In some ophthalmology studies, severity of an eye condition is graded according to an ordered categorical scale. For example, in a study of diabetic retinopathy, eyes are graded as one of the following ordered categories; no condition, moderate, severe, and proliferative. The objectives of this proposal are to carry out statistical research that will help improve our ability to analyze and interpret such ordered categorical data arising from ophthalmology studies. Firstly, the investigators propose to develop a general class of regression models for analyzing univariate ordered categorical data. Secondly, the investigators propose to develop a very general bivariate probit regression model to adjust for specific covariate structure in the ophthalmological data. This will allow easy modeling of bivariate ordered categorical data arising from some ophthalmology studies. Thirdly, the investigators propose to generalize the log-linear association models based on the global and the local cross-product ratio so that they can be used in analyzing data arising from ophthalmologic studies. Finally, the investigators propose to investigate the feasibility of using the generalized estimating equation approach in the analysis of ophthalmologic data in which no assumptions are needed to be made regarding the correlation structure between the left and right eyes.
| Williamson, J M; Lipsitz, S R; Kim, K M (1999) GEECAT and GEEGOR: computer programs for the analysis of correlated categorical response data. Comput Methods Programs Biomed 58:25-34 |
| Williamson, J; Kim, K (1996) A global odds ratio regression model for bivariate ordered categorical data from ophthalmologic studies. Stat Med 15:1507-18 |
| Kim, K (1995) A bivariate cumulative probit regression model for ordered categorical data. Stat Med 14:1341-52 |
| Lipsitz, S R; Kim, K; Zhao, L (1994) Analysis of repeated categorical data using generalized estimating equations. Stat Med 13:1149-63 |
