Abstract Background Cardiopulmonary exercise testing (CPET) is a frequently used tool for clinicians in the management of adults with congenital heart disease. Breathing patterns during CPET are typically assessed by visual inspection of curves smoothed over a specified number of breaths or time intervals. Sample entropy, a measure of randomness developed for physiological time series data, has the potential to be a more data driven measure of abnormal breathing. However, it has rarely been studied during CPET, and to our knowledge, never in patients with congenital heart disease. Purpose To compare the sample entropy of breathing frequency (BF), oxygen uptake (VO2), and tidal volume (VT) during the loaded phase of a ramp CPET between three groups: tetralogy of Fallot (ToF), single ventricle and healthy controls. Additionally, to examine whether sample entropy is related to the patients' peak oxygen consumption (VO2peak) or age. Method Data were collected from the first participants (aged 18–60 years) enrolled in the ongoing IMPROVE-ACHD study conducted in Sweden, in which a maximum effort ramp CPET was performed. The groups analysed consisted of 23 patients with ToF (12 females), 11 with single ventricle (7 females), and 11 healthy controls (6 females). Raw breath-by-breath data of BF, VO2 and VT from the exercise phase were extracted. Outliers were identified using the median absolute deviation (MAD) over 30 breaths, with breaths exceeding 3 times the average MAD replaced by linear interpolation. Data were made stationary using differencing (xt - xt-1) followed by standardisation (subtraction of mean and division by standard deviation). Due to frequent hyperventilation in the end stage of the test, the last minute of effort was excluded. Stationarity was confirmed via the Augmented Dickey-Fuller test. Sample entropy was calculated with an embedding dimension (m) of 2 and a similarity threshold (r) of 0.2 times the standard deviation. Group comparisons were performed using the Kruskal-Wallis and Mann-Whitney U-test. Correlation was assessed using Spearman's ρ. Results No significant differences in sample entropy were observed in BF, VO2 or VT across the groups (Figure 1). The calculated sample entropies from the ToF and single ventricle groups did not correlate significantly with either VO2peak (BF: ρ = -0.12, p = 0.47; VT: ρ = -0.31, p = 0.06; VO2: ρ = -0.21, p = 0.22) or age (BF: ρ = -0.07, p = 0.66; VT: ρ = 0.17, p = 0.30; s VO2: ρ = 0.08, p = 0.65). Conclusion Our findings indicate no significant difference in random variability of breathing variables among patients with ToF or single ventricle compared to healthy controls. Furthermore, sample entropy in the patient groups was not significantly correlated with VO2peak or age. However, as the sample size in the ongoing study is still small, new analyses performed after including more participants would be valuable to more robustly explore potential differences and correlations.
Karlsson et al. (Sat,) studied this question.