Evaluation of Visual Serum Indices Measurements and Potential False Result Risks in Routine Clinical Chemistry Tests in Addis Ababa, Ethiopia

BACKGROUND: Serum indices (SI) including hemolyzed, lipemic, and icteric samples, affects the accuracy of test result. The aim of this study was to evaluate SI values done by visual inspections and potential false result risks by comparing with actual measurements done by Cobas 6000 Chemistry analyzer at Ethiopian Public Health Institute (EPHI), Addis Ababa, Ethiopia. METHODS: An observational, cross-sectional study was conducted from April to May 2017 on samples referred to Clinical Chemistry laboratory of EPHI, Ethiopia. These samples SI values, after visual inspection by three trained observers, was analyzed again on Roche Cobas 6000 analyzer (RCA). The generated data was analyzed by using weighted kappa methods on STAT statistical software version 20. RESULTS: From a total of 1509 samples, SI values identified by the RCA as hemolysis, icteric, and lipemic were 933 (62%), 74(5%) and 59(4%) respectively. The SI average weighted kappa between RCA and visual inspection were: 0.1870, 0.3421, and 0.1259 for hemolysis, icteric, and lipemic samples, respectively. Combined inter-observers variability among observers for hemolysis, Icterus, and lipemic samples were 0.4758, 0.3258, and 0.3628 respectively. The best agreement among observers was in the case of hemolysis (0 grades), while the lowest agreement was observed in the case of icterus (+3 grades). In addition, test parameters, such as CK-MB (22%), and LDH (20%) were falsely accepted, whereas Cl(-) and Na+ (up to 25%) were falsely rejected tests by observers. On the other hand, results rejected by Cobas SI assessments included CK-MB (22%), LDH (20%), and BIL-D (4%). CONCLUSION: Visual inspection of SI showed poorly agreement with automated system. Thus, there is genuine need for more training of Laboratory professionals on identification of SI, and as much as possible SI should be done by automated system to improve quality of test results.