0739 Does Benzodiazepine Receptor Agonist Use, Insomnia, and OSA Explain Variation in Cognitive Performance?

Abstract Introduction Older adults have multiple risk factors for cognitive impairment including insomnia, benzodiazepine receptor agonist (BZRA) use, and obstructive sleep apnea (OSA). Simultaneously addressing all risk factors is ideal but impractical. Clarifying the relative contribution of each risk factor could help prioritize strategies to reduce risk of cognitive impairment. We sought to quantify the effects of sleep risk factors on cognitive performance above and beyond other common risk factors for cognitive impairment. Methods We used clinical trial baseline data from outpatients taking a BZRA hypnotic. In four nested regression analyses, we modeled predictors of Mini-Mental State Examination (MMSE), Trail Making Test (TMT) A, TMT B, and Digit Symbol Substitution Test (DSST). Base Models included age, sex, ethnicity, race, education, vascular risk, depression, traumatic brain injury, and alcohol use. Nested models were: Model 2 = base model + Insomnia Severity Index score (ISI); Model 3 = Model 2 + BZRA hypnotic use; Model 4 = Model 3 + OSA risk; and Model 5 = Model 4 + OSA-BZRA interaction term. We compared changes in R-squared values and used F-tests to examine the fit of nested models. Results Of 348 participants (mean age 68.8), 313 were included for MMSE (m=28.2), 298 for TMT-A (m=48.2 seconds), 287 for TMT-B (m=130.2 seconds), and 285 for DSST (m=8.0) analyses. Mean ISI was 14.2 (SD 6.6). Mean 7-day BZRA total dose was 29.5 diazepam equivalents milligrams (SD 31.4). 44% were high-risk for OSA. Base Model R-squared was 0.139 (MMSE), 0.241 (TMT-A), 0.229 (TMT-B), and 0.189 (DSST). Change in R-square from Model 1 to Model 5 was 0.003 (MMSE), 0.005 (TMT-A), 0.010 (TMT-B), and 0.001 (DSST), with p-values >.145 for F-tests comparing changes in R-squared. Change in R-squared for Models 2 to 3, 3 to 4, and 4 to 5 were not statistically significant. Conclusion Although our base models account for more than 13% of the variation in cognitive performance, sleep variables did not significantly explain additional variation in cognitive performance. These results suggest that in older adults using BZRAs, interventions targeting common risk factors should be prioritized to address cognitive impairment. Support (if any) VAIIR 17-234, NIAR01AG057929, UL1TR001881