Dependence of premature ventricular complexes on heart rate—it’s not that simple

Figures

• 

  Figure 1.. PVC-HR relationship for 3 different patients over a 24-hour period computed using different time interval durations. Rows show individual patients with positive (A-D), neutral (E-H), and negative (I-L) classifications. Columns show the PVC-HR relationship using 1-hour, 10-minute, and 1-minute time intervals, and 1-minute time intervals aggregated by heart rate. PVC count is scaled to the expected number of PVCs in 1 hour to facilitate comparison across methodologies. Lines show the linear regression. Correlation of PVC count with heart rate is positive (blue), neutral (gray, P>.05), or negative (red). Inset shows Pearson’s correlation coefficient (r), P-value (P), and the slope of the linear regression (B). The PVC-HR relationship on a log-linear scale is shown in Figure S1.

• 

  Figure 2.. Patients with complex PVC-HR relationships show variable linear classifications across different 24-hour periods and different methodologies. Panels show the PVC-HR relationship for a single patient (MK993) computed using the third day (A-D), sixth day (E-H), and all days (I-L) of the recording. This patient shows a nonlinear PVC-HR relationship with highest PVC frequency at intermediate heart rates. The PVC-HR relationship on a log-linear scale is shown in Figure S2. Further details are in the caption of Figure 1.

• 

  Figure 3.. Daily variability of PVC-HR classification. (A-I) Pearson correlation coefficient and PVC-HR classification as positive (blue), negative (red), or neutral (gray) on different days for a sample of 9 patients using 1-hour intervals. (J) Variation in PVC-HR classification for each record (row) using different methodologies (columns). Variation is quantified using (normalized) entropy which goes from 0 (yellow) to 1 (black). Patients with low entropy have a consistent classification (eg, A). Patients with high entropy have an inconsistent classification (eg, C). Classifications using a log-linear scale are shown in Figure S3.

• 

  Figure 4.. Entropy of PVC-HR classifications as a function of methodology and class. Patients with a positive classification on a given day typically have lower entropy scores (the same classification on the other days). Box center line is the median, box edges are the upper and lower quartiles, and whiskers capture the range up to 1.5 times the interquartile range.

• 

  Figure 5.. Distribution of classification sets for each patient using different methodologies. A classification set contains each 24-hour classification of a patient and is positioned on the triangle according to the ratios between each classification frequency. A patient with the same classification on each day would be placed in a corner. Records with different classifications on each day are placed in the interior according to the relative frequencies of each classification. The distribution is shown using a ternary density plot with red (white) representing a high (low) concentration of patients. The distributions for classifications using a log-linear scale are shown in Figure S5.

• 

  Figure 6.. PVC count against heart rate for records with a positive (A, B) and negative (C) classification using 1-minute intervals. Intervals with a repeating rhythm are colored. Intervals without a repeating rhythm are gray and transparent. Repeating rhythms include no PVCs, and a fixed number of intervening regular beats (NIB) between PVCs.