How To Do The Data Analysis Of The IVD Clinical Trial For Quantitative Detection?

In clinical trials of quantitative detection reagents, the results of the investigational reagent will be compared with the results from a similar NMPA-approved product or other reference methods. Regression analysis combined with Bland-Altman method can be used for consistence analysis of the results of quantitative detection.

Methods commonly used in regression analysis include Passing- Bland-Altman regression and Deming regression. In the statistical analysis, the regression coefficient and intercept, with confidence intervals, should be calculated. Hypothesis test can also be performed. Other evaluation parameters should be estimated at the same time, such as correlation coefficient or determination coefficient. If necessary, it is recommended to consider the difference in consistence performance for different concentration ranges, so as to perform stratified statistics.

Along with the regression analysis, the Bland-Altman method can be used to evaluate the consistency of the two test results by calculating the consistency limit. In particular, analysis is performed for consistency of test results near the level of medical decision. Consistency limits should be within clinically acceptable level.

In clinical trials of quantitative detection reagents, if the investigational reagent is compared with clinical reference standards (qualitatively), the statistical analysis of receiver operating characteristic curve (ROC) can be used. The area under the ROC curve reflects the diagnostic value of the investigational kit. Or it can also be chosen to compare the diagnostic value of the investigational reagent with one of the NMPA-approved similar products.

For clinical trials of IVD reagents, when using ROC curve for data analysis, it is still necessary to further evaluate the sensitivity, specificity and other parameters (with confidence intervals) with the recommended cut off.