Sleep Apnea Linked to Poorer Memory and Higher Dementia Risk in Middle Age

Obstructive sleep apnea affects millions of adults worldwide and has long been suspected as a contributor to cognitive decline and dementia. Yet evidence from community-based samples has been inconsistent, partly because most prior studies used insensitive cognitive tests, failed to control adequately for vascular confounders, or focused on older or clinically recruited populations. This study addressed those gaps by examining a large, well-characterized cohort of middle-aged Australians at elevated dementia risk.

The Healthy Brain Project enrolled 2,795 cognitively unimpaired adults aged 40–70 residing in Australia, recruited between 2016 and 2022. OSA status was determined by self-reported clinical diagnosis, with participants classified as OSA+ (n = 382, approximately 13.7%) or OSA−. Cognition was measured with the Cogstate Brief Battery — a validated online tool comprising four subtests assessing psychomotor speed, visual attention, working memory, and visual learning — organized into composite attention and memory scores. Dementia risk was quantified using the CAIDE score (0–15), which incorporates age, sex, education, BMI, physical activity, hypertension, and hypercholesterolemia. A modified CAIDE (0–7) isolating only the four modifiable vascular risk factors was also computed. APOE genotyping was available for a subset of participants.

Participants with OSA were significantly older (mean 57.1 vs. 54.7 years), more likely to be male (72% vs. 42%), had higher BMI (30.2 vs. 26.5 kg/m²), greater rates of hypertension (41% vs. 19%) and hypercholesterolemia (43% vs. 28%), lower physical activity, and worse sleep quality across multiple measures including the Pittsburgh Sleep Quality Index, Epworth Sleepiness Scale, and Insomnia Severity Index (all p < 0.001). In unadjusted models, OSA+ participants performed significantly worse on the memory composite (Cohen's d ≈ 0.15–0.20; p < 0.05) but not the attention composite. After covariate adjustment (age, sex, education, treatment status), the memory difference remained significant. However, when vascular risk factors were additionally controlled, the OSA-memory association was attenuated to non-significance, suggesting that shared vascular burden explains a meaningful portion of the cognitive effect.

For dementia risk, OSA+ participants had significantly higher CAIDE scores than OSA− participants even after adjustment, and this held for the modified CAIDE isolating modifiable vascular factors (p < 0.001). Planned comparisons across four APOE × OSA subgroups (OSA−/ε4−, OSA+/ε4−, OSA−/ε4+, OSA+/ε4+) showed that OSA+/ε4− and OSA+/ε4+ groups both had significantly higher CAIDE scores than OSA−/ε4−, but the OSA × APOE ε4 interaction was non-significant for both cognition and CAIDE, indicating APOE ε4 does not synergize with OSA in this midlife sample.

These findings carry important practical implications. OSA is a modifiable condition — treatable with CPAP and lifestyle changes — and its association with elevated dementia risk through vascular pathways (obesity, hypertension, dyslipidemia, inactivity) suggests that treating OSA could simultaneously address multiple dementia risk factors. The authors argue for routine OSA screening in primary care settings during midlife, well before clinical cognitive decline becomes apparent, as this represents the most promising window for dementia prevention. However, the cross-sectional design precludes causal inference, and self-reported OSA diagnosis lacks the precision of polysomnography.