Analytical variability and uncertainty in canine leukocyte ratios obtained with manual counts

This study aimed to determine the analytical imprecision of calculated neutrophil to lymphocyte (NLR) and lymphocyte to monocyte ratios generated from manual differential white blood cell (WBC) counts in peripheral blood smears, and to describe how to report the uncertainty around a single WBC ratio result. No information on the analytical imprecision of WBC ratios in dogs is available.

Coefficient of variations (CVs) of paired readings of one operator on 105 smears (intraoperator variability) and of three operators on 301 smears (interoperator variability) were calculated. The interoperator agreement was examined with the Fleiss' kappa coefficient (κ). Observed total errors (TEos), expanded measurement of uncertainty (EMU) and reporting intervals (RIs) were also calculated.

Median CVs ranged from 3.14 to 28.28 (intraoperator) and from 5.39 to 53.85 (interoperator). No agreement among operators was found around the cut-offs. TEos were higher than allowable total errors in 32%-88% of smears. EMU ranged from 0.10 to 1.13. According to the RI, the calculated WBC ratios should be rounded to the nearest 10.

WBC ratios should be interpreted cautiously in dogs. The EMU should be reported to make the clinician aware of the uncertainty of these parameters. For example, an NLR result of 17 is needed to have high confidence that the result is above a cut-off of 6.