Veterinary Epidemiology

Veterinary Epidemiology

Fourth Edition

Michael ThrusfieldVeterinary Clinical SciencesRoyal (Dick) School of Veterinary StudiesUniversity of Edinburgh

With Robert ChristleyEpidemiology & Population HealthInstitute of Infection & Global Health, andInstitute of Veterinary ScienceUniversity of Liverpool

And Helen Brown, Peter J. Diggle, Nigel French, Keith Howe, Louise Kelly, Annette O’Connor,Jan Sargeant and Hannah Wood

This edition first published 2018 © 2018 by John Wiley & Sons Ltd

Edition HistoryFirst edition © 1986 by Butterworth & Co. (Publishers) LtdSecond edition © 1995 by Blackwell Science LtdThird edition © 2005, 2007 by Blackwell Science Ltd, a Blackwell Publishing company

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means,electronic, mechanical, photocopying, recording or otherwise, except as permitted by law. Advice on how to obtain permission to reusematerial from this title is available at http://www.wiley.com/go/permissions.

The right of Michael Thrusfield and Robert Christley to be identified as the authors of the editorial material in this work has been asserted inaccordance with law.

Registered OfficesJohn Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, USAJohn Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester, West Sussex, PO19 8SQ, UK

Editorial Office9600 Garsington Road, Oxford, OX4 2DQ, UK

For details of our global editorial offices, customer services, and more information about Wiley products visit us at www.wiley.com.

Wiley also publishes its books in a variety of electronic formats and by print-on-demand. Some content that appears in standard print versionsof this book may not be available in other formats.

Limit of Liability/Disclaimer of WarrantyThe contents of this work are intended to further general scientific research, understanding, and discussion only and are not intended andshould not be relied upon as recommending or promoting scientific method, diagnosis, or treatment by physicians for any particular patient. Inview of ongoing research, equipment modifications, changes in governmental regulations, and the constant flow of information relating to theuse of medicines, equipment, and devices, the reader is urged to review and evaluate the information provided in the package insert orinstructions for each medicine, equipment, or device for, among other things, any changes in the instructions or indication of usage and foradded warnings and precautions. While the publisher and authors have used their best efforts in preparing this work, they make norepresentations or warranties with respect to the accuracy or completeness of the contents of this work and specifically disclaim all warranties,including without limitation any implied warranties of merchantability or fitness for a particular purpose. No warranty may be created orextended by sales representatives, written sales materials or promotional statements for this work. The fact that an organization, website, orproduct is referred to in this work as a citation and/or potential source of further information does not mean that the publisher and authorsendorse the information or services the organization, website, or productmay provide or recommendations it maymake. This work is sold withthe understanding that the publisher is not engaged in rendering professional services. The advice and strategies contained herein may not besuitable for your situation. You should consult with a specialist where appropriate. Further, readers should be aware that websites listed in thiswork may have changed or disappeared between when this work was written and when it is read. Neither the publisher nor authors shall beliable for any loss of profit or any other commercial damages, including but not limited to special, incidental, consequential, or other damages.

Library of Congress Cataloging-in-Publication Data

Names: Thrusfield, M. V., author. | Christley, Robert, 1968- author.Title: Veterinary epidemiology / by Michael Thrusfield, Veterinary ClinicalSciences, Royal (Dick) School of Veterinary Studies, University ofEdinburgh ; with Robert Christley and [8 others].

Description: Fourth edition. | Hoboken, NJ : Wiley, 2018. | Includesbibliographical references and index. |

Identifiers: LCCN 2017051658 (print) | LCCN 2017053201 (ebook) | ISBN9781118280263 (pdf) | ISBN 9781118280270 (epub) | ISBN 9781118280287(paperback)

Subjects: LCSH: Veterinary epidemiology. | MESH: EpidemiologicMethods–veterinary

Classification: LCC SF780.9 (ebook) | LCC SF780.9 .T48 2018 (print) | NLM SF780.9 | DDC 636.089/44–dc23

LC record available at https://lccn.loc.gov/2017051658

Cover images: (Top: from left to right) © holbox/Shutterstock; © Paul Looyen/Shutterstock; © Lorado/Gettyimages;© Lisa Van Dyke/Gettyimages; (Map) © yukipon/Gettyimages; (Bottom: from left to right) © Palenque/Gettyimages;© l i g h t p o e t/Shutterstock; © Seiji/Shutterstock; © claire norman/ShutterstockCover design by Wiley

Set in 10/12 pt Warnock by SPi Global, Pondicherry, India

10 9 8 7 6 5 4 3 2 1

In memory of George

Contents

Contributors xviiiFrom the preface to the first edition xixFrom the preface to the second edition xxFrom the preface to the third edition xxiPreface to the fourth edition xxiiAbout the companion website xxiv

1 The development of veterinary medicine 1Michael Thrusfield

Historical perspective 1Domestication of animals and early methods of healing 1Changing concepts of the cause of disease 2Impetus for change 5Quantification in medicine 10Contemporary veterinary medicine 12Current perspectives 12The fifth period 19Recent trends 20Further reading 25

2 The scope of epidemiology 28Michael Thrusfield

Definition of epidemiology 28The uses of epidemiology 29Types of epidemiological investigation 32Epidemiological subdisciplines 33Components of epidemiology 35Qualitative investigations 35Quantitative investigations 36Epidemiology’s locale 39The interplay between epidemiology and other sciences 39The relationship between epidemiology and other diagnostic disciplines 40Epidemiology within the veterinary profession 40Further reading 41

3 Causality 42Michael Thrusfield

Philosophical background 42Causal inference 43Methods of acceptance of hypotheses 44

vii

Koch’s postulates 45Evans’ rules 45Variables 46Types of association 46Non-statistical association 46Statistical association 46Confounding 49Causal models 50Formulating a causal hypothesis 53Methods of deriving a hypothesis 53Principles for establishing cause: Hill’s criteria 55Further reading 56

4 Describing disease occurrence 58Michael Thrusfield

Some basic terms 58Basic concepts of disease quantification 61The structure of animal populations 62Contiguous populations 62Separated populations 65Measures of disease occurrence 67Prevalence 67Incidence 67The relationship between prevalence and incidence rate 70Application of prevalence and incidence values 72Mortality 72Survival 73Example of calculation of prevalence, incidence, mortality, case fatality and survival 75Ratios, proportions and rates 76Mapping 80Geographic base maps 80Further reading 84

5 Determinants of disease 86Michael Thrusfield

Classification of determinants 86Host determinants 89Genotype 89Age 90Sex 91Species and breed 92Behaviour 93Other host determinants 93Agent determinants 94Virulence and pathogenicity 94Gradient of infection 97Outcome of infection 98Microbial colonization of hosts 100Environmental determinants 101Location 101Climate 101Husbandry 104

Contentsviii

Stress 105Interaction 106Biological interaction 108Statistical interaction 109The cause of cancer 110Further reading 112

6 The transmission and maintenance of infection 115Michael Thrusfield

Horizontal transmission 115Types of host and vector 115Factors associated with the spread of infection 118Routes of infection 121Methods of transmission 123Long-distance transmission of infection 125Vertical transmission 129Types and methods of vertical transmission 129Immunological status and vertical transmission 129Transovarial and trans-stadial transmission in arthropods 130Maintenance of infection 131Hazards to infectious agents 131Maintenance strategies 132Transboundary diseases 135Further reading 136

7 The ecology of disease 138Michael Thrusfield

Basic ecological concepts 139The distribution of populations 139Regulation of population size 142The niche 148Some examples of niches relating to disease 150The relationships between different types of animals and plants 152Ecosystems 155Types of ecosystem 156Landscape epidemiology 158Nidality 159Objectives of landscape epidemiology 161Landscape characteristics determining disease distribution 164Further reading 165

8 Patterns of disease 168Michael Thrusfield

Epidemic curves 168Kendall’s Threshold Theorem 168Basic reproductive number (R0) 169Dissemination rate 172Common-source and propagating epidemics 172The Reed–Frost model 173Kendall’s waves 175Trends in the temporal distribution of disease 177Short-term trends 177

Contents ix

Cyclical trends 178Long-term (secular) trends 179True and false changes in morbidity and mortality 180Detecting temporal trends: time series analysis 180Trends in the spatial and temporal distribution of disease 186Spatial trends in disease occurrence 186Space–time clustering 186Further reading 187

9 Comparative epidemiology 189Michael Thrusfield

Types of biological model 189Cancer 191Monitoring environmental carcinogens 191Identifying causes 192Comparing ages 193Some other diseases 196Diseases with a major genetic component 196Some non-infectious diseases 197Diseases associated with environmental pollution 198Reasoning in comparative studies 199Further reading 199

10 The nature of data 201Michael Thrusfield

Classification of data 201Scales (levels) of measurement 201Composite measurement scales 204Data elements 205Nomenclature and classification of disease 205Diagnostic criteria 207Sensitivity and specificity 208Accuracy, refinement, precision, reliability and validity 209Bias 210Representation of data: coding 210Code structure 211Numeric codes 212Alpha codes 213Alphanumeric codes 214Symbols 215Choosing a code 215Error detection 216Further reading 217

11 Data collection and management 219Michael Thrusfield

Data collection 219Questionnaires 219Quality control of data 228Data storage 229Database models 229Non-computerized recording techniques 231

Contentsx

Computerized recording techniques 232Veterinary recording schemes 232Scales of recording 232Veterinary information systems 234Some examples of veterinary databases and information systems 237Geographical information systems 244Further reading 248

12 Presenting numerical data 251Michael Thrusfield and Robert Christley

Some basic definitions 251Some descriptive statistics 252Measures of position 253Measures of spread 254Statistical distributions 254The Normal distribution 254The binomial distribution 255The Poisson distribution 255Other distributions 256Transformations 256Normal approximations to the binomial and Poisson distributions 257Estimation of confidence intervals 257The mean 257The median 258A proportion 258The Poisson distribution 259Some epidemiological parameters 260Other parameters 261Bootstrap estimates 261Displaying numerical data 262Displaying qualitative data 262Displaying quantitative data 263Monitoring performance: control charts 266Further reading 269

13 Surveys 270Michael Thrusfield and Helen Brown

Sampling: some basic concepts 270Types of sampling 272Non-probability sampling methods 272Probability sampling methods 272What sample size should be selected? 275Estimation of disease prevalence 275Detecting the presence of disease 284The cost of surveys 290Calculation of confidence intervals 290Further reading 294

14 Demonstrating association 296Michael Thrusfield

Some basic principles 296The principle of a significance test 296The null hypothesis 297

Contents xi

Errors of inference 297Multiple significance testing 298One- and two-tailed tests 298Independent and related samples 299Parametric and non-parametric techniques 299Hypothesis testing versus estimation 300Sample-size determination 300Statistical versus clinical (biological) significance 300Interval and ratio data: comparing means 302Hypothesis testing 302Calculation of confidence intervals 303What sample size should be selected? 304Ordinal data: comparing medians 304Hypothesis testing 304Calculation of confidence intervals 308What sample size should be selected? 309Nominal data: comparing proportions 309Hypothesis testing 310Calculation of confidence intervals 313What sample size should be selected? 314χ2 test for trend 314Correlation 316Multivariate analysis 317Statistical packages 318Further reading 318

15 Observational studies 319Michael Thrusfield

Types of observational study 319Cohort, case-control and cross-sectional studies 319Measures of association 321Relative risk 321Odds ratio 323Attributable risk 325Attributable proportion 327Interaction 328The additive model 328Bias 330Controlling bias 332What sample size should be selected? 335Calculating the power of a study 336Calculating upper confidence limits 337Further reading 338

16 Design considerations for observational studies 339Robert Christley and Nigel French

Descriptive observational studies 339Analytical observational studies 340Design of cohort studies 340Design of case-control studies 346Design of cross-sectional analytical studies 352Overview of other study designs 354Further reading 359

Contentsxii

17 Clinical trials 361Michael Thrusfield

Definition of a clinical trial 361Design, conduct and analysis 364The trial protocol 364The primary hypothesis 364The experimental unit 367The experimental population 368Admission and exclusion criteria 368Blinding 369Randomization 369Trial designs 370What sample size should be selected? 372Losses to follow-up 373Compliance 373Terminating a trial 374Interpretation of results 374Meta-analysis 375Goals of meta-analysis 376Components of meta-analysis 377Sources of data 377Data analysis 378Further reading 380

18 Validity in epidemiological studies 383Robert Christley and Nigel French

Types of epidemiological error 383Accuracy, precision and validity in epidemiological studies 384Background factors 385Interpretation bias 385Selection bias 386Examples of selection biases 387Information bias 390Examples of information biases 390Statistical interaction and effect-measure modification 392Confounding 392Criteria for confounding 393Confounding and causal diagrams 394Controlling confounding 394Errors in analysis 395Communication bias 395Further reading 396

19 Systematic reviews 397Annette O’Connor, Jan Sargeant and Hannah Wood

Evidence synthesis 397Overview of systematic reviews 397Differences between systematic reviews and narrative reviews 398Questions that are suitable for systematic reviews 398Types of review questions suitable for systematic reviews 399Extensive search of the literature 399Assessment of risk of bias in a systematic review 400

Contents xiii

Steps of a systematic review 400Step 1: Define the review question and the approach to conduct of the review (i.e., create a protocol) 402Step 2: Comprehensive search for studies 403Step 3: Select relevant studies from the search results 406Step 4: Collect data from relevant studies 407Step 5: Assess the risk of bias in relevant studies 409Step 6: Synthesize the results 412Step 7: Presenting the results 416Step 8: Interpret the results and discussion 419Further reading 419

20 Diagnostic testing 421Michael Thrusfield

Serological epidemiology 421Assaying antibodies 421Methods of expressing amounts of antibody 421Quantal assay 423Serological estimations and comparisons in populations 424Antibody prevalence 424Rate of seroconversion 425Comparison of antibody levels 426Interpreting serological tests 427Refinement 427Accuracy 429Evaluation and interpretation of diagnostic tests 430Sensitivity and specificity 430Youden’s index 433Diagnostic odds ratio 434Predictive value 434Likelihood ratios 436ROC curves 441Aggregate-level testing 443Multiple testing 444Diagnostic tests in import risk assessment 446Guidelines for validating diagnostic tests 447Validating diagnostic tests when there is no gold standard 448Agreement between tests 450Practical application of diagnostic tests 456Further reading 456

21 Surveillance 457Michael Thrusfield

Some basic definitions and principles 457Definition of surveillance 457Goals of surveillance 458Types of surveillance 459Some general considerations 461Sources of data 464Mechanisms of surveillance 471Surveillance networks 475Surveillance in less-economically-developed countries: participatory epidemiology 475Principles of participatory epidemiology 477Techniques of data collection 478Strengths and weaknesses of participatory epidemiology 481

Contentsxiv

Some examples of participatory epidemiology 483Companion-animal surveillance 483Wildlife surveillance 485Aquatic-animal surveillance 485Assessing the performance of surveillance systems 486Improving the performance of surveillance: risk-based surveillance 486Further reading 488

22 Statistical modelling 492Robert Christley and Peter J. Diggle

Simple linear regression models 492Key assumptions of linear regression models 495Modelling more than one input variable 499Handling categorical input variables 500Non-linear modelling of quantitative input variables 502Additive models 502Categorization of the input variable 502Transformation of the input and/or output variable 504Piece-wise regression 504Modelling interactions 505Model selection 506Modelling binary outcomes 509Generalized linear models 511The multiple logistic regression model 511Model selection for logistic regression models 512Diagnostic checking of logistic regression models 513Generalized additive models 514Modelling clustered data 514Further reading 519

23 Mathematical modelling 520Michael Thrusfield

Types of model 521Modelling approaches 521Deterministic differential calculus modelling 521Stochastic differential calculus modelling 525Empirical simulation modelling 526Process simulation modelling 527Monte Carlo simulation modelling 528Matrix population modelling 530Network population modelling 532Contact-network modelling 533Systems modelling 534The rational basis of modelling for active disease control 534Available knowledge, and the functions of models 534From theory to fact 535Model building 536Further reading 538

24 Risk analysis 540Michael Thrusfield and Louise Kelly

Definition of risk 540Risk analysis and the ‘precautionary principle’ 543Risk analysis in veterinary medicine 543

Contents xv

Components of risk analysis 545Hazard identification 546Risk assessment 546Risk management 548Risk communication 551Qualitative or quantitative assessment? 551Semi-quantitative risk assessment 551Qualitative risk analysis 552Framework for qualitative risk assessment 552Qualitative risk assessment during epidemics 554Quantitative risk analysis 556Framework for quantitative risk assessment 556What level of risk is acceptable? 560Further reading 563

25 Economics and veterinary epidemiology 565Keith Howe and Michael Thrusfield

General economic concepts 565Production functions 565Disease and animal production functions 566Value and money 567Money and prices 567Opportunity cost 568Technical and economic efficiency 568Positive and normative economics 569Levels of aggregation 569Disease contained at farm level 569Disease not contained at farm level 570Zoonotic disease 570Disease at international level 571Evaluating disease-control policies 575Components of disease costs 576Optimum control strategies 577Partial budgets 579Social cost–benefit analysis (CBA) 579Summary of methods 582Further study 582Further reading 584

26 Health schemes 586Michael Thrusfield

Private health and productivity schemes 586Structure of private health and productivity schemes 586Dairy health and productivity schemes 588Pig health and productivity schemes 591Sheep health and productivity schemes 592Beef health and productivity schemes 594National schemes 597Accredited/attested herds 597Health schemes 598Companion-animal schemes 599Further reading 603

Contentsxvi

27 The control and eradication of disease 604Michael Thrusfield

Definition of ‘control’ and ‘eradication’ 604Strategies of control and eradication 605Important factors in control and eradication programmes 616Outbreak investigation 623Cause known: foot-and-mouth disease 623Cause unknown: chronic copper poisoning 625The epidemiological approach to investigation of outbreaks 626Veterinary medicine in the 21st century 628Livestock medicine 628Companion-animal medicine 629Further reading 630

General reading 633Appendices 635

Appendix I: Glossary of terms 636Appendix II: Basic mathematical notation and terms 641Appendix III: Some computer software 643Appendix IV: Veterinary epidemiology on the Internet 648Appendix V: Student’s t-distribution 650Appendix VI: Multipliers used in the construction of confidence intervals based on the Normal distribution,

for selected levels of confidence 651Appendix VII: Values of exact 95% confidence limits for proportions 652Appendix VIII: Values from the Poisson distribution for calculating 90%, 95% and 99% confidence intervals

for observed numbers from 0 to 100 658Appendix IX: The χ2 distribution 660Appendix X: Technique for selecting a simple random sample 661Appendix XI: Sample sizes 663Appendix XII: The probability of detecting a small number of cases in a population 669Appendix XIII: The probability of failure to detect cases in a population 671Appendix XIV: Sample sizes required for detecting disease with probability, p1, and threshold number of

positives 672Appendix XV: Probabilities associated with the upper tail of the Normal distribution 676Appendix XVI: Lower- and upper-tail probabilities for Wx, the Wilcoxon–Mann–Whitney rank-sum

statistic 678Appendix XVII: Critical values of T + for the Wilcoxon signed ranks test 683Appendix XVIII: Values of K for calculating 95% confidence intervals for the difference between

population medians for two independent samples 685Appendix XIX: Values of K∗ for calculating 95% confidence intervals for the difference between

population medians for two related samples 688Appendix XX: Common logarithms (log10) of factorials of the integers 1–999 689Appendix XXI: The correlation coefficient 691Appendix XXII: The variance-ratio (F) distribution 692

References 694Index 841

Contents xvii

Contributors

Michael ThrusfieldVeterinary Clinical SciencesRoyal (Dick) School of VeterinaryStudiesUniversity of EdinburghEdinburgh, UK

Robert ChristleyInstitute of Infection and Global HealthInstitute of Veterinary ScienceUniversity of LiverpoolLiverpool, UK

Helen BrownThe Roslin InstituteUniversity of EdinburghEdinburgh, UK

Peter J. DiggleCHICASLancaster Medical SchoolLancaster UniversityLancaster, UK

Nigel FrenchInstitute of Veterinary Animal andBiomedical SciencesMassey UniversityPalmerston North, New Zealand

Keith HoweCentre for Rural Policy ResearchUniversity of ExeterExeter, UK

Louise KellyUniversity of StrathclydeGlasgow, UK

Annette O’ConnorCollege of Veterinary MedicineIowa State UniversityAmes, Iowa, USA

Jan SargeantOntario Veterinary CollegeUniversity of GuelphGuelph, Canada

Hannah WoodYork Health Economics ConsortiumYork, UK

xviii

From the preface to the first edition

The common aim of the many disciplines that com-prise veterinary medicine is an increase in the healthof animal populations, notably of domestic livestockand companion animals. This goal has traditionallybeen achieved by individual diagnosis and treatment:procedures that evolved contemporaneously in veter-inary and human medicine, when infectious diseases,which had predominantly single causes and clearlyidentifiable signs, were commonplace.Four major changes in the veterinarian’s apprecia-

tion of and approach to disease problems haveoccurred over the past 20 years. First, despite tradi-tional control techniques, for example slaughter andvaccination, some diseases remain at refractory levelsand now require continuous scrutiny to detect chan-ging levels of occurrence associated with ecologicaland management factors. An example is the detectionof ‘pockets’ of bovine tuberculosis in England in areaswhere infection of badgers is recorded. Secondly, thecontrol of infectious disease has freed animals frommajor causes of death, thereby facilitating the emer-gence of non-infectious diseases as major problems:examples are the cardiac, dermal and renal diseasesof dogs. Many of these diseases have a poorly under-stood, often complex (i.e., multifactorial) cause.Thirdly, the intensification of animal industries hashighlighted new ‘diseases of production’, often mani-fested as poor performance, rather than clinical

disease, and frequently with multifactorial causes.Fourthly, economic evaluation has become important:the economic advantages of disease control, which areobvious with the major animal plagues such as rinder-pest, can be difficult to identify when overt disease anddramatic changes in levels of performance are notinvolved. These four changes in the approach to,and appreciation of, disease have added momentumto the emergence of veterinary epidemiology as a dis-cipline concerned with the measurement of theamount of disease and its economic effects, the iden-tification and quantification of the effects of factorsassociated with disease, and the assessment of theeffects of prevention and treatment of disease ingroups of animals.A knowledge of elementary statistics is essential for

an understanding of the full range of epidemiologicaltechniques. Hitherto, most epidemiology books eitherhave assumed a knowledge of statistics or have avoideda description of the mathematical manipulations thatare commonly used in epidemiology. However, theextent of statistical teaching varies widely between vet-erinary schools. Two chapters therefore are includedas an introduction to basic statistics, and are intendedto make this book statistically self-sufficient (thoughnot comprehensive). Similarly, a chapter includes anintroduction to computers, which are now used widelyin the recording and analysis of epidemiological data.

xix

From the preface to the second edition

Since publication of the first edition, veterinary med-icine has faced several new problems, and has beenforced to evaluate established ones more critically.Bovine spongiform encephalopathy emerged as a seri-ous problem in the United Kingdom. Rinderpest is stillthe subject of a global eradication campaign. There isan increasing demand for comprehensive, high-qualitytechnical and economic information on animal diseaseand productivity at the national and internationallevel; and information systems, such as the UnitedStates’ National Animal Health Monitoring System,have been designed to suit these requirements. Themoves towards an open market, both in the EuropeanUnion and internationally following the Uruguayround of the General Agreement on Tariffs and Trade,highlight the need for information on animal diseasestatus in trading nations. Multifactorial diseases con-tinue to predominate in intensive production systems,

and many companion animal diseases are similarlycomplex. The solving of these problems and fulfillingof these tasks rely heavily on epidemiological princi-ples and techniques.All chapters of the first edition have been revised.

Chapter 11 has been modified to take account of theincreasing popularity of microcomputers, and therapid development of veterinary information systems.New chapters on clinical trials and comparative epide-miology have been added in response to suggestionsfrom colleagues. More statistical methods are includedin Chapters 12–15 and 17. The goal of this edition nev-ertheless remains the same as that of the first: to pro-vide an introduction to veterinary epidemiology forveterinary undergraduates, postgraduates who havereceived limited or no training in epidemiology, andpractising veterinarians and members of other disci-plines with an interest in the subject.

xx

From the preface to the third edition

The ten years since publication of the second editionhave witnessed further expansion in the applicationof veterinary epidemiology. Quantitative methodshave increased dramatically, both in their develop-ment and use. Evidence-based clinical veterinary med-icine is now widely appreciated, and relies heavily ofthe results of epidemiological analyses, includingobservational studies, clinical trials and the quantita-tive interpretation of diagnostic tests. The successfulemployment of epidemiological techniques has wit-nessed continued progress in rinderpest eradication,whereas the epidemic of foot-and-mouth disease inEurope (notably in the UK) in 2001 presented freshchallenges.

All chapters from the second edition have beenrevised.Numerous topics (e.g., causality,nowexclusivelyconsidered in Chapter 3) have been expanded inresponse to questions raised by undergraduates, post-graduates and professional colleagues, with the aim ofconsolidating the more enduring principles and con-cepts of epidemiology. Surveillance is now addressedspecifically in Chapter 10. Chapter 17 has been enlargedin response to increased interest in diagnostic-test vali-dation and performance. The opportunity also has beentaken to correct several typographical errors that werepresent in the second edition. The text continues to bean introduction to veterinary epidemiology, directed atall with an interest in the subject.

xxi

Preface to the fourth edition

Since the publication of the third edition, veterinaryepidemiology has continued to contribute to the suc-cessful control of animal disease; notably, in the globaleradication of rinderpest. However, some diseasesremain refractory, both at the international level(e.g., rabies) and nationally (e.g., bovine tuberculosisin England). Additionally, new microbes, such asSchmallenberg virus, pose fresh challenges. Evi-dence-based veterinary medicine, which relies heavilyon epidemiological techniques, has continued toexpand in both livestock and companion-animal prac-tice. Similarly, epidemiology is central to the practiceof ‘One Health’, which recognizes the link betweenhuman, animal and ecosystem health, and whichrecently has gained increased attention.All chapters of the third edition have been revised.

Although the subject coverage remains essentiallythe same as earlier editions, the order of some chaptershas been changed. The first part of the book now con-cerns mainly qualitative material. This is followed bylargely quantitative topics. The last part addressesthe application of epidemiology to disease control.Risk analysis, which was included as an appendix tothe paperback re-issue of the third edition, is nowincorporated into the main body of the text. Anupdated version of the bibliography of observationalstudies given in Appendix XXII of the third editionhas been excluded from this edition on the groundsthat, with the substantial increase in the number ofsuch studies, it is unlikely to be comprehensive, whilstalso adding considerably to the size of the book. Someold references have been retained, and some added, togive the reader a perspective on the historical develop-ment of knowledge and ideas. New, more advanced,chapters on the design of observational studies, valid-ity in epidemiological studies, systematic reviews, andstatistical modelling have been included, should thereader wish to pursue these topics in depth. However,the goal of this edition remains the same as that of thefirst three: to provide an introduction to veterinaryepidemiology for veterinary undergraduates, post-graduates who have received little or no training in vet-erinary epidemiology, and practising veterinarians and

members of other disciplines with an interest in thesubject.My gratitude is again due to colleagues whose

scholarship and comments during the writing of thisfourth edition have been invaluable. Ian Handel wasvery generous with his time in engaging in discus-sions on test validation without a gold standard,and in providing data for the Bland–Altman analysisin Chapter 20. David Argyle supplied comments onthe aetiology of cancer, discussed in Chapter 5. AlexDonaldson criticized aspects of disease control in thelast chapter. Helen Brown, Louise Kelly and KeithHowe are recognized for their contributions to Chap-ters 13, 24 and 25, respectively. I am also indebted toRobert Christley, who collaborated with Nigel Frenchin writing Chapters 16 and 18 and with Peter Digglein writing Chapter 22, and who supplied additions tosome of the other chapters. Annette O’Connor, JanSargeant and Hannah Wood wrote Chapter 19, andMarshall Dozier passed valuable comments on thischapter.Eoghan Clarkson and Brian Mather skilfully con-

verted rough diagrams into reproducible figures.Pat Fairbairn, Doreen Graham, Bogusia McRoberts

and Lisa Norman, librarians at the Royal (Dick) Schoolof Veterinary Studies, University of Edinburgh, effi-ciently obtained many of the references cited inthe text.Justinia Wood at Wiley continued her earlier

involvement with previous editions of the book, and,latterly, Erica Judisch and Purvi Patel took over herrole. Jane Andrew, Nick Morgan, James Schultz andKathy Syplywczak assisted with conversion of themanuscript to published text.Finally, I am grateful to the Literary Executors of the

late Sir Ronald A. Fisher FRS and the late Dr FrankYates FRS, and to the Longman Group Limited, Lon-don, for permission to reprint Tables III, IV, V and VIIfrom their book, Statistical Tables for Biological, Agri-cultural and Medical Research (6th edition, 1974).

Michael ThrusfieldEdinburgh

xxii

Over the years since publication of the third edition,veterinary science, veterinary epidemiology and ani-mal health more generally have faced new challengesand opportunities. The eradication of rinderpest hasseen the successful elimination of only the second dis-ease and the first animal disease. However, the contin-ued emergence of new animal diseases, the increasedrecognition of the role of zoonotic diseases in theemergence of diseases of people, and increasing appre-ciation of the impacts of global change, including cli-mate change, global movement of people, animals andproducts has led to greater appreciation of the need tounderstand these phenomena and the changes inhealth with which they are associated. Simultaneously,new methods have been developed and older methods‘rediscovered’. From techniques for using ‘big data’ toparticipatory approaches to engaging with commu-nities, veterinary epidemiology continues to developacross a broad front. It is an exciting time.This fourth edition sees incorporation of additional

chapters on epidemiological study design, validity ofepidemiological studies, statistical modelling and sys-tematic reviews. These new chapters provide up todate information on these key issues, complementing,it is hoped, the material in the revised and updatedchapters from the third edition.I am indebted to many people who have contribu-

ted to development of this book. In particular,Michael Thrusfield has provided essential guidanceand scrutiny. At Wiley, Justinia Wood was a sourceof support and advice that has been seamlessly con-tinued by Erica Judisch. I am indebted to NigelFrench and Peter Diggle for agreeing to co-author

chapters with me, and to Annette O’Connor, Jan Sar-geant and Hannah Wood who have authored thechapter on systematic reviews. Numerous peoplehave provided guidance and critique on sections ofthe text, including, but not limited to, Gina Pinch-beck, Dan O’Neill, Dave Brodbelt, Dai Grove White,Cathy McGowan, Alan Radford, Navneet Dhand andPeter Cripps.Many colleagues at the University of Liverpool have

provided advice, including Sarah O’Brien, MatthewBaylis, Malcolm Bennett and Jo Turner. I have been(and continue to be) lucky to work with many tal-ented postgraduate students who have challengedme to think again about things I might otherwise havetaken for granted: David Sutton, Nick Malikides, JoelHotchkiss, Emily Brook, Maz Behnke, Carri West-garth, Chris Jewell, Susie Robinson, Jane Murray,Marnie Brennan, Lutfi Al Tunesi, Bryony Parsons,Richard Smith, John Mehers, Andy Stringer, ClaireScantlebury, Alex Berriman, Puteri Nohuddin, JudyBettridge, Zelalem Gutu, Lisa Jameson, KwankamonDittakan, Gabby Laing, Elsa Sandoval Barron, SophieBeale, Charlotte Rôbin, Margaux Mesle, StephanieBegemann, Sara Owczarczak-Garstecka, MohamedAL-Zeyadi, Arturo Hernandez Colina, Tamzin Fur-tado, John Tulloch, Catherine McLeonard and Han-nah Brindle.Finally, my heartfelt thanks and love go to Rosa and

Jake, who have, more than is reasonable, ridden thisrollercoaster with me.

Robert ChristleyLiverpool

Preface to the fourth edition xxiii

About the companion website

This book is accompanied by a companion website:

www.wiley.com/go/veterinaryepidemiology

The website material was produced by Robert Christley and includes:

• Interactive multiple-choice questions

xxiv

1

The development of veterinary medicine

Veterinary epidemiology is concerned with disease inanimal populations. Its evolution has spanned severalcenturies and has been central to the successful con-trol of many animal diseases. This introductory chap-ter traces the development of veterinary medicine ingeneral (including relevant aspects of human medi-cine), showing that it has been inseparably linked tothat of veterinary epidemiology.Although man’s association with animals began in

prehistoric times, the development of scientific veter-inary medicine is comparatively recent. A milestone inthis growth was the establishment of the first perma-nent veterinary school at Lyons, France, in 1762. Earlydevelopments were governed largely by economicrather than humanitarian motives, associated withthe importance of domestic stock as a source of foodand as working animals; and there are still importanteconomic reasons for concern about disease in animalpopulations. Later, with the advent of the industrialrevolution and the invention of the internal combus-tion engine, the importance of draft animals declinedin the more-economically-developed countries.Although dogs and cats have been companion animalsfor several thousand years, it is only relatively recentlythat they and other pets have increased in importanceas components of human society.Until the last half of the 20th century, the emphasis

of veterinary medicine had been on the treatment ofindividual animals with clearly identifiable diseasesor defects. Apart from routine immunization and pro-phylactic treatment of internal parasites, restrictedattention had been given to herd health and compre-hensive preventive medicine, which give proper con-sideration to both infectious and non-infectiousdiseases.Currently, the nature of traditional clinical practice

is changing in the more-economically-developedcountries. The stock owner is better educated, and,among livestock, the value of individual animals rela-tive to veterinary fees has decreased. Therefore, con-temporary large-animal practitioners, if they are to

meet modern requirements, must support herd healthprogrammes designed to increase production by pre-venting disease, rather than just dispensing traditionaltreatment to clinically sick animals.In the less-economically-developed countries, the

infectious diseases still cause considerable loss of ani-mal life and production. Traditional control techni-ques, based on identification of recognizable signsand pathological changes, cannot reduce the level ofsome diseases to an acceptable degree. Different tech-niques, based on the study of patterns of disease ingroups of animals, are needed.Similarly, contemporary companion-animal practi-

tioners, like their medical counterparts, are becomingincreasingly involved with chronic and refractory dis-eases which can be understood better by an investiga-tion of the diseases’ characteristics in populations.This chapter outlines the changing techniques of

veterinary medicine by tracing man’s attempts at con-trolling disease in animals, and introduces some cur-rent animal disease problems that can be solved byan epidemiological approach.

Historical perspective

Domestication of animals and early methods ofhealing

The importance of animal healers has been acknowl-edged since animals were initially domesticated, whenthey were already likely to have been chronicallyaffected by various infections (McNeill, 1977). Thedog, naturally a hunter, was probably the first animalto be domesticated over 14 000 years ago when itbecame the companion of early hunters, with evidenceof close proximity to humans as early as 31 000 yearsago (Germonpré et al., 2009); and differentiation fromits ancestor, the wolf, was likely to have occurred atleast 10 000 years ago, as hunter-gatherer societiesgradually evolved into sedentary agricultural

1

Veterinary Epidemiology, Fourth Edition.© 2018 John Wiley & Sons Ltd. Published 2018 by John Wiley & Sons Ltd.Companion website: www.wiley.com/go/veterinaryepidemiology

populations (Vilà et al., 1997). Sheep and goats weredomesticated by 9000 BC in the fertile Nile valley andwere the basis of early pastoral cultures. A few of thesesocieties have lasted (e.g., Pfeffer and Behera, 1997),but many were superseded by cattle cultures; in some,the pig increased in importance (Murray, 1968). AnEgyptian cattle culture evolved from4000 BC, and farm-ing spread from the Near East into Europe (Figure 1.1).There is archaeological evidence of cattle shrines inAnatolia dating back to 6000 BC (Mellaart, 1967). Thisrecord illustrates that animals had religious, as well aseconomic, significance in early civilizations. Theaurochs was central to the religion of the Sumerians,who migrated throughout Asia, North Africa andEurope in the third millennium BC taking their animalsand beliefs with them. India is the largest cattle culturethat remains. Cattle cultures also persist in north-eastAfrica, the result of interaction between the AncientEgyptians and early Nilotic tribes. Cattle still playimportant roles in these cultures: they are food, com-panionship, and status and religious symbols to theSuk (Beech, 1911) andDinka (Lienhardt, 1961) of SouthSudan.The first extensive colonization of the Eurasian

steppe and semi-arid areas occurred in the third mil-lennium BC. The horse provided the key to successfulexploitation of the area north of the Black Sea, theCaucasus, and the Taurus and Zagros mountains(Barraclough, 1984), and a Eurasian horse culture,associated with warrior tribes, emerged (Simpson,1951). Some of these tribes overran the older cattlecultures. The horse is represented in Iranian, Greekand Celtic pantheons. It has become a symbol of vet-erinary medicine in the form of a centaur, one of

which, Chiron, was considered to be the mythologicalfounder of Greek medicine.There have been several movements of animals with

concomitant social and agricultural modificationssince the early changes. The camel was introduced intoSaharan Africa in the first century BC, and into the Sub-Saharan region around AD 400 (Spencer and Thomas,1978; Phillipson and Reynolds, 1996), the latter alreadyhaving well established domestic cattle and goat popu-lations (Cain, 1999; Tefera, 2004). The Spanish intro-duced cattle, sheep, pigs and goats to North Americain the 16th century. Haired sheep were introducedto Africa by European slave traders. The Spanishbrought turkeys to Europe from North America.The early Egyptian healers combined religious and

medical roles by being priest-healers, often associatedwith temples. Their therapeutic techniques arerecorded in the veterinary Papyrus of Kahun (c. 1900BC). Literary records of similar age, describing veteri-nary activities, are extant from other parts of the world,such as Indian Sanskrit texts from the Vedic period(1800–1200 BC).

Changing concepts of the cause of disease

Concepts of the cause of disease have changed andevolved1. A method of treatment used by earlyEgyptians was incantation. This was partly ritual, butalso reflected their belief in supernatural spirits as apossible cause of disease. Approaches to treatmentand prevention are the direct result of theories ofcause. There have been five main theories up to themiddle of the last century2. One theory was oftensuperseded by another, but traces of each still can beseen in different parts of the world.

DemonsEarly man attributed disease to supernatural powers,the product of animism, which imbued all movingthings with a spirit. In this ‘spirit-world’, disease couldbe produced by witches3, superhuman entities andspirits of the dead (Valensin, 1972). Treatment there-fore included: placation, for example by sacrifice; exor-cism (forcible expulsion); evasion, for instance

3000

4000

5000

Figure 1.1 A generalized map to show the spread of farmingfrom the Near East to Europe in years BC. (Adapted fromDyer, 1990.)

1 Causality is outlined in this chapter specifically in the contextof disease. A more general discussion is presented in Chapter 3.2 Theories of the cause of disease also have similarities withtheories of the origin of species, and both have rationalistic andtheological dimensions (Bullock, 1992).3 A witch was originally defined as ‘one who by commerce withthe Devil has a full intention of attaining his (or her) own ends’(Bodin, 1580).Witchcraft becamewidespread in Europe between the12th and 18th centuries. In the depositions of witch trials, there aremany examples of the supposed induction of disease and death inman and domestic animals by witches (L’Estrange Ewen, 1933).

1 The development of veterinary medicine2

scattering millet seeds to avoid vampires (Summers,1961); and transference, often to human and animal‘scapegoats’4, probably the best known single exampleof which is the Gadarene swine (the Bible: Mark 5,i–xiii). The techniques included: ritual ceremonies;material objects that could be suspended (amuletsand periapts), carried (talismans), hung in a building(fetishes and icons) or displayed in the community(totems); the use of special people such as witch doc-tors; and incantations5. The meaning of the Indianword ‘brahmin’ originally was ‘healer’ because theBrahmin were a class of healers. In the Neolithicperiod (4200–2100 BC), trepanation (the removal of a

bone disc from the skull) may have been practised torelease evil spirits from sick people (Buckland, 1882;Wakefield and Dellinger, 1939). The practice also isrecorded in veterinary texts as early as the 16th cen-tury, but may be much older (Binois, 2015); and theGreek physician, Galen, practised trepanation on apesin the second century AD (Arnott et al., 2003).During the 19th century, many European peasants

still believed that diseases of cattle were caused by evilspirits, which could be kept at bay by fire (Frazer,1890), and the African Nuer tribe occasionally stilluses incantations during ritual sacrifice when cattleepidemics occur (Evans-Pritchard, 1956)6. Moreover,sacrifice was practised in England as late as the 19thcentury (Baker, 1974).

Divine wrathThe demonic theory involved many spirits; the nextdevelopment, monotheistic in origin, argued thatdisease was the product of a displeased supremebeing: disease was punishment for sinful behaviour(Figure 1.2). This belief is prominent in the Old Testa-ment, for example, the animal plague of Egypt (the

Figure 1.2 Plague allegory: the fateof gamblers and lechers. Giovannide Paolo: Allegory of the Plague,Sienese, 1436–1437. (Reproducedwith permission of bpk/Kunstgewerbemuseum, StaatlicheMuseen zu Berlin/Saturia Linke.)

4 The scapegoat had the dual purpose of averting and magicallytransferring guilt and evil, both generally and at a specific time ofcrisis, such as plague or failure of crops. It takes its name from theHebrew rites of the Day of Atonement when a goat was driven intothe wilderness after the High Priest had ritually confessed the sins ofthe people and transferred them to the goat. The custom occursuniversally from Ancient Babylonian times to modern times, wherehuman sacrificial scapegoats have been known in some tribalsocieties (Cooper, 1990).5 Historical examples of the use of amulets, talismans and ‘whitewitches’ to prevent and control diseases of livestock in England andthe British Colonies are given by Baker (1974) and, as late as the 19thcentury, ‘white witches’ (‘charmers’) were recommended by theMinistry of Agriculture’s veterinarians (St Leger-Gordon, 1994).

6 More recently, there has been a trend towards a contemporaryunderstanding of disease (Hutchinson, 1996).

Historical perspective 3

Bible: Exodus 9, iii) and is also evident in Persian andAztec writings. The only effective treatment of diseaseinduced in this way was placation because exorcismand evasion would not be effective against a supremebeing. The English veterinary surgeon, WilliamYouatt, writing in 1835, supported the practice ofburning crosses on the heads of cattle to cure and pre-vent disease. In 1865, Queen Victoria, believing thatthe current British rinderpest (cattle plague) outbreakwas the result of divine displeasure, ordered that aprayer should be used in each church in England whilethe epidemic continued. Beliefs in causation involvingdivine wrath are still evident today; for example, insome interpretations of the HIV/AIDs epidemic aspunishment of the sinful (Kopelman, 2002).

Metaphysical medicineThe next development did not assume the existence ofa supreme being, either demonic or divine, butassumed the presence of occult forces beyond thephysical universe. This ‘metaphysical’medicine embo-died a theory of natural laws but excluded scientificprinciples such as observation and the repeatabilityof phenomena. The moon, stars and planets were con-sidered to affect health (Whittaker, 1960), these con-cepts being obvious predecessors of astrology.Several outbreaks of rinderpest in Dark Age Europewere ascribed to earthquakes, floods and comets.Treatment frequently included particularly foulmed-

icines and practices that persisted for many centuries.A recommended 17th-century cure for broken windin horses comprised toads, swallows and moles roastedalive and mixed with shoe soles7. Divination, practisedby the Babylonians using sheep livers, and the ‘Doctrineof Signatures’ which suggested a similarity between thedisease and its cure – for example, using toads to treatwarts – were notable metaphysical developments.

The universe of natural lawA major intellectual revolution began in Greece in thesixth century BC in which the universe was rationalizedwithout either demonic or metaphysical influences.The Greeks thought that disease was the result ofderangement of four ‘humours’ of the body, whichwere associated with four properties (heat, moisture,dryness and cold) and with four elements (air, earth,water and fire) (Figure 1.3)8. Diseases were considered

to be caused by external forces, including climatic andgeological changes that affected the population. Localoutbreaks of disease were thought to be the result oflocal eruptions of noxious air: miasmata (miasmas)9.The word ‘malaria’ literally means ‘bad air’ and hints atthe 19th-century belief that the disease was caused bystale air around swamps. This belief led to Europeancolonists preferring to settle away from such areas,often choosing higher altitudes. This strategy was ben-eficial because it moved the colonists away from thebreeding sites of the actual vectors of the disease,Anopheles spp. mosquitoes.The concept of humoral derangement was reim-

ported into mediæval Europe, via Sicily, during theCrusades, and food was imbued with the same prop-erties as the humours (Tannahill, 1968). The conceptpersists in several cultures. In indigenous IndianAyurvedic human and veterinary medicine, basedon the Hindu Scriptures (Vedas), there are threehumours (tridosa): vata (wind), pitta (bile) and kapha(phlegm); derangement of vata, for example, causingasthma and diarrhoea. This concept is also central tomodern Mahayana Buddhist medicine. However, inEurope, the popularity of the miasmatic theorydeclined at the beginning of the 20th century, bywhich time the microbial theory of infectious diseasewas adequately supported.

Humour = Blood

Associated = Airelement

Source = Heart

Humour = Phlegm

Associated = Water

elementSource = Pituitary

gland

Humour = Yellow bile

Associated = Fireelement

Source = Liver

Humour = Black bile

Associated = Earth

elementSource = Spleen

Excess

Excess

Sanguine

temperament

Choleric

(bilious)temperament

Melancholictemperament

Phlegmatictemperament

Excess

Excess

CHARACTERISTIC

Moisture

Hea

tC

old

Dryness

CH

AR

AC

TE

RIS

TIC

Figure 1.3 Components of humoral pathology.

7 Such mixtures were essentially witches’ concoctions, which couldbe formulated for either good or evil (Fletcher, 1896).8 These elements were originally posited by the PresocraticGreek philosopher, Empedocles, who termed them ‘roots’ andassociated them with the mythical individuals, Hera, Aidoneus(Hades), Nestis (Persephone) and Zeus (Kirk et al., 1983).

9 This explains why urban Victorians draped thick curtains intheir windows in an attempt to keep out disease. Miasmata weredocumented as being responsible for the initial development of theEuropean plague epidemics (the Black Death) throughout thesixth and seventh centuries (Maddicott, 1997) and during1347–1350 (De Smet, 1856).

1 The development of veterinary medicine4