A PSI Training Course on Missing Data
Presented by Mike Kenward (GlaxoSmithKline Professor of Biostatistics) and James Roger (Honorary Professor of Biostatistics) Department of Medical Statistics
London School of Hygiene and Tropical Medicine
There has been much recent activity concerning the problem of handling missing data in clinical trials. In 2010, a new set of guidelines was produced by the European regulators, and a major report was produced by the US National Research Council Panel on Handling Missing Data in Clinical Trials, at the behest of the FDA. The current course has two main threads that reflect this activity. In the first, the conceptual issues surrounding missing data in clinical trials are explored, reflecting the debate that has been taking place over the last ten years. In the second, the relevant statistical methodology is introduced and developed In particular there will be an introduction to the roles of the so-called selection and pattern-mixture frameworks, and to multiple imputation. For completeness and for purposes of comparison there will also be a brief treatment of other approaches, including ad hoc methods such as Last Observation Carried Forward (LOCF) and more principled approaches like inverse probability weighting. The two main threads will be brought together in a thorough exploration of sensitivity analyses that can be applied in this setting. Methodology will be illustrated with examples from real longitudinal clinical trials, using SAS procedures and macros. The course will consist of lectures. There will not be any computer exercises.
The following key topics will be addressed:
• The documents from US and European regulators.
• Definitions: missing value mechanisms (MCAR, MAR, MNAR), ignorability, estimands (de jure, de facto); other jargon: intention to treat, per protocol, efficacy, effectiveness.
• The distinction between missingness as a nuisance and as part of the outcome.
• Ad hoc methods: completers analyses, last observation carried forward, simple imputation, worst case analyses.
• Model based analyses under MAR for continuous and categorical data.
• Sensitivity analyses: selection and pattern-mixture models, multiple imputation, controlled imputation
Use of Computers:
There will not be any workshops where course participants do their own computing, but there will be extensive examples throughout the course, and example code will be supplied showing how to implement the preferred methods. All such code is either in the public domain or is made freely available to the participants for them to copy and use within their own organisations.
About the presenters:
Mike Kenward, GSK Professor of Biostatistics, Department of Medical Statistics, London School of Hygiene and Tropical Medicine
Mike Kenward has worked in Iceland, Finland and the UK, in both research institutes and universities. He has a broad interest in modelling in biostatistics, with particular experience in longitudinal data and cross-over trials, as well as the general problem of missing data. He has co-authored three textbooks, The Design and Analysis of Cross-Over Trials (with Byron Jones), Missing Data in Clinical Studies (with Geert Molenberghs) and Multiple Imputation and its Application (with James Carpenter). He has been a consultant, principally for the pharmaceutical industry, for over 25 years, and has given many short courses throughout the world on various areas of biostatistics, especially missing data.
James Roger, Honorary Professor of Biostatistics, Department of Medical Statistics, London School of Hygiene and Tropical Medicine
James Roger has a long career as university lecturer and statistician within with pharmaceutical industry including periods with J&J and GSK. Collaboration with Mike Kenward has spanned most of that career. James’ interest in missing data stemmed from shared research on linear mixed models and small sample approximations. Recent collaboration has centred on methods that address alternative estimands to those associated with classic MAR. Collaboration with GSK has allowed the development of an implementation of these approaches within SAS using multiple imputation.
Course runs from: 10:00 – 17:30 (registration from 9:30) on Day 1 and 09:00 – 16:00 on Day 2.
Please register online at www.psiweb.org and click on Events; payment now available online.
Registration costs (includes lunch and refreshments)
Registration before 15 April 2014
PSI Members: £495 plus vat
Non-members: £540 plus vat
Registration on or after 15 April 2014
PSI Members: £595 plus vat
Non-members: £640 plus vat
Accommodation can’t be guaranteed after the early bird deadline
PSI aims to be fully inclusive and endeavours to accommodate delegates with disabilities wherever possible. Please help us to help you by letting us know if you require additional facilities or have any special requirements. Please contact us on +44 (0)845 1800 349 or at PSI@mci-group.com for further information.
Contact: Emma Lovett, Tel: +44 (0)845 1800 349