Automatic filtering of outliers in RR intervals before analysis of heart rate variability in Holter recordings: a comparison with carefully edited data

Table 2 Equivalence test of preprocessing methods.

	*a) Edited vs* *b) Edited and filtered*	*a) Edited vs* *c) Unedited and filtered*	*b) Edited and filtered vs* *c) Unedited and filtered*
ΔP _tot (ms ² , log)
Subjects 1-14 years (N = 25)	(0.014, 0.042)*	(0.012, 0.039)*	(-0.006, 0.001)
Subjects 15-24 years (N = 32)	(0.020, 0.030)*	(0.014, 0.025)*	(-0.009, -0.002)
Subjects 25-49 years (N = 40)	(0.017, 0.026)*	(0.008, 0.019)	(-0.012, -0.004)
Subjects 50-75 years (N = 43)	(0.035, 0.067)*	(0.026, 0.059)*	(-0.012, -0.005)
ΔP _HF (ms ² , log)
Subjects 1-14 years (N = 25)	(0.036, 0.085)*	(0.041, 0.089)*	(-0.001, 0.011)
Subjects 15-24 years (N = 32)	(0.015, 0.030)*	(0.007, 0.026)*	(-0.011, -0.001)
Subjects 25-49 years (N = 40)	(0.011, 0.027)*	(0.002, 0.021)	(-0.013, -0.003)
Subjects 50-75 years (N = 43)	(0.042, 0.096)*	(0.035, 0.085)*	(-0.014, -0.003)

Data are given as 90% confidence interval for difference. * p < 0.05 in TOST test using ±5% as acceptable difference, corresponding to (Ө_L, Ө_U ) = (-0.022, 0.0212) in log-transformed units.
Summary of differences in HRV indices when calculated from differently pre-processed R-R intervals: a) edited data, b) edited and filtered data, and c) unedited and filtered data. Filtering of RR-intervals was performed using the age-based threshold.

ISSN: 1475-925X