How to Deal with Continuous and Dichotomic Outcomes in Epidemiological Research: Linear and Logistic Regression Analyses

Abstract

Because of some limitations of stratification methods, epidemiologists frequently use multiple linear and logistic regression analyses to address specific epidemiological questions. If the dependent variable is a continuous one (for example, systolic pressure and serum creatinine), the researcher will use linear regression analysis. Otherwise, if the dependent variable is dichotomic (for example, presence/absence of microalbuminuria), one could use logistic regression analysis. In both linear and logistic regression analyses the independent variables may be either continuous or categorical. In this paper we will describe linear and logistic regression analyses by discussing methodological features of these techniques and by providing clinical examples and guidance (syntax) for performing these analyses by commercially available statistical packages. Furthermore, we will also focus on the use of multiple linear and logistic regression analyses to control for confounding in etiological research.