Efficiency Improvement of The Workflow for Nucleated Red Blood Cell (NRBC) Enumeration by Beckman Coulter Unicel®DxH Hematology Analyzer

Abstract

Manual blood  smear examination is a standard method for nucleated red blood cells (NRBC) reporting. However, it is a time-consuming, labor-intensive, and costly method. Modern hematology  analyzers can automatically quantify NRBC but with some limitations. This study aimed to establish a workflow for NRBC reporting to maximize its accuracy as well as to enhance its  efficiency.  NRBC  measurement of  241  blood  samples was  carried out  using  Beckman Coulter Unicel®DxH800.  The results were compared against manual examination. The receiver operating  characteristic  (ROC)  analysis  and a  decision  tree  model  (Gini  impurity)  were conducted to construct workflows  which incorporated NRBC (%) and related flags from the analyzer. The effectiveness  of workflows was evaluated by sensitivity, specificity, positive (PPV) and negative predictive values (NPV) while efficiency  was assessed by the rate of manual slide review. The strong agreement between the automated analyzer and the manual NRBC count was found (r = 0.98; p < 0.0001,  bias -0.7).  To reduce clinical errors by taking NPV into account, the findings were as followed. By using the cutoff at the limit of detection of the analyzer (<0.1%) alone,100% NPV could be achieved but the efficiency was compromised due to the high slide review rate (59.3%).  On the other hand, the cut-off  derived from the ROC analysis (<0.4% and <1.8%) could drastically reduce the slide review rate (29.0% and 7.9% respectively) but the false-negative rate was escalated (7 and 32 samples) with compromised NPV (95.9% and 85.6%). Alternatively,   100%  NPV  could  be  achieved  using  a decision   tree  that  incorporated   both  a numerical value and certain flags with acceptable efficiency  (48.1% manual slide review rate). Reporting  NRBC  from  the  automated  hematology   analyzer  improves  the  workflow  and efficiency   of  the  laboratory.  However,  due  to  the  variation  of  analyzers  and  patient characteristics, an individual laboratory should construct and verify its own workflow before implementation to ensure acceptable accuracy and efficiency.