Poor accuracy of noninvasive cardiac output monitoring using bioimpedance cardiography [PhysioFlow®] compared to magnetic resonance imaging in pediatric patients

BACKGROUND: Identification of low cardiac output (CO) states in anesthesia is important because preoperative hemodynamic optimization may improve outcome in surgery. Accurate real-time CO measurement would be useful in optimizing "goal-directed" therapy. We sought to evaluate the reliability and accuracy of CO measurement using bioimpedance cardiography (PhysioFlow®, NeuMeDx, Bristol, PA) in pediatric patients with and without cardiac disease undergoing anesthesia for magnetic resonance imaging (MRI). METHODS: All consenting patients undergoing anesthesia for cardiac MRI were enrolled. After equilibration of anesthesia for ≥10 minutes, 6 PhysioFlow electrodes were applied to the patient's chest for continuous real-time monitoring for 10 minutes. Data were stored in 15-second epochs and later averaged offline to obtain CO. Phase contrast MRI measurements of flow volumes in the superior vena cava and ascending and descending aorta were made from a single imaging plane through all 3 vessels at the level of the right pulmonary artery. Both CO measurements were indexed to body surface area. The anesthetic technique was the same for both measurements. Agreement was assessed using Bland-Altman analysis. RESULTS: Thirty-one patients were enrolled and 23 were analyzed. The median age at study was 2.8 years (range, 0.02-8.02 years) and median body surface area was 0.54 m (range, 0.21-1.00 m). Eleven of the 23 patients (48%) were males. Patients were grouped into those with univentricular physiology, 6 of 23 (26%); biventricular physiology with shunt, 3 of 23 (13%); biventricular without shunt, 10 of 23 (43%); and no structural heart disease, 4 of 23 (17%). The mean bias was -0.34 ± 1.50 L/min/m (P = 0.29). The 95% limits of agreement were -3.21 to +2.69 L/min/m. Only 8 of 23 measurements (35%) were within 20% and 14 of 23 measurements (61%) were within 30% of each other. CONCLUSION: PhysioFlow performance was not sufficiently accurate in this population. Modifications of the algorithm and further testing are required before this device can be recommended for routine clinical use in pediatric patients.