Feasibility and data quality of long-term wearable-based heart rhythm monitoring in atrial fibrillation patients with obstructive sleep apnea: data from the pre-treatment phase of the STAROSA study

Background - Obstructive sleep apnea (OSA) is a prevalent risk factor among atrial fibrillation (AF) patients and may influence AF burden. Long-term heart rhythm monitoring remains challenging in clinical practice. The STAROSA study integrates a structured OSA screening and treatment program with semi-continuous rhythm monitoring to assess the impact of OSA and its treatment on AF episodes and burden. Purpose - This analysis evaluates the feasibility, data quality, AF detection performance and patient satisfaction of smartwatch-based rhythm monitoring combined with smartphone-based spotchecks in AF patients included in the STAROSA study. Methods – Symptomatic AF patients participated in at-home screening for clinically relevant OSA using polygraphy. Patients with a positive OSA screening entered a 3-month pre-treatment period of heart rhythm monitoring and were referred for polysomnography. Photoplethysmography-based rhythm monitoring was performed automatically every 9 minutes for 1 minute via a smartwatch. Patients also performed 2 daily manual spotchecks with their smartphone, and in case of symptoms. The recordings were classified as green (sinus rhythm), red (suggestive of AF), orange (mild irregularity), or blue (insufficient quality). Red and green recordings identified AF episodes, defined as consecutive red recordings terminated by at least one green. Patient satisfaction with the monitoring protocol was evaluated via a questionnaire. Results – Of the 44 patients who screened positive for OSA, the smartwatch was successfully installed in 95.5%. The median monitoring duration was 88 days (IQR:77-90). Of the smartwatch data, 58.8% were of high quality, with a median of 68 (IQR:50–76) high-quality measurements per day. Compliance, defined as the number of measurements performed divided by the recommended measurements, was 93% (IQR:52-109) for the spotcheck and 92% (IQR:81-96) for the smartwatch protocol. Motivation, calculated as the proportion of days with ≥2 daily spotchecks and ≥90% of the expected smartwatch measurements, was only 26.7% (IQR:11.6–61.1%; Figure 1) for the spotcheck and 70% (IQR:48-84) for the smartwatch method. AF was detected in 80% of patients with the smartwatch, compared to 43% with spotchecks (p<0.001). The smartwatch identified 9.5 times more AF episodes than the spotcheck method (626 vs. 66, p=0.029). The median AF burden was 0.36% (IQR:0.03-6.45) in those with AF. High patient satisfaction with the monitoring was observed (median score: 54 out of 60 (IQR:46-58, n=37; Figure 2)). The majority of patients (n=22) were willing to wear the smartwatch at all times in the long term, and most (n=20) preferred the combined smartwatch-spotcheck approach. Conclusion – Long-term rhythm monitoring using a smartwatch-based approach is both practical and provides reliable, high-quality data in AF patients. Wearable-based monitoring is well accepted by patients and can be integrated into structured AF care pathways.