Nocturnal hypoxemia biomarker predicts sleepiness in patients with severe obstructive sleep apnea

Purpose This study aims to assess the association between excessive daytime sleepiness (EDS) and variables extracted from the pulse-oximetry signal obtained during overnight polysomnography. Methods A cross-sectional design was used to study the relation between four hypoxemia variables and EDS as determined by Epworth Sleepiness Scale scores (ESSS) in 200 consecutive patients, newly diagnosed with obstructive sleep apnea (OSA), as defined by an apnea-hypopnea index (AHI)>= 15. Hypoxemia measurements were compared between sleepy (ESSS >= 10) and nonsleepy (ESSS<10) patients before and after dichotomizing the cohort for each hypoxemia variable (and for AHI) such that there were 35 (165) patients in each of the corresponding higher (lower) subcohorts. The hypoxemia variables were combined into a biomarker, and its accuracy for predicting sleepiness in individual patients was evaluated. We planned to interpret prediction accuracy above 80% as evidence that hypoxemia predicted EDS. Results Hypoxemia was unassociated with sleepiness in OSA patients with AHI in the range of 15 to 50. In patients with AHI>50, the hypoxemia biomarker (but not individual hypoxemia variables) predicted sleepiness with 82% accuracy. Conclusion Nocturnal hypoxemia as determined by a polyvariable biomarker reliably predicted EDS in patients with severe OSA (AHI>50), indicating that oxygen fluctuation had a direct role in the development of EDS in patients with severe OSA.