Evaluation of the Reynolds Pathfinder II system for 24 h heart rate variability analysis

Heart rate variability (HRV) in 24-h Holter recordings is normally determined using commercially available Holter equipment. However, short-term speed variations in tape-based Holter systems have not been thoroughly investigated and recommendations for speed control are vague. In this study, overall accuracy and precision of RR interval measurement was assessed for two types of tape recorder replayed on the same Reynolds Pathfinder II system. Artificial QRS complexes spaced at exact intervals were recorded simultaneously by the two recorders. In addition, records of 118 healthy subjects were evaluated for false-positive normal-normal intervals. During replay, digital output from the pathfinder was processed by a locally developed personal computer-based HRV system. Using the recorded crystal generated timing track (TT) for speed control and the TT dependent speed surveyor (SS), set to allow speed errors up to +/- 3.1%, mean RR per 5 min.h-1.24 h-1, for test tapes, corresponded exactly to true RR, for both recorders. Beat-to-beat variation, for 1000 ms test intervals, calculated as standard deviation (SD) and root mean square of successive differences (RMSSD) was 3 ms, 3 ms versus 5 ms, and 8 ms for Oxford and Tracker tapes respectively. No difference between sequential intervals exceeded +/- 10 ms versus +/- 20 ms for Oxford and Tracker respectively. Reproducibility of mean RR and HRV parameters for patient records, expressed as a coefficient of variation, was 0.1-1.7%. False positive normal-normal intervals greater than 2.3 s, due to artifacts, were found in 99 of 118 records, and these increased mean SD and RMSSD values significantly, whereas percent successive RR interval differences greater than 50 ms were unaffected. A crystal generated TT (60/16 Hz) and an SS allowing a maximum error of +/- 3.1%, together correct speed during recording and replay to a degree that is acceptable for HRV analysis. A simple estimate of HRV counts of successive RR interval differences greater than 50 ms is a reliable parameter as it is relatively insensitive to artifact.