In 2014 a group of clinicians published a case series describing something they kept seeing in military patients and could not fit into any existing category: nightmares that replayed a real event, movement and vocalisation during sleep, and a body that stayed switched on through the night. They gave it a name — trauma associated sleep disorder — and were careful to call it proposed, because a handful of cases in one unusual population is not a population finding. That caution turned out to matter for more than a decade. The pattern was described, refined, argued about, and studied almost entirely in people who had already sought help for it. Nobody could say how common it was in ordinary life, because nobody had counted.

A Pattern Described in Soldiers

The original 2014 paper in the Journal of Clinical Sleep Medicine proposed trauma associated sleep disorder as a distinct parasomnia: trauma-related nightmares occurring alongside disruptive nocturnal behaviours — movement, vocalisation, dream enactment — together with signs of an activated stress response during sleep, in people who had survived trauma.

Later work characterised it further using overnight recordings. What none of that work could do was estimate how often the pattern occurs in the general population, because clinical samples are made of people who came to a clinic. A study of patients tells you about patients. It cannot tell you what fraction of a population is affected, and it cannot tell you what that fraction looks like in people who never sought care at all.

This is a familiar limitation and an important one. A disorder characterised only in the people who present with it tends to be understood in its most severe form, and its true frequency stays unknown in either direction — it may be far rarer than clinical impression suggests, or far more common.

Counting It Across a Country

The 2026 study, published in Communications Medicine in June 2026 by Unnarsdottir and colleagues at the University of Iceland, took the opposite approach. Rather than starting with a clinic, it started with a population: the SAGA Cohort, a nationwide study of Icelandic women.

Of 30,403 women who gave informed consent, 27,938 were included in the analysis, spanning ages 18 to 69. Participants completed assessments of their trauma history, of trauma-associated sleep disturbances, and of other mental health outcomes. The researchers identified past-month symptoms using established symptom criteria, and estimated prevalence ratios using modified Poisson regression, adjusted for demographic, socioeconomic, behavioural and trauma-related factors.

The scale is the point. At nearly twenty-eight thousand participants, subgroups that would be invisible in a study of a few hundred people are large enough to compare — which is what makes the distribution of the finding more interesting than the headline figure.

6.9 Percent, Unevenly Distributed

Across the whole cohort, 6.9% of women reported trauma-associated sleep disturbances in the past month (Unnarsdottir et al., Communications Medicine, 2026).

Read carefully, that number says two things at once, and the second is the one most coverage drops. Roughly one woman in fourteen reported the pattern — enough that it is not a rarity. And roughly thirteen in fourteen did not. Trauma exposure across a lifetime is common; this specific night-time pattern was not the typical outcome of it. Both halves are findings.

The distribution was not even. Prevalence was highest among young women, and among women who had been exposed to physical or sexual violence. That the pattern clusters around interpersonal violence rather than distributing evenly across all difficult experiences is a substantive result, not a footnote.

Repetition and Timing Tracked the Pattern

Two other factors were associated with higher prevalence: experiencing the worst life stressor more than once, and close temporal proximity to it — that is, repetition and recency.

This is worth sitting with, because it suggests the association is not only with what happened but with how often and how recently. A single category label — “trauma” — flattens experiences that this data appears to separate.

It is an association measured at one point in time, and the direction cannot be read from it. Higher prevalence following recent events is equally compatible with disturbance that fades as time passes and with disturbance that is simply more likely to be reported while the event is fresh. The study measures how common the pattern is and what it travels with; it is not built to trace what causes what.

What Sleep May Be Showing

Why sleep, specifically? The reasoning behind studying it is that sleep is a poor place to hide an activated stress response. During waking hours a person can compensate — stay busy, stay occupied, stay in company. Those options disappear at night.

The study also found that these disturbances were strongly associated with symptoms of post-traumatic stress, depression, anxiety and general sleep problems. That co-occurrence is informative and it is also a complication: when four things travel together this closely, isolating any one of them from a single survey is not possible. It tells you where to look. It does not tell you which came first.

Sleep is not a passive gap in the day — it is an active period with its own architecture, which is why disruption to it has consequences well beyond feeling tired. A nervous system that stays in a state of readiness has effects on how clearly a person thinks under pressure, and patterns of night-time arousal are difficult to separate from the daytime patterns that repeat often enough to become habitual.

What This Study Cannot Establish

Four limits are worth stating plainly, because a large number invites more confidence than the design supports.

It is a snapshot, not a sequence. Past-month prevalence and adjusted prevalence ratios describe how common something is and what it occurs alongside. They do not establish that one thing caused another, and no arrangement of statistical adjustment turns a single time point into a chronology.

It is based on reported symptoms, not overnight recording. Established symptom criteria applied to a large sample are the only practical way to study something at this scale, and they are not the same instrument as a night in a sleep laboratory. The trade-off buys scale and gives up precision.

It is Icelandic women aged 18 to 69. Iceland is one country with its own demography and health system, and the study by design says nothing about men. Extending the 6.9% figure to another population is an assumption, not a result.

The disorder itself is still proposed. The 2026 paper describes trauma-associated sleep disorder as “a proposed sleep disorder” — twelve years after it was named. Prevalence estimates depend on the criteria used to define a thing, and those criteria are not yet settled.

And the limit that matters most to any individual reader: 6.9% is a fact about a population, not a prediction about a person. A prevalence figure cannot tell anyone whether they have this pattern, will develop it, or would recover from it. It describes a group of twenty-eight thousand women, none of whom it can speak for individually.