Sleep onset latency

Sleep onset latency is the gap between settling down with the intention of sleeping and the first sleep that actually arrives. In healthy adults it usually runs 10 to 20 minutes, and the line research and clinics draw for a problem is about 30 minutes, on three or more nights a week.

Two things about that number catch people out. It cuts in both directions: a latency near zero is not a better score, and in a tired population it is one of the most reliable signs that sleep is in short supply. And almost nobody reports their own figure accurately — the size of the error is large, and its direction depends on how you sleep.

The 10-to-20-minute band is a central tendency and not a target. Healthy sleepers scatter either side of it, and the scatter within one person across a fortnight is wider than the difference between two people's averages. A single fifty-minute night is noise, which is why every definition in this area counts nights per week rather than minutes per night.

Age moves it less than people expect. The large meta-analysis of normative values by Ohayon and colleagues found latency rising gradually across adulthood, by a margin measured in a few minutes — far smaller than the change in how fragmented the middle of the night becomes. If falling asleep has gone from ten minutes to an hour, that is not ageing.

Half an hour is a convention with work behind it rather than a law. When Lichstein's group went looking for a quantitative cut-off across two decades of trial data, roughly thirty minutes was where ordinary bad nights stopped and clinical patterns started, and it has been the working line since.

One thing worth watching more than the average: the spread. Twenty minutes every night and an average of twenty minutes made of five, ninety, ten and twenty are different situations, and only the second one tells you something is unstable.

The spread also usually has content in it. Ninety minutes on the night after a row is a different object from ninety minutes every night, and why can't I sleep after an argument is about the first kind.

Speed of falling asleep is the standard laboratory measure of sleep pressure. That is the whole premise of the Multiple Sleep Latency Test: a person is given repeated chances to nap across the day, and the average time they take to fall asleep is read as how much drive to sleep they are carrying. Short latency is the symptom, not the achievement — a mean of eight minutes or less across the day is the threshold for pathological sleepiness, and under five minutes is severe.

The dose-response has been visible since Carskadon and Dement restricted volunteers to five hours a night in 1981: cut sleep, and latency falls further every day of the restriction.

So "I'm out in two minutes" is a report about something. Usually one of three things. There is not enough opportunity, the question in the entry on sleep debt. Or the night is being broken up by something that stops the pressure discharging, with untreated apnoea the common offender. Or bedtime is now far enough past the body clock's own window that sleep has been waiting for hours.

The honest caveat: the number means nothing by itself. Someone who falls asleep in five minutes, wakes unprompted, and is fine at four in the afternoon is a person whose timing is right. It is the pairing of the figure with the following day that carries the meaning, which is also why the day is written into the definition of insomnia.

Subjective and measured latency come apart reliably, and they come apart in opposite directions depending on who is reporting.

People with insomnia overestimate how long sleep took, often substantially. Good sleepers tend to underestimate it. Means and colleagues put both groups through the comparison in 2003 and found exactly that asymmetry, which means the gap between the two numbers is itself a feature of the complaint rather than a measurement failure.

The reason is that falling asleep is not a switch being thrown. The first minutes of light sleep feel, from the inside, like lying there thinking — wake someone out of them and they will tell you they were awake. Harvey and Tang's review assembled the pieces: impaired time estimation while lying in the dark, attention tuned to any evidence of not sleeping, and a transition that genuinely does not feel like one. None of that is imagination. It is what the boundary is like.

Clock-watching makes both halves worse. It adds arousal, and it manufactures a figure that then gets rehearsed the next day.

A diary is still the right instrument, for three reasons: a consistent bias is usable, the perception is part of what needs treating, and every assessment in this area is built from a fortnight of self-report rather than from a laboratory. The standard form is the Consensus Sleep Diary. Fill it in the morning, estimate, and do not look at the clock to do it.

Of everything a person can write down about their night, latency is the one a clinician will reach for first.

It is one of the two inputs to sleep efficiency, the ratio that sets how much time in bed a course of treatment allows — the mechanism is in the entry on CBT for insomnia and is not repeated here.

It is also the trigger for the instruction that does most of the work. Stimulus control asks you to leave the bed once you have been lying there awake and frustrated, and it deliberately does not give a number, because a rule phrased as twenty minutes puts you back to watching the clock to obey it.

Practically, it is the variable that moves first. Latency shortens early in treatment, usually before total sleep time increases at all, and knowing that is worth something in the first fortnight, which is when people quit.

And it discriminates. A long latency on work nights that collapses to ten minutes when nothing is scheduled is a timing problem rather than insomnia, a fork worked through in insomnia or a late chronotype.

What it cannot do is diagnose anything. A rough fortnight is not a disorder. Three or more nights a week for three months or more, with the following day visibly paying for it, is the recognised shape, and the treatment for that shape is not a tablet. MoodyWonder's Insomnia Check scores onset separately from the rest of the night for the reason this page exists: the components behave differently, and adding them together hides which one is failing.