Group Sequential Methods for Longitudinal Data with Ordinal Responses
| Autor: | Jian-Jhih Lin, 林鍵志 |
|---|---|
| Rok vydání: | 2007 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 95 For ethical, economical and administrative considerations, interim analyses are often conducted to allow for possibly early termination of a clinical trial. Group sequential methods are essentially used for a correct application of interim analyses. Three common group sequential methods are proposed by Pocock (1977), O''Brien and Fleming (1979) and Lan and DeMets (1983). Those classical group sequential methods are applied for cross-sequential data as well as based on the assumption of independent increment structure (IIS) between the successive test statistics. For longitudinal data, the IIS assumption between the successive test statistics is violated due to the correlation between the measurements from the same subject. However, Scharfstein{et al}. (1997) prove that the IIS holds in parametric and semi-parametric models when efficient test statistics are employed. In the article, we propose group sequential methods based on GLMM (generalized linear mixed model) and GEE (generalized estimating equations) model for analysing ordinal longitudinal data. These two methods are compared with respect to the probability of type I error and power by simulation studies. The testing procedures are illustrated by a clinical trial for ordinal responses. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|