Inaccuracy and imprecision of reticulocyte counting.

Reticulocyte count is an essential clinical test to detect erythropoietic activity. The test has been done manually with the use of one of three dyes: new methylene blue, brilliant cresyl blue, and azure B. When reticulocyte counts with the different dyes were compared, correlation was good. When reticulocyte counts obtained by use of the dyes were compared with the count obtained by an automated counter, correlation was high. In a sampling study, the specimen prepared by a spinner method gave smaller variations than a wedge method. Intertechnologist bias was higher than inter-specimen or intra-specimen bias: The variation of the reticulocyte count was examined with from 200 to 2,000 cells; with 1,000 cells or more, variation was low. From these results, we concluded that more than 1,000 erythrocytes on one slide prepared by the spinner method should be analyzed. Clear standards for cell identification are also needed.