Chronic insomnia is prevalent in older adults, affecting up to 70%, and has been linked in prior studies to cognitive decline and dementia. However, results are inconsistent, partly due to confounders such as obstructive sleep apnea (OSA), psychiatric comorbidities, and hypnotic use. Emerging evidence suggests that insomnia combined with short sleep duration may represent a particularly high-risk phenotype. Yet, few longitudinal, population-based studies have examined insomnia in relation to both cognitive outcomes and brain biomarkers of neurodegeneration and cerebrovascular disease. This study aimed to determine whether chronic insomnia predicts cognitive decline and increased risk of cognitive impairment (CI), and whether it is associated with neuroimaging markers—amyloid burden (amyloid-PET) and white matter hyperintensities (WMH).

Data were drawn from the Mayo Clinic Study of Aging, a large, population-based longitudinal cohort of older adults in Olmsted County, Minnesota. Eligible participants were cognitively unimpaired adults =50 years with at least two valid neuropsychological assessments and complete covariate data. A sample of 2,750 participants was included in mixed-effects models (mean age 70.3, 49% female), 2,814 in Cox models, 1,027 with MRI WMH data, and 561 with amyloid-PET data. Insomnia was defined using repeated ICD-9/10 diagnostic codes =30 days apart. Sleep duration changes were based on self-reported BDI-II responses (“reduced,” “more,” or “no change”). Annual neuropsychological testing produced global z-scores; consensus panels determined cognitive status (normal, mild cognitive impairment, dementia). Imaging outcomes included serial amyloid-PET scans (centiloid units) and MRI-derived WMH (percent of intracranial volume). Mixed-effects regression and Cox proportional hazards models were conducted to assess associations between insomnia, cognitive outcomes, and imaging biomarkers, adjusting for demographic, vascular, psychiatric, genetic (APOE e4), and sleep-related factors. Exploratory analyses tested hypnotic use.

Chronic insomnia was associated with faster decline in global cognition (β = -0.011 per year; p = 0.028). Insomnia conferred a 40% higher risk of incident CI (HR 1.40, 95% CI 1.07–1.85) and reduced sleep was linked to worse baseline cognitive performance (β = -0.21; p = 0.012). Insomnia with reduced sleep correlated with higher baseline WMH burden (β = 0.147; p = 0.005) and greater amyloid-PET load (ß = 10.5 centiloid; p = 0.039). Insomnia with increased sleep duration (possibly remission) was associated with lower WMH burden (β = -0.142; p = 0.028). Insomnia was not linked to rates of WMH or amyloid accumulation over time. Hypnotic use was not significantly associated with cognitive decline, CI risk, WMH, or amyloid burden overall. Exploratory analyses hinted at possible protective effects in subgroups with reduced sleep, but results were inconsistent.

In this large, longitudinal, community-based study, chronic insomnia independently predicted accelerated cognitive decline and greater risk of incident CI in older adults. Those with insomnia and reduced sleep exhibited the worst baseline brain health, marked by poorer cognition, higher WMH burden, and elevated amyloid deposition, highlighting this subgroup as particularly vulnerable. Conversely, sleeping more than usual among those with insomnia was associated with lower WMH, suggesting potential resilience or remission effects.

These findings underscore the clinical importance of recognizing and treating chronic insomnia—especially in patients reporting short sleep—as part of strategies to preserve cognitive health. While hypnotics showed no clear adverse or protective associations, nonpharmacologic approaches such as cognitive behavioral therapy for insomnia (CBT-I) may be critical in mitigating long-term neurocognitive risks.

Reference:

Carvalho DZ et al. Neurology 2025;105(7):e214155. Abstract