室内临床生化不同检测体系间检测结果的可比性及偏倚评估

目的对临床生化内部三种不同检测体系Cobas6000，Hitachi7600及Vitros350（干化学）间相同项目的测定结果进行可比性及偏倚评估分析，为实验室认可及标准化提供实验数据。方法参考美国临床和实验室标准化委员会（CLSI）的EP9A2文R2﹤0.95，并有5项ALT、AST、ALP、Crea、Ca方法间偏倚超出允许误差范围。结论不同检测体系间存在一定偏差，特别是干化学与湿化学之间主要指标差异显著，相对偏差大于30%。当同一项目在2个或以上的检测系统件，以Hitachi7600作为参比方法，Cobas6000和 Vitros350作为待评方法，对相同项目结果进行线性回归分析、并计算医学决定水平处的方法间偏差，以美国临床实验室修正法规（CLIA88）规定的室间质评允许误差范围的1/2为标准，判断偏倚的临床可接受性。结果 Cobas6000与Hitachi7600间相同检测项目40项，其中4项：钙、镁、氯和二氧化碳相关系数R2﹤0.95，并有7项：ALT、AST、ALP、ADA、β2MG、Phos及Lac方法间偏倚超出允许误差范围；Hitachi7600与Vitros350相同项目16项，其中3项：ALB、Na＋、CL＋相关系数测试时，应定期进行结果可比性及偏倚评估，判断临床可接受性，必要时分别建立参考值系统，以满足临床需求。

Bias estimation and comparability between different results of clinical biochemistry detection systems

Objective to analyse the comparasity and bias estimation of results among three different biochemistry decetion systems including Cobas6000,Hitachi7600 and Vitros350 （ dry-chemistry ）, and to provide experimental data for laboratorys’ standardization and identification. Methods Reference to the file of EP9A2 from National committee for clinical laboratory （ CLSI ）, firstwe took Hitachi7600 as a way of comparison so as to evaluate Cobas6000 and Vitros350, then we analysed the results of the same tests by linear regression and to calculate the deviation of medical decision level between different methods, and took 1/2 of allowable error of provision of Clinical Laboratory Improvement Amendment （ CLIA88 ） as standard to estimate the clinical acceptability of bias. Results There were 40 same detections between Cobas6000 and Hitachi7600, for 4 of them （ Calcium, Magnesium, Chlorine and Carbon dioxide ）, the correlation coefficient R2﹤0.95, and for 7 of them （ALT、AST、ALP、ADA、β2MG、Phos and Lac ）, method biases excessed the allowable error range;there were 16 same detections between Hitachi7600 and Vitros350, for 3 of them （ALB、Na＋、CL＋ ）, the correlation coefficient R2﹤0.95, and for 5 of them （ALT、AST、ALP、Crea、Ca ）, method biases excessed the allowable error range. Conclusion There was a certain deviation between different detection system, especially the dry-chemistry and wet-chemistry, the relative deviation was more than 30%. When one detection is tested in 2 or more detection systems, analyse of the comparasity and bias estimation of results should be carried out at regular intervals, clinical acceptability should be estimated, and different reference systems should be established necessarily, so as to meet clinical needs.