Multicenter evaluation of blood cell counts on 8 automated hematology analyzers

The cellular analysis performed on hematology analyzers is based on the interaction of cells with electrical or optical signals. The heterogeneity of adopted methods and technologies by different analyzers can translate in a lack of homogeneity in analytical performance. This study compares 8 hematological analyzers vs. optical microscopy ( OM) and, where possible, also compares the analyzers among each other. Correlations were assessed by Pearson's coefficient of correlation, Passing and Bablock regression and Bland-Altman bias plot analysis. The comparison among analyzers regarding leukocyte differential counts showed a good level of agreement, except for the basophil cell count. For this "critical population", the bias ranged from -5,8% (Cell-Dyn Sapphire vs. XN-9000) to 30,6% ( Advia 2120i vs. XE-2100). The comparison between automated differential leukocyte counts and OM showed also a good level of agreement, with a bias ranging from -0,9% to 8.9%. The bias for basophil cell count was however very high (79.5%).