Comparison of cardiac indices during pregnancy between noninvasive automated device and two-dimensional echocardiography: a longitudinal study

Objective: To determine the correlation, precision, mean percentage difference and agreement between cardiac indices (stroke volume [SV], cardiac output [CO], SV index [SVI], and cardiac index [CI]) measured by noninvasive cardiac output monitor (NICOM^®, Cheetah Medical, Boston, MA, USA) and 2-dimensional transthoracic echocardiography (2D-TTE) across gestations in Chinese pregnant women.

Methods: This was a prospective longitudinal study performed in women with singleton pregnancy at 11-14⁺⁶ (n = 152), 19-24⁺⁶ (n = 152), 30-34⁺⁶ (n = 141), and 35-37⁺⁶ (n = 103). Cardiac indices, including CO and SV, were obtained by NICOM^®, which uses thoracic bioreactance, and 2D-TTE. CI and SVI were calculated from CO and SV adjusted for body surface area. The measurements of cardiac indices obtained using NICOM^® were assessed relative to that of 2D-TTE by calculating correlation coefficient, bias, precision, mean percentage difference, and 95% limits of agreement, adjusted for repeated measurements.

Results: Comparison of the SV and SVI measurements by the two approaches showed significant moderate correlation in the first trimester (r = 0.2-0.3; p = .01). Overall, the SV and SVI measurements obtained using NICOM^® relative to that obtained by 2D-TTE revealed a bias of -12.1 mL and -6.1 mL/m² (95% confidence interval [CI]: -44.5 to 20.2 and -24.8 to 12.5), respectively. Comparison of the CO and CI measurements by the two approaches showed significant moderate correlation in the first trimester (r∼0.2; p = .01). Overall, the CO and CI measurements obtained using NICOM^® relative to that obtained by 2D-TTE revealed a bias of -0.50 L/min and -0.18 L/min/m² (95% CI: 2.26-3.27 and -1.79 to 1.39), respectively. Mean percentage difference for all cardiac parameters in all three trimesters were more than 30%.

Conclusions: In Chinese pregnant women, NICOM^® has underestimated cardiac indices (SV, CO, SVI, and CI) compared to that measured by 2D-TTE. The mean percentage differences for all cardiac indices are more than 30%, which are higher than the clinically acceptable limit. Future research is needed to determine whether adjustment factors should be applied to the proprietary algorithms used by the NICOM^® for the estimation of SV and CO in the Chinese pregnant population.