Nonlinear measures of heart period variability: Decreased measures of symbolic dynamics in patients with panic disorder

Time series of heart period are not linens and recent studies illustrated t he importance Of using nonlinear methods to quantify the complexity of thes e time series. We compared different techniques to quantify, the nonlinear complexity of these time series ka patients with panic disorder and normal controls aid correlated these measures with spectral powers in different ba nds of interest. Twenty-four hour ECG was recorded in 23 normal controls an d 29 patients with panic disorder by using Holter records. Time series of h eart period were analyzed by using approximate entropies, slopes of 1/f sca ling, two algorithms to calculate fractal dimension, and word sequences usi ng symbolic dynamics. Measures using symbolic dynamics, especially worn cou nt (WC-100), showed highly significant differences between the two groups s imilar to some of the frequency normal (spectral) measures, while the other techniques were relatively ineffective to distinguish between the two grou ps. Different nonlinear techniques may relate to different aspects of nonli near complexity of the time series. These nonlinear techniques were also no t uniform in showing the differences between awake and sleep periods. Some correlate with the measures of respiratory sinus arrhythmia and some measur es obtained from symbolic dynamics may reflect not only the nonlinear compl exity of the time series bat also the total variability in the 24 hr HP rim e ser ies, especially power in the ultra-low frequency band (<0.0033 Hz). H owever; word count (WC-100) had only weak correlations with other measures and discriminated best between the two groups and showed that this nonlinea r measure was of additional value to the linear measures in classifying the two groups. (C) 2000 Wiley Liss, Inc.