Abstracts

Rationale: Epilepsy and sleep are in a bidirectional relationship. While not fully understood, it is known that epileptic seizures worsen sleep quality, whereas poor sleep heightens seizure risk. A better understanding of this interaction could result in an improvement of the quality of life of patients with epilepsy (PWE). Although polysomnography is the gold standard for sleep studies, it is not always readily accessible, making wearable devices appealing for long-term sleep monitoring. The aim of this study was to compare the sleep macrostructure of nights followed by seizures in the 24 hours post-awakening (pre-seizure nights) and nights not followed by seizures in the 24 hours post-awakening (non pre-seizure nights) in PWE using the Hexoskin smart shirt (SS).

Methods: Fifty-four PWE hospitalized at the University of Montreal Hospital Center (CHUM) Epilepsy Monitoring Unit were recruited to wear the SS during their stay. Six patients with nightly seizures were excluded. Using cardiac, respiratory and movement data, the SS algorithm calculated the following sleep metrics for each night: sleep efficiency (SE), sleep latency and time spent in Wake, NREM, and REM. Wakefulness After Sleep Onset (WASO), total sleep duration, and percentage of time in each stage (%NREM, %REM, %WASO) were then computed. The presence of nocturnal seizure (NS) was noted as a dichotomous variable. Seizures were determined by video-EEG recordings. A total of 81 pre-seizure nights (24 with NS) and 217 non pre-seizure nights (17 with NS) were analyzed. Three statistical analyses were performed between pre-seizure and non pre-seizure nights using the Mann-Whitney U-test with Bonferroni correction: 1) with all 298 nights, 2) with only the 41 nights with NS, and 3) with only the 257 nights without NS. A fourth statistical analysis was performed between nights with and without NS.

Results: Significantly lower SE (μ=90.5% < 93.0%) was observed in pre-seizure nights compared to non pre-seizure nights (p < 0.0056). Differences in %REM (μ=29.2% < 33.1%), WASO (μ=26.0min >17.2min) and %WASO (μ=5.3% >3.7%) between groups were not significant after correction. No significant difference was found when analyzing nights with and without NS separately. Nights with NS had significantly lower SE (μ=89.8% < 92.7%) and %REM (μ=27.2% < 32.7%), and greater WASO (μ=36.2min >17.1min) and %WASO (μ=7.1% >3.7%) than nights without NS (p < 0.0056). Furthermore, the occurrence of NS was significantly greater in pre-seizure nights (OR 4.95, 95% CI: 2.49 to 9.85).