Thermodilution Cardiac Index Has Poor Agreement with That Measured by the Direct Fick Method in Low-Output States

Abstract

<h3>Purpose</h3> The thermodilution (Td) method is routinely used for measurement of cardiac output although its accuracy in low-output states remains uncertain. Criteria for designating heart transplant listing status in the contemporary allocation system is incumbent on meeting strict cardiac index (CI) cutoffs. We aimed to assess the agreement between Td and direct Fick (dFick) CI in patients with low CI. <h3>Methods</h3> In this retrospective analysis of 535 patients who underwent right heart catheterization with dFick CI measurement as part of clinically indicated hemodynamic evaluation using the gold-standard Douglas bag technique, 253 patients had both Td/dFick CI measurements. Three saline injections were typically used and averaged in Td CI measurements. We included 118 patients who specifically had a CI of </= 2.2 L/min/m² by dFick method. Agreement between Td and dFick method was assessed by median absolute percentage difference, Spearman's rank correlation, and Bland-Altman Method. <h3>Results</h3> Among the 118 patients included (Mean [SD] age 60 [12.2] years, 42.3% female), Median (interquartile range, IQR) overall dFick and Td CI values differed significantly (1.86 [1.65, 2.1] and 2.03 [1.78, 2.31] L/min/m², respectively; P<0.001). The two measurements weakly correlated (r_s=0.31, P<0.001), with a systematic error introduced by Td CI expressed a mean difference of 0.21 (-0.61, 1.03) L/min/m² (Figure). Median (IQR) absolute percentage difference between Td and dFick was 16.9 (7.9, 27.9)%. Td and dFick CI's differed by >25% in 33.1% of patients. <h3>Conclusion</h3> The agreement of Td with the gold-standard dFick CI measurements in patients with low CI was suboptimal. These data are important consider in in clinical scenarios such as heart transplant listing status where strict hemodynamic cutoffs guide urgency.