0723 Slow wave activity surrounding K-complexes is associated with long-term all-cause mortality in a large community-dwelling cohort

Abstract

Abstract Introduction Sleep fragmentation is thought to be associated with adverse long-term consequences including mortality. However, conventional metrics of sleep fragmentation such as time spent in non-REM, REM sleep stages have been inconsistently associated with long-term consequences of sleep fragmentation. Recent studies suggest that morphological features of K-complexes might better describe sleep fragmentation as they are sensitive to effects of sleep disruption and better correlate with short-term outcomes such as daytime sleepiness. We previously demonstrated that the slow wave activity (SWA) surrounding K-complexes (∆SWAK) was a strong predictor of sleepiness in a cohort of sleep apnea subjects, mediated in part by its association with sleep fragmentation, however its association with long-term consequences of sleep fragmentation is not well known. Methods We analyzed nocturnal polysomnography (NPSG) data from Sleep Heart Health Study (1995-1998), a well-characterized cohort examining the cardiovascular consequences of sleep-disordered breathing. The primary quantitative EEG (C3-A2) metrics were: SWA (%relative power in 0.5-4 Hz), K-complex density (number/min of N2), and ∆SWAK. K-complexes were detected automatically using previously published open-source method (DETOKS, Parekh et. al., 2015) during stage N2 of sleep only. All EEG metrics were divided into quartiles to explore potential non-linear associations. Death from any cause up until 2011 (mean follow up of 11 years) was the primary outcome. Cox-regression models were used to assess the relationship between EEG metrics and all-cause mortality. Results After accounting for missing data, 3,909 NPSG’s with at least 6h of usable EEG were available for analysis (age=64±11 yrs., 53.4% female). In multivariate Cox-regression models adjusted for age, gender, race, severity of sleep apnea, total sleep time, time spent in stage N2, and smoking, all-cause mortality was significantly associated with the highest quartile of SWA (HR_q4=0.77 [0.6-0.9], p=0.01), highest two quartiles of ∆SWAK (HR_q3=0.8 [0.6-0.9], p=0.04; HR_q4=0.6 [0.5-0.8], p< 0.001). Conclusion Quantitative EEG measures predict mortality in a large community-dwelling cohort. Specifically, slow-wave activity surrounding K-complexes appear to be strongly associated with all-cause mortality beyond the effect of known covariates. Combined with our previous studies, our data suggest that slow-wave specific EEG measures are predictive of short- and long-term consequences of sleep fragmentation. Support (if any) AASM Foundation BS-233-20, NIH-K25HL151912, R21HL165320