Accuracy of WatchPat Portable Sleep Monitoring and Sleep Assessment in Patients with Atrial Fibrillation

Document Type

Conference Proceeding - Restricted Access

Publication Date


Publication Title



Introduction: Obstructive sleep apnea(OSA) is an established risk factor for atrial fibrillation(AF), necessitating early diagnosis and management. Home-based sleep-monitoring technology has become a mainstream diagnostic modality. Peripheral arterial tonometry (PAT) device is increasingly being used to screen for OSA in patients with AF. Our study aimed to examine the accuracy of Watch-PAT (WP) in OSA evaluation, and night-to-night variability of sleep characteristics as measured by WP when used during consecutive night sleep studies. Methods: Patients with history of AF undergoing clinically indicated PSG were prospectively enrolled and had concurrent WP while undergoing PSG. Patients then were studied again using WP over two consecutive days at home. We compared agreement of OSA severity(defined as no OSA[AHI (apnea hypopnea index)<5], mild OSA[15>AHI³5], moderate OSA[30>AHI³15], severe OSA[AHI>30]) and total sleep time(TST) using Cohen’s Kappa(K) for categorical and Bland-Altman plots for continuous variables. To further characterize PSG versus WP, the cohort was stratified into paroxysmal versus persistent AF types. 1A/1B hypopnea scoring criteria was defined as per AASM. Results: Our cohort included 24 patients with AF(80% male, mean age 68y). Most patients had clinically defined OSA(AHI³5). Patients with persistent AF had more severe OSA than those with paroxysmal AF (severe OSA present in 60% vs. 29%). Comparison of PSG to concurrently conducted WP in the lab showed substantial agreement in OSA severity by both 1A(K=0.623) and 1B(0.706) criteria. Percent difference in TST between PSG versus WP in the paroxysmal AF versus persistent groups was not statistically significant(p=0.387). Comparing two consecutive at-home WP tests showed substantial agreement in OSA severity measures(1A=0.872,1B=0.889). Bland-Altman plots for TST and sleep architecture showed ³95% of residuals within 2 standard deviations, suggesting intertester agreement. Confidence intervals were broad in these plots reflecting our small sample size. Conclusion: Our findings demonstrate that WP is a reasonable alternative to PSG, particularly if using the 1B criteria to diagnose OSA. Additionally, our results show that persistent versus paroxysmal AF does not seem to affect the results of WP tests. Night to night variability of OSA measures and TST was small. Future studies should verify the results of our study in a larger cohort.




Supplement 1

First Page


Last Page