Epidemiology provides the scientific basis for much of public health practice, and the revolution in health care and disease prevention indicates that the demand for valuable results from this field will continue to grow. Sound epidemiologic research requires a solid statistical basis for both study design and data analysis. As knowledge about the underlying causes of disease increases, we often see that they have multiple causes, so it is often not possible to limit conclusions to a single factor. Therefore a multivariate approach to data analysis is an essential part of epidemiologic researc ... More

Epidemiology provides the scientific basis for much of public health practice, and the revolution in health care and disease prevention indicates that the demand for valuable results from this field will continue to grow. Sound epidemiologic research requires a solid statistical basis for both study design and data analysis. As knowledge about the underlying causes of disease increases, we often see that they have multiple causes, so it is often not possible to limit conclusions to a single factor. Therefore a multivariate approach to data analysis is an essential part of epidemiologic research. The multivariate methods considered in this book involve the simultaneous analysis of the association between multiple attributes of an individual and the risk of a disease. This book aims to address the practical questions of how to approach data analysis. The book is organized in four parts. Part I introduces ways of thinking quantitatively about the disease process. Part II explores some of the computationally direct methods that have long been a part of classical epidemiological data analysis. Separate chapters deal with the analysis of proportions, rates, and semi-parametric approaches from time to failure data. Formal model fitting is considered in much more detail in Part III, which considers not only the analysis of proportions, but also parametric and semi-parametric approaches to the analysis of hazard rates. Part IV deals with special problems that arise when one incorporates aspects of the study design into the analysis, along with approaches to designing a study that will be the right size for addressing the study aims.