Veterinary Epidemiologic Research

About this book

Veterinary Epidemiologic Research is a comprehensive text covering the key principles and methods used in veterinary epidemiologic research. It is written primarily for researchers and graduate students in veterinary epidemiology, but the material is equally applicable to those in related disciplines (human epidemiology, public health etc).

The first 13 chapters are devoted to issues related to the design and execution of observational studies and controlled trials.

Chapters 14 through 24 cover the statistical (multivariable) methods commonly used in the analysis of epidemiologic studies, including extensive coverage of mixed (random effects) models.

Veterinary Epidemiologic Research concludes with chapters dealing with spatial data, infectious disease epidemiology, meta-analysis, and ecologic studies. Extensive use is made of worked examples to demonstrate the principles being covered. All datasets referred to in Veterinary Epidemiologic Research are described in the book (Chapter 31) and on the book's website. Listings of program files (primarily Stata -do- files) used in all examples are provided on the book's website.

Contents

1. INTRODUCTION AND CAUSAL CONCEPTS
1.1     Introduction     2
1.2     A brief history of multiple causation concepts     2
1.3     A brief history of scientific inference     5
1.4     Key components of epidemiologic research     8
1.5     Seeking causes     9
1.6     Models of causation     10
1.7     Counterfactual concepts of causation for a single exposure     17
1.8     Experimental versus observational evidence of causation     20
1.9     Constructing a causal diagram     21
1.10     Causal criteria     23

2. SAMPLING
2.1     Introduction     34
2.2     Non-probability sampling     34
2.3     Probability sampling     37
2.4     Simple random sample     37
2.5     Systematic random sample     38
2.6     Stratified random sample     38
2.7     Cluster sampling     38
2.8     Multistage sampling     40
2.9     Targeted (risk-based) sampling     41
2.10     Analysis of survey data     42
2.11     Sample-size determination     46
2.12     Sampling to detect disease     53

3. QUESTIONNAIRE DESIGN
3.1     Introduction     58
3.2     Designing the question     60
3.3     Open question     61
3.4     Closed question     61
3.5     Wording the question     64
3.6     Structure of questionnaires     65
3.7     Pre-testing questionnaires     66
3.8     Validation     67
3.9     Response Rate     67
3.10     Data-coding and editing     68

4. MEASURES OF DISEASE FREQUENCY
4.1     Introduction     74
4.2     Count, proportion, odds and rate     74
4.3     Incidence     75
4.4     Calculating risk     76
4.5     Calculating incidence rates     77
4.6     Relationship between risk and rate     79
4.7     Prevalence     80
4.8     Mortality statistics     81
4.9     Other measures of disease frequency     81
4.10     Standard errors and confidence intervals     83
4.11     Standardisation of risks and rates     85

5. SCREENING AND DIAGNOSTIC TESTS
5.1     Introduction     92
5.2     Attributes of the test per se     92
5.3     The ability of a test to detect disease or health     100
5.4     Predictive values     103
5.5     Interpreting test results that are measured on a continuous scale     105
5.6     Using multiple tests     111
5.7     Evaluation of diagnostic tests     114
5.8     Evaluation when there is no gold standard     116
5.9     Other considerations in test evaluation     122
5.10     Sample size requirements     123
5.11     Herd-level testing     123
5.12     Use of pooled samples     127

6. MEASURES OF ASSOCIATION
6.1     Introduction     136
6.2     Measures of association     137
6.3     Measures of effect     139
6.4     Study design and measures of association     143
6.5     Hypothesis testing and confidence intervals     143
6.6     Multivariable estimation of measures of association     148

7. INTRODUCTION TO OBSERVATIONAL STUDIES
7.1     Introduction     152
7.2     A unified approach to study design     154
7.3     Descriptive studies     156
7.4     Observational studies     157
7.5     Cross-sectional studies     158
7.6     Repeated cross-sectional versus cohort studies     162

8. COHORT STUDIES
8.1     Introduction     168
8.2     Study group     169
8.3     The exposure     171
8.4     Ensuring exposed and non-exposed groups are comparable     174
8.5     Follow-up period     175
8.6     Measuring the outcome     175
8.7     Analysis     176
8.8     Reporting of cohort studies     177

9. CASE-CONTROL STUDIES
9.1     Introduction     182
9.2     The study base     182
9.3     The case series     183
9.4     Principles of control selection     184
9.5     Selecting controls in risk-based designs     185
9.6     Selecting controls in rate-based designs     187
9.7     Other sources of controls     190
9.8     The number of controls per case     193
9.9     The number of control groups     193
9.10     Exposure and covariate assessment     193
9.11     Keeping the cases and controls comparable     193
9.12     Analysis of case-control data     194
9.13     Reporting guidelines for case-control studies     195

10. HYBRID STUDY DESIGNS
10.1     Introduction     200
10.2     Case-crossover studies     200
10.3     Case-case studies     203
10.4     Case-series studies     204
10.5     Case-cohort studies     206
10.6     Case-only studies     208
10.7     Two-stage sampling designs     209

11. CONTROLLED STUDIES
11.1     Introduction     214
11.2     Stating the objectives     215
11.3     The study group     216
11.4     Allocation of study subjects     221
11.5     Specifying the intervention     225
11.6     Masking (blinding)     225
11.7     Follow-up/compliance     226
11.8     Measuring the outcome     227
11.9     Analysis     227
11.10     Clinical trial designs for prophylaxis of communicable organisms     230
11.11     Ethical considerations     233
11.12     Reporting of clinical trials     235

12. VALIDITY IN OBSERVATIONAL STUDIES
12.1     Introduction     244
11.2     Examples of selection bias     244
12.3     Examples of selection bias     249
12.4     Reducing selection bias     254
12.5     Information bias     255
12.6     Bias from misclassification     257
12.7     Validation studies to correct misclassification     263
12.8     Measurement error     264
12.9     Errors in surrogate measures of exposure     265
12.10     The impact of information bias on sample size     266

13. CONFOUNDING: DETECTION AND CONTROL
13.1     Introduction     272
13.2     Control of confounding prior to data analysis     275
13.3     Matching on confounders     276
13.4     Matching using propensity scores     281
13.5     Detection of confounding     283
13.6     Analytic Control of Confounding     288
13.7     Other approaches to control confounding and estimate causal effects     295
13.8     Multivariable modelling to control confounding     301
13.9     Instrumental variables to control confounding     302
13.10     External adjustment and sensitivity analysis for unmeasured confounders     304
13.11     Understanding causal relationships     306
13.12     Summary of effects of extraneous variables     315

14. LINEAR REGRESSION
14.1     Introduction     324
14.2     Regression analysis     324
14.3     Hypothesis testing and effect estimation     326
14.4     Nature of the X-variables     333
14.5     Detecting highly correlated (collinear) variables     338
14.6     Detecting and modelling interaction     340
14.7     Causal interpretation of a multivariable linear model     341
14.8     Evaluating the least squares model     344
14.9     Evaluating the major assumptions     349
14.10     Assessment of individual observations     356
14.11     Time-series data     360

15. MODEL-BUILDING STRATEGIES
15.1     Introduction     366
15.2     Steps in building a model     367
15.3     Building a causal model     367
15.4     Reducing the number of predictors     368
15.5     The problem of missing values     374
15.6     Effects of continuous predictors     375
15.7     Identifying interaction terms of interest     381
15.8     Building the model     383
15.9     Evaluate the reliability of the model     388
15.10     Presenting the results     390

16. LOGISTIC REGRESSION
16.1     Introduction     396
16.2     The logistic model     396
16.3     Odds and odds ratios     397
16.4     Fitting a logistic regression model     398
16.5     Assumptions in logistic regression     399
16.6     Likelihood ratio statistics     400
16.7     Wald tests     401
16.8     Interpretation of coefficients     402
16.9     Assessing interaction and confounding     405
16.10     Model-building     408
16.11     Generalised linear models     408
16.12     Evaluating logistic regression models     410
16.13     Sample size considerations     421
16.14     Exact logistic regression     421
16.15     Conditional logistic regression for matched studies     422

17. MODELLING ORDINAL AND MULTINOMIAL DATA
17.1     Introduction     428
17.2     Overview of models     429
17.3     Multinomial logistic regression     431
17.4     Modelling ordinal data     436
17.5     Proportional odds model (constrained cumulative logit model)     437
17.6     Adjacent-category model     441
17.7     Continuation-ratio model     443

18. MODELLING COUNT AND RATE DATA
18.1     Introduction     446
18.2     The Poisson distribution     447
18.3     Poisson regression model     448
18.4     Interpretation of coefficients     449
18.5     Evaluating Poisson regression models     451
18.6     Negative binomial regression     454
18.7     Problems with zero counts     461

19. MODELLING SURVIVAL DATA
19.1     Introduction     468
19.2     Non-parametric analyses     473
19.3     Actuarial life tables     473
19.4     Kaplan-Meier estimate of survivor function     475
19.5     Nelson-Aalen estimate of cumulative hazard     478
19.6     Statistical inference in non-parametric analyses     479
19.7     Survivor, failure and hazard functions     480
19.8     Semi-parametric analyses     485
19.9     Parametric models     503
19.10     Accelerated failure time models     507
19.11     Frailty models and clustering     510
19.12     Multiple outcome event data     517
19.13     Discrete-time survival analysis     518
19.14     Sample sizes for survival analyses     522

20. INTRODUCTION TO CLUSTERED DATA
20.1     Introduction     530
20.2     Clustering arising from the data structure     530
20.3     Effects of clustering     536
20.4     Simulation studies on the impact of clustering     540
20.5     Introduction to methods for dealing with clustering     542

21. MIXED MODELS FOR CONTINUOUS DATA
21.1     Introduction     554
21.2     Linear mixed model     555
21.3     Random slopes     560
21.4     Contextual effects     564
21.5     Statistical analysis of linear mixed models     565

22. MIXED MODELS FOR DISCRETE DATA
22.1     Introduction     580
22.2     Logistic regression with random effects     580
22.3     Poisson regression with random effects     584
22.4     Generalised linear mixed model     587
22.5     Statistical analysis of GLMMs     593
22.6     Summary remarks on analysis of discrete clustered data     603

23. REPEATED MEASURES DATA
23.1     Introduction to repeated measures data     608
23.2     Univariate and multivariate approaches to repeated measures data     611
23.3     Linear mixed models with correlation structure     616
23.4     Mixed models for discrete repeated measures data     624
23.5     Generalised estimating equations     627

24. INTRODUCTION TO BAYESIAN ANALYSIS
24.1     Introduction     638
24.2     Bayesian analysis     638
24.3     Markov chain Monte Carlo (MCMC) estimation     642
24.4     Statistical analysis based on MCMC estimation     647
24.5     Extensions of Bayesian and MCMC Modelling     651

25. ANALYSIS OF SPATIAL DATA: INTRODUCTION AND VISUALISATION
25.1     Introduction     664
25.2     Spatial data     664
25.3     Spatial data analysis     667
25.4     Additional topics     673

26. ANALYSIS OF SPATIAL DATA
26.1     Introduction     680
26.2     Issues specific to statistical analysis of spatial data     680
26.3     Exploratory spatial analysis     682
26.4     Global spatial clustering     690
26.5     Localised spatial cluster detection     697
26.6     Space-time association     700
26.7     Modelling     704

27. CONCEPTS OF INFECTIOUS DISEASE EPIDEMIOLOGY
27.1     Introduction     716
27.2     Infection vs disease     718
27.3     Transmission     719
27.4     Mathematical modelling of infectious disease transmission     721
27.5     Estimating R0 and other infectious disease parameters     725

28. SYSTEMATIC REVIEWS AND META-ANALYSIS
28.1     Introduction     740
28.2     Narrative reviews     740
28.3     Systematic Reviews     741
28.4     Meta-analysis – Introduction     745
28.5     Fixed- and random-effects models     746
28.6     Presentation of results     749
28.7     Heterogeneity     750
28.8     Publication bias     758
28.9     Influential studies     760
28.10     Outcome scales and data issues     760
28.11     Meta-analysis of observational studies     764
28.12     Meta-analysis of diagnostic tests     766
28.13     Use of meta-analysis     766

29. ECOLOGICAL AND GROUP-LEVEL STUDIES
29.1     Introduction     774
29.2     Rationale for group level studies     774
29.3     Types of ecologic variable     775
29.4     Issues related to modelling approaches in ecologic studies     776
29.5     Issues related to inferences     778
29.6     Sources of ecologic bias     778
29.7     Non-ecologic group-level studies     782

30. A STRUCTURED APPROACH TO DATA ANALYSIS
30.1     Introduction     790
30.2     Data-collection sheets     790
30.3     Data coding     791
30.4     Data entry     791
30.5     Keeping track of files     792
30.6     Keeping track of variables     792
30.7     Program mode versus interactive processing     793
30.8     Data-editing     794
30.9     Data verification     795
30.10     Data processing—outcome variable(s)     795
30.11     Data processing—predictor variables     796
30.12     Data processing—multilevel data     796
30.13     Unconditional associations     796
30.14     Keeping track of your analyses     797

31. DESCRIPTION OF DATASETS            799

Customer Reviews

Biography

Ian Dohoo has been teaching epidemiology and conducting epidemiologic research for more years than he is willing to admit. He has an international reputation as an excellent teacher and has been the recipient of numerous research and teaching awards, including honorary doctorates from the Swedish University of Agricultural Sciences and the University of Guelph. His particular interest is in the application of quantitative methods to animal-health research problems.

Wayne Martin is considered to be the father of modern veterinary epidemiology in Canada (where he has been working at it even longer than Ian). He is internationally recognised as one of the key contributors to the development of the discipline and was awarded the Calvin W Schwabe Award in 2006. He is a Professor Emeritus from the U. of Guelph. Wayne remains active in epidemiology and maintains his particular interest in study design principles and epidemiologic methods.

A native of Denmark, Henrik Stryhn moved to UPEI in 2001. He has established himself as one of the leading figures in veterinary biostatistics. He has an international reputation and is frequently asked to teach high level “methods” courses relevant to veterinary epidemiology. His particular research interest is in multilevel modelling, an area that is crucial in animal health studies.