eighty-five percent, sleep restriction is your most powerful tool. It's free, it's evidence-based, and it addresses the root cause of low homeostatic sleep drive.
Second, if you're considering medication, ask your prescriber specifically about DORAs or low-dose doxepin — not zolpidem, not trazodone. The mechanism matters more than the brand name or what's commonly prescribed.
Third, the cognitive shuffle technique is free, evidence-based, and something you can try tonight. When you wake up, pick a random category and slowly visualize items. It occupies working memory without triggering anxiety loops. And combine it with stimulus control — get out of bed after twenty minutes awake.
Fourth, if you can access CBT-I — whether in person or through a digital program like Somryst or Sleepio — it's the treatment with the strongest long-term evidence. Medication can be a bridge, but the behavioral changes are what stick.
The bottom line is that sleep maintenance insomnia is not just "insomnia lite" or a variant of the falling-asleep problem. It's a distinct condition with distinct mechanisms, and treating it effectively requires understanding what's actually happening in your brain at four AM. Once you know that, the treatment choices become much clearer.
Now: Hilbert's daily fun fact.
Hilbert: In the nineteen twenties, botanists on the island of Bioko in Equatorial Guinea documented a carnivorous plant called the Bioko sundew, which used sticky tentacle-like leaves to trap insects in nitrogen-poor volcanic soil. It was presumed extinct after a series of landslides in the nineteen forties buried its only known habitat, but a single population was rediscovered in twenty nineteen growing on a remote cliff face accessible only by rope.
A cliff face accessible only by rope. Of course it was.
The plant that fakes its own death and then hides somewhere you need climbing gear to reach. I respect the commitment.
This has been My Weird Prompts. Thanks to our producer Hilbert Flumingtop. If this episode was useful, share it with someone who wakes up at three AM — they probably think they're alone in it. Find us at myweirdprompts dot com.
