Agreement between cardiac outputs by four-dimensional echocardiography and thermodilution method is poor

Abstract

The objective of the study was to determine the agreement of cardiac output (CO) measured by four-dimensional echocardiography (4D echo) to simultaneously obtain CO from pulmonary artery catheter (PAC) using thermodilution technique. Sixty-three comparable readings from 27 patients scheduled for elective coronary artery bypass were included. All echocardiographic measurements were obtained by one experienced echocardiographer. All echo images were analyzed independently and blinded from PAC-obtained measurements. Analysis was primarily done by Bland and Altman plot. The collected data were further controlled for interobserver bias and image quality. Differences in CO measurements increased with higher CO, hence values were logarithmically transformed. On the logaritmic scale, the 4D echo underestimated CO by 0.37 l/min compared with PAC, indicating that PAC measurements were 1.45 times higher than the 4D echo (95% confidence interval 1.32-1.52) and limits of agreement 0.97-2.14). The interobserver bias of 4D echo measurement analysis was 0.29 l/min (95% confidence interval 0.16-0.42) and limits of agreement -0.8-1.38). No difference was seen in image quality between comparisons with good agreement compared with comparisons with poor agreement. The agreement between COs by 4D echo and standard PAC thermodilution technique was poor. 4D echo underestimates CO as compared with PAC. This is most likely caused by the analysis software or low frame rate inherent to the technique.