Characterisation of nocturnal arrhythmia avalanche dynamics: Insights from generalised linear model analysis

Nocturnal arrhythmia avalanche (NAA) episodes, characterised by transient non-sustained cardiac arrhythmias during sleep, have been demonstrated as a predictor of adverse cardiovascular events. However, their dynamics and association with sleep architecture and events remain unclear. While generalised linear models (GLM) have captured sleep-disordered breathing (SDB) dynamics, their application to NAA remains underexplored. This study explored whether changes in sleep architecture contribute to nocturnal arrhythmias and if the impact of sleep stages, SDB, and arousal events on these arrhythmias varies by demographic factors. We analysed 7341 ECG recordings from the multi-ethnic study of atherosclerosis (MESA) and the sleep heart health study (SHHS) datasets. R-R intervals were divided into 10-min periods to detect NAA, defined as a 30% drop from baseline followed by recovery to 90% of baseline. A GLM framework was developed to characterise NAA episodes as functions of SDB, sleep arousal events, sleep stages, and prior NAA episodes. The GLM analysis revealed that NAA occurrence was 18% and 30% higher during non-rapid eye movement (NREM) light sleep compared with deep sleep in SHHS (p < 0.001) and MESA (p < 0.001), respectively. SDB events increased the NAA risk in 34% of participants, and arousals in 29%. In SHHS, the impact of SDB on NAA was 5% greater in men (p = 0.018), while the arousal effects were more pronounced in those over 75, highlighting the role of demographic factors in modulating arrhythmia risk. These findings demonstrate the utility of the GLM framework in modelling the dynamics of nocturnal arrhythmias and their associations with sleep disruptions and architecture.