Testing the presence of non stationarities in short heart rate variability series

The study proposes a test to evaluate stationarity over short beat-to-beat variability series of heart period (about 300 samples). This test checks the steadiness of mean and variance (i.e. a restricted form of weak stationarity). Artificial series, in which changes of the mean and variance are simulated, are utilized to set the analysis parameters and to evaluate the performance of the method. Next, the test is applied to beat-to-beat series variability series of heart period derived from humans at rest, during controlled respiration at different breathing rates (10, 15 and 20 breaths/minute) and during 80° head-up tilt. The application proves the difficulty to find stationary periods in short term recordings of heart period variability and prompts for an assessment of the effect of non stationarities on traditional linear and non linear indexes assuming stationarity as a requirement.