摘要
Introduction. Bias and precision statistics have succeededregression analysis when measurement techniques are compared. However, whenapplied to cardiac output measurements, inconsistencies occur in reporting theresults of this form of analysis. Methods. A MEDLINE search wasperformed, dating from 1986. Studies comparing techniques of cardiac outputmeasurement using bias and precision statistics were surveyed. An error-gramwas constructed from the percentage errors in the test and reference methodsand was used to determine acceptable limits of agreement between methods.Results. Twenty-five articles were found. Presentation of statisticaldata varied greatly. Four different statistical parameters were used todescribe the agreement between measurements. The overall limits of agreementin studies evaluating bioimpedance (n = 23) was ±37%(15–82%) and in those evaluating Doppler ultrasound (n = 11)±65%(25–225%). Objective criteria used to assess outcome weregiven in only 44%of the articles. These were (i) limits of agreementapproaching ±15–20%, (ii) limits of agreement of less than 1L/min, and (iii) more than 75%of bias measurements within ±20%of themean. Graphically, we showed that limits of agreement of up to ±30%were acceptable. Conclusions. When using bias and precision statistics,cardiac output, bias, limits of agreement, and percentage error should bepresented. Using current reference methods, acceptance of a new techniqueshould rely on limits of agreement of up to ±30%.