应用CLSI EP9-A2-IR文件对两种降钙素原检测系统进行方法学比较及偏倚评估

A Method Comparison and Bias Evaluation of Two ProcalcitoninDetection Systems Using CLSI EP9-A2-IR Guidelines

目的评估两种降钙素原检测系统的偏倚及可比性。方法应用美国临床和实验室标准化协会(CLSI)EP9-A2-IR文件,评价德国罗氏(Roche)公司Cobas e601电化学发光免疫分析系统与美国贝克曼库尔特(Beckman Coulter)公司Unicel DXI800全自动顺磁性微粒子化学发光免疫分析系统检测降钙素原的偏倚及可比性,以Cobas e601检测系统作为参比方法,进行线性回归分析、偏倚分析,计算医学决定水平(X_(c))和临床常用具有指导意义的浓度(X_(c)=0.05、0.1、0.2、0.25、0.5、1、2、5、7、10ng/mL)时的预期偏倚(Bc)以及可接受性评估。结果两种检测系统测定结果之间呈良好的线性关系,相关系数r=0.997(R^(2)=0.994)(n=48,P<0.01),95%CI为0.995~0.998,线性回归方程为Y=1.21 X+0.332,斜率(b)95%CI为1.18~1.24,截距(a)95%CI为-0.11~0.77,线性回归的Cusum检验P<0.05,表示差异具有统计学意义。偏倚分析显示呈正偏倚趋势,平均绝对偏倚为1.8ng/mL,平均相对偏倚为24.9%。两种检测系统在PCT浓度临界值为0.05、0.1、0.2、0.25、0.5、1、2ng/mL时,预期偏倚可接受;当PCT浓度≥2.4ng/mL以上时两种检测系统就出现了明显差异,且差异不可接受,PCT浓度临界值为5、7、10ng/mL时预期偏倚不可接受。结论两种降钙素原检测系统的测定结果存在差异,当PCT浓度≥2.4ng/mL以上时可比性较差,临床常用的具有指导意义的cut-off值不能随意套用,必要时应重新建立新的适合的cut-off值为临床提供诊疗服务。

Objective To evaluate the bias and comparability of two procalcitonin detection systems.Methods The American Society for clinical and laboratory standards institute(CLSI)EP9-A2-IR guidelines was used to evaluate the bias and comparability of Cobas E601 electrochemiluminescence immunoassay system of Roche company in Germany and Unicel DXI800 automatic paramagnetic particle chemiluminescence immunoassay system of Beckman Coulter company in the United States.Using Cobas E601 detection system as the reference method,linear regression analysis and bias analysis were conducted.Expected bias(Bc)and acceptability assessment were obtained by calculating medical decision level(X_(c))and clinically commonly used guiding concentrations(X_(c)=0.05,0.1,0.2,0.25,0.5,1,2,5,7,10ng/mL).Results There was a good linear relationship between the measurement results of the two detection systems,with correlation coefficient r=0.997(R^(2)=0.994)(n=48,P<0.01),with 95%CI 0.995-0.998,and linear regression equation Y=1.21 X+0.332,slope(b)95%CI 1.18-1.24,intercept(a)95%CI-0.11-0.77.Cusum test of linear regression showed P<0.05,indicating that the difference was statistically significant.Bias analysis showed a positive bias trend,the average absolute deviation was 1.8ng/mL and the average relative deviation was 24.9%.The expected bias of the two detection systems was acceptable when the critical values of PCT concentration were 0.05,0.1,0.2,0.25,0.5,1 and 2ng/mL.When the PCT concentration was over 2.4ng/mL,the two detection systems had significant differences,and the differences were unacceptable when the critical values of PCT concentration were 5,7 and 10ng/mL.Conclusion The results of the two procalcitonin detection systems are different and the comparability is poor.When PCT concentration is over 2.4ng/mL,the comparability is poor,and the cut-off value with guiding significance commonly used in clinic cannot be applied arbitrarily.A new suitable cut-off value should be reestablished for clinical diagnosis and treatment services.