Propensity scoring is an increasingly popular analysis method for handling confounding in observational data. This two-day workshop will provide participants with a practical exploration of propensity score methods in health research.
What will I learn?
The course will introduce the basic principles underlying propensity score methodology, demonstrate the use of the key propensity score methods (stratification, matching, covariate-adjustment and inverse probability-of-treatment weighting), and discuss advantages and disadvantages of each approach.
Participants will have the opportunity to apply the methods to a simulated realistic dataset, using code provided in R or Stata.
During the second day, participants will learn about more complex aspects of propensity score analysis, including: how to handle missing data; estimating the propensity score in large datasets via the High Dimensional Propensity Score; and how to compare more than two treatment groups.
Who is this workshop for?
This course is aimed at epidemiologists and biostatisticians using observational data to estimate treatment effects. Participants will need to be familiar with standard regression modelling (particularly multivariable logistic regression).
The course includes a practical computer session. Participants will need to bring a laptop computer, with R or Stata already installed.
How much does it cost?
Attendance is free.
However, there is a £20 cancellation fee for no-shows or cancellations within 2 days of the course, to cover administrative costs.
When and where is the workshop?
The workshop will run from 9.30-5pm Mon 17th – Tues 18th September 2018. Participants can attend one or both days of the workshop. The programme will be delivered in central London at the UCL IHI/Farr Institute London, 222 Euston Road, London NW1 2DA.
Who are the presenters?
The course is delivered jointly by Health Data Research UK (HDR-UK) and the Centre for Statistical Modelling, London School of Hygiene & Tropical Medicine.
The presenters have considerable practical experience in using propensity scores in health research, and undertake methodological development in this area. Short biographies of some of the presenters can be found below.
|Elizabeth Williamson is an Associate Professor of Statistical Methodology at the London School of Hygiene & Tropical Medicine. She has longstanding interests in propensity score methodology. Her PhD in biostatistical methodology focused on the use of propensity score methods, exploring how to obtain correct confidence intervals after applying propensity scores.|
|Ian Douglas is an Associate Professor at the London School of Hygiene & Tropical Medicine. He is interested in pharmacoepidemiology, particularly in the issue of how to use large linked electronic healthcare record databases to investigate the effects of drugs – both harmful and beneficial. His work focuses on methodologies to minimise the biases inherent in such studies.|
|Clemence Leyrat is an Assistant Professor at the London School of Hygiene & Tropical Medicine. Her PhD research focused on the use of propensity score methods in cluster randomised trials at risk of confounding bias. More recently, she studied the optimal way to implement multiple imputation for propensity score analysis when some confounders are partially observed.|
|Sanni Ali is an Assistant Professor at the London School of Hygiene & Tropical Medicine. He is primarily interested in causal inference in clinical and pharmacoepidemiology, investigating methods dealing with confounding, mediation, effect modification and missing data. He uses large linked electronic healthcare record databases to generate evidence on comparative effectiveness and safety – benefits and risks – of medications, medical devices and interventions.|
|Karla Diaz-Ordaz is an Assistant Professor at the London School of Hygiene & Tropical Medicine. She is a biostatistician primarily interested in causal inference and missing data methodology, in both clinical trials and observational studies. Currently, her research focuses on how machine learning can be used to obtain valid causal inferences, especially when using Electronic Health records.|
Helen Blake is a doctoral student at the London School of Hygiene & Tropical Medicine. Her project investigates missing data methods in propensity score analysis, with a particular focus on methods based on the missingness pattern approach and missing indicator methods. Prior to starting the PhD, she completed an MSc Medical Statistics at LSHTM.
|John Tazare is a doctoral student at the London School of Hygiene & Tropical Medicine. His project investigates the use of propensity scores as a method for confounder adjustment in the analysis of electronic health records (EHRs), particularly focusing on issues surrounding the application of the high-dimensional propensity score to UK EHRs.|