The linear range evaluation data is an important basis for evaluating the effectiveness of the kits to be marketed, and it is also one of the important application materials required for the registration of diagnostic reagents. At present, the registration test of quantitative in vitro diagnostic reagents in China includes indicators such as linear range, accuracy, specificity, minimum detection limit, and precision.
The linear range evaluation method is divided into linear evaluation of the kit calibrator and the use of samples to evaluate the linear range of the kit.
Requirement: The concentration of the calibrator of the in vitro diagnosis reagents should cover the entire linear range declared by the manufacturer.
Draw scatter plots and discovery and treatment of outliers: the drawing of scatter plot is very important for linear range evaluation. The functions of in vitro diagnostic reagents include:
① Obtain the linear relationship of the data initially (straight line, parabola or S-curve, etc.) to initially verify the correctness of the linear regression equation declared in the corporate standard.
② Observe the precision of the data. When the precision of a certain concentration level is low compared with other concentrations, especially when it occurs at both ends of the declared linear range, attention should be paid to whether the linear range is too wide; find outliers. Obvious outliers can be easily found by observing the scatter plot of in vitro diagnostic reagents.
If there is only one outlier, it can be eliminated and there is no need to repeat the experiment, but when there is more than one outlier, careful analysis should be made to find out the problem and repeat the experiment, and then use new data for analysis.
Linear regression analysis: Use the equation declared by the enterprise to perform regression analysis on the data through appropriate software. The confidence level is generally two-sided 0.05. The correlation coefficient r or the determination coefficient R2 of the calculation equation should be calculated. In vitro diagnostic reagents should meet the requirements of the enterprise registration standard.
According to industry standards, the absolute value of the correlation coefficient r should generally not be less than 0.98. When the scatter plot clearly indicates that the fitted curve may cross the origin, or the zero-point sample reading has been deducted during data calculation, attention should be paid to the intercept term in the regression equation.
Requirements: The samples involved in the linear evaluation of the quantitative in vitro diagnostic reagent product registration test are usually clinical samples, internal calibrators, national reference products or international standards and so on.
There should be no less than 5 samples that can cover the entire linear range declared by the manufacturer. Each concentration level sample of the testing reagent should be measured no less than twice. Each sample can be identified by absolute concentration or relative value.
Ideally, high-value and low-value clinical samples should be mixed in proportion to obtain a series of concentration samples. This mainly considers the influence of matrix effect on the test results.
Draw a scatter plot and the discovery and treatment of outliers: evaluation of the linear range of in vitro diagnostic reagents and calibrators.