Abstract
Background: Assessment of cardiac output (CO) in pulmonary hypertension (PH) relies on the thermodilution (TD) and Fick methods using invasively derived data. Echocardiography (echo) enables quick, noninvasive assessment of CO. We assessed the correlation and agreement among TD, Fick, and echo methods.
Methods: Indirect Fick, TD, and echo CO were compared in 89 subjects (age 55 ± 16 yrs, 66% Group 1 PH) using linear regression and Bland-Altman methods. The agreement among the methods for classifying subjects with low cardiac index (CI < 2.5) and elevated pulmonary vascular resistance (PVR ≥ 3) was assessed. For assessment of PVR by echo, invasively derived pressures were used.
Results: Fick and TD demonstrated fair correlation (r = 0.72, p < 0.001) and modest agreement (bias of +0.31 l/min for Fick, limits of agreement –2.27 to 2.90). Echo CO had weak correlation with TD CO (r = 0.46, p < 0.001) and inferior agreement (bias of +0.81 l/min for echo, -3.15 to 4.77). Echo CO also had weak correlation and inferior agreement with Fick CO (r = 0.37, p < 0.001) (bias of +0.50 l/min for echo, -3.78 to 4.77). The agreement by kappa for identifying CI < 2.5 for Fick and TD, Fick and echo, and TD and echo were 0.55, 0.30, 0.27, respectively. TD classified more patients with CI < 2.5 than Fick or echo. The agreement by kappa for identifying PVR ≥ 3 between Fick and TD, Fick and echo, and TD and echo were 0.81, 0.57, 0.54, respectively. Ninety percent of those with PVR≥3 were still classified as such by substitution of echo CO for Fick or TD CO.
Conclusion: Fick and TD CO had fair correlation and agreement. Echo showed bias towards higher CO and had weaker correlation with Fick and TD. Echo and Fick classified fewer patients with low CI than TD, and all three methods had good agreement for distinguishing those with and without elevated PVR.
Keywords
Cardiac output, Pulmonary hypertension, Echocardiography, Fick, Thermodilution, Low cardiac index, Elevated pulmonary vascular resistance