Insomnia has a first-line treatment. It isn't a drug.

What counts as insomnia

A bad night is not insomnia. The clinical picture is difficulty falling or staying asleep, several nights a week, going on for months, and — the criterion people forget — costing something during the day: fatigue, irritability, attention gone soft, dread of the evening.

The decisive detail lies elsewhere. What triggers an insomnia — a bereavement, a move, a punishing stretch at work — is almost never what keeps it going. The trigger passes; the insomnia stays, held in place by what the bad period installed: staying in bed longer to catch up, checking the clock, dreading bedtime, going up early just in case. Treatment does not aim at the original cause. It aims at those maintaining mechanisms, because they are the part still present.

Why sleep hygiene isn't enough

Cool room, screens off, last coffee at two: the advice is sound, and Sillon carries several of these cards. But they describe the conditions for good sleep, not the mechanism that keeps an established insomnia running. Someone who has slept badly for eight months usually has a perfectly decent bedroom already. That is no longer the problem.

The problem is two things. First, sleep pressure has been diluted: spending nine hours in bed to get five hours of sleep spreads the sleep thin, fragments it, and strips it of density. Second, the bed has changed meaning. In a good sleeper it triggers sleepiness within minutes. In someone who has spent months turning over in it, checking a phone and worrying about not sleeping, it triggers the opposite: alertness. That is learning, and it can be unlearned.

The protocol, in order

1. Two weeks of diary Grade A

One minute on waking, every day, for fourteen days. A standardised format exists for this — the consensus sleep diary, published in Sleep in 2012 — precisely so the measurement would stop varying from clinic to clinic.

The diary does two things. It separates what you believe from what happens: most poor sleepers underestimate how long they slept and overestimate how long they lay awake. And it produces the one number everything downstream depends on, sleep efficiency: time asleep divided by time in bed. That figure, not a feeling, sets the window you apply next.

2. Stimulus control Grade A

The bed is for sleep and for sex. Nothing else. Not reading, not working, not a series, not scrolling. And above all: past roughly twenty minutes awake, you get up. Another room, low light, something dull, and back to bed only when sleepy. As many times as the night requires.

It is counter-intuitive and it is the core of the treatment. Lying awake waiting for sleep teaches the brain that the bed is a place for thinking, worrying and waiting. Getting out breaks that association. The clinical description of this mechanism is old and robust; it remains one of the two components the guidelines retain.

3. Compressing the window Grade A

This is the most effective component, and the most unpleasant for the first fortnight. The principle runs against instinct: spend less time in bed, so that more of that time is sleep.

In practice: the window in bed is set to the sleep your diary actually shows — never below five hours. Wake time is fixed first, seven days a week, no exceptions: that is the anchor, not bedtime. Push bedtime later until the window matches. Hold it a full week. If efficiency clears 85%, move bedtime fifteen minutes earlier; below 80%, fifteen minutes later.

A mechanistic trial published in Sleep in 2022 confirmed the route: the method works through increased sleep pressure and reduced arousal. It is not a placebo effect of discipline.

4. Stop watching the clock Grade B

Clock turned to the wall, phone out of reach. An experimental study compared nights with and without clock monitoring in people with insomnia: monitoring lengthens time to fall asleep and feeds the worry. The mental arithmetic of how much sleep is left is itself an arousing act.

5. No naps while you're repairing Grade B

The nap has good uses, but not here: during the compression weeks it spends exactly the sleep pressure the protocol needs in the evening. Bring it back later.

The components, summarised

Three things to rule out first

A behavioural protocol does not fix a problem that isn't behavioural. Three common causes are screened in minutes and change the plan entirely.

Obstructive sleep apnoea

Close to a billion adults worldwide have it, and the large majority don't know. It is the most common reason immaculate sleep hygiene changes nothing: the problem isn't behaviour, it's breathing. The STOP-Bang questionnaire, validated as a screening tool, takes five minutes; its high sensitivity means it misses few cases, at the cost of false alarms. That is the right trade for a screen. If you snore heavily, or you have been told you stop breathing at night, get that assessed before compressing anything.

Restless legs syndrome

An overwhelming urge to move the legs in the evening, relieved by movement. The clinical practice guidelines on iron treatment for this syndrome make ferritin a first-line test: relative deficiency is common and correctable. This is not behavioural insomnia.

What you already take

Sedating antihistamines, benzodiazepines, Z-drugs, and some treatments taken for something else entirely. The 2023 AGS Beers Criteria list these classes explicitly among medications to avoid in older adults, on cognitive effects and fall risk. The 2023 European Insomnia Guideline restricts hypnotics to short-term use and confirms CBT-I first-line. None of this is changed on your own: it is a conversation with a doctor or pharmacist.

What it delivers

A meta-analysis published in Annals of Internal Medicine in 2015 pooled the controlled trials of CBT-I in adults with chronic insomnia: shorter time to fall asleep, less time awake after sleep onset, better sleep efficiency. A broader meta-analysis in Sleep Medicine Reviews in 2018 finds the same effects across delivery formats.

The decisive argument isn't effect size, though. It's duration. A hypnotic works while you take it; CBT-I gains hold after treatment stops, because what changed is not a blood level but a piece of learning. That is why two independent guidelines put it ahead of medication.

How Sillon runs it

The app's sleep-repair path runs exactly this sequence over twenty-eight days: two weeks of diary, then stimulus control, then compression, with the weekly widening computed from your own efficiency. Every card carries its evidence grade and its references. Contraindications are not shown as warnings: they remove the practice from the engine. And the screening cards — apnoea, restless legs, medication review — come before the compression ones, not after.