0757 Cluster Analysis for Identifying Good CPAP Adherence Using the PSG Parameters and Patient Characteristics. (25th May 2022)

• Record Type: Journal Article
• Authors: Hamada, Eriko; Yamauchi, Motoo; Fujita, Yukio; Ikegami, Azusa; Shirahama, Ryutaro; Takaoka, Toshio; Nishijima, Tsuguo; Yoshikawa, Masanori; Muro, Shigeo
• Is Part Of: Sleep. Volume 45(2022)Supplement 1
• Page Start: A329
• Page End: A330
• Publication Date: 2022-05-25
• Subjects: Sleep -- Physiological aspects -- Periodicals; Sleep disorders -- Periodicals; Sommeil -- Aspect physiologique -- Périodiques; Sommeil, Troubles du -- Périodiques; Sleep disorders; Sleep -- Physiological aspects; Sleep -- physiological aspects; Sleep Wake Disorders; Psychophysiology; Electronic journals; Periodicals; 616.8498
• DOI: 10.1093/sleep/zsac079.753
• Languages: English
• ISSNs: 0161-8105

Abstract

Abstract: Introduction: CPAP is the standard treatment for obstructive sleep apnea (OSA). One of the important clinical issues to be solved is poor CPAP adherence. A growing body of studies has identified predictive factors for CPAP adherence including AHI, BMI, age, gender, symptoms, etc. When our sleep physicians prescribe CPAP, we would consider these known factors in multiple ways. One may want to know factors' combinations rather than each factor. In this case, cluster analysis might be useful since it is a powerful data-mining tool to sort various factors into meaningful groups. Recently, cluster analysis has been adopted for research of sleep breathing disorders. However, no one has adopted to predict CPAP adherence. In this study, we aimed to explore the usefulness of cluster analysis to predict CPAP adherence using the diagnostic PSG parameters and patients' characteristics. Methods: The study design was a retrospective observational multi-center study including 5 certified sleep centers in Japan. For 2 years from 2017, 1133 patients who were diagnosed with OSA with in-lab PSG and newly initiated CPAP therapy were enrolled. We performed cluster analysis using the K-means clustering. Variables for clustering were determined by several sleep physicians among PSG parameters and patients' characteristics. We assessed CPAP adherence for 90 days and 365 days after CPAP initiation in each created cluster. We adopted CMS criteria for good CPAP adherence, which is, more than four hours of use on 70% of nights. Results: Cluster analysis classified 5 clusters. A significant difference in CPAP adherence for 90 days and 365 days was seen among 5 clusters with a test of independence (p=0.001, p=0.005, respectively). The cluster presenting moderate obese, very high AHI and ODI, and apnea predominant indicated good adherence, whereas the cluster presenting morbid obese, very high AHI and ODI, sustained severe hypoxia, younger age, and daytime sleepiness indicated poor adherence according to the post-hoc Chi-square test. Conclusion: Cluster analysis successfully distinguished the different CPAP adherence and identified a combination of OSA patients' profiles. Thus, cluster analysis would be a useful tool for predicting long-term CPAP adherence. Support (If Any):