Tracking changes in cardiac output: methodological considerations for the validation of monitoring devices

Until now, tools for continuous cardiac output (CO) monitoring have been validated as if they were tools for snapshot measurements. Most authors have compared variations in cardiac output between two time-points and used Bland-Altman representations to describe the agreement between these variations. The impacts of time and of repetitive measurements over time are not taken into consideration. This special article proposes a conceptual framework for the validation of CO monitoring devices. Four quality criteria are suggested and studied: (1) accuracy (small bias), (2) precision (small random error of measurements), (3) short response time and (4) accurate amplitude response. Because a tolerance is obviously admitted for each of these four criteria, we propose to add as a fifth criterion the ability to detect significant CO directional changes. Other important issues in designing studies to validate CO monitoring tools are reviewed: choice of patient population to be studied, choice of the reference method, data acquisition method, data acceptability checking, data segmentation and final evaluation of reliability. Application of this framework underlines the importance of precision and time response for clinical acceptability of monitoring tools.