Veterinary Epidemiology

MICHAEL THRUSFIELD is Professor of Veterinary Epidemiology at the Royal (Dick) School of Veterinary Studies in the University of Edinburgh in Edinburgh, UK. ROBERT CHRISTLEY is Professor of Epidemiology and One Health at the Institute of Infection and Global Health and the Institute of Veterinary Science in the University of Liverpool in Liverpool, UK.

• 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 xxiv1 The development of veterinary medicine 1Michael ThrusfieldHistorical 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 252 The scope of epidemiology 28Michael ThrusfieldDefinition 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 413 Causality 42Michael ThrusfieldPhilosophical background 42Causal inference 43Methods of acceptance of hypotheses 44Koch’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 564 Describing disease occurrence 58Michael ThrusfieldSome 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 845 Determinants of disease 86Michael ThrusfieldClassification 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 104Stress 105Interaction 106Biological interaction 108Statistical interaction 109The cause of cancer 110Further reading 1126 The transmission and maintenance of infection 115Michael ThrusfieldHorizontal 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 1367 The ecology of disease 138Michael ThrusfieldBasic 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 1658 Patterns of disease 168Michael ThrusfieldEpidemic curves 168Kendall’s Threshold Theorem 168Basic reproductive number (R 0) 169Dissemination rate 172Common-source and propagating epidemics 172The Reed–Frost model 173Kendall’s waves 175Trends in the temporal distribution of disease 177Short-term trends 177Cyclical 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 1879 Comparative epidemiology 189Michael ThrusfieldTypes 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 19910 The nature of data 201Michael ThrusfieldClassification 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 21711 Data collection and management 219Michael ThrusfieldData collection 219Questionnaires 219Quality control of data 228Data storage 229Database models 229Non-computerized recording techniques 231Computerized recording techniques 232Veterinary recording schemes 232Scales of recording 232Veterinary information systems 234Some examples of veterinary databases and information systems 237Geographical information systems 244Further reading 24812 Presenting numerical data 251Michael Thrusfield and Robert ChristleySome 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 26913 Surveys 270Michael Thrusfield and Helen BrownSampling: 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 29414 Demonstrating association 296Michael ThrusfieldSome basic principles 296The principle of a significance test 296The null hypothesis 297Errors 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 31815 Observational studies 319Michael ThrusfieldTypes 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 33816 Design considerations for observational studies 339Robert Christley and Nigel FrenchDescriptive 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 35917 Clinical trials 361Michael ThrusfieldDefinition 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 38018 Validity in epidemiological studies 383Robert Christley and Nigel FrenchTypes 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 39619 Systematic reviews 397Annette O’Connor, Jan Sargeant and Hannah WoodEvidence 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 400Steps 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 41920 Diagnostic testing 421Michael ThrusfieldSerological 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 45621 Surveillance 457Michael ThrusfieldSome 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 481Some 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 48822 Statistical modelling 492Robert Christley and Peter J. DiggleSimple 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 51923 Mathematical modelling 520Michael ThrusfieldTypes 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 53824 Risk analysis 540Michael Thrusfield and Louise KellyDefinition of risk 540Risk analysis and the ‘precautionary principle’ 543Risk analysis in veterinary medicine 543Components 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 56325 Economics and veterinary epidemiology 565Keith Howe and Michael ThrusfieldGeneral 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 58426 Health schemes 586Michael ThrusfieldPrivate 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 60327 The control and eradication of disease 604Michael ThrusfieldDefinition 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 630General reading 633Appendices 635Appendix 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 χ 2 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, p 1 , 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 W X , 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 (log 10) of factorials of the integers 1–999 689Appendix XXI: The correlation coefficient 691Appendix XXII: The variance-ratio (F) distribution 692References 694Index 841