Assessment of ultra low frequency band power of heart rate variability: validation of alternative methods

• Published in: International journal of cardiology 1999-09, Vol.71 (1), p.1-6
• Main Authors: Bilge, Ali R., Stein, Phyllis K., Domitrovich, Peter P., Gérard, Paul L., Rottman, Jeffrey N., Kleiger, Robert E., Kulbertus, Henri E., Piérard, Luc A.
• Format: Article
• Language: English
• Subjects: Biological and medical sciences; Cardiology. Vascular system; Electrocardiography, Ambulatory - methods; Female; Heart; Heart failure; Heart Failure - physiopathology; Heart failure, cardiogenic pulmonary edema, cardiac enlargement; Heart Rate; Humans; Male; Medical sciences; Middle Aged; Myocardial Infarction - mortality; Predictive Value of Tests; Prognosis; Reference Values; Signal Processing, Computer-Assisted; Spectral analysis; Time domain analysis; Time Factors
• ISSN: 0167-5273; 1874-1754
• DOI: 10.1016/S0167-5273(99)00085-6

Alternative methods for assessing ULF spectral power using data from commercial Holter analysers were studied. Different heuristics for ULF calculation were compared with standard research software-based determination of ULF. Setting: University Hospital. Patients: 43 patients in NYHA classes I–IV heart failure and seven normals of similar ages. Methods: SDNN, SDANN, ULF, VLF, LF, HF calculated from 24 h Holter monitoring using Oxford scanner software (method 1). ULF power also calculated by subtracting the sum of VLF, LF and HF powers obtained from the Holter scanner from the total variance (method 2) from 2·ln(SDANN) (method 3), and by performing a standard, research-quality 24-h EFT analysis on the beat files (standard). Results of methods 1–3 were compared with standard using two-way ANOVA with repeated measures, regression analysis and a graphical technique. Results: ULF calculated by method 1 correlated r=0.66 with standard but means differed substantially. In contrast, ULF calculated by method 2 correlated r=0.99 with standard with no significant differences between means. ULF calculated from SDANN (method 3) correlated r=0.983 with standard but means, while similar, were significantly lower ( P=0.005). Conclusion: ULF reported by commercial holter software is not equivalent to ULF power derived from 24 h FFT analysis. ULF calculated by method 2 can be considered equivalent to the ULF derived by standard 24-h FFT. ULF estimated by method 3 offers direct ULF power estimation from a temporal measure of HRV and can be useful when spectral values are not available.