Application of EQA data in the measurement of bias uncertainty

Objective To investigate the minimum number of trials for external quality assessment( EQA) data in evaluating bias uncertainty. Methods The relative bias uncertainty( ubias,rel) consists of the relative uncertainty from the certified value( ucref,rel) and the relative bias from the laboratory( RMSbias,rel). The EQA data including 20 analysis items measured by Roche closed system were collected from nine hospitals in Shanghai during 2010 to 2011. The minimum number of EQA trials was evaluated by the stable ucref,rel and RMSbias,relbased on the analysis with 20% of the expected value. The effect of analysis items on reliable RMSbias,relwas analyzed by u-test. The relative biases of ubias,relbetween the sixth EQA trial and the last and between the eleventh and the last for all 20 items were analyzed. Results There was concordance change tendency of RMSbias,relfor all analysis items,and the concordance rate achieved to 94. 21%. The minimum numbers of trials for obtaining stable ucref,reland RMSbias,relwere 10 and 11,respectively. For all investigated hospitals,the ratios of the hospitals with the relative bias ≤10% between the sixth EQA trial and the last and between the eleventh and the last were 34% and 65%,respectively. Conclusion The EQA data with more than 11 trials are more suitable to evaluate the bias uncertainty in medical laboratories.