Automated flagging influences the inconsistency and bias of band cell and atypical lymphocyte morphological differentials

Abstract

AbstractThis study evaluated inter- and intra-observer variabilities of band cell and atypical lymphocyte differentials and the influence of instrument flagging information on resulting microscopic differentials. Five stained slides with a range of band cell counts and five with variable numbers of atypical lymphocytes were sent for morphological review by 30 technicians. No supplementary full blood cell count information was provided. Two months later, the same slides were sent, together with their corresponding analyzer reports comprising the full blood cell count, automated differentials and flags, to the same technicians. The first and second appraisals of band cells and variant lymphocytes both showed poor levels of inter-observer consistency. Observed values for all slides were very wide and suggested a high inherent predisposition to erroneous reporting practices. Analysis of category trends showed that analyzer left shift or immature granulocytes flags had no influence on observer band cell assessments as downward vs. upward category revisions were evenly balanced. The findings for atypical lymphocytes were, however, somewhat different. Two slides with no flags both showed balanced category revisions, whereas two of the three slides with atypical lymphocyte flags showed clear evidence of upward category revision. The third slide with an atypical lymphocyte flag did not show any overall category trend, but six of the seven observers who in the first examination recorded atypical lymphocyte estimates of ≤30% revised their estimates upward when the slides were examined the second time. These results suggest that morphologist access to an analyzer report and flagging information is unlikely to affect the “randomness” of band cell determinations but it may induce observer bias in variant lymphocyte estimates.