ObjectivesTo evaluate strategies for thermodilution-based measurement of cardiac output and right ventricular (RV) ejection fraction and to assess the effects of controlled mechanical ventilation in patients. Furthermore, to compare strategy-associated repro-ducibility with reference values obtained during long-term apnea.DesignCrossover trial in patients; reference values from apneic animals.SettingUniversity ICU and physiology laboratory.PatientsSix consecutive male ICU patients (48 to 70 yrs) after major abdominal vascular surgery. Animals: two adult female sheep.InterventionsThree ventilatory rates (8,16, and 24 cycles/min) and 15-sec periods of apnea were selected for measurements in patients. In animals, continuous apnea was achieved with extracorporeal CO2removal and apneic oxygenation.MeasurementsMeasurements were performed using an appropriate pulmonary artery catheter and an ejection fraction/cardiac output computer prototype. The thermal indicator was injected automatically at four defined points of the ventilatory cycle, but triggered manually during apnea.Main ResultsAt 8 cycles/min, there was a wide mean range of cyclic variable modulation, with a coefficient of variation of 11.6% and 23.2% for cardiac output and RV ejection fraction, respectively. Allowing for ventilatory phase or changing from 8 to 16 cycles/min reduced errors by half. Combining both procedures resulted in a coefficient of variation of 4.7% and 6.6% for cardiac output and RV ejection fraction, respectively. The best coefficient of variation values obtained during 15 sees of apnea in patients approached those variations in experimental apnea (coefficient of variation of 2.1% and 4.5% for cardiac output and RV ejection fraction, respectively).ConclusionsAt low ventilatory rates, best results are achieved by averaging four phase-selected measurements. One-point measurements were less accurate and random point measurements less reproducible.