Agreement between Electrical Cardiometry and Pulmonary Artery Thermodilution for Measuring Cardiac Output in Isoflurane-Anesthetized Dogs

Abstract

In animals, invasive pulmonary artery thermodilution (PATD) is a gold standard for cardiac output (CO) monitoring, but it is impractical in clinical settings. This study evaluates the agreement between PATD and noninvasive electrical cardiometry (EC) for measuring CO and analyzes the other EC-derived hemodynamic variables in six healthy anesthetized dogs subjected to four different hemodynamic events in a sequential order: (1) euvolemia (baseline); (2) hemorrhage (33% blood volume loss); (3) autologous blood transfusion; and (4) 20 mL/kg colloid bolus. The CO measurements obtained using PATD and EC are compared using Bland-Altman analysis, Lin's concordance correlation (LCC), and polar plot analysis. Values of p < 0.05 are considered significant. The EC measurements consistently underpredict the CO values as compared with PATD, and the LCC is 0.65. The EC's performance is better during hemorrhage, thus indicating its capability in detecting absolute hypovolemia in clinical settings. Even though the percentage error exhibited by EC is 49.4%, which is higher than the standard (<30%), EC displays a good trending ability. Additionally, the EC-derived variables display a significant correlation with the CO measured using PATD. Noninvasive EC may have a potential in monitoring trends in hemodynamics in clinical settings.

Keywords: anesthesia; blood transfusion; canine; colloids; electrical velocimetry; hemodynamics; hemorrhage; hypovolemia; monitoring; noninvasive.

Figure 1

Cardiotronic electrode placement in a Beagle dog placed in the right lateral recumbency while using the electrical cardiometry (EC) monitor (ICON; Osypka Medical Inc., La Jolla, CA, USA). The electrodes are attached to an adhesive patch. The area on the left side of the neck adjacent to the common carotid artery and the left lower aspect of the thorax are clipped, thoroughly cleaned and dried before the application of the adhesive patch. The electrodes are connected to the ICON EC monitor by a cable and the monitor is synced with the laptop using an external communication cable to provide easy data management.

Figure 2

The alignment of the erythrocytes inside the aorta during a cardiac cycle induces significant variations in the impedance. During diastole, prior to aortic valve opening (left), the erythrocytes in the aorta are randomly oriented (due to no flow inside the aorta), which causes the applied electrical current to follow the circumference of the erythrocytes during their passage through the aorta resulting in a higher voltage and impedance measurement. During systole, after aortic valve opening (right), pulsatile flow causes the erythrocytes to parallelly align in the direction of the blood flow and the electrical current resulting in a lower impedance.

Figure 3

Timeline of the study design and data collection in six, healthy, adult isoflurane-anesthetized Beagle dogs. After anesthetic induction and instrumentation, data were collected: (1) at the baseline; (2) after 33% blood volume loss (H); (3) after the autologous blood transfusion (T); and (4) after the infusion of 20 mL/kg 6% Hydroxyethyl Starch 130/0.4 in 0.9% sodium chloride solution (C). A 10 min hemodynamic stabilization period was provided between the manipulation of blood volume during H, T, and C, and before data collection.

Figure 4

Scatter plot representing the cardiac output (CO) measurements using electrical cardiometry (CO_EC) and pulmonary artery thermodilution (CO_PATD) for six anesthetized Beagle dogs across five timepoints during four hemodynamic events (baseline, hemorrhage, autologous blood transfusion, and colloid infusion), thus yielding 120 paired observations (circles). Regression analysis about the line Y = X (dashed line) resulted in a good fit (solid line) as shown by the slope (0.72) and r² (0.94).

Figure 5

Bland−Altman analysis for non-uniform differences using cardiac output (CO) values measured using electrical cardiometry (CO_EC) and pulmonary artery thermodilution (CO_PATD) techniques in anesthetized Beagle dogs (n = 6) across five timepoints during four hemodynamic events (baseline, hemorrhage, autologous blood transfusion, and colloid infusion), thus yielding 120 paired observations (circles). Each circle represents an individual difference value corresponding to an average value and mean shows a strong positive bias (slope = 0.48; intercept = −0.31) indicating underprediction. As displayed, the solid lines indicate the mean and upper and lower limits of agreement, and the dashed lines indicate the 95% confidence intervals around these values.

Figure 6

Polar plot displaying changes in cardiac output (CO) measured using electrical cardiometry and pulmonary artery thermodilution methods in anesthetized Beagle dogs (n = 6) across five timepoints during four hemodynamic events (baseline, hemorrhage, autologous blood transfusion, and colloid infusion), thus yielding 120 paired observations (circles). Dotted lines indicate 10% boundaries (i.e., 10% = 0.206 L/min as mean CO = 2.06 L/min). The distance from the center reflects the absolute values of the mean change in CO ($\left[∆{\mathrm{CO}}_{\mathrm{PATD}}+∆{\mathrm{CO}}_{\mathrm{EC}}\right]/2$), and the angle with the horizontal (0° radial axis) is indicative of a lack of agreement. The polar plot analysis exhibits good trending ability as only four points are located on the exterior of the limits of good agreement.

Figure 7

Scatterplot of cardiac output (CO) measured using pulmonary artery thermodilution (CO_PATD) versus stroke volume variation (SVV) and variation in pre-ejection period (ΔPEP) in six healthy, isoflurane-anesthetized Beagle dogs across five timepoints during four hemodynamic events (baseline, hemorrhage, autologous blood transfusion, and colloid infusion), thus yielding 120 paired observations. Th orange dotted line with orange solid diamonds represents the best-fit correlation for SVV and the blue dotted line with blue solid circles represents the best-fit correlation for ΔPEP.