To explore the feasibility of using the sigma metrics calculated with the data of internal quality control for the comparison of the analytical performance between different biochemical analyzers.

The internal quality control results of twenty-five biochemical assays in the biochemical analyzers of the department of clinical laboratory in Cancer Hospital from February 1, 2021 to July 31, 2021 were collected. The formula sigma =(TEa-Bias)/CV was used to calculate the sigma metrics of two different levels of the biochemical assays including albumin, alkaline phosphatase, alanine aminotransferase, aspartate aminotransferase, calcium, cholesterol, creatine kinase, chlorine, creatinine, γ- glutamyltranspeptidase, blood glucose, high density lipoprotein cholesterol, immunoglobulin A, immunoglobulin G, immunoglobulin M, potassium, lactate dehydrogenase, low density lipoprotein cholesterol, sodium, inorganic phosphorus, total bilirubin, triglyceride, total protein, urea, uric acid. The imprecision was obtained by the coefficient of variation of internal quality control. The bias was calculated by the deviation between the mean of internal quality control of the comparison instrument and the target instrument. The allowable total error (TEa) was based on People's Republic of China Health Industry Standard (WS/T403-2012) or EQA standard of National Center for Clinical Laboratories (NCCL). Compared the sigma values of the comparison instrument relative to the target instrument with the average percentage bias obtained by the traditional comparison method. Quality goal index was used to analyze the causes of poor performance and judge the results of comparison.

Compared with the target instrument Beckman AU5800-3, the comparison instrument Beckman AU5800-1 had 10 assays with σ>6, accounting for 40%, 23 assays with σ>3, accounting for 92%, and only albumin and blood glucose showed σ<3. Through statostical analysis, the comparisons of all assays were passed. The comparison instrument Beckman AU5800-2 had 8 assays with σ>6, accounting for 32%, 20 assays with σ>3, accounting for 80%. Only alkaline phosphatase, calcium, lactate dehydrogenase, total protein and urea showed σ<3. Through statostical analysis, the comparisons of GGT and IgM failed. For the traditional comparison method, the percentage bias between the comparison instruments and the target instrument were all within the range of the evaluation standard. But there was no significant correlation between the σ value and the average bias of the traditional comparison method, and the biases were correlated.

Using the sigma metrics calculated with the data of internal quality control for the comparison of different detection systems is a convenient and operable method. It can monitor the comparability between different detection systems in the laboratory at any time and be the supplement of the traditional comparison method.

Quality control; Sigma metric; Comparability; Detection system