Bias Analysis on Results of Total Protein Detected with Different Reagents

Hongbin Ni
2010-01-01
Abstract:Objective To investigate the comparability of total protein(TP) test results between two kinds of reagents.Methods Based on the guideline of EP9-A2,eight samples with different concentrations were randomly selected from the clinical samples and measured TP by two kinds of reagents for 5 d consecutively.All results were recorded and subjected to conduct the relative regression analysis.The anticipated bias (at medical decision level) and its 95% confidence interval were calculated.To estimate the acceptability of the bias.Results The correlation coefficient of using two kinds of TP reagents was 0.993 (r2=0.986),intercept(a):0.362,slope(b):0.991,the regression equation:Y=0.991X+0.362,without outliers of within-method and between-method,the bias of results were accepted in the range of the allowable error.Conclusion When analyzing the same test item by over two kinds of reagents,it is necessary to evaluate the clinical acceptability to insure the accuracy and comparability by method comparison and bias estimation based on EP9-A2 document.
What problem does this paper attempt to address?