Original article

Automated haematology analysis and blood smear evaluation in paediatric leukaemia: a method-comparison study

---

Abstract

Aim: To compare complete blood count and leucocyte differential results from the Yumizen H2500 and Sysmex XN-3000 in paediatric leukaemia samples and assess Yumizen H2500 differential counts against manual blood smear evaluation.

Methods: This cross-sectional study included 101 paediatric leukaemia blood samples. Samples were analysed on both automated haematology analysers within one hour of collection, followed by a 200-cell manual differential count. Passing–Bablok regression assessed constant and proportional bias, Lin’s concordance correlation coefficient (CCC) agreement, and Spearman’s correlation associations between Yumizen H2500 differential parameters and microscopy.

Results: Among 101 leukaemia samples, white blood cells showed the highest concordance (CCC = 0.999), followed by neutrophils (CCC = 0.985), lymphocytes (CCC = 0.982), eosinophils (CCC = 0.977) and platelets (CCC = 0.967), whereas nucleated red blood cells and mean corpuscular haemoglobin concentration showed low concordance (CCC = 0.279 and 0.170, respectively). No constant or proportional bias was detected for 11 parameters. Mean corpuscular volume and platelet distribution width showed proportional bias, eosinophils, monocytes and immature granulocytes constant bias, and mean corpuscular haemoglobin concentration, red cell distribution width standard deviation and platelets both types of bias. Manual counts correlated strongly with eosinophils (ρ = 0.716), segmented neutrophils (ρ = 0.853) and lymphocytes (ρ = 0.724) (all p < 0.001).

Conclusion: The analysers showed high concordance for several routine haematological parameters but were not universally interchangeable, supporting manual smear review for discordant or immature-cell findings.

Keywords: blood cell count, leukocyte count, microscopy, reproducibility of results