Heart rate variability in 24-hour Holter recordings. Comparative study between short- and long-term time- and...

Mean hourly parameters obtained from all beats (long series) were compared with those obtained from a sample of 512 beats extracted each hour (short series) in nine presumably normal subjects. For both the short and long series, the spectral components, very low frequency, (VLF), low frequency (LF), and high frequency (HF), and time-domain indices (such as the Ewing statistic [PNN50] and RR standard deviation [SD-RR]), have been estimated. The spectral components LF and HF, estimated from the short and long series, were not significantly different, whereas significant differences were found between VLF, SD--RR, and PNN50. In both the short and long series, a strong correlation was found between LF and SD-RR and between HF and PNN50. The results suggest that, over a period of 24 hours, hourly LF and HF spectral components can be obtained using a single series of 512 beats every hour, with a great advantage over the evaluation of the mean hourly parameters. This method would be particularly useful in the study of circadian heart rate spectral analysis in Holter recordings with multiple artifacts or ectopic beats, and in general, when analysis of the entire 24-hour series is not feasible.