Validation of stroke volume and cardiac output by electrical interrogation of the brachial artery in normals: assessment of strengths, limitations, and sources of error

Abstract

The goal of this study is to validate a new, continuous, noninvasive stroke volume (SV) method, known as transbrachial electrical bioimpedance velocimetry (TBEV). TBEV SV was compared to SV obtained by cardiac magnetic resonance imaging (cMRI) in normal humans devoid of clinically apparent heart disease. Thirty-two (32) volunteers were enrolled in the study. Each subject was evaluated by echocardiography to assure that no aortic or mitral valve disease was present. Subsequently, each subject underwent electrical interrogation of the brachial artery by means of a high frequency, low amplitude alternating current. A first TBEV SV estimate was obtained. Immediately after the initial TBEV study, subjects underwent cMRI, using steady-state precession imaging to obtain a volumetric estimate of SV. Following cMRI, the TBEV SV study was repeated. Comparing the cMRI-derived SV to that of TBEV, the two TBEV estimates were averaged and compared to the cMRI standard. CO was computed as the product of SV and heart rate. Statistical methods consisted of Bland-Altman and linear regression analysis. TBEV SV and CO estimates were obtained in 30 of the 32 subjects enrolled. Bland-Altman analysis of pre- and post-cMRI TBEV SV showed a mean bias of 2.87 % (2.05 mL), precision of 13.59% (11.99 mL) and 95% limits of agreement (LOA) of +29.51% (25.55 mL) and -23.77% (-21.45 mL). Regression analysis for pre- and post-cMRI TBEV SV values yielded y = 0.76x + 25.1 and r(2) = 0.71 (r = 0.84). Bland-Altman analysis comparing cMRI SV with averaged TBEV SV showed a mean bias of -1.56% (-1.53 mL), precision of 13.47% (12.84 mL), 95% LOA of +24.85% (+23.64 mL) and -27.97% (-26.7 mL) and percent error = 26.2 %. For correlation analysis, the regression equation was y = 0.82x + 19.1 and correlation coefficient r(2) = 0.61 (r = 0.78). Bland-Altman analysis of averaged pre- and post-cMRI TBEV CO versus cMRI CO yielded a mean bias of 5.01% (0.32 L min(-1)), precision of 12.85% (0.77 L min(-1)), 95% LOA of +30.20 % (+0.1.83 L min(-1)) and -20.7% (-1.19 L min(-1)) and percent error = 24.8%. Regression analysis yielded y = 0.92x + 0.78, correlation coefficient r(2) = 0.74 (r = 0.86). TBEV is a novel, noninvasive method, which provides satisfactory estimates of SV and CO in normal humans.

Keywords: Cardiac output; Impedance cardiography; Noninvasive; Stroke volume; Transbrachial electrical velocimetry.

Fig. 1

a A pictorial circuit diagram from which SV and CO are obtained. Shown are the upper and lower housings for the electrode patches on the arm. AC is injected via the outer electrodes and voltage electrodes are shown as the inner electrodes. Voltage is demodulated, amplified and filtered to obtain Z₀ and ∆Z(t), which then undergo analog to digital (A to D) conversion. Electronic differentiation of ∆Z(t) yields dZ/dt. dZ/dt_max/Z₀ undergoes square root transformation. T_SF (systolic flow time) is multiplied by the volume conductor (V_C) and ([dZ/dt_max/Z₀])^0.5 to obtain stroke volume (SV). SV is multiplied by heart rate (HR) to obtain cardiac output (CO). Details found in text. b Photograph of volunteer with ECG patch electrodes on upper chest, TBEV electrode housing on the upper brachium, and wiring to an electrode housing just proximal to the antecubital fossa. The impedance signals are relayed to a signal processor shown on the wrist. Also shown is a pulse oximeter on thumb

Fig. 2

a In the upper tracing is shown a dZ/dt waveform from the thorax. In the lower tracing is shown a dZ/dt tracing from the brachium. Note that for the transthoracic waveform, the end of ejection (flow) is the nadir after the first zero crossing after point C, dZ/dt_max. In the lower tracing is shown a dZ/dt waveform obtained from the brachium. Note that the end of flow is the second zero crossing after point C, dZ/dt_max. The waveforms were obtained simultaneously, but, because of the transmission delay, they were aligned synchronously in time to illustrate the same T_SF. b The transbrachial impedance pulse variation ∆Z (upper tracing) and its differentiated first time-derivative dZ/dt (lower tracing). In both tracings the onset of flow is marked as point B, the maximum forward slope of ∆Z and peak first time-derivative dZ/dt_max as point C, and termination of flow as point X [20]. The temporal interval from B to X is systolic flow time T_SF

Fig. 3

The transbrachial impedance (Z(t)) comprises four elements. The three static elements include the impedance path through the tissue (Z_t), the blood (Z_b), and the interstitial water (Z_w). The parallel connection of these static impedances constitutes the transbrachial base impedance, Z₀. The parallel connection between Z₀ and the pulsatile component, ∆Z_b(t), constitutes the total impedance of the brachium (Z(t)). Shown is the current input (I(t)) and time variable voltage output (U(t))

Fig. 4

a Bland–Altman analysis: average TBEV CO versus cMRI CO See Table 3 for mean bias precision, ±LOA, and percent error. b Correlation analysis for TBEV CO versus cMRI CO: the slope (0.92) and y-intercept (0.78) suggest little systematic bias. See Table 4