Can heart rate variability parameters derived by a heart rate monitor differentiate between atrial fibrillation and sinus rhythm?

Table 2 Area under the curve (AUC), cut off values and their sensitivity and specificity, and 95% confidence intervals (95% CI) for the use of RMSSD as discriminatory variable to distinguish atrial fibrillation from sinus rhythm in 14 warmblood horses

Pace	Method	Artifact correction	AUC	cut off	sens		spec
Pace	Method	Artifact correction	AUC	cut off		95% CI		95% CI
Rest	ECG_Man	none	0.99	215	1	0.77–1	1	0.77–1
Rest	ECG_Aut	none	0.96	383	0.93	0.66–1	0.93	0.66–1
Rest	ECG_Aut	1	1	188	1	0.77–1	1	0.77–1
Rest	ECG_Aut	2	0.99	131	1	0.66–1	0.93	0.66–1
Rest	ECG_Aut	3	0.94	111	1	0.77–1	0.93	0.66–1
Rest	ECG_Aut	4	0.77	65	0.93	0.66–1	0.69	0.42–0.92
Rest	ECG_Aut	5	0.37	24	0,5	0.23–0.77	0.38	0.13–0.65
Rest	HRM_Elec	Polar Flow	0.93	136	0.86	0.57–0.98	0.86	0.57–0.98
Walk	ECG_Man	none	0.97	121	0.93	0.66–1	0.93	0.66–1
Walk	ECG_Aut	none	0.96	164	0.93	0.66–1	0,86	0.57–0.98
Walk	ECG_Aut	1	1	119	1	0.77–1	1	0.77–1
Walk	ECG_Aut	2	1	111	1	0.77–1	1	0.77–1
Walk	ECG_Aut	3	1	92	1	0.77–1	1	0.77–1
Walk	ECG_Aut	4	0.99	57	1	0.77–1	0.93	0.66–1
Walk	ECG_Aut	5	0.74	22	0.78	0.49–0.95	0.57	0.29–0.82
Walk	HRM_Elec	Polar Flow	0.97	70	0.93	0.66–1	0.93	0.66–1
Trot	ECG_Man	none	1	36	1	0.77–1	1	0.77–1
Trot	ECG_Aut	none	0.77	90	0.86	0.57–0.98	0.72	0.42–0.91
Trot	ECG_Aut	1	1	51	1	0.77–1	0.93	0.66–1
Trot	ECG_Aut	2	1	42	1	0.77–1	0.93	0.66–1
Trot	ECG_Aut	3	1	40	1	0.77–1	1	0.77–1
Trot	ECG_Aut	4	1	33	1	0.77–1	1	0.77–1
Trot	ECG_Aut	5	0.93	15	0.93	0.66–1	0.93	0.66–1
Trot	HRM_Elec	Polar Flow	1	21	1	0.77–1	1	0.77–1

Two-minute recordings at rest, walk and trot were analyzed using 4 different methods of RR detection and HRV analysis. The automatically analyzed ECG followed by beat-to-beat manual correction of QRS detection (ECG_Man) with HRV analysis using Kubios HRV Software was considered the gold standard method. HRM Polar heart rate monitor (Equine H7) with plastic electrodes and Polar Flow Software. ECG_Aut ECG with automatic RR interval detection (Televet 100) and Kubios HRV software with 5 different artifact correction levels. The artifact correction algorithm identifies every RR interval that differs from the average RR interval of the 2-min recording more than a specified threshold value as artifacts and replaces the corrupted RR interval by interpolated RR values. The threshold values used in this study were 0.45 s (very low artifact correction), 0.35 s (low), 0.25 s (moderate), 0.15 s (high) and 0.05 s (very high)

ISSN: 1746-6148