Evaluation of Coulter S-Plus three-part differential in population with a high prevalence of abnormalities.

The authors evaluated the Coulter S-Plus three-part differential and compared it with the total of component cells obtained by manual slide differential. The patient population (N = 989) was strikingly abnormal, in that 109 samples contained nucleated red blood cells, 118 samples contained one or more abnormal cells (blasts, promyelocytes, or lymphoma cells), and 95 samples had greater than 10% monocytes. The GRAN and LYMPH fractions were accurately measured in the specimens that did not display region (R) flags. The MONO fraction correlated poorly with the manual differential and underestimated monocytes when they were increased. Not all samples with small numbers of abnormal cells or nucleated red blood cells were flagged by the three-part differential. Specimens with eosinophilia and reactive lymphocytosis likewise did not always show R-flags. The following findings (prevalence = 23.9%) were arbitrarily defined to constitute an "abnormal" manual differential: nrbc greater than 0%, blasts greater than 0%, promyel greater than 0%, lymphoma cells greater than 0%, baso greater than 5%, myel greater than 5%, meta greater than 5%, eos greater than 10%, reactive lymphs greater than 10%. Sensitivity of R-flagging for these abnormalities was 81.7% and specificity 73.3%. The predictive value of an R-flagged ("positive") result was 48.3% and of an unflagged ("negative") result, 93.0%. Efficiency (accuracy) was calculated to be 75.5%.