Night-time insomnia predicted no extra heart risk without daytime symptoms

Most of what is known about sleep and heart disease comes from studies that measured sleep once. A cohort analysis published in Sleep Medicine on September 22 measured it repeatedly, across eight years, in 24,223 Swedish adults — and reports that the association depends on a distinction single-timepoint studies mostly cannot make [s1].

Night-time insomnia symptoms, on their own, were not associated with higher cardiovascular risk. They were only associated with it when daytime symptoms came with them [s1].

What was done

The study used the Swedish Longitudinal Occupational Survey of Health, with biennial follow-up between 2010 and 2018 [s1]. Sleep characteristics were self-reported at each wave. Cardiovascular outcomes were identified by linkage to national registers rather than by self-report, which removes one common source of error in this literature [s1].

Cox proportional hazards models estimated associations between sleep variables and incident cardiovascular disease; mixed-effects models described how sleep duration changed over time [s1]. During follow-up, 1,687 participants developed a cardiovascular outcome [s1].

The findings

Night-time insomnia was not associated with increased cardiovascular risk unless it was accompanied by daytime symptoms. With daytime symptoms present, the hazard ratio was 1.22 (95% CI 1.03 to 1.44), against a reference group with neither insomnia nor daytime symptoms [s1].

Long sleep duration of more than eight hours followed the same pattern: associated with higher risk only in combination with daytime symptoms, hazard ratio 1.35 (95% CI 1.13 to 1.61), referenced against six to eight hours of sleep with no daytime symptoms [s1].

The trajectory analysis is the study's distinctive contribution. Among participants who were long sleepers at baseline, those who went on to develop cardiovascular disease showed a gradual increase in sleep duration over time (β for CVD >8 h × time = 0.06, 95% CI 0.04 to 0.07; β × time² = −0.005, 95% CI −0.01 to 0.00), while long sleepers who stayed free of cardiovascular disease had stable or declining sleep duration [s1].

How this sits against the existing literature

A 2024 meta-review and meta-analysis combining four meta-analyses of observational studies with nine Mendelian randomisation studies concluded that insomnia is an independent risk factor for several cardiovascular diseases, and reported Mendelian randomisation odds ratios including 1.14 for coronary artery disease (95% CI 1.10 to 1.19), 1.16 for hypertension (95% CI 1.13 to 1.18), and 1.09 for any ischaemic stroke (95% CI 1.03 to 1.14) [s2].

Those estimates and this cohort's are not in direct conflict, but they are asking different questions. The genetic instruments used in Mendelian randomisation studies of insomnia are built from symptom-based definitions that do not separate night-time complaint from daytime impairment. If the Swedish analysis is right that the risk sits with the combined phenotype, then a pooled estimate over an undifferentiated insomnia exposure will land somewhere between the two subgroups — which is roughly where the meta-analytic odds ratios sit relative to this study's 1.22 [s1][s2].

What the design supports, and what it does not

The direction of causation is the open question, and the authors treat it as such. Their own interpretation is that daytime symptoms, particularly alongside prolonged or lengthening sleep, may reflect early physiological changes preceding cardiovascular disease [s1]. That framing makes the sleep pattern a marker of incipient disease rather than a cause of it — which would be consistent with the trajectory finding, since rising sleep need is a plausible early consequence of declining cardiac function.

Reverse causation of this kind is very hard to exclude in an observational cohort even with eight years of repeated measurement. The study cannot say whether treating insomnia would change cardiovascular risk.

Sleep was self-reported throughout, with no actigraphy or polysomnography [s1]. Self-reported sleep duration is a poor proxy for measured sleep duration, and the error is not random.

The cohort is a Swedish working population, so the age range and employment status are restricted relative to the general population, and the findings may not carry to retired or unemployed adults.

What it changes in practice

Not much on its own, and that is worth saying plainly. What it does is sharpen a research question: insomnia trials and cohort studies that recruit on night-time complaints alone are mixing two groups that this analysis suggests carry different cardiovascular risk. Separating them costs little — daytime symptoms are already part of the diagnostic definition of insomnia disorder — and may explain some of the inconsistency across previous studies.

What to watch

Replication in cohorts with objective sleep measurement; whether the lengthening-sleep trajectory predicts cardiovascular events independently of baseline cardiac markers; and whether trials of insomnia treatment stratified by daytime symptom burden show differential effects on cardiovascular endpoints.

This article describes observational cohort research and is informational only. It is not medical advice and does not recommend any test or treatment.