Background: The aim of our study is to examine the performance of an algorithm used by autoverification middleware (LimasVfy; Hwasan System, Korea) in the diagnostic hematology laboratory of a single tertiary hospital in Korea.
Methods: In this study, we included manually verified complete blood count (CBC) results from June 2019 to January 2020, as well as CBC results from February 2020 to September 2021 that underwent both autoverification and manual verification. The autoverification rates and turnaround time (TAT) of both datasets were compared, thereby obtaining an effective comparison between the period before and after the introduction of the autoverification system.
Results: After introducing autoverification, the rates of autoverification and manual verification were 64.6% and 35.4%, respectively. Of the manually verified cases, those with delta check, panic value, flags, and critical values were 46.4%, 28.2%, 19.8%, and 5.6%, respectively. After the autoverification rule related to platelet count was modified in August 2020, the rate slightly increased from 63.9% to 64.8%, and critical value reporting rates decreased from 26.8% to 2.4%. Prior to autoverification, the median time from reception to production of results/from production of results to final reports was 9.6/35.1 minutes. After introducing autoverification, it was 10.0/39.9 minutes in manually verified cases and 3.2/0.0 minutes in autoverified cases. Total time, that is the product of total number of cases and average TAT, decreased from 350,000 to 200,000 minutes. Rate of excess TAT also decreased from 0.065% to 0.045%.
Conclusions: LimasVfy would be a useful tool in reducing TAT and verification workload in the CBC testing, and this study can provide useful information in determining the introduction of autoverification systems in the setting of diagnostic hematology.

Keywords: Autoverification, Middleware, Postanalytical, Turnaround time, Blood cell count