Application of ISO 13528 robust statistical method for external quality assessment of blood glucose measurements in China

External quality assessment (EQA) including extensive participants often result in abnormal distribution. Robust statistical methods are insensitive to slight deviation for a given probability model and can estimate the population parameters utilizing robust algorithm. The aim of this study was to evaluate the blood glucose testing performance of EQA in China using ISO 13528 robust statistical method. A total of 4571 participants measured two lots of materials at different concentrations and submitted the results through the Clinet-EQA reporting system V1.5. Data were collected and analyzed by Clinet-EQA evaluating system and SPSS 13.0. Participants were classified by method principles they adopted. Acceptability of relative differences between each result and robust mean of the group was calculated using biological variation desirable quality specification for allowable total error. The kernel density plots and Youden Plot were also applied to show the distribution of z-scores. The robust means of the two lots were 6.31 mmol/L and 15.71 mmol/L, and the robust standard deviations were 0.23 mmol/L and 0.5 mmol/L, respectively. Acceptability of relative differences was more than 90 % among all groups. About 91.5% and 89.7 % of the participants out of the two lots fulfilled the condition of |z| ≤ 2. In addition, 87.2 % participants fell inside the confidence ellipse for probability levels of 95 % in Youden Plot. According to the statistical results, we can conclude that most participants are qualified for performing blood glucose test and ISO 13528 robust statistical method is appropriate for EQA result analysis.