全线性范围内多个定量检测系统间的偏倚评估

Improvement of EP9-A2 for bias estimation among multiple quantitative detection systems in full range of AMR

目的改进EP9-A2指南使之可用于全线性范围内多个定量检测系统间的偏倚评估。方法在4套检测系统(A、B、C、D)上检测40份患者标本的血清总胆固醇浓度,以A系统为参考系统,按EP9-A2指南评估B、C、D系统的偏倚;在此基础上,计算两两系统间偏倚可接受距离DD(偏倚与分析质量目标之间的距离),以置信区间包含0为标准来识别系统间不可接受的偏倚。结果与A系统相比,B、C两系统呈负向偏倚,D系统呈正向偏倚,但B、C、D与A系统的偏倚均可接受;各系统间DD呈正态分布;除DDBD置信区间包含0,B、D两系统间偏倚不可接受外,其余系统间偏倚均可被接受;将D系统结果经A系统校正后,各系统间偏倚均可接受;作图可提供全线性范围内的偏倚评估。结论 DD均值的置信区间,可在全线性范围内有效评估3个或更多系统间的偏倚,可用于评估系统间的结果可比性。

Objective To improve EP9-A2 for Bias estimation among multiple quantitative detection systems within full range of AMR.Methods 40 patients specimens were determined twice for serum total cholesterol by four detection systems（A,B,C and D）.With system A served as comparative method,Bias between A and the other three was evaluated according to CLSI EP9-A2 separately.Furthermore,DD（distance from deviation to tolerable error）and its average confidence intervals between every two system were calculated and compared with zero.The confidence interval of greater than zero was served as criteria for accepting bias between every two system.Results Bias between A and the other three meet the analytical quality specification according to EP9-A2,although that of D system was positive,and those of B and C system were negative.DD between every two system obeyed normality distribution.All biases between every two system wes acceptable except that between B and D,causing of their interval containing zero.After correcting of results from system D,Biases between every two system were all acceptable.Plots of confidence interval could provide a full range bias assessment within AMR.Conclusion Comparability and Bias estimation in full range of AMR for results between every two system among 3 or more systems could be evaluated by confidence intervals.