Cardiac Output Monitoring: Validation Studies-how Results Should be Presented

Abstract

Purpose of review: Cardiac output monitors can be assessed by a variety of techniques, but a common principle is quantifying agreement between a reference standard and new monitor. The current standard analysis technique is a Bland-Altman plot. The Bland-Altman plot evaluates bias between mean differences of cardiac output, from which an agreement interval is derived. These limits are, however, statistical limits of agreement and the clinical acceptability will depend upon context and application. This article provides suggestions for understanding and presenting the results of cardiac output validation, using standard metrology alongside proposals for criteria used to accept new techniques.

Recent findings: Confusion about the appropriate way to report "precision" in method comparison studies stem from a lack of clarity on how single or repeated measurements should be interpreted. During serial measurements of cardiac output the true value changes, thus measurement should be considered as serial rather than repeated. Method agreement based upon precision achieved by cardiac output monitors needs to consider each method's general variability around true values obtained and this data should be generated and presented as part of each study design.

Summary: Studies should report serial measurements from two techniques for cardiac output monitoring. Results of similar techniques from other studies may not always be transferred and compared. Bias and intervals of agreement should be presented as Bland-Altman plots with dynamic cardiac output trends in polar plots. Percentage error should be calculated to allow appropriate comparison of techniques for study populations with different expected cardiac output values.

Keywords: Accuracy; Bland-Altman analysis; Cardiac output; Hemodynamic monitoring; Method comparison; Precision.

Fig. 1

Bland-Altman plot and polar plots for new techniques versus reference techniques, with representation of the limits of agreement and lines of identity respectively (dotted lines)