Statistical Models in Epidemiology

by David Clayton, Michael Hills

Overview

This self-contained account of the statistical basis of epidemiology has been written specifically for those with a basic training in biology, therefore no previous knowledge is assumed and the mathematics is deliberately kept at a manageable level. The authors show how all statistical analysis of data is based on probability models, and once one understands the model, analysis follows easily.

In showing how to use models in epidemiology the authors have chosen to emphasize the role of likelihood, an approach to statistics which is both simple and intuitively satisfying. More complex problems can then be tackled by natural extensions of the simple methods. Based on a highly successful course, this book explains the essential statistics for all epidemiologists.

Product Details

ISBN-13:	9780199671182
Publisher:	Oxford University Press
Publication date:	03/01/2013
Edition description:	Reprint
Pages:	384
Product dimensions:	6.10(w) x 9.10(h) x 0.90(d)

About the Author

David Clayton, Diabetes and Inflammation Laboratory, Cambridge Institute for Medical Research

Michael Hills, London School of Hygiene and Tropical Medicine

Table of Contents

I. Probability Models and Likelihood1. Probability models2. Conditional probability models3. Likelihood4. Consecutive follow-up intervals5. Rates6. Time7. Competing risks and selection8. The Gaussian probability model9. Approximate likelihoods10. Likelihood, probability, and confidence11. Null hypotheses and p-values12. Small studies13. Likelihoods for the rate ratio14. Confounding and standardization15. Comparison of rates within strata16. Case-control studies17. Likelihoods for the odds ratio18. Comparison of odds within strata19. Individually matched case-control studies20. Tests for trend21. The size of investigationsII. Regression Models22. Introduction to regression models23. Poission and logistic regression24. Testing hypotheses25. Models for dose-response26. More about interaction27. Choice and interpretation of models28. Additivity and synergism29. Conditional logistic regression30. Cox's regression analysis31. Time-varying explanatory variables32. Three examples33. Nested case-control studies34. Gaussian regression models35. PostscriptIII. AppendicesA. ExponentialsB. Some basic calculusC. Approximate profile likelihoodsD. Table of the Chi-squared distributionIndex

Editorial Reviews

3 Stars from Doody

This book provides a statistical foundation for performing epidemiologic analyses, beginning with a basic explanation of probability and likelihood function in early chapters and expanding on these concepts in later chapters to illustrate their application in epidemiology. The authors' stated objective is to present a basis for understanding statistical models commonly used in epidemiology. This goal is certainly laudable, because epidemiology students often have difficulty incorporating their knowledge of statistical theory with the more applied concepts taught in epidemiology courses. However, if viewed alone, this book falls somewhat short of achieving this goal, because both relevant theoretical and applied concepts are excluded as a result of combining the two disciplines. The authors certainly seem credible authorities on the subject matter; both have affiliations with respected institutions and together they have taught numerous courses in epidemiology. However, although the book is intended for masters degree students in epidemiology or biostatistics with no previous statistical knowledge, the ideas presented become increasingly complex throughout the book and would seem arduous for a student at a beginning level. This book contains many diagrams that are very useful in illustrating complicated ideas. The table of contents and index are complete and accurate. There are numerous short chapters that focus on one key concept, a style more effective than combining many intricate and often confusing ideas together into a single chapter. The ordering of topics, however, is better suited for a statistical than for an epidemiological mind set. Important concepts are not always introducedin a sequence consistent with planning and undertaking an epidemiologic analysis. For example, the concept of statistical interaction should always be considered when evaluating any stratified analysis, yet it is not presented until much later in the regression analysis section. This book is most useful for an intermediate course in epidemiology. This book has several shortcomings that could be overcome by using this book as a supplement to other epidemiology texts rather than as the sole text. Additionally, this book does not replace a solid background in statistical theory. It does, however, help to bridge the gap between statistics and epidemiology, and thus provides a unique perspective.

Laura A. Schieve

Reviewer: Laura A. Schieve, MS (University of Illinois at Chicago)
Description: This book provides a statistical foundation for performing epidemiologic analyses, beginning with a basic explanation of probability and likelihood function in early chapters and expanding on these concepts in later chapters to illustrate their application in epidemiology.
Purpose: The authors' stated objective is to present a basis for understanding statistical models commonly used in epidemiology. This goal is certainly laudable, because epidemiology students often have difficulty incorporating their knowledge of statistical theory with the more applied concepts taught in epidemiology courses. However, if viewed alone, this book falls somewhat short of achieving this goal, because both relevant theoretical and applied concepts are excluded as a result of combining the two disciplines.
Audience: The authors certainly seem credible authorities on the subject matter; both have affiliations with respected institutions and together they have taught numerous courses in epidemiology. However, although the book is intended for masters degree students in epidemiology or biostatistics with no previous statistical knowledge, the ideas presented become increasingly complex throughout the book and would seem arduous for a student at a beginning level.
Features: This book contains many diagrams that are very useful in illustrating complicated ideas. The table of contents and index are complete and accurate. There are numerous short chapters that focus on one key concept, a style more effective than combining many intricate and often confusing ideas together into a single chapter. The ordering of topics, however, is better suited for a statistical than for an epidemiological mind set. Important concepts are not always introduced in a sequence consistent with planning and undertaking an epidemiologic analysis. For example, the concept of statistical interaction should always be considered when evaluating any stratified analysis, yet it is not presented until much later in the regression analysis section.
Assessment: This book is most useful for an intermediate course in epidemiology. This book has several shortcomings that could be overcome by using this book as a supplement to other epidemiology texts rather than as the sole text. Additionally, this book does not replace a solid background in statistical theory. It does, however, help to bridge the gap between statistics and epidemiology, and thus provides a unique perspective.

Doody's Review Service