Impact on Goodness of Fit of Proportional Hazards model due to Non-compliance of Randomization

• Main Authors: Ting-Wei Liu, 劉庭維
• Other Authors: Shih-Chang Lee
• Format: Others
• Language: zh-TW
• Published: 2009
• Online Access: http://ndltd.ncl.edu.tw/handle/92541532819677350930

碩士 === 銘傳大學 === 應用統計資訊學系碩士班 === 97 === To enhance the research and development of global pharmaceutical industry to a harmonized manner, the government authorities and the medical industries of United States, European Union and Japan organize the International Conference on Harmonization (ICH). The ICH published many guidelines that make statistics as a method to draw the scientific evidence of the efficacy/safety of the study drug. The ICH E6 (Good Clinical Trial Practice, GCP) includes the topics in design of a clinical trial, the principles of data analysis, as well as conducting the trial. The ICH E9 (Statistical Principles for Clinical Trials) covers what statistical concerns should be implemented in clinical trials. However, the violation of pre-planned random allocation is not unusually seen deficiency in many clinical trials, which makes the clinical conclusions unreliable. This study will investigate the impacts due to the non-compliance with randomization. Firstly, the impact on sample sizes between treatment groups through the power function. Secondly, using risk set to study the impact of the statistical survival models. Building the partial likelihood function in the three types of non-compliance situations, explore the difference between the compliance with non-compliance. Thirdly, non-parametric Cox`s proportional hazard models will be used to explore the effects of parametric and explained variation.