WITHDRAWN: Statistical Methods to Check Agreement Between Two Coding Systems

• Author(s): Qingzhao Yu and Xiaocheng Wu and Patricial Andrews and Donald Mercante
• Keywords: Agreement ; Cancer Stage Distribution ; EM Algorithm ; Likelihood Ratio Test ; SEER

Abstract

Since 2004, the Collaborative Staging (CS) System has been used in place of the directly Surveillance Epidemiology and End Results Summary Staging (SS) and American Joint Committee on Cancer staging systems to code cancer stages. SS can be derived from CS, but CS and SS2000 have not been used to code the same cancer cases simultaneously. The absence of a double-coded dataset provides a challenge when we want to check the agreement between SS2000 and Derived SS2000. This talk proposes a method to solve this problem. We compare the cancer stage distributions over several years. If the stage distributions differ over the years, we build models trying to explain the differences by known factors such as the linear trends distributions. If the known factors satisfactorily explain the distribution differences, we may conclude that the two systems include no significant coding difference.