Unpredictable shifts track insomnia in Korea roughly twice as strongly as in Europe

Most occupational sleep research asks when people work — nights, rotating shifts, early starts. Two analyses of national working-conditions surveys published this year ask a different question: whether workers know their schedule in advance. Both find that unpredictability is associated with insomnia symptoms independently of the hours themselves [s1][s2].

The Korean cross-section

The first study analysed data from 20,534 wage workers aged 20–64 in the 2020 Korean Working Conditions Survey [s1]. Unpredictable work schedules were defined using three items: sudden callbacks, advance notice of schedule changes, and control over working hours [s1]. Insomnia symptoms were measured with the minimal insomnia symptom scale, with scores of 6 or above indicating insomnia [s1].

The raw contrast is large. Insomnia symptom prevalence was 16.10% (157 of 975) among workers with unpredictable schedules and 7.25% (1,418 of 19,559) among those without, a difference significant at p<0.001 [s1].

Adjustment shrinks the association but does not remove it. In unweighted analysis the crude odds ratio was 2.46 (95% CI 2.05–2.94); after adjustment for demographic and work-related covariates it fell to 2.15 (1.78–2.59), and in the fuller model to 1.64 (1.34–2.01) [s1]. In the weighted analysis the fully adjusted odds ratio was 2.30 (1.74–3.05) [s1].

That spread between 1.64 and 2.30 across model specifications is worth noticing rather than averaging away. It means the size of the association depends materially on which covariates and which survey weights are applied, even though the direction and significance do not. Sensitivity analyses using alternative definitions of unpredictability — sudden recall, combined, strict and lenient — produced a consistent, statistically significant association in both crude and adjusted models [s1].

The authors' framing is deliberately modest: unpredictable schedules were associated with higher odds of insomnia symptoms, improving schedule predictability and providing advance notice may help sleep health, and longitudinal studies are needed to clarify causal relationships [s1].

The cross-regional comparison

A second analysis, published in July, put the same construct side by side across two labour markets. It used the 2017 Korean Working Conditions Survey and the 2015 European Working Conditions Survey, assessed habitual schedule unpredictability by advance notice timing, and measured sleep disorders with the same Minimal Insomnia Symptom Scale [s2].

Unpredictable schedule changes were associated with higher odds of sleep disorders in both regions, but not equally: adjusted OR 2.544 (95% CI 2.195–2.948) in Korea against 1.787 (1.543–2.070) in Europe, with a formally confirmed regional difference (p for interaction <0.0001) [s2].

Two further features strengthen the case beyond a simple correlation. Risk increased progressively as advance notice shortened, which the authors describe as a clear dose-response relationship [s2]. And among subgroups, a statistically significant interaction was identified for long working hours in Korea, with stronger associations observed among those without overtime [s2].

That last result is counterintuitive enough to flag. If the mechanism were simply total workload, the association should be strongest in the most-worked group. It was not.

What a dose-response does and does not buy

A gradient by advance notice is the most persuasive element in either paper. Cross-sectional designs cannot establish direction — insomnia might plausibly push workers into jobs with worse schedule control, or into perceiving their schedules as less predictable — but a monotonic relationship with a graded exposure is harder to produce by reverse causation alone than a simple exposed/unexposed contrast.

It is not proof. Both studies are cross-sectional, both measure exposure and outcome by self-report in the same instrument, and both rely on a brief insomnia symptom scale rather than clinical diagnosis [s1][s2]. A worker sleeping badly may recall recent schedule disruption more readily, which inflates the association in a way no amount of covariate adjustment can correct.

The two surveys also do not measure the same thing at the same time. The Korean study uses the 2020 wave with a three-item construct [s1]; the comparison study uses the 2017 Korean and 2015 European waves with an advance-notice construct [s2]. The regional gap they report is a comparison of two national samples collected two years apart under different labour regulations, and the interaction test establishes that the coefficients differ, not why.

Why the regional difference is the interesting part

The comparison paper's own hypothesis was that the association would differ in magnitude between the two regions given their contrasting institutional contexts [s2]. It did, in the predicted direction. The Korean study makes the same point from the inside, noting that evidence from South Korea — where long hours and low schedule autonomy are prevalent — has remained scarce despite those conditions [s1].

Taken together the two papers point at something a sleep clinic cannot fix. If schedule predictability is a modifiable exposure with a dose-response relationship to insomnia symptoms [s2], the lever is advance-notice rules and scheduling practice rather than individual sleep behaviour. The comparison paper puts it in those terms explicitly, describing habitual schedule unpredictability as a distinct occupational risk factor and schedule predictability as a potential occupational health priority warranting regulatory and organisational attention [s2].

What would settle it

Neither study can rule out reverse causation, and both say so. The design that would is a longitudinal cohort tracking workers through a change in scheduling practice — a regulatory advance-notice requirement, or an employer switching rostering systems — with insomnia measured before and after. Several jurisdictions have introduced predictive-scheduling rules in recent years, which makes that a tractable study rather than a hypothetical one.

Until then, what exists is a consistent association in two national datasets, a dose-response by advance notice, and a measured difference between two labour markets [s1][s2] — enough to justify the question, not enough to answer it.

Sources

• Association between unpredictable work schedules and insomnia symptoms among wage workers in South Korea: the sixth Korean Working Conditions Survey, BMC Public Health, 18 May 2026
• Association between unpredictable work schedule changes and sleep disorders among wage workers: evidence from national Working Conditions Surveys in Korea and Europe, Journal of Occupational Health, published online 31 July 2026