Application of cluster analysis in OSA population

OSA is increasingly being identified as an important health issue. It is typified by repeated episodes of upper airway collapse during sleep leading to occasional hypoxaemia, sleep fragmentation and poor sleep quality. OSA is also being considered as an independent risk factor for hypertension, diabetes, cardiovascular diseases, leading to increased multimorbidity and mortality. Cluster analysis, a powerful statistical set of techniques, may help in investigating and classifying homogeneous groups of patients with similar OSA characteristics. This study aims to investigate the (possible) different groups of patients in an OSA population, and to analyze the relationships among the main clinical variables in each group to better understand the impact of OSA on patients. Our results show that in a general population of subjects afferent to the sleep centre three different communities of OSA patients can be identified. The first has a very severe disease (AHI 65.91 ± 22.47), and sleep disorder has a strong impact on daily life: low level of diurnal PaO2 (77.39 ± 11.64 mmHg), high prevalence of Hypertension (64%). The second with less severe disease (AHI 28.88 ± 17.13) in which sleep disorders seems to be less important for diurnal PaO2 and have a minimum impact on comorbidity; and the last one with very severe OSA (AHI 57.26 ± 15.09) but with low risk of nocturnal hypoxemia (T90 11.58 ± 8.54) less sleepy (ESS 10.00 ± 4.77).