Validation of body fluid analysis on the Coulter LH 750

Lab Hematol. 2003;9(3):155-9.

Abstract

The role of the hematology laboratory in the analysis of body fluid has been to provide accurate enumeration of red blood cells (RBCs), total nucleated cells (TNCs), and differentials by manual analysis. Three hospitals (London Health Sciences Centre, University of Pittsburgh Medical Center, and University of Michigan Health System) participated in the assessment of the performance of automated analysis of body fluid by the Beckman Coulter LH 750, an impedance-based hematology analyzer. We evaluated the accuracy of analysis results for both the TNCs and RBCs of 372 samples (158 serous fluid, 148 cerebrospinal fluid [CSF], 66 synovial fluid) run on the LH 750 compared to results obtained from manual chamber counting. Of the 372 samples, 152 were suitable for evaluation of accuracy of the automated TNC. The remaining 220 samples were either flagged for interfering substances or the reference results were < 0.2 x 10(9)/L, below the background limit of the analyzer. Correlation coefficients for serous fluid were 0.895, P = .88; for CSF, 0.993, P = .84; and for synovial fluid, 0.836, P = .94. Of the 372 samples, 106 had RBC counts greater than 0.01 x 10(12)/L and were used for method comparison. Correlation coefficients for serous fluid were 0.957, P = .66; for CSF, 0.849, P = .55; and for synovial fluid, 0.667, P = .81. Linearity and precision studies showed excellent agreement for both TNC and RBC parameters. Low-level sensitivity excluded the majority of cerebrospinal (119) and a small number of peritoneal dialysate fluid samples (8), which require accurate enumeration at clinical decision points between 0 to 100 cells/microliter. In the case of synovial and serous fluids, however, most clinicians are interested in TNC counts above 0.2 x 10(9)/L, and RBC counts are relevant only if they are significantly increased (> or = 0.05 x 10(12)/L). Adopting the criteria of reporting TNC counts as < 0.2 x 10(9)/L or accurate enumeration on counts > or = 0.2 x 10(9)/L, clinically relevant results could be provided by automated analysis in 93.8% of serous fluids and 85.8% of synovial fluids.

Publication types

  • Validation Study

MeSH terms

  • Blood Cells / cytology
  • Body Fluids / cytology*
  • Cell Count / instrumentation*
  • Cell Count / standards
  • Cerebrospinal Fluid / cytology
  • Erythrocytes
  • Humans
  • Leukocytes
  • Reproducibility of Results
  • Synovial Fluid / cytology