Flow dependency of error in thermodilution measurement of cardiac output during acute tricuspid regurgitation

Objective: To evaluate the effect of variable degrees of tricuspid regurgitation on thermodilution cardiac output measurements during changes in venous return. Design: Prospective, controlled animal study. Setting: University laboratory. Participants: Eight anesthetized mongrel dogs instrumented for simultaneous measurement of cardiac output by thermodilution and ascending aortic electromagnetometry. Interventions: Data were collected before and after induction of moderate and severe tricuspid regurgitation. Under each condition, measurements were obtained at baseline and after opening 2 peripheral arteriovenous shunts to increase venous return. Measurements and Main Results: Baseline electromagnetic flow ranged from 1.74 to 3.62 L/min (median 2.73 L/min). Moderate and severe regurgitation reduced median electromagnetic flow values by 31% and 51%. Applying generalized estimating equations to model thermodilution cardiac output as a function of electromagnetic flow, arteriovenous shunt, and severity of tricuspid regurgitation revealed that (1) moderate and severe regurgitation changed the slope and intercept of the thermodilution/electromagnetic regression, but the differences between them were not significant, and (2) arteriovenous shunt alone had no effect under any condition. A simplified model independent of shunt and containing just 2 levels of tricuspid regurgitation (none or present) crossed with electromagnetic flow was applied. This analysis showed that regurgitation caused thermodilution to significantly underestimate electromagnetic flow at cardiac outputs >2.27 L/min (99 mL/kg/min) and overestimate it at flows <1.02 L/min (44 mL/kg/min). Conclusion: These data show that acute tricuspid regurgitation may produce underestimation of cardiac output by thermodilution when flow is relatively high, produce overestimation when flow is relatively low, or have minimal effect when flow is in the midrange.