Validation of continuous thermodilution cardiac output in patients implanted with a left ventricular assist device

Objective: To assess the accuracy of a continuous cardiac output (CCO) monitor against an independent, intravascular measurement of flow as can be performed in patients fitted with a left ventricular assist device (LVAD). Design: A prospective cohort study. Setting: Academic tertiary-care center. Participants: Adult patients (n = 15) presenting for LVAD placement. Interventions: Consenting patients presenting for LVAD placement for end-stage cardiac failure were anesthetized, and a CCO pulmonary artery catheter was placed (OptiQ, CCO/SvO₂; Abbott Critical Care, North Chicago, IL). Patients were monitored with transesophageal echocardiography and excluded from analysis if aortic regurgitation was found. Cardiac output was determined using a Q-Vue, CCO/SvO₂ computer with digital readout (Abbott Critical Care, North Chicago, IL). The LVAD was placed in standard fashion during cardiopulmonary bypass. The Thoratec vented electric Heartmate (Thoratec Co, Pleasanton, CA) incorporates an LVAD flow console, which computes LVAD flow within ± 5% (range, 1.8 to 10 L/min). Measurements and Main Results: Cardiac output flow measurements were made from both systems at the following time points: 5 minutes and 30 minutes after protamine administration, at chest closure, and after skin closure. Mean cardiac output for each device did not differ at any time point. Regression analysis (Pearson's) showed acceptable correlation (r² = 0.79, p < 0.0001), whereas a bias of 529 mL with limits of agreement of 1,208 mL were shown for CCO measurement compared with LVAD flow. Conclusion: The data indicate that the CCO system tends to overestimate cardiac output by approximately 500 mL/min when compared with LVAD flow. Nevertheless, this bias is within the range found by other less-invasive studies done to assess the accuracy of this system and further serves to confirm its relative accuracy.