1 025 EDTA-K₂ anti-coagulated venous whole-blood samples were collected from Ruijin Hospital(including outpatient and inpatient samples). Each sample was both analyzed on the analyzer in CAL 8000 auto sample processing line and performed manual differential. The analysis results and microscopic results were used to evaluate the drafted automatic verification rules. Furthermore, 40198 outpatient and inpatient sample results obtained from March to May in 2014 were used as experimental data by Mindray automatic verification software(LabXpert), where the percentage of samples passing the automatic verification was obtained and the samples failing in the verification were categorized. A comparison of verification time using the software and in manual way was also performed.

The false negative percentage of 1 025 samples tested was 0.78%, lower than the 5% required by the International Hematology Consensus Group. The false negative items mainly included erythrocyte morphology, platelet clump and toxic granules of neutrophils. By analysis of the 40 198 samples using the automatic verification software, 65.3% samples need automatic verification, and 34.7% need manual verification. Samples requiring manual verification included: PLT out-of-range samples, WBC out-of-range samples, samples from hematology department, samples with immature granulocyte flag, RBC and MCV out-of-range samples, samples with blast cell flag, HGB out-of-range samples, etc.

Using the automatic verification software increases clinical work efficiency, decreases false negative ratio, saves the report time and improves the patients' satisfaction.