Abstract 17046: Accuracy of Apple Watch for Detection of Atrial Fibrillation: A Systematic Review and Meta-Analysis

Abstract

Background: Apple Watch (AW) for the detection of silent atrial fibrillation (AF) has shown encouraging results due to its photoplethysmogram (PPG) sensors. However, clinical accuracy studies show varying results. To the best of our knowledge, this is the first meta-analysis conducted to ascertain the pooled sensitivity, and specificity in AF detection using the AW. Methods: PubMed, Google Scholar, and Scopus were queried till May 2023 for studies examining the accuracy of AW in AF detection. The outcomes of interest were the sensitivity and specificity of AW for the detection of AF. Sensitivities and specificities were calculated using true positives, true negatives, false positives, and false negatives wherever applicable. A random-effects meta-analysis was conducted using the generic inverse variance method to pool sensitivities and specificities. Further, sensitivity analysis was performed excluding studies using KardiaBand. Results: A total of 12 studies with a patient population of 421,156 were included with a mean age of 63 years. 6/12 studies used AW series 4 or higher while 3/12 studies evaluated the use of KardiaBand on top of the AW. AF was detected in 3,184 patients from a total of 420,312 patients. Following pooled analysis, an overall point estimate of sensitivity was observed to be [0.87 (0.76, 0.98); Figure 1A]. Ten studies evaluated specificity, with the pooled estimate being [0.91 (0.87, 0.95); Figure 1B]. After excluding studies using KardiaBand, the sensitivity was reported as 0.84 [0.68, 1.00] and specificity was observed to be 0.94 [0.91, 0.98]. Conclusion: AW proved to be effective in detecting AF in our analysis owing to high overall sensitivity and specificity with and without the addition of KardiaBand. However, cardiologists should still exercise caution before undertaking action based on the electrocardiographic diagnosis generated by this wrist-worn monitor.