Comparison of bioreactance non-invasive cardiac output measurements with cardiac magnetic resonance imaging.

Impedance cardiography measurement of cardiac output gained wide interest due to its ease of use and non-invasiveness. However, validation studies of different algorithms yielded diverging results. Bioreactance (BR) as a recent adaption differs fundamentally as the flow signal is derived from phase shifts. Our aim was to assess the accuracy and reproducibility of BR, as compared to the non-invasive gold standard--cardiac magnetic resonance imaging (CMR). We prospectively included 32 stable patients. BR was performed twice in the supine position and averaged over 30 seconds. Mean bias was 0.2 ± 1.8 l/minute (1 ± 28%, percentage error 55%) with limits of agreement ranging from  -3.4 to 3.7 l/minute. Reproducibility was acceptable with a mean bias of 0.1 ± 0.9 l/minute (1 ± 14%, 27%). Low cardiac output was significantly overestimated (-1.1 ± 1.5 l/minute), while high cardiac output was underestimated (1.5 ± 1.7 l/minute), (P=0.001), although reproducibility was unaffected. Bias and weight were moderately correlated in men (r = 0.50, P=0.02). No differences for accuracy were found in nine patients who had an arrhythmia (0.3 ± 1.4 versus 0.1 ± 2.0 l/minute, P=0.76), while clinically relevant differences were found in patients with mild aortic valve disease (1.9 ± 2.2 versus -0.3 ± 1.7 l/minute, P=0.02). Overall, BR showed insufficient agreement with CMR, overestimating low and underestimating high cardiac output states. Reproducibility was acceptable and not negatively affected by the circulatory condition. Consequently, absolute values acquired with BR should be interpreted with caution and must not be used interchangeably in clinical practice.