ABSTRACT
This chapter discusses the simultaneous analysis of several risk factors. It explains the procedure of stratification, and describes the Mantel—Haenszel algorithm for calculating an adjusted relative risk or odds ratio. Most epidemiological studies are analyzed with some kind of regression model. The strength of the evidence for the slope model depends partly on the steepness of the slope, and partly on the amount of variation around the average value of each year group. Computer programmes for regression usually provide a possibility to enter interaction variables in the analysis. This is done by indicating which two variables should interact, and then running the model again to see if this improved the regression fit significantly compared with a model with no interaction. In many situations one wants to analyze the association between several different factors and risk for disease: age, sex, vaccination status, family size, time of year, and nutritional status.
