Advanced epidemiological and statistical methods
Tutors
Prof David Gunnell and Prof Kate Tilling (course organisers), Prof Yoav Ben-Shlomo, Dr. Rachael Hughes, Prof Debbie Lawlor, Dr Richard Martin, Dr Margaret May, Dr Chris Metcalfe, Dr Tom Palmer, Prof Jonathan Sterne.
How to book on this course
The contributors are epidemiologists and medical statisticians from the University of Bristol's School of Social and Community Medicine.
Duration
Five days
Dates
25 -29 June 2012
Course fee
£1000
Course aims and objectives
- To provide a grounding in the concepts and analysis of life-course data, measurement error, clustered data, missing data and causal models. By the end of the course students should:
- Understand the concepts of additive and multiplicative interactions and use Stata to explore these concepts;
- Describe the impact of non-linearities on exposure-disease associations and conduct analyses to assess the effect of non-linearities on exposure-disease associations
- Understand the potential causes of measurement error, how to minimise and monitor these in study design, and how to incorporate information about measurement error into analyses;
- Understand how clustered data arise, the impact this may have on usual methods of analysis, and be able to carry out simple appropriate analyses using Stata;
- Understand the problems caused by missing data, and be able to carry out exploratory analyses, imputation and sensitivity analyses using Stata;
- Use a causal diagram to assess possible causal pathways, and understand models for exploring causality using observational data.
Who the course is intended for
This course is intended for researchers, applied statisticians and epidemiologists who are familiar with basic epidemiology (to at least the level covered by the "Basic Epidemiology" course), and have experience in analysing epidemiological data. Participants should have a knowledge of basic regression models and their implementation in Stata of at least the level achieved in the "Introduction to Regression Models" course.
Course outline
- Interactions: additive and multiplicative, cumulative exposures;
- Models for investigating non-linear associations
- Measurement error: description, design issues, monitoring measurement error, methods of analysis (including ICC, SIMEX, errors in binary variables and methods for dealing with inter-related errors);
- Clustered data: description, problems with simple analyses, methods of analysis including generalized estimating equations and multilevel models;
- Missing data: loss to follow-up, missing-data mechanism, simple methods of analysis, likelihood-based methods, imputation and sensitivity analyses;
- Causal models: causal diagrams, path analysis, structural equation models, propensity scores, time-dependent confounding and marginal structural models;
- An opportunity for participants to present their own planned research or research in progress.
Suggested pre-course reading
Ben-Shlomo Y, Kuh D. A life course approach to chronic disease epidemiology. Int J Epid 2002;31:285-293.
Hernan MA, Hernandez-Diaz S, Werler MM, Mitchell AA. Causal knowledge as a prerequisite for confounding evaluation. Am J Epidemiol 2002; 155:176-184.
Betty R Kirkwood and Jonathan AC Sterne. Essential Medical Statistics 2nd Edition (2003). Reprinted 2003, 2004, 2005.
Phillips A and Davey Smith G. The design of prospective epidemiological studies: more subjects or better measurements? J Clin Epi 1993;46:1203-1211.
Kenneth J Rothman and Sander Greenland. Modern Epidemiology 2nd Ed 1998.
For further information please contact short-course@bristol.ac.uk