4种血细胞分析检测系统的一致性评价

Evaluation of the consistency of four hematology systems

目的为儿童血细胞分析参考区间在不同检测系统的应用提供一致性数据依据。方法依据美国临床和实验室标准化协会(CLSI)EP9-A3文件要求,选取45份临床全血标本。以SysmexXN20-A1作为基准检测系统,分别与贝克曼-BECKMAN(DSH800)、西门子-ADVIA 2120i、迈瑞-BC5310检测系统做比对分析。用4个系统分别检测全血细胞计数及白细胞分类,采用广义极端学生化偏差(ESD)法对检测结果离群值进行检验,通过散点图、偏差图、频数分布图选择最佳回归模型。以选定的最佳回归模型拟合回归方程并计算参考区间及医学决定水平处偏移。全血细胞计数以《临床血液学检验常规项目分析质量要求》允许偏移作为可接受范围,白细胞分类以澳大利亚皇家学院室间质评的允许误差作为可接受范围。结果将离群值删除后,散点图显示,基准系统与另3个比对系统所有参数呈线性关系,偏差图显示差值均为非恒定变化,依据频数分布图选取Deming回归或Passing-Bablok进行回归分析。基准系统与另3个比对系统全血细胞计数及白细胞分类绝对值的相关系数R2在0.95~0.99。在参考区间的上、下限处除MONO#项目在0.12×109/L水平处XN-20A1与ADVIA 2120检测系统偏移不可接受外,其他项目偏移均在可接受范围内。疾病诊断层面的医学决定值所有项目偏移均在可接受范围内,PLT项目下限治疗及预后相关的医学决定值部分结果不可接受。结论基准系统与比对系统的全血细胞计数及白细胞分类结果在儿童参考区间范围内具有良好的一致性,为不同检测系统采用统一参考区间的可行性提供科学依据。

Objective To provide consistent data basis for the application of reference intervals for children blood cell analysis in different testing systems.Methods According to the requirements of American Institute for Clinical and Laboratory Standardization(CLSI)EP9-A3 document,45 samples were collected and Sysmex XN20-A1 were used as reference system.Beckman DxH800,Siemens ADVIA 2120i,and Mindray BC5310 were comparison systems.Complete blood count and leukocyte classification were performed by four systems.The outliers of the detection results were tested by the generalized extreme student deviate(ESD)method.An optimal regression model was selected by scatter diagram,deviation diagram and frequency distribution diagram,which was used to fit the regression equation and calculate the deviation at the medical decision level and reference interval.The acceptable range for blood count deviation was cited from the Analytical Quality Specifications for Routine Tests in Clinical Hematology.The acceptable range for leukocyte classification was based on the EQA program of Royal College of Pathologists of Australasia(RCPA).Results After the outliers were deleted,the scatter plot showed a linear relationship between the reference system and the three comparison systems.The deviation plot showed that the differences were variable.Deming regression or Passing-Bablok regression was selected according to the data distribution.The determination coefficient R2 of reference system and three comparison systems ranged from 0.95 to 0.99 in blood count and leukocyte classification.At the upper and lower limits of the reference interval,the deviations between XN-20A1 and ADVIA 2120 system were all acceptable,except for MONO#at 0.12×109/L.The deviations of all parameters at medical decision level were within acceptable ranges.The lower limit of PLT is partially unacceptable at the level of medical decision related to treatment and prognosis.Conclusions The results of complete blood count and leukocyte classification in reference system and the comparison system had good consistency within the children′s reference interval.Our study provided a scientific basis for the feasibility of adopting a unified reference interval for different detection systems.