GlaxoSmithKline Medicines Research Centre, Gunnels Wood Road, Stevenage, Hertfordshire, SG1 2NY, UK.
Anal Chem. 2009 Dec 15;81(24):9849-57. doi: 10.1021/ac901945f.
Method equivalence assessments should be considered when analytical methods are either modified or substituted. The TOST (two one sided tests) approach provides a sounder data driven method for testing equivalence than a simple comparative intermediate precision study which is typically performed as part of method validation. Prior to designing an equivalency study, an acceptance criterion (an acceptable bias between original and modified/changed method) must be chosen. The choice of acceptance criteria requires the identification of the smallest mean difference or bias between methods that is practically important. Equivalence testing in this manner is used to prove that the new method can generate data which continues to support previously established specifications. Once the acceptance criterion is decided, other aspects of the study can be designed following a set of design principles. When the design and acceptance criteria have been established, the collection of the data can commence. Demonstration of equivalence should not start until the validity of the observations has been confirmed such as assessment for outliers, normality, and comparison of variances. Once the suitability of the data is confirmed, the mean difference between the two data sets can be calculated along with a +/-90% confidence interval using the TOST approach. It can then be established whether equivalence of the two methods has been demonstrated.
当分析方法被修改或替换时，应考虑方法等效性评估。与通常作为方法验证一部分进行的简单比较中间精密度研究相比，TOST（双单边检验）方法为测试等效性提供了更可靠的数据驱动方法。在设计等效性研究之前，必须选择接受标准（原始方法和修改/更改方法之间可接受的偏差）。接受标准的选择需要确定方法之间实际重要的最小平均差异或偏差。以这种方式进行等效性检验是为了证明新方法可以生成数据，这些数据仍然支持先前建立的规范。一旦确定了接受标准，就可以根据一组设计原则设计研究的其他方面。一旦确定了设计和接受标准，就可以开始收集数据。只有在确认了观测值的有效性后，例如离群值、正态性和方差比较的评估，才能开始等效性验证。一旦确认了数据的适用性，就可以使用 TOST 方法计算两个数据集之间的平均差异以及 +/-90%置信区间。然后可以确定两种方法是否具有等效性。
