Received: 01.05.2014 Accepted: 06.05.2014 Available as AoP: 26.05.2014

higher fluctuations of R-R intervals for the slow average HR than for the fast one. Moreover, the fluctuations of R-R inter-vals for fast HR may not be as high as for slow HR because the R-R intervals should have become negative (Fig. 1) [13–15, 17–20]. Due to these facts, standard HRV analysis may be mathematically biased, particularly if patients exhibit different average HRs. To overcome this problem, one should calculate the variability of R-R intervals with respect to the average R-R interval, i.e. normalise the oscillations with respect to the mean value. One can do that by dividing the sequence of R-R intervals by the corresponding average R-R interval [15, 17–20]. Or, one may divide standard deviation of R-R inter-vals by average R-R interval (i.e. calculation of coefficient of variation), or divide HRV power spectrum (or its components) by the average R-R interval squared [15]