St. James Ballroom
A Practical Approach to Evaluate the Change in Precision Statistics (303879)
*Ming Lu, Avid Radiopharmaceuticals, Inc.
Michael J. Pontecorvo, Avid Radiopharmaceuticals, Inc.
Haiqing Shen, Avid Radiopharmaceuticals, Inc.
Keywords: precision, kappa, analysis method, simulation
When assessing performance of a diagnostic test, not only accuracy, but also the precision of the test are required. Kappa statistics is usually the statistics in practice for precision assessment when the endpoints are categorical variables. In diagnostic pharmaceutical development, there are often needs to evaluate the precision improvement after modifying an existing diagnostic test. Unfortunately there are no readily available statistics to assess the statistical significance of change in Kappa values. This study proposes a straightforward simulation method which not only calculates the confidence interval for the change in Kappa values, but also provides a p-value as a common statistic would do to indicate the statistical significance of this change. This method was used in a clinical trial and the study results were published in European Journal of Nuclear Medicine and Molecular Imaging. A simple and practical method like this is useful for analysis of data relevant to the pharmaceutical industry. It may provide an approach for other statisticians facing similar issues in real-life statistical practice.