Comparison of non-calibrated pulse-contour analysis with continuous thermodilution for cardiac output assessment in patients with induced hypothermia after cardiac arrest

Aim: Induced mild hypothermia after cardiac arrest interferes with clinical assessment of the cardiovascular status of patients. In this situation, non-invasive cardiac output measurement could be useful. Unfortunately, arterial pulse contour is altered by temperature, and the performance of devices using arterial blood pressure contour analysis to derive cardiac output may be insufficient. Methods: Mild hypothermia (32-34 degrees C) was induced in eight patients after out-of-hospital cardiac arrest and successful resuscitation. Cardiac output (CO) was measured simultaneously by continuous thermodilution using a pulmonary artery catheter and a cardiac output monitor (Vigilance II, Edwards Lifesciences) and by pulse contour analysis using an arterial line and the Vigileo monitor (Edwards Lifesciences) during both normothermia (> 36 degrees C) and hypothermia. Continuous CO from both monitors was compared (Bland-Altman) and concordance of changes measured in consecutive 8-min intervals was measured. Results: Mean cardiac output was 3.9 +/- 1.2 l/min during hypothermia and 6.1 +/- 2.6 l/min during normothermia (p < 0.001). During hypothermia (normothermia), bias was 0.23 (0.77) l/min, precision (1 SD) was 0.6 (0.72) l/min, and the limits of agreement were -1.06 to 1.51 (-0.64 to 2.18) l/min, corresponding to a percentage error of +/- 34% (+/- 24%). Concordance of directional CO changes > 10% was 53.9% in hypothermia and 51.4% in normothermia. Conclusion: Induced hypothermia was not associated with increased bias or limits of agreement for the comparison of Vigileo and continuous thermodilution, but percentage error was high during normothermia and increased further during hypothermia. Less than 50% of clinically relevant CO changes during hypothermia were concordant. (C) 2011 Elsevier Ireland Ltd. All rights reserved.