The Bias Evaluation And Comparative Research Of Bilirubin Between Different Biochemical Analysis Systems

Objective:By evaluating the comparability of the total bilirubin and direct bilirubin value determined between different biochemical analysis systems under the same laboratory conditions, to investigate the acceptability of the bias between two analysis systems.Methods:According to the EP9-A2 file of Clinical and Laboratory Standards Institute(CLSI; formerly:National committee for clinical laboratory, NCCLS), both the BECKMAN DXC-800 biochemical analyzer(reference method) and HITACHI-7600(test method) were used to determine the total bilirubin and direct bilirubin, tested 8 samples each day for 5 consecutive days. The data was recorded and the separate spots were detected. The linear regression equation and the coefficient of correlation were calculated to evaluate the bias between two methods.Results:The linear regression equation for total bilirubin was:Y=0.843X-0.593, R2=0.9988, SYX=2.597.The linear equation for direct bilirubin was:Y=1.215X+2.97, R2=0.9948, SYX=4.590. The expected bias of total bilirubin in medical decision levels with 25,50 and 340(μmol/L),were 4.158,8.443 and 53.97;the 95% reference ranges of expected bias were -5.15181～-3.88419,-9.03252～-7.85347 and -55.72050～-52.21951.The expected bias of direct bilirubin in medical decision levels with 1,6 and 35.5(μmol/L),were 3.185,4.26 and 10.6;the 95% reference ranges of expected bias were 2.03631～4.33369,3.49101～5.02899 and 9.56045-11.63955.Conclusion:The results of total bilirubin and direct bilirubin between two different analysis systems have high correlation. When the concentration increases, the correspondent prospective bias also increase, which was greater than the allowed bias of 1/2CLIA88 in various levels of medical decision, the permitted acceptable bias of two bilirubins were not in the 95% reference ranges of expected bias. So do not have comparability. Applying different biochemical analysis system to detect total bilirubin and direct bilirubin in one laboratory, method comparison and bias estimation should be employed.