APPLICATION OF GEE IN THE FIT OF ORDINALMARGINAL REGRESSION MODEL FOR TREATMENTRESPONSE OF HYPERTENSION

Abstract

This study adopts the ordinal logistic marginal model to fit the response to treatment of hypertensive patients with a view to ascertain any importance in incorporating panels when studying such clinical data. The Generalized Estimating Equation (GEE) provides an appropriate approach for estimating the model parameters. Two versions of ordinal regression models in conjunction with two forms of covariance estimators under three working correlation structures are considered. Results obtained revealed the importance of panels when studying the risk factors and response to treatment of hypertensive patients. With the Quasi-likelihood Information Criterion obtained for ordinal regression and other fitness criteria, the exchangeable working correlation is recommended as most adequate for the given data set.