epi.pdf - Authors Kenneth J Rothman Vice President Epidemiology Research RTI Health Solutions Professor of Epidemiology and Medicine Boston University

epi.pdf - Authors Kenneth J Rothman Vice President...

This preview shows page 1 out of 896 pages.

You've reached the end of your free preview.

Want to read all 896 pages?

Unformatted text preview: Authors Kenneth J. Rothman Vice President Epidemiology Research, RTI Health Solutions; Professor of Epidemiology and Medicine, Boston University, Boston, Massachusetts Sander Greenland Professor of Epidemiology and Statistics University of California Los Angeles, California Timothy L. Lash Associate Professor of Epidemiology and Medicine Boston University, Boston, Massachusetts Contributors James W. Buehler Research Professor Department of Epidemiology, Rollins School of Public Health, Emory University, Atlanta, Georgia Jack Cahill Vice President Department of Health Studies Sector, Westat, Inc., Rockville, Maryland Sander Greenland Professor of Epidemiology and Statistics University of California, Los Angeles, California M. Maria Glymour Robert Wood Johnson Foundation Health and Society Scholar Department of Epidemiology, Mailman School of Public Health, Columbia University, New York, New York, Department of Society, Human Development and Health, Harvard School of Public Health, Boston, Massachusetts Marta Gwinn Associate Director Department of Epidemiology, National Office of Public Health Genomics, Centers for Disease Control and Prevention, Atlanta, Georgia Patricia Hartge Deputy Director Department of Epidemiology and Biostatistics Program, Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health, Rockville, Maryland Irva Hertz-Picciotto Professor Department of Public Health, University of California, Davis, Davis, California 1 C. Robert Horsburgh Jr. Professor of Epidemiology, Biostatistics and Medicine Department Epidemiology, Boston University School of Public Health, Boston, Massachusetts Jay S. Kaufman Associate Professor Department of Epidemiology, University of North Carolina at Chapel Hill, School of Public Health, Chapel Hill, North Carolina Muin J. Khoury Director National Office of Public Health Genomics, Centers for Disease Control and Prevention, Atlanta, Georgia Timothy L. Lash Associate Professor of Epidemiology and Medicine Boston University, Boston, Massachusetts Barbara E. Mahon Assistant Professor Department of Epidemiology and Pediatrics, Boston University, Novartis Vaccines and Diagnostics, Boston, Massachusetts Robert C. Millikan Professor Department of Epidemiology, University of North Carolina at Chapel Hill, School of Public Health, Chapel Hill, North Carolina Hal Morgenstern Professor and Chair Department of Epidemiology, University of Michigan School of Public Health, Ann Arbor, Michigan Jørn Olsen Professor and Chair Department of Epidemiology, UCLA School of Public Health, Los Angeles, California Keith O'Rourke Visiting Assistant Professor Department of Statistical Science, Duke University, Durham, North Carolina, Adjunct Professor, Department of Epidemiology and Community Medicine, University of Ottawa, Ottawa, Ontario, Canada Charles Poole Associate Professor Department of Epidemiology, University of North Carolina at Chapel Hill, School of Public Health, Chapel Hill, North Carolina Kenneth J. Rothman 2 Vice President, Epidemiology Research RTI Health Solutions, Professor of Epidemiology and Medicine, Boston University, Boston, Massachusetts Clarice R. Weinberg National Institute of Environmental Health Sciences, Biostatistics Branch, Research Triangle Park, North Carolina Noel S. Weiss Professor Department of Epidemiology, University of Washington, Seattle, Washington Allen J. Wilcox Senior Investigator Epidemiology Branch, National Institute of Environmental Health Sciences/NIH, Durham, North Carolina Walter C. Willett Professor and Chair Department of Nutrition, Harvard School of Public Health, Boston, Massachusetts 3 Preface and Acknowledgments This third edition of Modern Epidemiology arrives more than 20 years after the first edition, which was a much smaller single-authored volume that outlined the concepts and methods of a rapidly growing discipline. The second edition, published 12 years later, was a major transition, as the book grew along with the field. It saw the addition of a second author and an expansion of topics contributed by invited experts in a range of subdisciplines. Now, with the help of a third author, this new edition encompasses a comprehensive revision of the content and the introduction of new topics that 21st century epidemiologists will find essential. This edition retains the basic organization of the second edition, with the book divided into four parts. Part I (Basic Concepts) now comprises five chapters rather than four, with the relocation of Chapter 5, “Concepts of Interaction,” which was Chapter 18 in the second edition. The topic of interaction rightly belongs with Basic Concepts, although a reader aiming to accrue a working understanding of epidemiologic principles could defer reading it until after Part II, “Study Design and Conduct.” We have added a new chapter on causal diagrams, which we debated putting into Part I, as it does involve basic issues in the conceptualization of relations between study variables. On the other hand, this material invokes concepts that seemed more closely linked to data analysis, and assumes knowledge of study design, so we have placed it at the beginning of Part III, “Data Analysis.” Those with basic epidemiologic background could read Chapter 12 in tandem with Chapters 2 and 4 to get a thorough grounding in the concepts surrounding causal and non-causal relations among variables. Another important addition is a chapter in Part III titled, “Introduction to Bayesian Statistics,” which we hope will stimulate epidemiologists to consider and apply Bayesian methods to epidemiologic settings. The former chapter on sensitivity analysis, now entitled “Bias Analysis,” has been substantially revised and expanded to include probabilistic methods that have entered epidemiology from the fields of risk and policy analysis. The rigid application of frequentist statistical interpretations to data has plagued biomedical research (and many other sciences as well). We hope that the new chapters in Part III will assist in liberating epidemiologists from the shackles of frequentist statistics, and open them to more flexible, realistic, and deeper approaches to analysis and inference. As before, Part IV comprises additional topics that are more specialized than those considered in the first three parts of the book. Although field methods still have wide application in epidemiologic research, there has been a surge in epidemiologic research based on existing data sources, such as registries and medical claims data. Thus, we have moved the chapter on field methods from Part II into Part IV, and we have added a chapter entitled, “Using Secondary Data.” Another addition is a chapter on social epidemiology, and coverage on molecular epidemiology has been added to the chapter on genetic epidemiology. Many of these chapters may be of interest mainly to those who are focused on a particular area, such as reproductive epidemiology or infectious disease epidemiology, which have 4 distinctive methodologic concerns, although the issues raised are well worth considering for any epidemiologist who wishes to master the field. Topics such as ecologic studies and meta-analysis retain a broad interest that cuts across subject matter subdisciplines. Screening had its own chapter in the second edition; its content has been incorporated into the revised chapter on clinical epidemiology. The scope of epidemiology has become too great for a single text to cover it all in depth. In this book, we hope to acquaint those who wish to understand the concepts and methods of epidemiology with the issues that are central to the discipline, and to point the way to key references for further study. Although previous editions of the book have been used as a course text in many epidemiology teaching programs, it is not written as a text for a specific course, nor does it contain exercises or review questions as many course texts do. Some readers may find it most valuable as a reference or supplementary-reading book for use alongside shorter textbooks such as Kelsey et al. (1996), Szklo and Nieto (2000), Savitz (2001), Koepsell and Weiss (2003), or Checkoway et al. (2004). Nonetheless, there are subsets of chapters that could form the textbook material for epidemiologic methods courses. For example, a course in epidemiologic theory and methods could be based on Chapters 1,2,3,4,5,6,7,8,9,10,11 and 12 with a more abbreviated course based on Chapters 1,2,3 and 4 and 6,7,8,9,10 and 11. A short course on the foundations of epidemiologic theory could be based on Chapters 1,2,3,4 and 5 and Chapter 12. Presuming a background in basic epidemiology, an introduction to epidemiologic data analysis could use Chapters 9, 10, and 12,13,14,15,16,17,18 and 19, while a more advanced course detailing causal and regression analysis could be based on Chapters 2,3,4 and 5, 9, 10, and 12,13,14,15,16,17,18,19,20 and 21. Many of the other chapters would also fit into such suggested chapter collections, depending on the program and the curriculum. Many topics are discussed in various sections of the text because they pertain to more than one aspect of the science. To facilitate access to all relevant sections of the book that relate to a given topic, we have indexed the text thoroughly. We thus recommend that the index be consulted by those wishing to read our complete discussion of specific topics. We hope that this new edition provides a resource for teachers, students, and practitioners of epidemiology. We have attempted to be as accurate as possible, but we recognize that any work of this scope will contain mistakes and omissions. We are grateful to readers of earlier editions who have brought such items to our attention. We intend to continue our past practice of posting such corrections on an internet page, as well as incorporating such corrections into subsequent printings. Please consult < ; to find the latest information on errata. We are also grateful to many colleagues who have reviewed sections of the current text and provided useful feedback. Although we cannot mention everyone who helped in that regard, we give special thanks to Onyebuchi Arah, Matthew Fox, Jamie Gradus, Jennifer Hill, Katherine Hoggatt, Marshal Joffe, Ari Lipsky, James Robins, Federico Soldani, Henrik Toft 5 Sørensen, Soe Soe Thwin and Tyler VanderWeele. An earlier version of Chapter 18 appeared in the International Journal of Epidemiology (2006;35:765–778), reproduced with permission of Oxford University Press. Finally, we thank Mary Anne Armstrong, Alan Dyer, Gary Friedman, Ulrik Gerdes, Paul Sorlie, and Katsuhiko Yano for providing unpublished information used in the examples of Chapter 33. Kenneth J. Rothman Sander Greenland Timothy L. Lash 6 CONTENTS 1.Introduction.....................................................................................................................1 Section Ⅰ-Basic Concepts 2.Causation and Causal Inference...................................................................................... 4 3.Measures of Occurrence................................................................................................ 43 4.Measures of Effect and Measures of Association .........................................................68 5.Concepts of Interaction .................................................................................................94 Section Ⅱ-Study Design and Conduct 6.Types of Epidemiologic Studies.................................................................................. 113 7.Cohort Studies.............................................................................................................129 8.Case-Control Studies...................................................................................................142 9.Validity in Epidemiologic Studies...............................................................................162 10.Precision and Statistics in Epidemiologic Studies ....................................................186 11.Design Strategies to Improve Study Accuracy..........................................................210 12.Causal Diagrams ....................................................................................................... 227 Section Ⅲ-Data Analysis 13.Fundamentals of Epidemiologic Data Analysis ........................................................259 14.Introduction to Categorical Statistics ........................................................................288 15.Introduction to Stratified Analysis ............................................................................310 16.Applications of Stratified Analysis Methods ............................................................339 17.Analysis of Polytomous Exposures and Outcomes...................................................362 18.Introduction to Bayesian Statistics............................................................................390 19.Bias Analysis.............................................................................................................409 20.Introduction to Regression Models ...........................................................................452 21.Introduction to Regression Modeling .......................................................................491 Section Ⅳ-Special Topics 22.Surveillance...............................................................................................................535 23.Using Secondary Data............................................................................................... 561 24.Field Methods in Epidemiology................................................................................ 574 25.Ecologic Studies........................................................................................................596 26.Social Epidemiology .................................................................................................620 27.Infectious Disease Epidemiology..............................................................................640 28.Genetic and Molecular Epidemiology ......................................................................660 29.Nutritional Epidemiology .........................................................................................682 30.Environmental Epidemiology ...................................................................................703 31.Methodologic Issues in Reproductive Epidemiology ...............................................730 32.Clinical Epidemiology ..............................................................................................756 33.Meta-Analysis ........................................................................................................... 768 References.........................................................................................................................806 Chapter 1 Introduction Kenneth J. Rothman Sander Greenland Timothy L. Lash Although some excellent epidemiologic investigations were conducted before the 20th century, a systematized body of principles by which to design and evaluate epidemiology studies began to form only in the second half of the 20th century. These principles evolved in conjunction with an explosion of epidemiologic research, and their evolution continues today. Several large-scale epidemiologic studies initiated in the 1940s have had far-reaching influences on health. For example, the community-intervention trials of fluoride supplementation in water that were started during the 1940s have led to widespread primary prevention of dental caries (Ast, 1965). The Framingham Heart Study, initiated in 1949, is notable among several long-term follow-up studies of cardiovascular disease that have contributed importantly to understanding the causes of this enormous public health problem (Dawber et al., 1957; Kannel et al., 1961, 1970; McKee et al., 1971). This remarkable study continues to produce valuable findings more than 60 years after it was begun (Kannel and Abbott, 1984; Sytkowski et al., 1990; Fox et al., 2004; Elias et al., 2004; ). Knowledge from this and similar epidemiologic studies has helped stem the modern epidemic of cardiovascular mortality in the United States, which peaked in the mid-1960s (Stallones, 1980). The largest formal human experiment ever conducted was the Salk vaccine field trial in 1954, with several hundred thousand school children as subjects (Francis et al., 1957). This study provided the first practical basis for the prevention of paralytic poliomyelitis. The same era saw the publication of many epidemiologic studies on the effects of tobacco use. These studies led eventually to the landmark report, Smoking and Health, issued by the Surgeon General (United States Department of Health, Education and Welfare, 1964), the first among many reports on the adverse effects of tobacco use on health issued by the Surgeon General ( ). Since that first report, epidemiologic research has steadily attracted public attention. The news media, boosted by a rising tide of social concern about health and environmental issues, have vaulted many epidemiologic studies to prominence. Some of these studies were controversial. A few of the biggest attention-getters were studies related to Avian influenza Severe acute respiratory syndrome (SARS) Hormone replacement therapy and heart disease 1 Carbohydrate intake and health Vaccination and autism Tampons and toxic-shock syndrome Bendectin and birth defects Passive smoking and health Acquired immune deficiency syndrome (AIDS) The effect of diethylstilbestrol (DES) on offspring Disagreement about basic conceptual and methodologic points led in some instances to profound differences in the interpretation of data. In 1978, a controversy erupted about whether exogenous estrogens are carcinogenic to the endometrium: Several case-control studies had reported an extremely strong association, with up to a 15-fold increase in risk (Smith et al., 1975; Ziel and Finkle, 1975; Mack et al., 1976). One group argued that a selection bias accounted for most of the observed association (Horwitz and Feinstein, 1978), whereas others argued that the alternative design proposed by Horwitz and Feinstein introduced a downward selection bias far stronger than any upward bias it removed (Hutchison and Rothman, 1978; Jick et al., 1979; Greenland and Neutra, 1981). Such disagreements about fundamental concepts suggest that the methodologic foundations of the science had not yet been established, and that epidemiology remained young in conceptual terms. The last third of the 20th century saw rapid growth in the understanding and synthesis of epidemiologic concepts. The main stimulus for this conceptual growth seems to have been practice and controversy. The explosion of epidemiologic activity accentuated the need to improve understanding of the theoretical underpinnings. For example, early studies on smoking and lung cancer (e.g., Wynder and Graham, 1950; Doll and Hill, 1952) were scientifically noteworthy not only for their substantive findings, but also because they demonstrated the efficacy and great efficiency of the case-control study. Controversies about proper case-control design led to recognition of the importance of relating such studies to an underlying source population (Sheehe, 1962; Miettinen, 1976a; Cole, 1979; see Chapter 8). Likewise, analysis of data from the Framingham Heart Study stimulated the development of the most popular modeling method in epidemiology today, multiple logistic regression (Cornfield, 1962; Truett et al., 1967; see Chapter 20). Despite the surge of epidemiologic activity in the late 20th century, the evidence indicates that epidemiology remains in an early stage of development (Pearce and Merletti, 2006). In recent years epidemiologic concepts have continued to evolve rapidly, perhaps becaus...
View Full Document

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture

Stuck? We have tutors online 24/7 who can help you get unstuck.
A+ icon
Ask Expert Tutors You can ask You can ask You can ask (will expire )
Answers in as fast as 15 minutes