Reliability of short-term measurements of heart rate variability (HRV) in patients with non-permanent atrial fibrillation.

Short-term measurements of heart rate variability (HRV) are practical, noninvasive markers of autonomic activity, and reduced HRV has been associated with onset and recurrence of non-permanent atrial fibrillation. However, clinical use is limited by strict standardization requirements and uncertain reliability in clinical populations. We evaluated the reliability of short-term HRV derived from 48-h Holter recordings during sinus rhythm in patients with non-permanent AF. Fifty participants underwent two standardized outpatient recordings at Holter attachment and removal, and additional, manually selected, out-of-hospital daytime and night-time presumed-sleep segments, with low editing burden, were extracted. HRV was assessed using SDNN and RMSSD. Reliability was evaluated using Bland-Altman limits of agreement (LoA) and intraclass correlation coefficients (ICC). For standardized outpatient recordings, LoA ranged from 43 to 242% for SDNN and 38% to 280% for RMSSD, with ICC values of 0.67 and 0.43, respectively. Consecutive night-time presumed-sleep segments showed higher relative reliability, with ICC values of 0.79 for SDNN and 0.89 for RMSSD, but absolute agreement remained limited (LoA 53%-173% and 59%-170%). These findings indicate poor absolute reliability in this workflow, limiting individual-level clinical interpretation, and likely reflecting a combination of measurement-process variability, inherent to a clinically pragmatic workflow and biological within-person variability in HRV.