Hematopathology / ELIMINATION OF REFLEX MANUAL DIFFERENTIAL COUNTS

Automated peripheral blood leukocyte differential counts (LDCs) are widely accepted in routine practice. However, many laboratories still reflexively perform manual LDCs based solely on abnormal automated results or instrument “flags,” before any manual triage step. We describe our transition to a procedure that uses manual methods to validate, rather than to replace, automated LDCs (an approach recommended early in the development of automated methods, but still not used in many clinical laboratories). Manual microscopic scans were performed in lieu of manual LDCs. Each scan that revealed cell types not quantifiable by the instrument triggered a manual LDC. However, if the manual scan simply confirmed the cell types seen on automated LDC, then the automated result was released, even if clinically significant quantitative abnormalities were present. This policy reduced manual LDCs by more than 70% and was validated by a manual retrospective audit. Patient care and laboratory operations can be optimized by using manual microscopic examination as a validation procedure rather than as a reflexive substitute for automated methods. There is no clinical rationale for reflex performance of manual LDCs based solely on instrument warnings. The accuracy and clinical usefulness of automated leukocyte differential counts (LDCs) have been validated in numerous studies.1-9 In most automated laboratories, manual differential counts are limited to cases in which instrument “flags” indicate the potential presence of cells not reliably identifiable by automated methods (such as blasts or other immature granulocytes) or of findings that may interfere with automated analysis (such as overlap in the distribution of different cell types or interference from matrix components such as cryoglobulin).3,10,11 Since the inception of automated differential counting methods, investigators have recommended the use of manual blood smear review as a validation of, rather than as a replacement for, automated methods.11-14 Although some clinical laboratories adhere to this principle,11 many laboratories continue to participate in the bias toward manual LDCs by building rules by which manual counts replace automated counts based solely on the presence of certain automated findings, before any manual scanning or triage.15-17 In essence, many laboratories use manual counts as reflexive substitutes for automated methods, rather than as validation steps. Reflexively ordering manual LDCs based on instrument flags can be clinically problematic. For instance, an instrument flag may be issued for suspicion of cell types not actually in the blood sample, resulting in the manual LDC as a simple validation of the automated result, but with the release of the less accurate and less precise manual result. Alternatively, quantitative flags may trigger manual LDCs when the only finding at issue is confirmation of the correct qualitative identification of the quantitatively abnormal cell type. This arbitrary replacement of automated LDCs by manual LDCs can have clinical repercussions, particularly when clinical decisions are heavily dependent on accurate and precise absolute neutrophil counts or other LDC results.17 Hematopathology / ORIGINAL ARTICLE Am J Clin Pathol 2003;119:656-662 657 657 DOI: 10.1309/VH1KMV8WB7GB7R14 657 © American Society for Clinical Pathology We describe our transition to a system that eliminated the reflexive performance of manual LDCs based on instrument flags and replaced reflex manual LDCs with a manual microscopic triage system. Manual LDCs were reserved for those cases in which the microscopic triage resulted in the discovery of cell types or findings not quantifiable by the instrument. Our purpose was to design a system that directed the use of automated and manual LDCs toward greatest clinical benefit and optimized laboratory operations by fully using the potential of automated methods and permitting medical technologists to focus on tasks not achievable by these methods. Materials and Methods The University of Michigan (Ann Arbor) Medical Center hematology laboratory performs more than 340,000 CBC counts and more than 180,000 LDCs per year. Samples are run on 1 of 4 Beckman-Coulter GenS analyzers (Hialeah, FL). As with other systems, the GenS system provides a 5part LDC (including neutrophils, lymphocytes, monocytes, eosinophils, and basophils) and uses an array of flags that serve as signals of the potential need for manual review of automated results.18 “Suspect” flags are built into the instrument technology and generally are based on aberrations in the distribution of cell populations or the detection of potential interference with these distributions (eg, suspect blasts, suspect immature granulocytes). “Definitive” flags are userdefined and generally are based on quantitative thresholds defined by each individual laboratory (eg, lymphocytosis, macrocytosis). Previous Manual Review Policy Before implementing our revised manual review policy, our standard procedure included instrument criteria by which blood smears should be microscopically scanned before determining the appropriateness of a manual LDC and instrument criteria by which a manual LDC should be ordered reflexively, without initial microscopic scan. The process is illustrated in ❚Figure 1❚, and specific criteria are outlined in ❚Table 1❚. LDC order