Accuracy and Precision of Non-invasive Cardiac Output Monitoring Devices in Perioperative Medicine: A Systematic Review and Meta-analysis

The main purpose of this article is to provide a systematic review of all medical literature (37 studies; 1543 patients) comparing cardiac output (CO) measured with a commercially available completely noninvasive CO monitoring device against bolus thermodilution (TD) in various clinical settings such as the intensive care unit, emergency department, and the operating room. In addition, the authors provide a meta-analysis of the data extracted from the systematic review to calculate the precision of these noninvasive CO monitoring devices when used perioperatively by following pooled estimates. The authors gathered data from the systematic review in order to calculate the following 4 variables: (1) the pooled estimate of the mean difference between the tested method of bolus TD and the noninvasive reference method (bias), (2) the pooled estimate of the standard deviation (precision) of the bias, (3) the pooled estimate of the 95% confidence intervals of the bias, and (4) the pooled estimate of the percentage of error (PE). An acceptable agreement between the tested and the reference method was defined as a pooled estimate (PE) of 30% or greater. Despite it being crucial in critical care management, CO, the basis of measuring oxygen delivery to end organs, has been demonstrably undermonitored in clinical settings. New noninvasive techniques and older methods abound for monitoring CO including pulse wave transit time (PWTT), noninvasive pulse contour analysis (niPCA), partial CO2 rebreathing (CO2r), and thoracic electrical bioimpedance (TEB), yet the “go-to” in the clinical setting is the traditional invasive technique of bolus TD. Mean CO was 4.78 L/min in both methods. The overall random-effects pooled bias was −0.13 L/min (−2.38 to 2.12 L/min) and PE 47% with high interstudy sensitivity heterogeneity (I = 83%, P < 0.001). Despite the many advances in medical technology since the early 2000s, both TEB and CO2r have not shown significantly increased agreement when compared with bolus TD (PE for TEB was 37% in 1999, 43% in 2010, and 42% in 2016 in this metaanalysis; CO2r PE was similar to 2010: 44.5% against 40% in this meta-analysis). Recent techniques, such as PWTTand the niPCA, have a PE of 62% for PWTT and 45% for niPCA. Completely noninvasive technologies for the measurement of CO did not reach an acceptable level of agreement. The PEs of all noninvasive techniques were above the preset threshold of 30%, and the