Interchangeability of Electrocardiography and Blood Pressure Measurement for Determining Heart Rate and Heart Rate Variability in Free-Moving Domestic Pigs in Various Behavioral Contexts

Abstract

This study assessed the interchangeability between heart rate (HR) and heart rate variability (HRV) measures derived from a series of interbeat intervals (IBIs) recorded via electrocardiogram (ECG) and intra-arterial blood pressure (BP) in various behavioral contexts. Five minutes of simultaneously recorded IBIs from ECG and BP signals in 11 female domestic pigs during resting, feeding, and active behavior were analyzed. Comparisons were made for measures of HR, the standard deviation of IBIs, and the root mean of the squared distances of subsequent IBIs derived from ECG and BP signals for each behavior category using statistical procedures with different explanatory power [linear regression, intraclass correlation coefficient (ICC), Bland and Altman plots, and analysis of variance (ANOVA)]. Linear regression showed a strong relationship for HR during all behaviors and for HRV during resting. Excellent ICCs [lower 95% confidence intervals (CI) >0.75] and narrow limits of agreement in all behavior categories were found for HR. ICCs for HRV reached the critical lower 95% CI value of 0.75 only during resting. Using Bland and Altman plots, HRV agreement was unacceptable for all of the behavior categories. ANOVA showed significant differences between the methods in terms of HRV. BP systematically overestimated HRV compared with ECG. Our findings reveal that HR data recorded via BP agree well those recorded using ECG independently of the activity of the subject, whereas ECG and BP cannot be used interchangeably in the context of HRV in free-moving domestic pigs.

Keywords: Bland and Altman plot; blood pressure; domestic pig; electrocardiogram; heart rate variability; interchangeability.

Figure 1

Locations of the four components of the telemetric system exemplary for one pig 5 days after surgical procedure: positions of the transmitter body (1), positive electrode (2), catheter (3), and negative electrode (4). (Upper left) Transmitter body with according wires [positive electrode (black), catheter (white), and negative electrode (red)].

Figure 2

Linear regression of HR, SDNN, and RMSSD obtained via ECG in function of those obtained using BP (n = 11) with according 95% confidence limits (gray area) and 95% prediction limits (dotted line) during active behavior, feeding, and resting.

Figure 3

Bland and Altman plots of differences between ECG and BP measurements in terms of HR (A–C), SDNN (D–F), and RMSSD (G–I). The plots illustrate, in beats per minute [HR (bpm)] and in milliseconds [SDNN and RMSSD (ms)], the differences between values derived from ECG and BP on the y-axis (BP–ECG) against their average values on the x-axis for every behavior category [active (A,D,G), feeding (B,E,H), and resting (C,F,I)].

Figure 4

Mean values (±SE) of HR, SDNN, and RMSSD derived from ECG and BP during active behavior, feeding, and resting. Calculation was based on 5 min segments per behavior and averaged over individuals. Significance is given as follows: ***P < 0.001; **P < 0.01; *P < 0.05; and ^†P < 0.1.