Front Page
Latest News
Research
Training
Researchers
Software
External links
HOST
Imperial College
Epidemiology and Public Health
Centre for Biostatistics
SPONSORS
National Centre for Research Methods
Economic and Social Research Council
External links
HOST
SPONSORS
Roger Bannister Theatre
Faculty of Medicine
Imperial College London
St. Mary's Campus, London
9th March, 2009
The econometric literature covers many approaches to identifying the causal effect of some treatment. These are not only limited to average effects but explore a number of alternatives. Typically in public health the effect estimate of interest is the average treatment effect (ATE) and little research has gone into exploring methods that allow for alternative measures of causal effect. The aim of the seminar is to explore the potential for using econometric causal theory in the context of public health and epidemiology.
Mixing econometrics and epidemiology Presentation slides
Evaluation methods in Economics: Could they be used in epidemiology? Presentation slides
Beware of the DAG! Presentation slides
Conditional independence (CI) is a basic property of a probability distribution. It has a number of important general properties, which allow it to be manipulated in a formal manner. One popular way of doing this is to create a graphical representation: for example (but not the only one) a directed acyclic graph (DAG). There is a clear formal semantics by means of which we can interrogate a DAG to determine just what CI properties of a distribution it represents.
But there are many differences between the properties of probability distributions and those of DAGs. For example, probabilistic CI is a symmetric relationship, whereas the directionality of arrows in a DAG appears to embody a non-symmetric relationship. Although such features of the DAG contribute only in very indirectly to its probabilistic interpretation, there is an almost irresistible temptation to read more into them than the formal semantics require: in particular, to interpret the arrows as having a causal interpretation. Such "reification" of entirely incidental ingredients of the model underlies the enterprise of causal discovery. How, when, and to what extent can it be justified?
Mixing Econometrics and Epidemiology: the perfect job for Health Economics? Presentation slides
In the recent years, health economics has become an increasingly important discipline in medical research, even more so with the ongoing transition from the paradigm of evidence based medicine to that of translational research. Since the late 1970s, methods like cost-effectiveness and cost-utility analysis have been established in the health care arena, and in the last ten years health economic evaluations have built on more advanced statistical decision-theoretic foundations, effectively becoming a branch of applied statistics, increasingly often under a Bayesian statistical approach. The process of economic evaluation in healthcare calls for the integration of clinical findings and econometric models to describe the associated costs. Moreover, with the increasing importance of post-marketing research and clinical surveillance, methods for causal inference from observational studies play a fundamental role in the further development of health economics.
SMM estimation with binary outcomes Presentation slides
Mendelian randomisation and causal inference in epidemiology Presentation slides
Inferring causality from observational data is difficult as it is not always clear which of two associated variables is the cause, which the effect, or whether both are common effects of a third unobserved variable, or confounder. Instrumental variable (IV) methods are widely used in econometrics and can be used to test for or estimate causal effects when confounding is believed to be present but is not fully understood. Mendelian randomisation refers to the situation where the instrument is a genetic variant and it has received a lot of attention in the epidemiological literature recently.
Testing for the presence of a causal effect is generally straightforward but point estimates are only obtainable under additional parametric and distributional assumptions. Moreover, there are several IV estimators to choose from, all requiring different assumptions and targeting different causal effects. As is usual in causal inference, such assumptions are difficult to check and have to be supported by background knowledge. Problems arise particularly when the outcome of interest is a binary indicator of disease status, for example, although there are special cases where these can be addressed.
This talk will introduce Mendelian randomisation and will focus on the context of binary outcome, in particular with regard to the different assumptions underpinning the various IV estimators and how violations of these underlying assumptions may bias results. Questions and comments on what these different causal effects might mean in an epidemiological context will be raised for discussion.
The course will take place in the SMMS-Clinical Lecture Theatre at St Mary's hospital. Note that this is in St Mary's Hospital, across from the Faculty of Medicine, Imperial College London at St. Mary's Campus, and NOT in the main campus in South Kensington.
The complete address is
Faculty of Medicine
Imperial College London
St. Mary's Campus, Norfolk Place
W2 1PG London - UK
More information on how to arrive can be found at http://www1.imperial.ac.uk/medicine/contacts/campuses/stmarys/
REGISTER: email Sara Geneletti: s.geneletti@imperial.ac.uk