Analytical method validation imprecision

Ion-selective electrodes (ISEs) represent the current primary methodology in the quantification of S-Li [11-13], Moreover, ISE modules are parts of large and fully automated clinical chemistry analysers. In practice, the validation parameters are most often chosen in terms of judging the acceptability of the new measurement system for daily use. For this reason, the first approach was to study whether the detected imprecision fulfilled the desired analytical quality specifications. Secondly, proficiency testing (PT) results from past samples were of great value in predicting future bias. The identity of the three ISE methods was evaluated using patient samples. The analytical performance was checked after 6 months routine use. Without any exception, method validations always mean an extra economical burden. Therefore, the validation parameters chosen and processed have to be considered carefully. 

This method validation study indicates that the choice of validation parameters is crucial. Checking the method performance in accordance to the obtained validation results should not be omitted. The preliminary imprecision results do not necessarily predict the analytical variation as the method is taken into routine use. The suitability of system controls have to be considered carefully when patient results are accepted according to the general internal quality control principles. Further, surprisingly high variations should be clarified with the manufacturer. The use of past PT samples is highly recommended to predict the future inaccuracy, whenever... 

Abstract Validation of analytical methods of well-characterised systems, such as are found in the pharmaceutical industry, is based on a series of experimental procedures to establish selectivity, sensitivity, repeatability, reproducibility, linearity of calibration, detection limit and limit of determination, and robustness. It is argued that these headings become more difficult to apply as the complexity of the analysis increases. Analysis of environmental samples is given as an example. Modern methods of analysis that use arrays of sensors challenge validation. The output may be a classification rather than a concentration of analyte, it may have been established by imprecise methods such as the responses of human taste panels, and the state space of possible responses is too large to cover in any experimental-design procedure. Moreover the process of data analysis may be done by non-linear methods such as neural networks. Validation of systems that rely on computer software is well established. 

Automation is an essential part of the sample preparation procedures. The added imprecision that results from the increased manipulation of the sample during extraction can be minimized by the use of robotics and automated sample preparation systems. For these reasons, it is essential that the validation of analytical methods include sample preparation. The validation process defines the method parameters and the impact that each parameter has on the effectiveness of the extraction method. In addition, the use of manual techniques is not practical... 

Theoretically IQC should be the front-line approach to quality. If a method has been adequately validated and shown to meet the requirements of the user and kept in analytical control with IQC to detect intrusion of bias or imprecision, then the EQA needs to provide the occasional, independent, objective reassurance. In practice however, the EQA is likely to play an equal role with IQC, both in confirming problems brought to the attention of the analyst by the IQC and in stimulating further action. 

Figure 17.1 illustrates how precision may vary as a function of analyte concentration. The %RSD values of ethanol quantification by GC increased significantly as the concentration decreased from 1000 ppm to 10 ppm. Higher variability is expected as the analyte levels approach the detection limit for the method (8). The analyst must determine during the validation of the method at what concentration the imprecision becomes too great for the intended use of the method. 

---