Sleep apnea phenotypes based on multi-night assessment of sleep perception.

Abstract

Objectives: Sleep time discrepancy is associated with a lower quality of life and is common in individuals with obstructive sleep apnea (OSA) during polysomnography (PSG), but little is known about multi-night sleep perception patterns in the habitual sleeping environment. We conducted multi-night measurements using a validated wrist-worn photoplethysmography sensor and a digital sleep diary, to identify OSA phenotypes based on sleep perception patterns.

Method: Sleep time discrepancy was evaluated using the misperception index (MI). K-means clustering was used to identify sleep perception patterns using two features: the median and median absolute deviation of a participant's multi-night MI values. Clusters were compared across clinical and demographic parameters.

Results: Based on 1485 recording nights of 120 individuals with OSA we identified four sleep perception clusters: accurate estimation (n = 64), underestimation (n = 33, with more insomnia symptoms and lower self-reported sleep quality), severe underestimation (n = 6, similar to underestimation, with high pre-sleep cognitive arousal) and variable overestimation (n = 17, older age, with higher apnea-hypopnea index and more disrupted sleep).

Conclusions: Distinct OSA phenotypes can be identified based on multi-night sleep perception patterns, and they present with diverging clinical characteristics. This highlights the heterogeneity of the OSA population and may facilitate the development of more personalized treatment approaches.