Short term heart rate variability is preserved in Parkinson’s disease under atomoxetine

Figures

• 

  Figure 1. Subject-level values for HRV metrics. Boxplots show the median (horizontal line) and interquartile range, with data points representing individual participants (N = 22 Controls; N = 15 people with Parkinson’s disease who underwent both placebo and atomoxetine sessions), for (A) Time domain, (B) Frequency domain and (C) Non-linear domain. Grey bars indicate metrics for which the Bayesian regression models reported in Figs 2 and 3 provided evidence of a group difference (i.e. the posterior mean fell outside the ROPE and the 95% CI excluded zero). SD1 = standard deviation of ellipse width on the Poincaré plot; SD2 = standard deviation of ellipse length on the Poincaré plot; SD1:SD2 = SD1-to-SD2 ratio; PD = Parkinson’s disease.

• 

  Figure 2. HRV comparisons on placebo versus atomoxetine in Parkinson’s disease. Distributions (thin black lines) show the Bayesian posterior estimates of HRV scores for participants with Parkinson’s disease on placebo versus atomoxetine (N = 15 participants), for (A) Time domain, (B) Frequency domain, (C) Non-linear domain. Estimates were obtained from Bayesian linear mixed-effects regression models with weakly informative Student’s t priors centred on the placebo-group mean for each metric. The dotted vertical line indicates the placebo-group mean; thick horizontal bars show the 75% CI and thin bars the 95% CI. The shaded orange column denotes the ROPE, defined as ±0.1 × σ of the control group for each metric. Standardized effect sizes (β/σ) and 95% CIs for each metric are reported in the Results. SD1 = standard deviation of ellipse width on the Poincaré plot; SD2 = standard deviation of ellipse length on the Poincaré plot; SD1:SD2 = SD1-to-SD2 ratio; PD = Parkinson’s disease.

• 

  Figure 3. HRV comparisons between Parkinson’s disease (on placebo) and healthy controls. Distributions (thin black lines) show the Bayesian posterior estimates of HRV scores for participants with Parkinson’s disease on placebo (n = 15) and healthy controls (N = 22), for (A) Time domain, (B) Frequency domain, (C) Non-linear domain. Estimates were obtained from Bayesian linear regression models with weakly informative Student’s t priors centred on the control-group mean for each metric. The dotted vertical line indicates the control-group mean; thick horizontal bars show the 75% CI and thin bars the 95% CI. The shaded purple column denotes the ROPE, defined as ±0.1 × σ of the control group for each metric. Standardized effect sizes (β/σ) and 95% CIs for each metric are reported in the Results. SD1 = standard deviation of ellipse width on the Poincaré plot; SD2 = standard deviation of ellipse length on the Poincaré plot; SD1:SD2 = SD1-to-SD2 ratio; PD = Parkinson’s disease.

• 

  Figure 4. Relationship between locus coeruleus integrity and HRV. (A) We extracted the CNR from the caudal portion of the locus coeruleus as a measure of integrity in Parkinson’s disease participants. (B) Relationship between vLF power and caudal locus coeruleus integrity in Parkinson’s disease, tested using Bayesian linear correlation, median r = 0.38, BF = 2.65. (C) Relationship between SD2 and caudal locus coeruleus integrity in Parkinson’s disease, tested using Bayesian linear correlation, median r = 0.41, BF = 3.08. Data points represent individual participants (N = 15 people with Parkinson’s disease). LC = locus coeruleus.