Sleep apnea plus insomnia raised hypertension risk — but only in one insomnia subtype

Insomnia and obstructive sleep apnea co-occur far more often than chance would predict, and the combination — abbreviated COMISA in the sleep literature — has been associated with worse cardiovascular outcomes than either condition alone. What has been unclear is whether that holds for insomnia in general, or only for a particular version of it.

A prospective analysis published in the Journal of Clinical Sleep Medicine on January 9 argues for the second [s1].

What was done

The study drew on the Penn State Adult Cohort. Of 1,741 participants, 1,395 were reassessed after 7.5 years, and 786 of those were free of hypertension at baseline and therefore eligible for the incidence analysis [s1].

The measurement design is what distinguishes this from most work on the question. Obstructive sleep apnea was defined by polysomnography at five or more events per hour, and sleep duration was measured objectively in the laboratory, with short sleep defined as under six hours [s1]. Insomnia symptoms were defined as chronic complaints lasting a year or more, or difficulty with sleep initiation, maintenance, or quality [s1]. Hypertension was defined by self-reported treatment for high blood pressure [s1].

That combination — subjective symptoms, objective sleep duration, objective apnea — is what allows the study to split insomnia into phenotypes rather than treating it as one exposure.

The result depends entirely on the split

The highest risk of incident hypertension was in participants with sleep apnea plus insomnia with objectively short sleep duration: odds ratio 4.25 (95% CI 1.52–11.90) [s1].

Sleep apnea alone carried an odds ratio of 3.31, and insomnia with short sleep alone 2.27 [s1]. Insomnia with normal objective sleep duration showed no significant association with incident hypertension [s1].

The authors' conclusion is that the additive effect of COMISA on hypertension risk is associated with the short-sleep phenotype, which they describe as the most severe biological phenotype of insomnia, and that objective sleep duration measurement would improve how COMISA is diagnosed and treated [s1].

What the numbers can and cannot support

The confidence interval on the headline estimate runs from 1.52 to 11.90 [s1]. That is a wide interval, and it is wide because the subgroup with both conditions and objectively short sleep is small. The interval excludes 1, so the association is unlikely to be chance alone; it does not support treating 4.25 as a precise multiplier.

Hypertension was ascertained by self-reported treatment rather than measured blood pressure [s1]. That will miss undiagnosed and untreated hypertension, and the direction of the resulting bias is not obvious — undiagnosed hypertension could be differentially distributed across sleep phenotypes.

This is an observational cohort. The design establishes temporal order — participants were free of hypertension at baseline and developed it over 7.5 years — which is more than a cross-sectional survey can offer. It does not establish that treating the sleep disorder would lower the risk. Confounding by obesity is the obvious concern in any sleep apnea analysis, and the abstract's summary does not detail the adjustment set.

Single-cohort findings also need replication. The Penn State Adult Cohort is a well-characterised population sample with laboratory polysomnography, which is precisely why it can address this question, and also why the result has not been easy to reproduce elsewhere: most large sleep datasets lack objective sleep duration.

Why the phenotype distinction matters

If the finding holds, it changes what a clinical question about insomnia is for. Asking a patient whether they have trouble sleeping does not distinguish the two groups this study separates, and the two groups had substantially different hypertension risk — one elevated, one not significantly so [s1].

It also complicates the interpretation of trials. Insomnia treatment trials that recruit on symptoms alone will mix phenotypes, and if cardiovascular risk is concentrated in one of them, an average treatment effect across both will understate the effect in the group that carries the risk.

The practical obstacle is that objective sleep duration currently requires a laboratory night or a validated equivalent. Whether consumer or clinical-grade home devices can identify the short-sleep phenotype accurately enough to be used this way is an open question, and an important one for whether this finding can be acted on at scale.

What to watch

Replication in other cohorts with objective sleep duration; whether the short-sleep insomnia phenotype predicts other cardiovascular endpoints in the same way; and whether trials of insomnia or apnea treatment stratified by phenotype show differential effects on blood pressure.

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

Sources

• [s1] Athanasiou N, Pejovic S, Vgontzas AN, Fernandez-Mendoza J, Li Y, Karataraki M, Bixler EO, Obstructive sleep apnea comorbid with insomnia symptoms and objective short sleep duration is associated with incident hypertension, Journal of Clinical Sleep Medicine, 2026;22(1):11, published 2026-01-09.