Performance of the imPulse device for the detection of atrial fibrillation in hospital settings
Sinéad T.J. McDonagh, PhD , Shelley Rhodes, PhD, Fiona C. Warren, PhD, Sam Keenan, BSc, Claire Pentecost, PhD, Philip Keeling, PhD, Martin James, MBBS, Rod S. Taylor, PhD), Christopher E. Clark, PhD
Background Atrial fibrillation (AF) increases thromboembolism and stroke risk; this can be reduced by oral anticoagulation, but only if AF is detected. A portable, point-of-care device, capable of accurately detecting and identifying AF, could reduce workload and diagnostic delay by minimizing need for follow-up 12-lead electrocardiogram (ECGs). Objective To assess the diagnostic performance of the Plessey imPulse lead I ECG device compared with a 12-lead ECG in detecting AF. Methods Cross-sectional diagnostic accuracy study. Participants underwent simultaneous 12-lead ECG and imPulse device recordings. The imPulse device reports AF to be “probable,” “possible,” “unlikely,” or “uncontrolled AF unlikely.” imPulse and ECG reference results were cross-tabulated; sensitivity, specificity, positive/negative predictive values, and positive/negative likelihood ratios with 95% confidence interval (CI) were estimated based on different imPulse device report categorizations and heart rate subgroups. Results A total of 217 participants were recruited (mean age 70.2 [standard deviation 12.7]), 56% male, 57% outpatients, 43% inpatients) and 199 were included in analyses. AF was diagnosed on ECG for 41 of 199 (20.6%) participants and reported by imPulse as possible, probable, or uncontrolled AF unlikely present for 49 of 199 (24.6%). Sensitivity and specificity for imPulse detection of possible, probable, or uncontrolled AF unlikely vs unlikely, compared with ECG, were 80.5% (95% CI, 65.1%–91.2%) and 89.9% (84.1%–94.1%), respectively. When probable or uncontrolled AF unlikely were compared vs possible or unlikely AF, sensitivity and specificity were 63.4% (46.9%–77.9%) and 98.1% (94.6%–99.6%), respectively. Conclusion The imPulse device has moderate sensitivity and good specificity compared with ECG AF detection in a hospital setting.