Preliminary evaluation of the sources of variability of the platelet estimate on canine blood smears

Background The platelet estimate (PE) is based on platelet count on stained blood smears and it is usually employed in routine hematology as an indirect evaluation of the platelet mass, independently on the instrumental platelet count. No information is available on observerdependent variability of PE or on the intrinsic variability of PEs on routinely performed blood smears. Objective Aim of this study is to preliminarily assess the magnitude of the aforementioned sources of variability of PEs. Methods Fifteen blood smears were microscopically read by two independent observers on three different areas of the smear (lateral edge - LE - central monolayer - CM - feathered edge -FE). In each area the number of platelets in 10 HPF was counted by each observer. Intra- and inter- observer coefficient of variations (CVs) were calculated, and the results recorded in the three areas were compared to each other using an ANOVA test for paired measurement. Results Results of the two observers were significantly different in the FE but the intra-observer CVs were always high (from 18% to 72%) but significantly different for the two observers only in the CM. For both the observers, median and mean values were higher in the CM than in the LE and lower in FE than in the other two areas and the CVs were higher in the CM than in the two other areas. However, all the differences above were statistically significant only for one of the two observers. Conclusions The PE suffers from an observer-dependent and from an intrinsic variability that should be taken into account in routine practice. Further studies are needed to evaluate the possible impact of this variability on clinical classification of patients with low instrumental platelet counts.