Purpose
To determine obstructive sleep apnea (OSA) phenotypes using cluster analysis including variables of sleep perception and sleep quality and to further explore factors correlated with poor sleep quality in different clusters.
Methods
This retrospective study included patients with OSA undergoing polysomnography (PSG) between December 2020 and April 2022. Two-step cluster analysis was performed to detect distinct clusters using sleep perception variables including discrepancy in total sleep time (TST), sleep onset latency (SOL), and wakefulness after sleep onset (WASO); objective TST, SOL, and WASO; and sleep quality. One-way analysis of variance or chi-squared tests were used to compare clinical and PSG characteristics between clusters. Binary logistic regression analyses were used to explore factors correlated with poor sleep quality.
Results
A total of 1118 patients were included (81.6% men) with mean age ± SD 43.3 ± 13.1 years, Epworth sleepiness score, 5.7 ± 4.4, and insomnia severity index 3.0 ± 2.4. Five distinct OSA clusters were identified: cluster 1 (n = 254), underestimated TST; cluster 2 (n = 158), overestimated TST; cluster 3 (n = 169), overestimated SOL; cluster 4 (n = 155), normal sleep discrepancy and poor sleep quality; and cluster 5 (n = 382), normal sleep discrepancy and good sleep quality. Patients in cluster 2 were older, more commonly had hypertension, and had the lowest apnea–hypopnea index and oxygen desaturation index. Age and sleep efficiency were correlated with poor sleep quality in clusters 1, 2, and 5, and also AHI in cluster 2.
Conclusion
Subgroups of patients with OSA have different patterns of sleep perception and quality that may help us to further understand the characteristics of sleep perception in OSA and provide clues for personalized treatment.
