Agreement between direct Fick and bolus thermodilution measurements of cardiac output during right heart catheterisation and the impact on haemodynamic classifications

Abstract

Abstract Background/Introduction When measuring cardiac output (CO) during right heart catheterisation (RHC), both direct Fick (DF) and bolus thermodilution (TD) methods are endorsed by pulmonary hypertension (PH) guidelines (1). Agreement between these techniques in contemporary practice is not well understood, nor are the diagnostic consequences of disagreement. Purpose To investigate the real-world agreement between DF and TD techniques for measuring CO during contemporary RHC, and evaluate the impact of disagreement on patients’ haemodynamic classification. Methods This was a single-centre study including 182 patients who had CO measured simultaneously by DF and TD methods during RHC. Oxygen consumption (VO2) was measured with an indirect calorimeter. The primary outcome was the agreement between DF and TD derived CO using Bland-Altman analysis. Secondary outcomes included the impact of specific clinical variables on CO agreement, and the frequency of disagreement in determining patients’ haemodynamic classification. Results Disagreement between DF and TD measures of CO was noted, with the DF technique tending to overestimate CO relative to the TD measurement (median DF-CO 5.42 L/min [interquartile range (IQR) 3.90 – 7.41] vs median TD-CO 4.10 L/min [IQR 3.47 – 5.10], p<0.001). For CO values between 3-7 L/min, the error between DF and TD measurements ranged from -4.75% to +44.5%, and limits of agreement were between -69.9% and +101% (Figure 1). Disagreement was increased by 19.2% in patients with at least moderate-to-severe tricuspid regurgitation, and by 16.0% in those with irregular heart rhythms (Table1). The agreement between DF and TD derived pulmonary vascular resistance (PVR) was equally poor. Nonetheless, in 86.3% of patients, the techniques agreed in classifying the PVR as abnormally elevated (> 2 Wood Units), with disagreement occurring more frequently in patients with less severe PH. Conclusions In this real-world study there was poor agreement between DF and TD derived CO, resulting in more than 10% of patients having discrepant haemodynamic classifications. This data highlight that both methods should be utilised in clinical practice, particularly in borderline cases or when the haemodynamic assessment influences treatment eligibility or operative risk assessment. The two techniques should not be used interchangeably for serial surveillance, and the impact of our findings on invasive assessments of valvular pathologies needs further investigation.