Background: We elucidated the effects of various body positions on the agreement of cardiac output (CO) measurement between pulse contour analysis with the PiCCO monitor and thermodilution with pulmonary artery catheterization.
Methods: Fifteen anesthetized and mechanically ventilated pigs (40 ± 2 kg) were sequentially placed in various positions to facilitate simultaneous CO measurement. Between-methods agreement was assessed using the Bland-Altman method. Trending ability was assessed using Pearson product-moment correlation coefficient analysis.
Results: In supine, reverse Trendelenburg, Trendelenburg, and left lateral decubitus (lateral) positions, CO measured by these two methods was comparable (4.9 ± 1.5 versus 4.6 ± 1.6 L/min, 4.6 ± 2.2 versus 4.8 ± 1.8 L/min, 5.1 ± 2.1 versus 4.9 ± 2.1 L/min, and 5.4 ± 1.8 versus 5.0 ± 1.6 L/min; all P > 0.05). Mean bias between methods and limits of agreement (percentage error) were 0.3 ± 2.9 L/min (61%), -0.3 ± 3.3 L/min (71%), 0.1 ± 4.1 L/min (77%), and 0.5 ± 3.7 L/min (71%). Directional changes of paired CO revealed 66% (reverse Trendelenburg), 57% (Trendelenburg), and 66% (lateral) concordance. The correlation coefficient (r(2)) was 0.199, 0.127, and 0.108. For paired CO ≤6 L/min, mean bias between methods and limits of agreement (percentage error) were 0.2 ± 1.0 L/min (25%), -0.1 ± 1.0 L/min (28%), 0.2 ± 1.1 L/min (29%), and 0.5 ± 0.9 L/min (23%). Directional changes of paired CO revealed 84% (reverse Trendelenburg), 76% (Trendelenburg), and 65% (lateral) concordance. The correlation coefficient (r2) was 0.583, 0.626, and 0.213.
Conclusions: The mean CO measured by pulse contour analysis and thermodilution did not agree well in various body positions. Moreover, the measurements tended to trend differently in response to positional changes. For paired CO ≤6 L/min, however, the between-methods agreement and the trending ability improved significantly.
Copyright © 2013 Elsevier Inc. All rights reserved.