Cardiovascular implications of sleep characteristics using real-world objective sleep data

点击翻译按钮获取中文摘要

Project Summary Cardiovascular disease (CVD) is the leading cause of mortality in the U.S. Impaired sleep is recognized as a strong risk factor of CVD. Sleep disorders such as obstructive sleep apnea (OSA), insomnia, abnormal sleep duration, and poor sleep quality have each been associated with CVD-related morbidity and mortality. Major limitations of existing studies on sleep and CVD is the lack of objective sleep measurement and the lack of understanding of the multi-dimensional nature of sleep and its complex interactions with CVD. In addition, emerging evidence also suggests potential relationships between sleep disorders and preclinical CV conditions, which often occur before the clinical manifestation of CVD. Blood pressure parameters including systolic blood pressure variability (SBPV) and mean systolic blood pressure (SBP) are examples of preclinical CVD that have prognostic value for future CV events. However, no research has yet explored if sleep disorders beyond OSA (such as impaired sleep quality and abnormal sleep duration) are risk factors attributable to preclinical CV conditions. This is a critical area of inquiry since understanding the complex relationships between sleep and preclinical and clinical CV conditions will allow healthcare providers to implement targeted interventions to reduce CVD. The proposed study will examine whether PSG-derived objective measures of sleep obtained in the clinical setting would be predictive of CVD and preclinical CVD. Given that hypertension is one of the major important CV risks and has been most well studied CV risk factor in relation to sleep, SBPV, mean SBP, and other blood pressure metrics will be the focus of the preclinical CVD. Our main aim is to examine the relationships between multidimensional sleep characteristics (in terms of duration, efficiency, quality, and disordered sleep breathing) and clinical CV conditions, after adjusting for personal, clinical, and other confounders. We will use sleep data collected from more than 7,000 individuals who completed a diagnostic sleep study at the University of Virginia Health System in 2010 - 2018. Machine learning (ML) models will be used to analyze the multidimensional PSG measures. This proposed study represents the largest real-world dataset on objective sleep measures, which will allow us to simultaneously examine the entire spectrum of sleep and advance our understanding about the impact of sleep on CV outcomes.