Indicator amount, temperature, and intrinsic cardiac output affect thermodilution cardiac output accuracy and reproducibility

Abstract

Objective: To determine the accuracy and reproducibility of four thermodilution indicators (5-mL room temperature, 10-mL room temperature, 5-mL iced, and 10-mL iced injectates) at clinically relevant flow rates.

Design: Quasi-experimental study.

Setting: Animal research laboratory of a health sciences university.

Subjects: Six virgin western-breed ewes.

Interventions: Data were collected from six ewes that had ascending aorta electromagnetic flow probes and inferior vena cava occluders. Cardiac output was manipulated by inferior vena cava occlusion and isoproterenol infusion. Four thermodilution indicators were tested at high and low levels of cardiac output and compared with the electromagnetic flowmeter measurements of cardiac output.

Measurements and main results: The indicator amounts were determined from both injectate volume and temperature difference between the injectate and blood. Using 5-mL room temperature injectate as a reference, 10-mL room contained 2 x, 5-mL iced 2.1 x, and 10-mL iced 4.1 x the indicator amount of 5-mL room temperature injectate. Approximately 210 simultaneous thermodilution and electromagnetic flow measurements were made for each injectate over a flow range of 1.5 to 15.7 L/min. For the entire cardiac output range, systematic error was not present. However, the r2 value (.92) for the 10-mL iced injectate group was greater (p < .05) than that value (.79) for the 5-mL iced injectate group, while r2 values were .79 for the 10-mL room temperature group and .49 for the 5-mL room temperature group. At flow rates of < 4.7 L/min, r2 was not different among injectates, but reduced indicator amounts progressively overestimated output (p < .05), reaching 21% for the 5-mL room temperature group. At flow rates of > 7.7 L/min, the r2 value (.81) for the 10-mL iced group was greater (p < .05) than that value (.45) for the 5-mL iced group, while r2 values were .24 for the 10-mL room temperature group and .08 for the 5-mL room temperature group. Systematic error was not present.

Conclusions: At low cardiac output levels, reduced indicator impairs accuracy but not reproducibility, a phenomenon that is perhaps related to indicator loss. At high cardiac output rates, reduced indicator impairs reproducibility. This phenomenon is probably related to low signal-to-noise ratio. Thermodilution indicator amounts should be tailored to the output range.