两种总蛋白试剂测定结果的偏倚分析

Bias analysis on results of total protein detected with different reagents

目的探讨两种总蛋白(TP)试剂测定结果间是否具有可比性。方法依据EP9-A2文件要求,每天随机从临床样本中抽取8份不同浓度水平的血清样本,分别用两种试剂进行总蛋白测定,共测定5d,记录检测结果,将结果进行相关回归分析。计算在其医学决定水平处的预期偏倚(B^c)和预期偏倚的95%可信区间,并判断偏倚是否可以接受。结果相关系数(r)=0.993,r2=0.986,截距(a)=0.362,斜率(b)=0.991,回归方程:Y=0.991X+0.362。无方法内、间的离群点,测定结果的偏差在允许误差范围内。结论在实验室用两种以上试剂检测同一项目前,可参照EP9-A2文件进行方法学比对和偏倚分析,判断其临床可接受性能,以保证实验室检测结果的准确性和可比性。

Objective To investigate the comparability of total protein（TP） test results between two kinds of reagents.Methods Based on the guideline of EP9-A2,eight samples with different concentrations were randomly selected from the clinical samples and measured TP by two kinds of reagents for 5 d consecutively.All results were recorded and subjected to conduct the relative regression analysis.The anticipated bias （at medical decision level） and its 95% confidence interval were calculated.To estimate the acceptability of the bias.Results The correlation coefficient of using two kinds of TP reagents was 0.993 （r2=0.986）,intercept（a）：0.362,slope（b）：0.991,the regression equation：Y=0.991X＋0.362,without outliers of within-method and between-method,the bias of results were accepted in the range of the allowable error.Conclusion When analyzing the same test item by over two kinds of reagents,it is necessary to evaluate the clinical acceptability to insure the accuracy and comparability by method comparison and bias estimation based on EP9-A2 document.