ORDINAL LOGISTIC-REGRESSION IN MEDICAL-RESEARCH

Medical research workers are making increasing use of logistic regress ion analysis for binary and ordinal data. The purpose of this paper is to give a non-technical introduction to logistic regression models fo r ordinal response variables. We address issues such as the global con cept and interpetation of logistic models, the model building procedur e from a practical point of view, and the assessment of the model adeq uacy. For illustrative purposes we apply these methods to real data of a study investigating the association between glycosylated haemoglobi n and retinopathy. We give some recommendations for the use and assess ment of ordinal logistic regression models in medical research.