Sleep Apnea, Insomnia, and Dementia Risk: What We Know

Persistent insomnia and obstructive sleep apnea deserve more attention than the usual advice to "get better sleep." Prospective studies link both conditions with a higher later risk of dementia and cognitive decline. But the evidence supports a warning-light interpretation, not a guarantee that treating a sleep disorder will prevent dementia.

A 2025 meta-analysis in GeroScience pooled 39 cohort studies and found that people with insomnia or obstructive sleep apnea (OSA) had higher rates of several later cognitive outcomes. Those are meaningful associations. They are not proof that disturbed sleep causes neurodegeneration, or that a CPAP machine and cognitive behavioral therapy for insomnia can erase dementia risk.

That distinction is more than academic. Sleep disorders are treatable, often miserable, and tied to cardiovascular and metabolic problems that also affect the brain. They warrant clinical care on their own merits. The dementia-prevention sales pitch, however, has outrun the trials.

Both sleep disorders tracked with later cognitive trouble

In the meta-analysis, OSA was associated with a 33% higher relative rate of all-cause dementia (pooled hazard ratio [HR], 1.33), a 45% higher rate of Alzheimer's disease (HR, 1.45), and a 44% higher rate of cognitive decline (HR, 1.44). OSA did not show a statistically significant association with vascular dementia: the pooled HR was 1.35, but the 95% confidence interval ranged from 0.99 to 1.84 and included 1.0.

Insomnia showed similar or larger associations: HR 1.36 for all-cause dementia, 1.49 for Alzheimer's disease, 1.59 for vascular dementia, and 1.63 for cognitive decline. These are relative estimates, not an individual forecast. The studies do not provide one universal absolute risk increase, because participants, follow-up periods, sleep-disorder definitions, and dementia assessments differed across cohorts.

The direction of the findings matters. Both conditions repeatedly appeared before later cognitive outcomes in prospective research, which makes the relationship harder to dismiss as a cross-sectional coincidence. Yet prospective observation still cannot fully separate cause from company kept: obesity, diabetes, depression, cardiovascular disease, sedative use, socioeconomic conditions, and differences in medical detection can all affect both sleep and brain-health outcomes.

The pooled estimates come with substantial variation

Meta-analysis can summarize a large literature, but it cannot turn unlike measurements into identical ones. The 2025 review combined self-reported symptoms, questionnaires, clinical diagnoses, administrative records, and objective sleep testing. Dementia outcomes also came from mixed methods, including formal diagnostic criteria and medical records.

Several results varied sharply across studies. Heterogeneity was 93% for OSA and all-cause dementia, 77% for insomnia and all-cause dementia, 93% for insomnia and vascular dementia, and 66% for OSA and cognitive decline. High heterogeneity means the studies did not cluster neatly around a single effect size. It may reflect real differences in age, disease severity, comorbidity, follow-up, measurement, or adjustment for confounders.

The review also found signals of small-study effects for OSA and Alzheimer's disease and for insomnia and cognitive decline. That does not erase the overall pattern, but it limits confidence in any precise percentage. The sensible reading is that these disorders identify populations with elevated later cognitive risk, not that every person with snoring or a rough month of sleep has acquired a fixed dementia trajectory.

Objective sleep studies support OSA as a marker, with limits

Self-report can blur a sleep question. A 2023 JAMA Network Open consortium analysis addressed that problem by pooling five cohorts that used polysomnography, the overnight sleep study used to measure OSA. The analysis included 5,946 adults, with mean cohort ages ranging from 58.5 to 89.1 years.

Over follow-up within five years, prevalent OSA was associated with poorer global cognition. Compared with an apnea-hypopnea index (AHI) below 5, both AHI of at least 5 and AHI of at least 15 had pooled beta coefficients of -0.06 for global cognition. The result supports the broader risk-marker story when OSA is measured objectively rather than inferred from symptoms alone.

It also blocks an overly tidy mechanism. In the same analysis, sleep-stage percentages and time spent below 90% oxygen saturation were not associated with global cognition. OSA is not one number wearing a trench coat. Repeated airway obstruction, sleep fragmentation, intermittent hypoxia, vascular health, inflammation, and the conditions that travel with OSA may all contribute differently across people.

Biology makes the link plausible, not proven

Several human studies connect OSA-related physiology with pathways relevant to cognitive decline. In older adults with sleep apnea, a 2023 Neurology study linked polysomnographic sleep abnormalities with neuroimaging biomarkers of cerebrovascular disease. Vascular brain injury offers one plausible route between disordered sleep and later cognitive impairment.

A 2022 study in Journal of Sleep Research associated intermittent hypoxia and sleep disruption in adults with OSA with higher serum tau and amyloid-beta levels. A separate 2023 study connected amyloid burden, hypoxemia, sleep architecture, and cognition in OSA. These findings are biologically interesting, but tau, amyloid-beta, imaging changes, and blood markers are not dementia diagnoses. An association with a biomarker cannot establish that changing the biomarker will prevent a patient outcome.

The picture grows more complicated in a 2026 study of 299 non-demented OSA patients. The investigators associated unstable sleep architecture and intermittent hypoxia with cognitive impairment and examined inflammatory and Alzheimer's-related biomarkers. Another study of 317 people without dementia linked neuronal glutamate-transporter measures to cognitive impairment among people with OSA. These studies widen the list of plausible pathways; they do not establish the direction of causation.

Could early brain disease be disturbing sleep?

Reverse causation remains a serious possibility. Neurodegeneration can alter sleep years before a dementia diagnosis, particularly through fragmented sleep and insomnia symptoms. If early brain changes cause some sleep disruption, sleep can function as a prodromal marker rather than an upstream driver.

A 2021 prospective study found that sleep profiles predicted cognitive decline over 12 months in 125 patients who already had mild to moderate Alzheimer's disease. That study does not answer whether sleep problems cause dementia in cognitively intact adults. It does show that sleep characteristics can track disease trajectory after Alzheimer's disease is established, which makes reverse causation a live explanation rather than a technical footnote.

Treat the disorder; do not promise dementia prevention

Standard treatment still makes sense. CPAP can improve executive function and memory in OSA, and cognitive behavioral therapy for insomnia can improve sleep quality and reduce systemic inflammation. Those benefits matter to daily functioning and health. They do not answer the larger question of whether treatment reduces incident dementia decades later.

A 2025 systematic review of multidomain prevention trials supports early, tailored risk reduction across the Alzheimer's disease continuum. It does not provide direct randomized proof that treating OSA or insomnia alone lowers dementia incidence. Long-term trials with dementia or durable cognitive decline as outcomes would change the conclusion most.

Persistent insomnia and likely OSA are credible brain-risk signals. Current evidence supports evaluation and standard treatment for the sleep disorder itself, while dementia prevention remains an unproven bonus.

If you snore loudly, stop breathing during sleep, wake unrefreshed, struggle with daytime sleepiness, or have insomnia that persists, bring it to a clinician rather than treating it as a personality trait. The evidence gives a solid reason to take the symptom seriously. It does not yet justify a promise that fixing sleep will prevent dementia.