How many heart beats are enough? A systematic assessment of data length reduction on measures of heart rate variability and complexity

Indices of heart rate variability (HRV) and complexity have become popular for the assessment of autonomic function and the presence of various disease states. Standard recommendations for R-R interval (RRI) analysis include a minimum of 256 beats for HRV and 800 beats for heart rate complexity. However, the sensitivity and reliability of these indices to reduced data length have not been fully delineated. We therefore systematically determined minimum data length requirements for a variety of common HRV and complexity indices. ECG was recorded in resting, supine, healthy human subjects (n=19). Measurements were made from the same ectopy-free ECG recording using 900, 800, 700, 600, 500, 400, 300, 256, 200, 150 and 100 heart beats by methodical truncation of the data set. Time domain variables assessed included RRISD, RMSSD, pNN50, and the Poincaré plot descriptors of SD1, SD2 and SD2/SD1. Frequency domain (LF, 0.04–0.15; HF, 0.15–0.4 Hz) variables were calculated via fast Fourier transform. Complexity analyses included entropy (ApEn and SampEn), fractal nature (FD-L and FD-DA), stationarity (StatAv), symbol dynamics (SymDyn) and detrended fluctuation analysis (DFA). Time and frequency domain variables were not significantly altered until the data were truncated to at least 150 beats. With the exception of ApEn and DFA, all complexity indices were unchanged until reaching 100 beats; ApEn and DFA were decreased (p≤0.034) at 600 and 250 beats. These data indicate that measurements of commonly used indices of HRV and complexity, with the exception of ApEn and DFA, are robust in data sets containing at least 150 heart beats.