Polysomnographic Markers of Clinical Sleep Apnea Outcomes
2016 Strategic Research Grant
Robert Thomas, MD
Beth Israel Deaconess Medical Center
Key Project Outcome
A common problem in care of patients with obstructive sleep apnea (OSA) is that use of continuous positive airway pressure (CPAP) is inadequate. An important question is if this can be predicted-who is at risk and why, of not using treatment well. It is now known that there are features of sleep and sleep breathing that are different person to person.
This grant targeted the predictors of use and efficacy of CPAP, derived from the polysomnograms (sleep studies) using several methods of data analysis. The types of analysis include (based on mathematical analysis) of sleep depth, rhythm of breathing, interaction of breathing and heart function, and the severity/duration of brief awakenings (arousals). Treatment was followed for a period of one year using information from the CPAP machines. In total over 700 patients were analyzed, including 350 with atrial fibrillation. The following has been found so far: 1) The accuracy of automatic detection of abnormal breathing during CPAP treatment by the CPAP machines is poor, missing about 60% of the abnormality. 2) Central apneas during the laboratory sleep study (pauses of breathing effort) predict high degrees of apnea on treatment. 3) Stable and unstable breathing was also scored, and was related to central apneas during laboratory treatment. 4) Arial fibrillation patients have poor sleep quality relative to those who do not, and do worse (less use, more persistent apnea on treatment) than those without. 5) When a special type of PAP was used, called an adaptive ventilator, patterns that predicted success or failure which is not detected by usual scoring methods was described. 6) Two new methods of assessing sleep-breathing were evaluated and fine-tuned, using cardiopulmonary coupling and a self-similarity measure. 7) A website through the Clinical Data Automation Center at the Massachusetts General Hospital (CDAC, http://cdac.mgh.harvard.edu/) will go online by the end of 2020, which will allow anyone to upload sleep data and perform the types of analysis we have used, and other new types of analysis which any developer/researcher may provide. 8) Several more detailed and unique types of analysis will be applied to the data in the near future. These results should improve the precision of treatment of sleep apnea.
Journal Articles
SLEEP MEDICINE
SLEEP
Algorithm for automatic detection of self-similarity and prediction of CPAP failure
ANNALS OF THE AMERICAN THORACIC SOCIETY
Automated apnea hypopnea index from oximetry and spectral analysis of cardiopulmonary coupling