Uso de modelos de regressão logística ordinal em epidemiologia: um exemplo usando a qualidade de vida

Abstract

The subject quality of life has been increasing emphasis in the last years. Typically, the results of the quality of life are measured by means of ordinal scales. In these situations, specific statistical methods are necessary and procedures as the dichotomization of the responsevariable, the loss of proportionate information for the data has as consequence, being, therefore not recommended. The models of ordinal logistic regression are appropriate in many situations. However, precautions in the use of these models must be taken, had to the premises for demanded them.. The objective of this work is to present a revision of the models of ordinal logistic regression, in particular of the model the proportional odds model, the partial proportional odds model, the continuation-ratio model and the stereotype model. Theadjustment and the comparison of the models are illustrated with data of a transversal study on quality of life, with 273 patients with schizophrenia diagnosis. The results indicate that all the tested models had shown a good adjustment, but the proportional odds model or partialproportional odds model provided a better goodness-of-fit because the character of the used data and the easiness of the interpretation of the results.