Epidemiology course

Course duration: 8 double lessons

Anyone who plans, organises or evaluates epidemiological studies must also make use of quantitative-statistical methods. But even those who want or need to understand and evaluate first-hand studies cannot do so without statistical knowledge. Contentious issues about epidemiological studies in the public may arise from the fact that either inadequate statistical methods are used or statistical statements are misinterpreted.

The aim of this course is to impart methodological skills necessary for epidemiological studies from scratch and to explain it by means of examples.

Speakers

Prof. Dr. Wolfgang Ahrens: Head of the Department of Epidemiological Methods and Etiological Research at the Leibniz Institute for Prevention Research and Epidemiology - BIPS GmbH, Bremen

Prof. Dr. Lothar Kreienbrock: Director of the Institute of Biometry, Epidemiology and Information Processing at the University of Veterinary Medicine, Hannover, Germany

Dr. Jürgen Wellmann: Scientific Assistant at the Institute for Epidemiology and Social Medicine, University Medical Centre Münster

Topic Overview

Part I

• Measures of disease incidences (prevalence, cumulative incidence, years at risk, incidence density)
• Standardisation of mortality rates
• Fundamentals of probability and statistics
• Measures of association between diseases and risk factors (relative risk, odds ratio, attributable risks)
• Types of epidemiological studies (cohort, case-control, cross-sectional)

Course duration: 8 double lessons

Anyone who plans, organises or evaluates epidemiological studies must also make use of quantitative-statistical methods. But even those who want or need to understand and evaluate first-hand studies cannot do so without statistical knowledge. Contentious issues about epidemiological studies in the public may arise from the fact that either inadequate statistical methods are used or statistical statements are misinterpreted.

The aim of this course is to impart methodological skills necessary for epidemiological studies from scratch and to explain it by means of examples.

Speakers

PD Dr. Amely Campe: Research Associate at the Institute of Biometry, Epidemiology and Information Processing at the University of Veterinary Medicine Hannover

Prof. Dr. Lothar Kreienbrock: Director of the Institute of Biometry, Epidemiology and Information Processing at the University of Veterinary Medicine Hannover

Dr. Hermann Pohlabeln: Research Associate in the Department of Biometry and Data Management at the Leibniz Institute for Prevention Research and Epidemiology - BIPS GmbH, Bremen

Dr. Jürgen Wellmann: Scientific Assistant at the Institute of Epidemiology and Social Medicine, University Medical Centre Münster

Topic Overview

Part II

•  Simple epidemiological evaluation procedures
• Case number calculations for epidemiological studies
• Bias in epidemiological studies (selection, misclassification and confounding)
• Stratified evaluation procedures
• Risk modeling using logistic regression

Course duration: 8 double lessons

In clinical veterinary medicine, epidemiological studies, and animal disease control, statistical quality measures for diagnostic tests have long been of central importance. In the context of the COVID-19 pandemic, these have also received media attention in human medicine. Since error-free diagnostic classification is not usually achieved, test errors occur, the probability of which is estimated during test evaluation and should be taken into account when assessing test results in diagnostic practice. A prerequisite for efficient diagnostic testing is the optimization of cut-off values and test procedures, especially when testing herds and other animal groups or when using multiple tests.

The aim of this course is to provide participants with knowledge of study design, sampling, evaluation, and critical interpretation of the various types of studies used to evaluate diagnostic tests. Further learning objectives include the principles and methods of optimizing diagnostic cut-off values, ROC analysis for the global assessment of tests and for comparing two tests, the use of herd testing, and multiple testing strategies (including test evaluation without a gold standard). Finally, participants will acquire knowledge of methods for adjusting for test errors when interpreting test results. The necessary statistical and epidemiological concepts are explained using examples. COVID-19 tests are also referred to for illustrative purposes.

The course is divided into thematic units and is supported by various media. Course participants are asked to contribute their own knowledge and experience to the discussion. Practical exercises are carried out on PCs under supervision in the core topics. Participants have the option of using their own data sets for evaluation. Basic knowledge of applied statistics and good practical knowledge of Microsoft Excel are recommended. Participants receive the course slides as well as exercises and answer sheets for further consolidation of the learning content.

The target audience for this course is anyone involved in the development, optimization, evaluation, application, or interpretation of diagnostic tests as part of their work. In similar courses in the past, the diverse professional backgrounds of the participants (from developers to decision-makers, but also veterinary students and consultants in the field of applied statistics) have always had a positive effect. Since the methods presented are not specific to veterinary medicine, the course also appeals to scientists and practitioners from other disciplines.

Participants who would like to prepare for the course in terms of content are recommended to read the following literature:

Bastos, M. L., Tavaziva, G., Abidi, S. K., Campbell, J. R., Haraoui, L. P., Johnston, J. C., ... & Khan, F. A. (2020). Diagnostic accuracy of serological tests for covid-19: systematic review and meta-analysis. bmj, 370.

Greiner, M., D. Sohr, and P. Göbel. A modified ROC analysis for the selection of cut-off values and the definition of intermediate results of serodiagnostic tests. J.Immunol.Methods 185:123-132, 1995.

Greiner, M. Serodiagnostische Tests: Evaluierung und Interpretation in der Veterinärmedizin und anderen Fachgebieten (Statistik und ihre Anwendungen), Berlin; Heidelberg; New York:Springer, 2003.

Greiner, M. and I. A. Gardner. Epidemiologic issues in the validation of veterinary diagnostic tests. Prev.Vet.Med. 45:3-22, 2000.

Greiner, M. and I. A. Gardner. Application of diagnostic tests in veterinary epidemiologic studies. Prev.Vet.Med. 45:43-59, 2000.

Greiner, M., D. U. Pfeiffer, and R. D. Smith. Principles and practical application of the receiver operating characteristic analysis (ROC) for diagnostic tests. Prev.Vet.Med. 45:23-41, 2000.

Kostoulas P, S.S. Nielsen, A.J. Branscum, W.O. Johnson, N. Dendukuri, N.K. Dhand, N. Toft, I.A. Gardner. STARD-BLCM: Standards for the Reporting of Diagnostic accuracy studies that use Bayesian Latent Class Models. Prev. Vet. Med. 138: 37–47, 2017

Watson, J., Whiting, P. F., & Brush, J. E. (2020). Interpreting a covid-19 test result. Bmj, 369.

Speakers

Prof. Dr. Matthias Greiner: Head of Department at the Federal Institute for Risk Assessment (BfR), Berlin and Professor at the University of Veterinary Medicine Hannover (TiHo)

Prof. Dr. Marcus G. Doherr: Head of Institute of Veterinary Epidemiology and Biometry, Free University Berlin

Course units

    Objectives of evaluation studies
    Quality measures for diagnostic tests
    Design, parameter estimation, and bias errors in pre-stratified studies for the evaluation of diagnostic tests
    Design, parameter estimation, and bias errors in cross-sectional studies with complete and incomplete verification
    Threshold optimization in theory and practice
    ROC analysis (threshold-independent quality measures, test comparison of two quantitative diagnostic tests)
    Herd testing in theory (principle, optimization, evaluation) and practice (HerdAcc program)
    Multiple diagnostic tests (combination strategies, conditional independence)
    Test evaluation without gold standard (latent class models)
    Adjustments for test errors in clinical diagnostics, prevalence estimation, and epidemiological studies
    Application of diagnostic tests in the context of animal disease surveillance

Course duration: 8 double lessons

Disease events in individual hosts, herds, and entire host populations are now increasingly understood as changing dynamic systems and managed accordingly. The resulting abundance of decision-relevant mechanisms and processes quickly complicates the consistent interpretation of observations or the comprehensive testing of causal hypotheses. Evidence-based decisions, e.g., in the control context, appear difficult. Increasingly, dynamic epidemiological models are being developed to represent the complex interplay of mechanisms and processes involved in the spread of disease in host populations. Such models can be used to recreate possible consequences of decisions and to weigh alternatives. For example, what impact would a change in a biological assumption have on the expected success of a control measure? This immediate function of models as a deduction and decision support tool is the focus of the course.

During the seminar, background knowledge and practical experience in the development and use of models will be provided. It will be demonstrated that it can be useful to use models, regardless of the mathematical background of the person seeking an answer. In addition, it will be shown how very sophisticated models can be created through many years of experience and thanks to modern information technology.

Models are jointly developed that reproduce the transmission dynamics of diseases in the computer, and their statements are interpreted. No previous knowledge is required. Guided experimentation and freedom for own creative ideas are taken into account in the conception as well as the exchange among each other. The background knowledge to be imparted will be illustrated by examples from the 20 years of modeling activity of the lecturers in the environment of science and politics of animal welfare.

The course is aimed at non-modelers who deal with diseases in populations in the course of their work and who (might) be confronted with the statements of epidemiological models. The seminar is equally suitable for scientists and practitioners in health and animal health agencies and research institutions, as well as for students of human and veterinary medicine and natural sciences.

Speakers

Hans-Hermann Thulke & Martin Lange: Scientists at the Department of Ecological Systems Analysis of the Helmholtz Centre for Environmental Research (UFZ) in Leipzig. The Leipzig Project Group Ecological Epidemiology (EcoEpi - www.ecoepi.eu) has a proven track record in providing model-based advice to national and international decision-making bodies on various problems in animal health policy, disease management and epidemiology.

Course Units

The course is conducted over four half-day sessions of two double hours each. The parts cannot be taken individually. The course includes illustrative lectures and practical exercises. Exercises are done on provided computers or, optionally, on the student's own laptop (program installation required).

Content

•  Classical approach to disease description
• Early models of infectious diseases in host groups
• Guided NetLogo-based experimentation with different model concepts
• The step from the deterministic compartmental view to the reflection of biological diversity of host-pathogen interactions
• Guided NetLogo-based development of an individual-based infection spread model
• Guided independent model application to reproduce model-based answers to selected problems in animal disease control
• Example-oriented outlook on modern applied disease modeling to solve real-world decision problems
• Translating model-based insights into recommendations for action on an actual disease situation