Comparing cardiac output monitors and defining agreement: A systematic review and meta-analysis.

BACKGROUND: Measuring cardiac output is common in critical care and perioperative medicine. Different monitoring systems are often judged against others in comparative studies. There is no agreed standard or definition on which to base the conclusions of such studies.

OBJECTIVES: To review comparative studies of cardiac output monitors using an agreement:tolerability index (ATI) as a measure of monitor precision. To compare the ATI of a monitor with the conclusions of authors regarding agreement and clinical utility.

DESIGN: Systematic review of comparative studies of cardiac output monitoring systems. The precision of each monitor was standardised against an ATI using a tolerability interval based on the normal range for cardiac index. The conclusions of each study were described as positive, neutral or negative, depending on whether authors reported the monitor to be acceptably precise and/or clinically useful. Comparison was made between the precision of a monitor and the likelihood of it being favoured by authors.

DATA SOURCES: PubMed was searched up to March 2012.

ELIGIBILITY CRITERIA: Studies published in English that compared two or more methods for measuring cardiac output in adult humans.

RESULTS: A total of 213 papers documenting 409 separate comparisons of two methods of measuring cardiac output were included. ATIs for the different comparisons varied from 0.07 to 6.84 (where an ATI < 1 indicates acceptable agreement, 1-2 marginal and >2 unacceptable agreement). Thirty-one percent of authors defined their own terms for acceptable agreement. ATI was only moderately correlated with the conclusions of the authors (Spearman rho = 0.47, P < 0.0001).

CONCLUSIONS: Authors should define what constitutes acceptable agreement a priori when reporting comparative studies of cardiac output monitors. The ATI and the tolerability interval may be a useful basis for helping define acceptable precision.