Accuracy of continuous thermodilution cardiac output monitoring by pulmonary artery catheter during therapeutic hypothermia in post-cardiac arrest patients

Abstract

PurposeThermodilution continuous cardiac output measurements (TDCCO) by pulmonary artery catheter (PAC) have not been validated during therapeutic hypothermia in post-cardiac arrest patients. The calculated cardiac output based on the indirect Fick principle (FCO) using pulmonary artery blood gas mixed venous oxygen saturation (FCO-BG-SvO2) is considered as the gold standard. Continuous SvO2 by PAC (PAC-SvO2) has also not been validated previously during hypothermia. The aims of this study were (1) to compare FCO-BG-SvO2 with TDCCO, (2) to compare PAC-SvO2 with BG-SvO2 and finally (3) to compare FCO with SvO2 obtained via PAC or blood gas. MethodsWe analyzed 102 paired TDCCO/FCO-BG-SvO2 and 88 paired BG-SvO2/PAC-SvO2 measurements in 32 post-cardiac arrest patients during therapeutic hypothermia. ResultsTDCCO was significantly although poorly correlated with FCO-BG-SvO2 (R2 0.21, p<0.01) without systematic bias (−0.15±1.76l/min). Analysis according to Bland and Altman however showed broad limits of agreement ([−3.61; 3.45]l/min) and an unacceptable high percentage error (105%). None of the criteria for clinical interchangeability were met. Concordance analysis showed that TDCCO had limited trending ability (R2 0.03). FCO based on PAC-SvO2 was highly correlated with FCO-BG-SvO2 (R2 0.72) with a small bias (−0.08±0.72l/min) and slightly too high percentage error (44%). ConclusionOur results show an extreme inaccuracy of TDCCO by PAC in post-cardiac arrest patients during therapeutic hypothermia. We found a reasonable correlation between BG-SvO2 and PAC-SvO2 and subsequently between FCO calculated with SvO2 obtained either via blood gas or PAC. The decision to start or titrate inotropics should therefore not be guided by TDCCO in this setting.