0429 Hypnotic Use Does Not Increase the Risk of Incident Mild Cognitive Impairment/Dementia: A Retrospective Cohort Study

Abstract Introduction Observational studies have reported associations between hypnotic use and increased risk of cognitive decline. These findings may reflect confounding from insomnia severity, healthcare utilization, and comorbidity burden. We evaluated whether hypnotic exposure is independently associated with incident mild cognitive impairment (MCI) or dementia after rigorous adjustment, including a comprehensive propensity score (PS) and Apolipoprotein E (APOE) ε4 status. Methods We conducted a retrospective cohort study using a biobank sample with linked electronic health records and genetic data. Hypnotic exposure was derived from orders data and categorized as unexposed (n=27,055), single prescription (n=4243), shorter-term use (n=3999), long-term use (≥6 orders for ≥1 year; n=3214), and by drug class: benzodiazepines (n=376), Z-drugs (n=6222), trazodone (n=2395), and mixed exposure (≥2 classes; n=2435). Incident MCI/dementia diagnoses (ICD-9/10) occurring ≥12 months after initial prescription constituted outcomes. The PS modeled likelihood of receiving hypnotics based on demographics; sleep-related conditions; healthcare utilization (record length, visit frequency, diagnostic burden); psychiatric comorbidities; alcohol use and chronic pain. Hypnotic users were 1:2 matched to unexposed individuals on age, sex, healthcare utilization, and PS. Cox proportional hazards models with a 60-month lag tested associations of exposure category, duration, and class with outcomes. Models adjusted for all PS covariates, cardiovascular risk score, and APOE ε4 status. Results Matched but unadjusted analyses suggested slightly higher hazards of MCI/dementia among hypnotic users versus unexposed (HR=1.08; CI=0.99–1.18). However, after PS adjustment and APOE ε4 inclusion, hypnotic users showed significantly lower hazards (HR=0.81; CI=0.74–0.89). Class-specific analyses indicated reduced hazards for Z-drugs (HR=0.83; CI=0.74–0.93), trazodone (HR=0.71; CI=0.60–0.86), and mixed users (HR=0.79; CI=0.68–0.91), with all classes showing lower hazards than benzodiazepine-only users. Long-term hypnotic use was associated with lower hazards than no prescription (HR=0.69; CI=0.61–0.80), single prescription (HR=0.81; CI=0.68–0.97), or short-term use (HR=0.76; CI=0.65–0.90). Conclusion The elevated cognitive risk among hypnotic users appears largely attributable to underlying clinical/utilization factors rather than hypnotic exposure itself. After comprehensive confounding control, all hypnotic classes except benzodiazepines demonstrated lower risk of MCI/dementia. These findings underscore the need for rigorous adjustment in studies of insomnia pharmacotherapy and suggest potential protective effects of certain hypnotics that warrant further investigation. Support (if any) None