Evaluations Of Enzyme Linked Immunosorbent Assays: A Method Of Data Analysis

It is important to evaluate ELISA detection reagent in science research and application. But there are not a standard for reagent evaluation. Especially, there are only sensitivity and specificity to be used in some research for the evaluation. This method is available to some reagents in which there are distinguished differences results and not available to the reagents in which there is little difference results. In this circumstance, it is difficult to select a better reagent.For solving this problem, we should use more methods for evaluation of ELISA detection reagent. While we looked up many related documents and books, some new methods have been used in our research. Those methods are included value and graph of Log(C. O. I.) frequency to considerate the distribution of data for detection results, Kappa value test for comparing the concordance of two groups of number data, and ROC curve analysis for general evaluating of reagents efficacy. By above methods, we should penetratingly study the inherent difference of the reagents, and then select the better reagent for our routine work.We select HBsAg ELISA detected reagent for the studied object. For reagents form market has been purchased, those are the reagent for detecting HBsAg, antiHBs and antiHBc. The serum specimens are constituent by 63 known referred specimens from national drug product department and 323 unknown serum specimens from a general hospital. The whole procedure of the test are completed under standard operating procedure(SOP) and be set the golden standard of the result for the two part of the test. SAS statistical analysis soft is use in the analysis for the detection result. The study results are listed below:The known reference specimen test:1. Four domestic HBsAg ELISA detected reagents all have better detectable to the best lowest limited serum of sensitivity.2. Four domestic HBsAg ELISA detected reagents all have better detectable to the high value HBsAg positive serum. Although there is "Hook" effective in that serum, it's not contribution to the detected results for the qualitative test.3. In precision test, for B reagent there is better than others and A is defect which precision is so high that it is exceed the highest limited value for national standard (15%).4. About general evaluation index for HBsAg detection, there are higher detectable efficacy for A and B reagent than C and D.5. In analysis of the linear relationship for HBsAg dose response, the equations of linear regression fitted by power function are significance and in residual error analysis, there is well practical efficacy within the equations for the four reagents.6. By 8 value and graph of Log(C. O. I.) frequency for HBsAg detected result, A and B reagent is better than C and D.The unknown specimen test:1. About general evaluation index for HBsAg detection, there is the best detectable efficacy for HBsAg detection reagent and lower for antiHBs and antiHBc detection reagent. For B reagent the detectable efficacy is the highest in the all reagents.2. There is well concordance between the four reagents and the Loche HBsAg detection reagent by Kappa value test. The order of concordance is: HBsAg detection reagent, antiHBc detection reagent and antiHBs reagent as the last one.3. In analysis byδvalue and graph of Log(C. O. I.) frequency, the detectable efficacy of B reagent is better than others for detection reagent of HBsAg, antiHBs and antiHBc.4. In ROC curve analysis, the result of B is the best in all four reagents for detecting HBsAg, antiHBs and antiHBc. Otherwise, C reagent is not good in the curve analysis, and it is high significance between C and A or B reagent. In our study, four domestic HBsAg ELISA detection reagents have be selected. By detecting HBsAg, antiHBs and antiHBc, we try to utilize some new methods to evaluate ELISA detected reagent. We thought the study would do some contribution to elevate level of evaluation FLISA detection reagent.