Exploration of statistical methods of analysis of inter-laboratory comparison results / 中国药学杂志

ABSTRACT

OBJECTIVE: To explore rational statistical methods for analysis of the results of inter-laboratory comparisons, which will be helpful to improve the statistical analysis ability of organizers of proficiency testing project against pharmaceutical laboratories. METHODS: Robust statistical Z-score method and iterative computation method were used to analyze the assay result of cefminox sodium powder for injection respectively, and the critical statistical parameters of the two methods were compared. RESULTS: The critical statistical parameters of robust statistical Z-score method anf iterative computation method were as follows: the robust average were 77.71% and 77.63%, the robust standard deviation were 0.66 and 0.67, and the coefficients of variation were 0.85% and 0.86%, respectively. CONCLUSION: It is very important to select the correct statistical approach to evaluate the proficiency of pharmaceutical testing laboratories. By comparing the steady average and the median value of Z-score, the range of best estimates for the true value can be found, and the judgment for the outliers can be more accurate.