Multicenter Study Verifying a Method of Noninvasive Continuous Cardiac Output Measurement Using Pulse Wave Transit Time

Abstract

Many technologies have been developed for minimally invasive monitoring of cardiac output. Estimated continuous cardiac output (esCCO) measurement using pulse wave transit time is one noninvasive method. Because it does not require any additional sensors other than those for conducting 3 basic forms of monitoring (electrocardiogram, pulse oximeter wave, and noninvasive (or invasive) arterial blood pressure measurement), esCCO measurement is potentially useful in routine clinical circulatory monitoring for any patient including low-risk patients. We evaluated the efficacy of noninvasive esCCO using pulse wave transit time in this multicenter study. We compared esCCO and intermittent bolus thermodilution cardiac output (TDCO) in 213 patients, 139 intensive care units (ICUs), and 74 operating rooms (ORs), at 7 participating institutions. We performed electrocardiogram, pulse oximetry, TDCO, and arterial blood pressure measurements in patients in ICUs and ORs; a single calibration was performed to measure esCCO continuously. TDCO measurement was performed once daily for ICU patients and every hour for OR patients, and just before the removal of the pulmonary arterial catheter from patients in both the ICU and OR. We evaluated esCCO against TDCO with correlation analysis and Bland and Altman analysis and also assessed the change of bias over time. Furthermore, we inspected the impact of change in systemic vascular resistance (SVR) on change in bias because abnormal SVR was assumed to be a factor contributing to the change of the bias. From among 588 esCCO and TDCO datasets (excluding calibration points), 587 datasets were analyzed for 213 patients. The analysis results show a correlation coefficient of 0.79 (P < 0.0001, 95% confidence limits of 0.756-0.819), a bias (mean difference between esCCO and TDCO) of 0.13 L/min (95% confidence interval of bias 0.04-0.22 L/min), and a precision (1 SD) of 1.15 L/min (95% prediction interval was -2.13 to 2.39 L/min). There were no significant differences among 3 defined time intervals over 48 hours after calibration (repeated-measures analysis of variance P = 0.781) in the ICU. The influence of SVR on esCCO analysis showed a correlation coefficient between SVR and an error of 0.37 (P < 0.0001, 95% confidence interval 0.298-0.438). The efficacy of noninvasive esCCO technology was compared with TDCO in 213 cases. Five hundred eighty-seven datasets comparing esCCO and TDCO showed close correlation and small bias and precision, which were comparable to current arterial waveform analysis technologies.