Insomnia and the Nervous System

The event has gone and the nights have not

What starts insomnia is almost never what maintains it. By the time most people ask for help, the precipitating event is history, and what remains is a nervous system reorganized around a problem that no longer exists.

Something happened. A bereavement, a hospital stay, a newborn, a job that went wrong, a diagnosis, a court date. Sleep broke, which is what sleep does under load, and at the time nobody thought much of it.

Then the thing resolved. The baby sleeps through. The contract was signed two years ago. And the nights did not come back.

That gap between cause and condition is the most useful single fact available to anyone who has been sleeping badly for years, because it names the right target: the factors present now.

The reorganization is measurable and it is durable. In Quebec, the sleep researcher Charles Morin put the same sleep questions to 3,073 adults once a year for five years. He was counting incidence, persistence and remission, meaning how many people acquire the condition, how many keep it, and how many get free of it.

Among people who already had an insomnia syndrome, 37.5 percent still had it at every annual check. Among those who improved, relapse was more likely than continued improvement. Among the good sleepers, 13.9 percent developed a syndrome during the five years.

The sleep page holds the physiology of ordinary sleep: what happens when you fall asleep, the two forces that decide when, the switch that throws the state, and what the night is for.

An episode that mends is ordinary. A condition that installs itself is a different kind of thing. It behaves like a state the system has settled into rather than an injury waiting to heal, and that difference decides what can move it.

The sleep patterns that need a physician first

Most insomnia is not dangerous. Sleep apnea, sudden severe insomnia beside new symptoms, and thoughts of self-harm are the exceptions, and each outranks everything else about sleeplessness.

Loud snoring, breathing that stops and restarts in the night, gasping or choking awake, or heavy daytime sleepiness are the pattern of obstructive sleep apnea. That needs a physician and a sleep study. Sleepiness is the word carrying the weight there. Insomnia usually produces exhaustion without the ability to fall asleep. Struggling to stay awake at a red light is a different problem, and it is dangerous on the road. The sleep apnea page covers it in full.

Insomnia that arrives suddenly and severely alongside new symptoms needs assessment before any schedule work. Breathlessness lying flat, chest pain, drenching night sweats, unexplained weight loss, a racing pulse, new confusion, severe pain, or a sharp change in mood are all reasons to be seen. Broken sleep sits downstream of a long list of medical conditions, and downstream of a long list of medications.

Which sets a boundary the model keeps. Not every disorder of sleep is a disorder of regulation. Thyroid disease, heart failure, pain, restless legs, an untreated airway, a prostate, a drug, and a shifted body clock all break nights by direct mechanisms, and each is treated on its own terms. The model here does not replace that assessment.

Its claim runs alongside it. Every one of those diseases also has a tonal expression, and many broken nights are started, held in place or amplified by a failure of regulation and not by the lesion alone. That is why the same airway, the same thyroid result and the same tablet cost one person a wrecked year and another a bad fortnight.

Nothing on this page is advice about any medicine. Do not start, stop, or change a prescribed sleep medication because of anything written here. Stopping some of them abruptly is hazardous in its own right, and the decision belongs to you and the person who prescribed them.

Insomnia and low mood travel together, and the traffic runs in both directions. If you are thinking about harming yourself, contact emergency services or the 988 Suicide and Crisis Lifeline.

Predisposing, precipitating, perpetuating, and why only one is still in the room

Arthur Spielman split the causes of insomnia into three kinds in the 1980s, and the split still runs treatment forty years later. It points the clinician at the factors present now rather than the event that is over.

In the mid-1980s a clinical psychologist named Arthur Spielman put a question to the field that sounds obvious once someone has said it out loud. The divorce is two years finished. So why is this person still awake at three?

His answer split the causes of insomnia into three kinds that behave nothing alike, and it still organizes the treatment of the condition forty years later. He set it out in a behavioral perspective on insomnia treatment.

Three kinds of cause that behave nothing alike

Predisposing factors are what a person carried in before anything happened. A tendency to run alert. A temperament that keeps working on a problem after the lights go out. A family history. Predisposition does not cause insomnia. It sets how much of an event it takes.

Precipitating factors are the event itself. Bereavement, illness, pain, shift work, a newborn, a deadline, a move. These are not rare. Almost everyone meets several, sleeps badly for a while, and mends.

Perpetuating factors are what happens next. They are the coping behaviors the person adopts, and the associations the nervous system forms while they are adopted. This is the category that decides the outcome, for one blunt reason. It is the only one of the three still present when the person finally asks for help.

The field's shorthand for this is the three-P model. Spielman's paper argues explicitly for the perpetuating half and for conditioning as the mechanism, and the tidy three-word list is what the line of work produced afterward. The model earns the attribution. The slogan came later.

The unstudied middle: acute insomnia

Now the hole in it, stated straight. In 2012 Jason Ellis, Philip Gehrman, Colin Espie, Dieter Riemann and Michael Perlis reviewed what is known about acute insomnia, meaning the first days and weeks. Chronic insomnia has been studied exhaustively. The ordinary bad fortnight after a bereavement has barely been studied at all. Their conclusion was that the field lacked even a working definition of the acute form, and so lacked the data to say who converts and why.

Every account of how an event becomes a condition is inference across a middle nobody has recorded.

A trait you can score before anything has gone wrong

Sleep reactivity, the tendency to lose sleep under stress, is a measurable property of the receiving system. It can be scored before any insomnia exists, and the score forecasts who converts, who stays well, and who becomes depressed two years on.

Predisposition is a hand-wave until somebody measures it. In Detroit, the sleep researcher Christopher Drake wanted to know whether some people are simply built to lose sleep under pressure, and whether that could be scored in advance.

Working with the sleep researcher Thomas Roth and others, he wrote 27 questions about how a person sleeps in stressful situations and reduced them to nine. They called the result the Ford Insomnia Response to Stress Test. Then they tested it against a stressor. A first night in a sleep laboratory is a strange bed with wires on, which is mild but real. They put 104 people through one, in their study of vulnerability to stress-related sleep disturbance and hyperarousal.

High scorers slept worse and took longer to drop off. The second half of the result is the one that matters. High scorers placed in a dark room during the day and invited to nap were still harder to put to sleep than low scorers. That is the signature of arousal rather than of sleep debt. The differences survived statistical control for current and past insomnia, so the trait was not a residue of having been an insomniac.

The score forecasts conversion a year out

Ten years later the same group turned the score into a forecast. In a prospective test of the stress-diathesis model, 2,316 adults with no history of insomnia and none of depression were scored first and followed after. Over the next year, high sleep reactivity raised the odds of developing insomnia by about 1.78 times.

It did so by amplifying the effect of stress-driven mental intrusion rather than by acting separately from it. Two years out it predicted depression as well, and the insomnia that had appeared in between accounted for that link.

A property of the receiving system, recorded before the event arrives, that changes what the same event does to the person it arrives at.

The genetics point at arousal, not sleep machinery

Two more layers sit under that. An international consortium of geneticists pooled data on 1,331,010 people, with Philip Jansen of Amsterdam as first author. They asked which stretches of DNA differ between people who sleep badly and people who do not. Their genome-wide analysis of insomnia found 202 regions of the genome, implicating 956 genes. The brain cells those genes point at sit in the striatum, in the hypothalamus, and in a thin sheet of cortex called the claustrum.

The hypothalamus is the small structure at the base of the brain that runs appetite, temperature, hormones and sleep. The other two belong to the machinery of arousal and of deciding what matters. None of the three is sleep machinery. All 202 regions together account for 2.6 percent of the differences between people on this trait, which is a very small share.

The predisposition is real and it is very diffuse.

Five subtypes, separated by mood and personality

At the level of the whole person, the Dutch sleep researcher Tessa Blanken refused to treat insomnia as one condition. Working in Amsterdam with a large volunteer registry, she ran 34 questionnaires through a method that hunts for natural groupings without being told what to look for.

It produced five insomnia subtypes, separated by mood and personality and not by anything about sleep itself. The classification replicated in 251 new participants and held when 215 were reassessed nearly five years later. The subtypes differed in course, in treatment response, and in depression risk by up to fivefold.

Notice where the differences live in all three results. Not in the sleep. In the system the day arrives at.

Every behavior that maintains insomnia is a reasonable response to it

The cruelty at the center of chronic insomnia is that its causes are defenses. Going to bed earlier, lying in, napping, and guarding the schedule all dilute sleep pressure and teach the bed to mean vigilance.

You slept four hours, so you go to bed an hour earlier to give yourself a better chance. You slept badly, so you lie in. You are hollowed out by four in the afternoon, so you nap. You guard the schedule, cancel the evening, decline the invitation. Every one of those decisions is defensible on its own. Together they do two specific things to the machinery.

The first is arithmetic. Sleep pressure builds across the hours you are awake and is spent by sleeping, and the sleep page covers how it is built and how it is discharged. What matters here is that the supply is finite. Spread the same pressure across ten hours in bed instead of seven and it is thinner in every one of them. The night becomes shallower and more broken, which is exactly the complaint.

The second is learning, and it is the more important of the two. A nervous system is a prediction machine before it is anything else. It learns what a place means from what reliably happens there.

Spend two hundred nights lying in a bed rehearsing tomorrow, checking a clock and failing, and the bed becomes a dependable cue for vigilance. That learning is not an opinion and cannot be talked out of anybody. It is the same machinery that makes a kitchen smell produce hunger before you have decided to be hungry.

Effort is the sharpest instance. Sleep belongs to a small class of functions that cannot be performed on purpose, alongside blushing and forgetting a tune. Trying recruits attention, monitoring and arousal. Arousal is the precise opposite of the state being aimed at. So the harder the person works at it, the further away it goes, and the working is entirely rational.

What results is a loop that no longer needs the event that started it. Poor sleep drives compensation, the compensation dilutes pressure and teaches the bed to mean wakefulness, the nights get worse, and the compensation intensifies. This model calls that a stable state rather than an ongoing injury. Nothing is being damaged. Something is being maintained.

You are not sleeping, and the instruction is to spend less time in bed

Sleep restriction therapy cuts the window for sleep down to roughly the amount actually being slept, then returns time only as the night consolidates. It is the strongest single component of insomnia treatment, and it works by concentrating pressure and reteaching the bed.

If the perpetuators are the target, the treatment follows.

In 1987 Spielman and two colleagues published treatment of chronic insomnia by restriction of time in bed. Thirty-five people, average age 46, who had slept badly for an average of 15 years. The window available for sleep was cut down hard, to roughly the amount they were actually sleeping.

Then it was handed back in small increments, only as the proportion of time in bed spent asleep improved. Total sleep time went up. Time to fall asleep dropped, and so did time awake in the night. It was still holding about 36 weeks later in the 23 people reassessed.

The study was small, uncontrolled, single-armed and self-reported, and the authors said plainly that some patients find the schedule hard to keep. Take it as the origin rather than the proof.

Why restricting the window works

Two things are happening at once, and the first explains why the procedure looks backwards. The body holds wake and sleep in a balance it defends actively, and the amount of sleep a person gets is one of the values under defense. Restriction does not argue with that defense by handing it more opportunity.

It changes the terms of it. Pressure is concentrated instead of diluted, so the same night has to be assembled inside a shorter window. The second thing is that the bed is being taught a new meaning, because the person is now only in it when sleep is imminent.

Stimulus control makes the bed mean sleep

The companion procedure does that second job directly. Stimulus control has one aim, which is to make the bed a cue for sleep and nothing else. The instructions told patients to go to bed only when sleepy and to leave the bed when awake and frustrated. They reserved the bed for sleep and sex, and they fixed the rising hour whatever the night had been.

Both procedures belong with a clinician who can set a schedule and adjust it. The founding description was a 1972 conference proceeding by the psychologist Richard Bootzin that never entered the indexed literature, which is why it is absent from the reference list below. What carries the component is the evidence that came after it, and the next two results name it directly.

Restriction and stimulus control lead the rankings

The first ranked the components. Charles Morin and two colleagues pooled 59 treatment studies covering 2,102 patients in a meta-analysis of nonpharmacological interventions. Average therapy time was five hours. The effect on time to fall asleep was 0.88 and on time awake after falling asleep 0.65, putting the treated patient ahead of about 81 and 74 percent of untreated controls.

Stimulus control and sleep restriction came out as the strongest single procedures. Sleep hygiene education, used alone, did not work. The sleep medicine guideline in the next section names those same two procedures as the recommended single-component therapies.

For thirty years everyone assumed that cutting time in bed was the active ingredient, and nobody had tested it. Leonie Maurer, working in the Oxford sleep and circadian group, ran the dismantling trial. She randomized 56 people either to true sleep restriction or to a control that regularized time in bed without reducing it, so the schedule was fixed in both arms and only the amount differed.

Restriction won by about 4.5 points on the insomnia severity index at four weeks and held the gain at twelve. That index runs from 0 to 28, with higher scores meaning worse insomnia. Restriction also won on objective sleep continuity and on quality of life.

The daytime cost, measured

The cost is real and it has been measured. Simon Kyle, then a sleep researcher in Manchester, wanted to know what the schedule does to the daytime while it is working. He followed 16 people through four weeks of restriction with laboratory recording and a reaction-time task, and reported reduced objective sleep, increased daytime somnolence and impaired vigilance.

Recorded total sleep fell by 91 minutes on the first night. Reaction times slowed and lapses of attention rose. Everything returned to baseline by three months. Anyone who drives needs that fact.

The model reads the whole sequence as hysteresis. Leaving the state costs more than entering it did, and the way out does not retrace the way in.

The first-line treatment barely changes how much you sleep

Cognitive behavioral therapy for insomnia, CBT-I, is the first-line recommendation of both American guideline bodies. Across 20 randomized trials it cut time to fall asleep by 19 minutes and lifted sleep efficiency by 9.9 points.

The package puts restriction, stimulus control, cognitive work on the beliefs and the effort, and some education and relaxation into one treatment.

Sleep hygiene alone is not a treatment

Sleep hygiene is the advice everybody gets before any of that. The dark room, the cool room, the fixed bedtime, no late caffeine. Ka-Fai Chung, a psychiatrist at the University of Hong Kong, wanted to know what that advice achieves on its own, and pooled 15 studies of sleep hygiene education as a treatment of insomnia.

It produced small to medium improvements, was consistently and substantially beaten by full CBT-I, and gained about 5 percent in diary sleep efficiency, with only the subjective measures moving at all. The sharpest finding is a hole. Not one of the 15 studies compared it against a sham or against no treatment, so its standalone effect is unknown rather than merely small.

What the guidelines actually grade

The guidelines are blunt about the ranking. The American College of Physicians, through a panel led by the physician Amir Qaseem, reviewed the trials and recommended CBT-I as the initial treatment for all adults with chronic insomnia. That is a strong recommendation on moderate-quality evidence.

The same guideline treats medication as a shared decision, graded weak on low-quality evidence. The American Academy of Sleep Medicine graded the pieces separately in its guideline on behavioral and psychological treatments. Only the full multicomponent package earned a strong recommendation. Stimulus control alone and sleep restriction alone earned conditional ones.

Ten points of efficiency, 7.6 minutes of sleep

Now the number that reorganizes the condition. James Trauer, a clinical epidemiologist in Melbourne, set out to quantify what the therapy actually changes on the clock. He pooled 20 randomized trials covering 1,162 people in a systematic review and meta-analysis of CBT for chronic insomnia. Time to fall asleep improved by about 19 minutes. Time awake in the night improved by about 26 minutes.

Sleep efficiency, meaning the share of time in bed spent asleep, rose by about 9.9 percentage points. Total sleep time gained about 7.6 minutes, with a confidence interval that crossed zero. A confidence interval is the range the true answer most likely sits in, so crossing zero means the range includes no gain at all.

The sleep page pairs those numbers as a paradox. The model reads them as a defended value: the amount of sleep is held where it is held, and the treatment that fixes the condition barely shifts it. Ten points of efficiency and no extra minutes is an organization rebuilt around a value staying put.

A later analysis pushed harder on the same question. Wai Sze Chan, a psychologist at the University of Hong Kong, asked it directly. Does the therapy give people more sleep? She pooled 43 randomized trials, in a meta-analysis of whether the therapy improves sleep duration.

On sleep diaries it added roughly 30 minutes. Overnight recording in adults agreed with the diaries in direction. Wrist actigraphy, which infers sleep from how still the wrist is, subtracted roughly 30 minutes. Older participants gained least, and there was publication bias favoring positive diary results at follow-up.

Two instruments say the treatment added half an hour of sleep. The third says it took half an hour away. The person recovers either way.

That disagreement is a finding rather than a failure. Diaries record the experience. Actigraphy reads stillness, and a patient told to leave the bed when awake is producing less stillness on purpose. A treatment that deliberately compresses time in bed should cut recorded sleep while it is doing so. What none of the three instruments disputes is the recovery.

Durability, and the head-to-head with zopiclone

Durability fades on a measured curve. Tanja van der Zweerde and Annemieke van Straten, clinical psychologists in Amsterdam who study how long psychological treatments hold, pooled 30 trials of long-term effects in controlled studies. Against inactive controls the effect on insomnia severity was 0.64 at three months, 0.40 at six, and 0.25 at twelve. Onset latency and efficiency held better, around 0.3 to 0.5 at a year. Real, fading, and measured against non-active comparators, which inflates all of it.

The cleanest head-to-head is Norwegian. Borge Sivertsen, a clinical psychologist in Bergen, randomized 46 older adults three ways: to the therapy, to nightly zopiclone, or to a placebo tablet. The two medication arms carried a genuine double blind, and polysomnography was run at every assessment point.

The therapy beat zopiclone on three of four outcomes at six weeks and at six months. For most outcomes zopiclone did not differ from placebo. Sleep efficiency in the therapy group went from 81.4 to 90.1 percent while the zopiclone group drifted from 82.3 to 81.9. The therapy group also spent markedly more time in deep slow-wave sleep.

Forty-six people across three arms is a small trial, and that last detail is worth the caution. More deep sleep after a conversation about a schedule, and none after a sedative, is a statement about organization rather than about sedation.

First-line care that most people cannot get

The recommended treatment for chronic insomnia mostly cannot be obtained. The world supply of specialist providers has been counted, and the two answers to the shortage, digital delivery and nurse delivery, have both been through randomized trials.

A recommendation is only as good as the supply, and somebody eventually counted the clinicians. Arthur Thomas and Michael Grandner, working at the University of Pennsylvania, wanted to know whether the recommended treatment could physically be obtained.

They assembled a geographic assessment of behavioral sleep medicine providers and found 752 in the world. Eighty-eight percent were in the United States. Nineteen percent of those were in two states. Four states had none. Of 167 American cities with more than 150,000 people, 105 had no provider of any kind.

It is a directory count, so it misses clinicians who deliver the treatment without registering anywhere. Even generously adjusted, 752 people is the entire supply standing against a guideline that recommends this treatment first for every adult with chronic insomnia. Two lines of work have gone at that.

Digital CBT-I in 1,711 people

The first moved the therapy onto a screen. Colin Espie, an Oxford sleep researcher who has spent his career on how the treatment reaches people, randomized 1,711 people with insomnia symptoms to digital CBT-I or to sleep hygiene education. He made the primary outcomes daytime ones, because the day is what people actually complain about. His trial of effects on health, psychological well-being and sleep-related quality of life found all three improved, with a large gain on quality of life.

Then the trial asked how the daytime gain arrived. There is a statistical test for that, called mediation analysis, and it asks whether a treatment reached a second outcome by way of the first or by some other route entirely. Between 45.5 and 84.0 percent of the daytime gain traveled through the improvement in insomnia itself.

Two cautions. The outcomes were self-reported and the comparator was sleep hygiene rather than an active treatment. The paper also lists an affiliation with the company that makes the program, which is the kind of thing a reader should be told.

Nurse-delivered restriction in general practice

The second line of work took the core component out of the specialist clinic entirely. Simon Kyle, by then in Oxford, ran a trial across 35 English general practices. It randomized 642 adults to four brief sessions of nurse-delivered sleep restriction therapy plus a booklet, or to the booklet alone.

At six months the nurse-delivered group scored 3.05 points lower on insomnia severity, a moderate to large effect. The health service paid 2,076 pounds for each extra year of good-quality life the treatment bought. Eight participants in each group had serious adverse events and none were judged related to the intervention. The trial was open-label and the primary outcome was self-reported.

An ordinary practice nurse, four appointments, and the effect holds. That is a fact about how much of this treatment lives in the schedule rather than in the therapist.

A sedated night has a different shape, and that is a measurement

Classical sleeping tablets amplify the brain's own inhibitory signal, GABA, at one receptor subtype. The night they produce carries less deep slow activity and more fast activity, with changes still visible the following drug-free night.

Start with the brake. The nervous system runs on a balance between signals that make a cell more likely to fire and signals that make it less likely. The main inhibitory signal in the brain is a molecule called GABA. Where it lands, activity falls.

Classical sleeping tablets do not press that brake themselves. They sit alongside it. The drug binds to a site on the GABA receptor and makes the receptor answer the body's own GABA more strongly, so the inhibition already present goes further. Benzodiazepines and the newer agents commonly called Z drugs both work this way.

Which brake, exactly? In 1999 a group of pharmacologists including Uwe Rudolph answered that by changing a single amino acid. They engineered mice in which one subtype of the GABA receptor, the alpha-1, could no longer be modulated by benzodiazepine-site drugs, while its response to the body's own GABA was left intact.

In their study of benzodiazepine actions mediated by specific receptor subtypes, those mice lost the sedative and amnesic effects of diazepam and kept the anxiety-reducing, muscle-relaxing and motor effects. Sedation lives at one receptor subtype, concentrated in cortex and thalamus. It is a separable thing.

What sedation does to the night's architecture

Now what sedation does to a night. Alexander Borbely is the Zurich pharmacologist whose two-process rule is taught on the sleep page. In 1985 his group gave eight healthy volunteers single bedtime doses of three different benzodiazepine hypnotics, then took each whole night apart frequency by frequency. Their study of the effect of benzodiazepine hypnotics on all-night sleep EEG spectra reported that the deepest slow activity fell and spindle-band activity rose.

A spindle is a short burst of fast rhythm the sleeping brain fires in bouts during lighter sleep, and it is one of the events the sleeping brain uses to work on memory. Fast activity rose in light sleep and in dreaming sleep. Stage shifts became fewer. Some of the changes were still visible on the following, drug-free night. The authors noted that the spectral analysis showed the drug's effect more sensitively than the conventional stage scores did.

That is the precise sense in which sedation and sleep are not the same organization. It is a statement about the shape of a night. It is not a judgment about anyone who takes a tablet.

A drug-shaped night is altered, not uniformly worse

And the tidy version of that story is wrong. Sara Mednick, a sleep scientist in California who studies what naps do to memory, used zolpidem as a research tool and not as a treatment. People took daytime naps under zolpidem, under sodium oxybate, or under placebo, in a pharmacology study of sleep spindles and memory. Zolpidem raised spindle density and cut dreaming sleep. The naps with more spindles produced significantly better verbal memory than control naps, and significantly worse perceptual learning.

Healthy adults, daytime naps, and not a patient population. What it establishes is that a drug-shaped night is not uniformly worse. Something is gained and something else is given up. The organization has been altered, and alteration is not the same as loss.

What the tablets are measured to do, and what they are measured to cost

The effect of sleeping tablets is best sized from the complete dossiers companies file with regulators, where the unpublished trials live. The benefit is real and small, the placebo share of it is large, and the costs concentrate in older adults.

Tania Huedo-Medina and Irving Kirsch went after the complete dossier a company files with a regulator, rather than the flattering subset that reaches print. They obtained the data submitted to the Food and Drug Administration for the approval of eszopiclone, zaleplon and zolpidem.

Across 13 studies and 4,378 participants, their meta-analysis of non-benzodiazepine hypnotics found recorded time to fall asleep cut by 22 minutes against placebo. The people on placebo also improved substantially. Drug and placebo together produced a reasonably large clinical response, and the drug's own share of it was small enough that the reviewers called its clinical importance questionable.

The pharmacological guideline concedes the same thing from the inside. Michael Sateia, Daniel Buysse and Andrew Krystal, three sleep physicians, graded each drug individually for the American Academy of Sleep Medicine in a clinical practice guideline for the pharmacologic treatment of chronic insomnia. Every pharmacological recommendation in it is weak, and the evidence behind each is low or very low quality. Several widely used agents were recommended against.

The placebo reaches the recording

Then the finding that belongs to the model. Alexander Winkler and Winfried Rief are clinical psychologists in Marburg who study what a placebo does to a body. They pooled 32 studies, 82 treatment conditions and 3,969 participants in which the placebo arms had been recorded with polysomnography.

They were asking what a dummy tablet does to a sleeping brain and not what it does to a questionnaire. Their meta-analysis of placebo conditions on polysomnographic parameters found the placebo moving the objective measures: latency, total sleep time, wake after sleep onset, efficiency. Their headline figure is that 63.56 percent of the drug response was achieved in the placebo groups.

Regression to the mean cannot be fully excluded when a group is measured before and after with nothing to compare it against, and the pooled trials were heterogeneous. Even heavily discounted, that number says something a purely pharmacological account has no room for. Expectation, ritual and being taken seriously are inputs to the same system the drug acts on, and they reach the recording.

The costs in people over 60

Now the costs, precisely and without alarm. Jennifer Glass, working at the University of Toronto, pooled 24 randomized trials covering 2,417 people aged 60 and over in a meta-analysis of risks and benefits in older people. Total sleep time rose by about 25 minutes and awakenings fell.

Against that, adverse cognitive events were 4.78 times more common than on placebo and daytime fatigue 3.82 times, both statistically significant. Adverse psychomotor events were 2.61 times more common, and that last estimate did not reach significance. Psychomotor is the category that carries falls, which is why this trade is weighed differently at 75 than at 35. These were short trials in older adults free of psychiatric illness.

Tolerance and rebound belong to particular agents and not to the class. A meta-analysis of 75 sleep laboratory studies covering 1,276 people found clear tolerance with triazolam and marginal tolerance with midazolam and zolpidem. Rebound insomnia on the first night after stopping was intense with triazolam and mild with zolpidem. Data were insufficient to estimate either for brotizolam or zopiclone.

That rebound is worth understanding rather than fearing. A worse night on stopping is what a system does when a chemical prop it had adapted to is removed, and it says nothing about whether the underlying insomnia got worse. It does mean that coming off a hypnotic is a clinical process with a shape to it, and one to plan with the prescriber.

Complex sleep behaviors, rare and serious

Finally the rare and serious one. Analysts at the Food and Drug Administration, led by Kelly Harbourt, searched 26 years of adverse event reports for complex sleep behaviors, meaning things people did while not awake. Their review of complex sleep behaviors resulting in serious injuries, including death found 66 cases meeting criteria.

Twenty involved death, from drowning, hypothermia, motor vehicle collisions and carbon monoxide poisoning. Forty-six involved serious injury. In 22 cases the same person had had a previous episode on the same drug before the reported event. The regulator responded with a boxed warning and a contraindication.

Spontaneous reporting gives a signal and never a rate, and these events are rare. The 22 repeat episodes are the detail worth holding, because they say the first episode was information.

Blocking wakefulness is a different act from forcing sleep

Orexin, discovered twice in 1998, is the hypothalamic peptide that holds wakefulness on. Its loss is human narcolepsy, and blocking its receptors produced the first hypnotics that release the accelerator rather than leaning on the brake.

In January 1998 a group of molecular biologists went hunting for messenger RNA found only in the hypothalamus, the base-of-brain controller introduced earlier. They were not looking for a sleep signal. They were cataloging what the hypothalamus makes that nothing else makes.

They found a gene encoding the precursor to two peptides. The peptides sat in neurons of the lateral hypothalamus, their fibers reached across brainstem and thalamus, and one of them excited cultured hypothalamic neurons. Luis de Lecea and his coworkers named them the hypocretins, from hypothalamus and secretin. Nobody knew what they were for.

Six weeks later a second laboratory published the same two molecules. It had arrived from the opposite direction, hunting for the natural partners of two orphan receptors, meaning receptors whose signaling molecule was unknown. Injected into rats, the peptides made the rats eat. So Takeshi Sakurai, in a laboratory led by Masashi Yanagisawa, named them orexins, from the Greek for appetite. One molecule, two names, two laboratories, and a first interpretation that turned out to be secondary.

Narcolepsy revealed the function

The function came from dogs. A long-established colony of dogs collapses when the animals get excited, their muscles giving way while they stay conscious. That collapse is cataplexy, and together with daytime sleepiness and sudden entries into dreaming sleep it makes the colony an animal model of narcolepsy.

In 1999 a team used positional cloning, which means tracking a trait to its address on a chromosome without knowing the gene in advance, to find the recessive mutation responsible. It landed on the gene for hypocretin receptor 2, with Emmanuel Mignot as senior author. The peptide nobody had a function for was the thing that holds wakefulness together.

Then the humans. Seiji Nishino and Emmanuel Mignot measured the peptide in the spinal fluid of people with narcolepsy. In seven of nine patients it was undetectable. Human narcolepsy is, in most cases, the loss of the wake-holding signal itself. Nine patients, and it has been replicated widely since.

That discovery split sleep pharmacology into two principles.

Daridorexant in two phase 3 trials

The second principle became a class of drugs. Two phase 3 trials at 156 sites in 17 countries tested daridorexant, an orexin receptor antagonist, against placebo. At 50 milligrams the drug cut recorded wake time after sleep onset and time to fall asleep, added self-reported sleep, and improved a validated measure of daytime sleepiness.

At 25 milligrams the night improved and the day largely did not. At 10 milligrams almost nothing separated it from placebo. Adverse event rates were comparable to placebo across the groups, in two multicentre randomised phase 3 trials.

The absolute numbers are modest. At 25 milligrams, latency to persistent sleep improved by 6.5 to 9 minutes over placebo. The trials were industry-funded. And the precise reading is the interesting one. Blocking wake is a genuinely different act from forcing sedation, and it is still not a restoration of the regulator.

When the night you lived and the night on the machine disagree

A person with insomnia can lie awake most of a night the recording scores as six hours of sleep, and both reports are faithful. The disagreement decomposes into a discrimination deficit, a bias toward reporting wake, and fragmented dreaming sleep experienced as wakefulness.

The first thing to fix is the received version. Jack Edinger and a colleague recorded 173 people with insomnia and compared machine against belief, in a study of the distribution of sleep time misperceptions. As a group they underestimated their sleep badly.

Individually the errors spread across a continuum, from gross underestimates to remarkable overestimates, and the direction of the error tracked what was actually wrong with that person's sleep. Underestimating is not universal in insomnia. Any blanket claim that people with insomnia think they slept less than they did is wrong.

Two faults: discrimination and bias

So what is going wrong? Jeremy Mercer, Richard Bootzin and Leon Lack took it apart. Fourteen people with insomnia and eight good sleepers spent five laboratory nights being interrupted by a recorded voice, which asked whether they had just been asleep or awake, and how confident they were.

Signal detection analysis separates how well a judgment discriminates from which way it leans, and it found two separate faults. The insomnia group was genuinely worse at telling sleep from wake. They also carried a separate bias toward answering wake. Both measures correlated with how much they underestimated their sleep at home.

Read those two faults precisely. One is a discrimination problem and one is a leaning. Neither is dishonesty and neither is imagination.

Fragmented dreaming sleep is read as wake

Bernd Feige and Christoph Nissen, sleep researchers in Freiburg, then found where the misread sleep is coming from. In 100 patients and 100 good sleepers, all medication-free, arousals within sleep were increased in the insomnia group, and the increase came mostly from dreaming sleep. When they modeled reported wake time against the recording, the amount of dreaming sleep predicted reported wakefulness over and above measured wake time, and did so more strongly in the insomnia group. Fragmented dreaming sleep is being experienced as lying awake.

Julio Fernandez-Mendoza and Edward Bixler, who run a large general-population sleep cohort at Penn State, added the arithmetic. They recorded 142 people with insomnia and 724 controls overnight, drawn from a general-population sample of 1,741. Two independent errors were at work. People who genuinely slept short over-reported their sleep. People with insomnia under-reported relative to controls who had slept the same amount. Underestimation appeared only where both errors pushed the same way, in insomnia with a normal recorded duration.

Michael Perlis and colleagues had proposed the mechanism years earlier. Their neurocognitive account argued that fast, thought-associated electrical activity around sleep onset interferes with the amnesia sleep normally produces for its own beginning. Memory keeps forming across the boundary, so the whole stretch is later reported as wakefulness.

The matching measurement came from the same group: more beta and gamma activity in primary insomnia, and the more of it a person had, the wider the gap between believed and recorded sleep. Nine people per group is a small series.

Perception as the system's account of itself

The model's reading is direct. Perception is never a readout of the world. It is the system's best account of its own state, assembled from noisy evidence and a running expectation. In insomnia the evidence is degraded, because the sleep is genuinely fragmented and shot through with waking-type activity. The expectation leans toward wakefulness, because that is what two hundred nights taught. The experience is faithful to the state that produced it. What is miscalibrated is the model doing the producing.

Insomnia runs loud at three in the afternoon as well as three in the morning

Hyperarousal in insomnia is a 24-hour finding. Metabolic rate, stress hormones and cortical fuel consumption are elevated at every hour, and the one convenient autonomic measure expected to confirm the picture has not. The sleep page sets out the core findings and the paradox they produce.

Michael Bonnet and Donna Arand are sleep researchers at a veterans hospital in Ohio, and they went at the condition with a metabolic question. Do people with insomnia burn more fuel than people who sleep well, and if so, do they burn it in the afternoon as well as at night? They kept ten people with objectively confirmed insomnia and ten matched good sleepers in a laboratory for 36 hours, sampling whole-body oxygen consumption repeatedly.

Their study of 24-hour metabolic rate in insomniacs and matched normal sleepers found the insomnia group burning more energy at every measurement point. Two details usually get left out. The nocturnal elevation survived after every stretch of the night containing wakefulness or an arousal was deleted from the data. And on daytime nap testing these people were less able to fall asleep than the controls, despite feeling worse. Ten per group, and the measurement is rarely attempted.

Two further measurements point the same way, and the sleep page narrates both. Alexandros Vgontzas and Edward Bixler monitored eleven young people with chronic insomnia and thirteen matched controls for four nights, sampling blood around the clock on the fourth day. Both ACTH and cortisol ran higher across 24 hours, with the largest elevations in the evening and the first half of the night, which is when they should be lowest.

In that report of nyctohemeral activation of the stress axis the cortisol difference reached only p equals 0.07, which by convention falls short of significance. Eric Nofzinger imaged brain glucose use in seven patients and twenty controls. He found more fuel burned awake and asleep, wake-promoting regions failing to power down on the way into sleep, and a relatively underactive prefrontal cortex by day. Seven patients is very small for an imaging study.

The vulnerability sits in arousal circuitry

The modern synthesis belongs to Eus Van Someren, a Dutch sleep neuroscientist in Amsterdam who spent years asking why no lesion turns up in the circuits that ought to be at fault. In brain mechanisms of insomnia he argues that the vulnerability does not sit in the circuits that time sleep or build sleep pressure at all. It sits in the circuits that regulate emotion and arousal.

His candidate is the locus coeruleus, a small brainstem cluster that supplies most of the brain's noradrenaline and works as its alarm bell. In insomnia it stays too answerable to the salience network, which is the circuit that decides which of the day's events matter.

It stays answerable even during dreaming sleep, when it should have fallen silent. The phrase he uses for the experience is sleeping with one eye open. That is a model of a vulnerability rather than a measurement of one, and he presents it as a model to be tested.

Rick Wassing, working in the same Amsterdam institute, supplied the loop it needs. Dreaming sleep is where the sleeping brain discharges the emotional charge of the day, and Wassing wanted to know what happens when that discharge fails. He measured insomnia severity, hyperarousal and self-conscious emotional distress in 1,199 people, separating distress that resolved overnight from distress that did not.

Distress that lasted overnight rose with insomnia severity. Distress that resolved did not. Modeling the relationships put 62.4 percent of the association between insomnia severity and hyperarousal through the slow dissolving of emotional distress. The data are cross-sectional and the causal direction is inferred rather than shown.

Put those together and the night stops being the site of the problem and becomes one place the problem shows. The night is where the previous day is supposed to be metabolized. If dreaming sleep is fragmented, yesterday's distress is still there in the morning, adding to the arousal that fragmented it. That is a loop with no external input, the same shape as the conditioned bed.

The autonomic null result

And now the null, carried straight. The body runs an automatic control system that nobody operates on purpose, with an accelerator that drives the heart faster and a brake that slows it. The gap between one heartbeat and the next varies from beat to beat, and the size of that variation is the standard reading of how accelerator and brake are balanced. Impaired variability in insomnia is close to received wisdom.

Kirsty Dodds, at the University of Sydney, gathered every study comparing patients with controls for a critical review of heart rate variability in insomnia. They found 22 papers, judged most at high risk of bias, and found them too heterogeneous in selection, protocol and processing to pool. Their conclusion is that the impairment is a widely accepted concept unsupported by the empirical evidence.

Metabolic rate, cortisol, cortical fast activity and regional glucose all point one way. The single most convenient autonomic measure does not, yet. The heart rate variability page covers what that measure can and cannot see. The null stands beside the four positives until better recordings decide it.

The arrow between insomnia and depression runs both ways

People with insomnia and no depression carry roughly twice the odds of becoming depressed later, and treating the insomnia prevented depression outright in a three-year randomized trial of 291 older adults.

Insomnia has been treated for a century as a symptom of something else, mostly of depression. The longitudinal evidence says it runs the other way at least as often.

Chiara Baglioni, a clinical psychologist in Freiburg, pooled 21 studies that had followed people forward in time, in a meta-analytic evaluation of insomnia as a predictor of depression. People who were not depressed but had insomnia carried roughly twice the odds of becoming depressed later, an odds ratio of 2.60. The included studies did not consistently control for intervening variables. Order in time is not cause.

Treating insomnia moves paranoia and prevents depression

The interventional test is stronger, and one trial dominates it. Daniel Freeman is an Oxford clinical psychologist who studies paranoia, and he suspected that broken sleep was feeding it. He randomized 3,755 university students with insomnia to digital CBT-I or to usual care. The authors record it as the largest randomised trial of a psychological intervention for a mental health problem then run.

Treating the insomnia reduced insomnia heavily, as expected. It also reduced paranoia and hallucinatory experiences, and the mediation analysis showed the change in insomnia carrying those reductions. The effects on paranoia and hallucinations were small, at 0.19 and 0.24 on the standard effect-size scale. That puts the average treated student at about the 58th and 59th percentile of the untreated ones. A student population, and the attrition typical of online trials. Here the route matters more than the size.

Michael Irwin runs a psychoneuroimmunology laboratory in Los Angeles, which means he studies how sleep, the nervous system and the immune system talk to one another. He wanted to know whether treating insomnia could prevent depression outright. He randomized 291 adults aged 60 and over who had insomnia and no current depression. One arm got insomnia therapy, the other an active comparator of sleep education therapy, and both were followed for three years.

Major depression occurred in 12.2 percent of the therapy group and 25.9 percent of the comparator group, in his trial on prevention of incident and recurrent major depression. Among those whose insomnia went into sustained remission, the likelihood of depression fell by 82.6 percent relative to comparator participants without sustained remission. Single site, and roughly half the sample did not complete the full follow-up.

The same treatment reaches inflammation

The same laboratory had already shown the same aim reaching a different system by a different route. Three terms first. C-reactive protein is a substance the liver puts into the blood when the body is running an inflammatory response, and it is the cheapest blood test of that state.

A monocyte is a kind of white blood cell, and cytokines are the chemical messages it sends. Gene expression means which of a cell's genes are being read at a given moment, as against which genes the cell merely carries.

Now the trial. Irwin randomized 123 older adults with insomnia to insomnia therapy, to tai chi, or to sleep education, and drew blood over 16 months. The therapy lowered C-reactive protein at four months and at sixteen. It cut monocyte production of inflammatory cytokines.

It shifted the inflammatory gene expression profile at four months. Tai chi reversed the same cellular and genomic markers through an entirely different door. An analysis of which switches the genes were answering to found reduced activity of two transcription factors in both arms, meaning two of the proteins that turn inflammatory genes on.

A conversation about a schedule and a slow movement practice, arriving at the same change in gene expression. That is what it looks like when two inputs reach the same regulatory system through different entrances. The depression page and the inflammation page take those systems on their own terms.

If you are thinking about harming yourself, contact emergency services or the 988 Suicide and Crisis Lifeline.

A timing disorder mistaken for chronic insomnia

Delayed sleep phase syndrome sent about one referral in fifteen to a 1981 insomnia clinic with normal sleep parked at the wrong hour. Restricting that sleep window treats a problem that was never there.

In 1981 a sleep laboratory reviewed 450 people seen for a primary complaint of insomnia and found 30 who did not have insomnia at all. Arthur Spielman was among the authors of the report.

These people could not fall asleep at the clock time they wanted. Left to their own schedule they slept normally, for a normal length, and woke refreshed. They were younger than the general insomnia population and carried no particular psychiatric profile. The team treated six of them by walking bedtime progressively later around the clock until it landed where they wanted it, and named the condition delayed sleep phase syndrome.

About one referral in fifteen. It is a case series from one clinic in 1981, and the diagnostic point has long outlived the method.

Read it as a defended value parked at the wrong hour. Nothing is wrong with the sleep itself. The architecture is normal, the duration is normal, the restoration is normal. What the system has fixed, and defends against every attempt to shift it, is where in the 24 hours sleep goes. The mismatch between that hour and a nine o'clock meeting is the entire illness. Insomnia defends the wrong balance. This defends a sound balance at the wrong time.

The distinction matters in practice, because the two conditions look identical from outside. Both people are awake at two in the morning and wrecked at eight. One is aroused and one is on a different clock. The question that separates them is what happens across a fortnight with no obligations. Treat a timing disorder as insomnia and you will restrict a sleep window that was never the problem.

Insomnia and sleep apnea travel together and blunt each other's treatment

Roughly 30 to 40 percent of people with chronic insomnia meet criteria for sleep apnea, having both is worse than either, and treating the insomnia first raised nightly airway-machine use by 61 minutes.

The two commonest sleep disorders occur together far more often than either is looked for in the other's clinic. Alexander Sweetman and Leon Lack, sleep researchers in Adelaide who work on the overlap between the two, reviewed the whole picture in a paper on bi-directional relationships between co-morbid insomnia and sleep apnea.

Roughly 30 to 50 percent of people with obstructive sleep apnea report clinically significant insomnia. Roughly 30 to 40 percent of people with chronic insomnia meet criteria for apnea. Having both is worse than either, harder to diagnose, and each condition blunts the response to the other's treatment.

The interference is easy to see once stated. Insomnia therapy restricts time in bed and asks a person to leave the bed when awake, which is hard for someone being woken repeatedly by their own airway. Airway pressure therapy asks a person who already dreads the bed to wear a mask in it.

Treating the insomnia first, tested

So the same group tested the order of operations. They randomized 145 people who had both conditions, either to four sessions of insomnia therapy before starting the machine, or to usual care. Treating the insomnia first raised average nightly machine use by 61 minutes over six months, and raised initial acceptance of the machine from 89 to 99 percent.

Nothing about the machine changed. Nothing about the airway changed. What changed was the state of the system that had to receive the treatment, and that is what the tonal expression of a structural disease looks like once somebody measures it.

What carrying both conditions costs

Bastien Lechat, a sleep researcher in the same Adelaide group, followed 5,236 participants in a long-running American sleep cohort for 15 years. Insomnia alone and apnea alone each raised the risk of hypertension but not of cardiovascular disease. Having both raised the odds of hypertension twofold, cardiovascular disease by 70 percent, and death from any cause by 47 percent. The cohort was observational and only 137 of the 5,236 had both conditions.

Which brings the red flags back. Loud snoring, witnessed pauses in breathing, gasping awake, or real daytime sleepiness in someone with insomnia is a reason to ask a physician about a sleep study. It is not a reason to work harder on the sleep schedule. The sleep apnea page covers the airway in full.

One argument the field has not closed, and one number that should reassure you

Whether insomnia with objectively short sleep marks a distinct biology is a live argument. Whether chronic insomnia shortens life is not. The best-powered answer, drawn from 36,938,981 people, says it does not.

Start with a test that has already been run. Christina Bathgate, Jack Edinger and Andrew Krystal re-analyzed 60 adults from a randomized trial, splitting them by whether they had actually slept over or under six hours on a recording made before treatment.

Six months after therapy, those with normal recorded duration did significantly better on remission, on actigraphic sleep efficiency and on night-time wakefulness than those with short recorded sleep. Same diagnosis, same treatment, different outcome, and the split was made in advance by a measurement of the receiving system.

The short-sleep phenotype proposal

That test exists because somebody made a proposal. Alexandros Vgontzas recorded a representative sample of 1,741 adults overnight and sorted them by how long they actually slept, and the sleep page narrates the result. Insomnia with under five hours of recorded sleep carried five times the odds of hypertension against normal sleepers over six hours.

Insomnia with five to six hours carried three and a half times. Insomnia in people who slept longer carried far less. From that he proposed that insomnia with objective short sleep duration marks out a biologically distinct kind of insomnia.

Sixty people from one academic center is hypothesis-supporting and not confirmatory, and the original claim rests on cross-sectional single-night recordings from one geographic sample. The phenotype remains actively contested across the field.

Insomnia does not shorten life

The second unsettled matter is more personal, and the answer is reassuring. In Adelaide, Nicole Lovato and Leon Lack pooled 17 studies covering 36,938,981 individuals with a mean follow-up of 11.6 years, testing the widely repeated claim that insomnia shortens life. It does not. The fully adjusted hazard ratio was 1.07, with a confidence interval crossing one.

A tendency toward higher mortality did appear in the data, and it attached to hypnotic use rather than to insomnia. That association is confounded by indication, because the people prescribed sleeping tablets are the people who are sicker and sleeping worst. It cannot be read causally and nobody should read it that way.

The reviewers' own conclusion is that patients can be reassured about longevity. Anyone who has lain awake at four in the morning calculating the cost to their lifespan now has the best-powered answer available. It is not the one you were braced for.

Insomnia read as one organization: gain, set point, prediction

Everything above is established science, and it arrives as a list. A trait scoreable in advance. Perpetuating behaviors. Diluted pressure. A conditioned bed. Raised metabolic rate around the clock. A stress axis elevated when it should be quiet. Fast cortical activity inside scored sleep. Fragmented dreaming sleep experienced as wakefulness. A locus coeruleus that will not go quiet. Distress that does not dissolve overnight.

Medicine files those in separate journals because they belong to separate specialties. The Unified Model of Tone reads them as one thing seen from several angles.

Tone is the integrated organization of the body's interacting state transitions. It is how the mechanical, electrical, chemical and neural processes stand in relation to one another at a given instant, taken as one bound state rather than as a list of parts. Oscillation carries it. The organization of the oscillations is what tone is. The chord, not the notes. Health is the breadth of the range that organization can move through, and disease is the narrowing of that range.

Sleep is the largest reorganization a nervous system performs each day. It asks the whole body to change key at once: cortical rhythm, autonomic balance, temperature, hormone release, muscle tone, breathing, blood pressure and the gating of sensation. Insomnia is what it looks like when that change of key is attempted and never completed, night after night. The system is not damaged. It is holding a state it can no longer leave.

Three aspects of tone dominate how that stuckness is organized, and each has already been measured above under a different name.

Gain is how loudly a system answers relative to the input it receives. In insomnia the volume is up around the clock and not only at night, which is what Bonnet and Arand measured in oxygen consumption, Vgontzas in cortisol, and Nofzinger in glucose. The daytime nap result is the tell.

A person short of sleep drops off quickly in a dark room. These people cannot, because the setting is high and not because the reserve is empty. Gain is also why the same life event costs one person a fortnight and another a decade, and the sleep reactivity score is that difference recorded in advance.

Set point is the value a system defends. The clearest evidence for it on this page is the result that looks like an anomaly. The treatment that fixes the condition adds 7.6 minutes of sleep, a figure whose confidence interval includes nothing at all, while sleep efficiency climbs about ten points. The defended amount holds. The organization around it is rebuilt.

Sleep restriction runs on the same logic, because it never argues for more sleep. It changes the terms on which the balance is defended. Wake is a defended state in its own right, which the orexin work shows by taking the signal away, and delayed sleep phase is the same value parked at the wrong hour. Set point is the aspect that explains why nothing is broken.

Prediction is a nervous system acting on its own model of the world. The bed comes to predict failure, the body prepares for what it predicts, and the preparation is arousal. The same machinery generates the experience of the night itself, which is why a fragmented night shot through with fast activity is reported, faithfully, as lying awake. Prediction is why the condition confirms itself.

None of the three is the condition. Insomnia is what those three sound together, which is why moving any one of them alone moves the chord only a little.

The puzzles the reading dissolves

Read this way, several puzzles dissolve at once. The condition outlives its cause, because a self-maintaining organization does not need its origin. It travels with anxiety, low mood, pain and inflammation, because those belong to the same aroused organization and are not separate illnesses that happened to arrive together.

Its best treatment works without adding sleep, because what was wrong was the organization and not the amount. And it varies enormously between people reporting the same complaint, because the same input met a different tone in each of them.

One more thing follows. Excessive tone in this model is excessive constraint: reduced variability, high maintenance cost, few available transitions, resistance to updating. That definition was written to describe any over-constrained system. It was not written about insomnia, and it fits one exactly.

Restoring a regulator against managing a night, and how to tell which one happened

Masking moves the output while the setting stays where it is. Restoring returns the loops' own ability to find the middle. The distinction decides what counts as help, and the model names in advance the results that would erase it.

A hypnotic masks, and does it honestly. It amplifies inhibition, the person sleeps sooner, and when the drug stops the system is where it was. Every number in the ledger section describes exactly that: 22 minutes of latency against placebo, a night with less deep slow activity in it, and rebound on the first night off for the fastest-clearing agents.

None of that is a criticism. Relief is worth having and this model grants it that place without qualification. A person whose insomnia is costing them their job, their driving or their marriage is carrying a regulatory load in its own right. Quieting it can return the reserve a system needs to reorganize at all. Surgery is the same principle at a larger magnitude. Nobody should apologize for taking what works, and medication decisions stay with the prescriber.

What restoration looks like in the data

Restoring means the regulatory loops recover their own ability to find the middle. The evidence for what that looks like is already above, and it is unusually clean. The therapy adds 7.6 minutes of sleep and still moves the disorder. It lifts sleep efficiency by about ten points, and in the Norwegian trial it added deep slow-wave sleep that a sedative did not.

It lowers C-reactive protein out to sixteen months and shifts inflammatory gene expression at four, and a movement practice reaches the same biology through another entrance. Those are the signatures of a system reorganizing rather than of a value being pushed.

The line runs across the instruments rather than between them, because it is drawn by aim. Two talking treatments on this page sit on opposite sides of it. Sleep hygiene advice, given alone, lifts what a person writes in a diary and leaves everything that was recorded where it was.

The full therapy moves the recording, the inflammatory markers and the gene expression. The instrument is a conversation in both cases and the achievement is not the same. The traffic runs the other way too. What reached the recording in the placebo arms was expectation, ritual and being taken seriously, delivered inside something shaped like a drug.

The bidirectional prediction

Now the claim that separates restoring regulation from masking a symptom.

A drug is directional by construction. It pushes one way in everybody who responds, regardless of where they started. A restored regulator is not directional. It moves a dysregulated value toward the middle of the healthy range from whichever side that person is on.

Insomnia is an unusually good arena for that test, because these patients are not uniform on the measurable variables even though they are uniform on the complaint. The gap between believed and recorded sleep runs in both directions, as Edinger's continuum showed. Some people grossly underestimate their sleep and some remarkably overestimate it. Time in bed splits the same way. Some patients are compressing an already thin window, and some are lying there eleven hours to collect six.

So the prediction, with the input fixed before any data arrive. Score a mixed group on a tone measure first. Let that measure name in advance where each person's input should be delivered, so that no match can be judged after the outcome is known.

The measures already on the table will serve: the sleep reactivity score, the gap between the diary and the actigraph, the recorded sleep duration. Stratify by which side of the middle each person sits on. Then randomize three ways. The input delivered where the measure pointed. The identical input delivered where it did not. And a control input matched for contact time, attention and expectation.

The model predicts two things at once. The matched arm should beat the mismatched arm substantially, because correspondence is the active ingredient and not force. And inside the matched arm the strata should converge. The misperception gap should shrink from below in the underestimators and from above in the overestimators.

Time in bed should fall in the people spending eleven hours there and rise in the people who have cut themselves to five. The stratum means should travel toward each other while the pooled mean barely moves. And the movement should appear across several separately regulated measures at once in the same person, which no account assembled from independent set points expects.

What would establish the reading

State the confirming findings plainly. A restoration that moves a value toward the middle from either starting side confirms the bidirectional claim. A matched arm that beats the mismatched arm establishes correspondence as the active ingredient. Convergence in the treated arms beyond what the control arm produces establishes the distinction. Clinical benefit that tracks the organization of the night rather than total sleep time gained confirms the organization-first reading here, and accounts for the 7.6-minute result.

Well-conducted 24-hour autonomic recordings that carry the arousal signature confirm the autonomic half of the arousal account. The heart rate variability review above is why that recording still has to be made. And the last one cuts deepest. Take any measure of the receiving system recorded before the event arrives, the sleep reactivity score among them. Such a measure carrying information about who converts and who mends is what establishes the input-meets-tone reading underneath all of this.

What can be recorded today

What a person can actually have recorded today is shorter than the theory. The gap between a sleep diary and a wrist actigraph, tracked across weeks rather than nights. A sleep reactivity score, taken before the next bad year arrives. Objective sleep duration on a proper recording, in anyone whose treatment has stalled. And an airway assessment in anyone carrying the red-flag pattern, because that one changes what should be done first.

None of those needs an instrument that does not exist. Every one of them reads the organization of a night rather than its total, which is the whole argument in a line.

How insomnia relates to the rest of the library

Insomnia sits where the foundations of tone meet the conditions of the night, and each neighboring page carries one piece of its story.

Four pages border this one.

• Sleep holds the healthy machinery: the two forces that time the night, the switch that throws the state, and what the night is for.
• Sleep apnea is the structural neighbor that shares 30 to 40 percent of insomnia's patients and blunts its treatment.
• Depression sits downstream often enough that treating insomnia halved its three-year incidence in older adults.
• Anxiety is the daytime voice of the same aroused organization, run through the same salience circuitry that will not stand down at night.

Two more pages complete the picture. Heart rate variability is the instrument that has not confirmed hyperarousal in insomnia, and its page explains what one autonomic channel can and cannot see. Inflammation is where insomnia therapy lowered C-reactive protein and shifted inflammatory gene expression out to sixteen months.

Frequently asked