Comparison of semiautomated method with official optical somatic cell counting method III for determining somatic cells in milk.

The new method specifying the Fossomatic-90 differs from the official method, 46.105-46.109, in that the modified instrument includes a halogen lamp; a semiconductor photoelectric detector; a less expensive, bench-top cabinet; manual injection of a larger sample, and a reduced capacity. The new instrument was compared with 2 optical somatic cell counters in routine use. On each of 3 days, 12 subsamples were prepared for each of 5 cell count levels from AM milk with half kept fresh and half preserved with 0.05% potassium dichromate. Subsamples were refrigerated and read 30+ h post-collection. Duplicate sets were read in random order on each machine daily (CV 0.77%). Two sets of slides read by 2 technicians each (strip reticle on 2 smears/slide) gave geometric mean direct microscopic somatic cell count (DMSCC) levels of 296, 526, 772, 930, and 1438 th/mL. Within-technician CV values (from day-level means) ranged from 1.68 to 2.28%. Geometric mean cells in th/mL on the new machine were significantly higher than those on the other two (674 vs 621) and were closer to the DMSCC (694). On the new machine, cell counts were 8.5% greater than on the original machines, were only 2.9% lower than the DMSCC, and showed no significant evidence of bias. Preserved samples averaged slightly greater than fresh (5.3%) but only on the original machines. Carryover by covariance analysis was insignificant. Except for cell levels, high machine precision (error CV value of 1.18%) gave differences with statistical but not practical significance, even for regulatory laboratories.