Skip to main content

Seminar Series

The Medical Statistics and Epidemiology Seminar Series presents research in applied statistics and epidemiology. Seminars are principally for Division of Health Sciences staff and research students, however, visitors are also welcome to attend. The seminar series runs approximately three times per term with a mix of internal and external speakers. This is a fairly informal programme in which staff and research students are encouraged to present their current work and to exchange ideas.

Meetings take place in the Medical School Building on the Gibbet Hill Campus unless otherwise specified. Lunch will usually be provided after seminars delivered by external speakers. If you wish to be added to the seminar mailing list or for any other enquiries regarding the seminars please email t.hamborg (at) warwick.ac.uk

Summer term seminars 2012
Tuesday 01 May 14:00 - 15:00
(Room A-011)
Thomas Jaki (Lancaster University) Designing Multi-Arm, Multi-Stage Clinical Studies We consider the problem of multiple comparisons with control in a clinical trial with treatment selection at the interim(s). Our approach is an extension of group sequential methodology that involves joint monitoring of the correlated, sufficient test statistics. We show the heavy computation involved can be dramatically reduced when the number of stages is small and the number of treatments is large. When designing the trial, sample size can be chosen to satisfy a power requirement based on the least favourable configuration of treatment effects. At each analysis, simultaneous confidence intervals are easily found such that the overall coverage is at the required level.
Tuesday 22 May 12:30 - 13:30
(Room A-150)
Helen Parsons (WMS) Patient reported outcome measures and the Rasch model Slides:
Tuesday 29 May 12:30 - 13:30
(Room A-117)
Richard Crossman (WMS) Methods for Classifying Ordinal Data using Entropy: a Quick Comparison
Friday 15 June 12:30 - 13:30
(Room A-117)
Cornelia Kunz (WMS) Adaptive treatment selection in seamless phase II/III trials using short-term endpoints Adaptive seamless phase II/III designs with treatment selection at an interim analysis have become increasingly more attractive due to their potential to save development costs and to shorten time-to-market of a new treatment. Different methods have been proposed for selection of the treatment group that will continue along with the control group to the second stage of the trial. If the primary endpoint is observed only after long-term follow-up it may be desirable to use short-term endpoint data at the interim analysis to select a treatment. In other cases, at least some long-term endpoint data might be available at the time of the interim analysis which might be used together with the short-term endpoint data to estimate the treatment effect upon which the treatment selection can be based. While appropriate methods to combine data from different stages of the trial ensure control of the family-wise type I error rate in the strong sense, the power of the different approaches for treatment selection differs depending on several assumptions. In this talk we present the results of a formal comparison of different methods for treatment selection based on analytical results and a simulation study, together with a summary of the strengths and weaknesses of the approaches. Based on these results, we show how existing methods can be improved, increasing both the probability of selecting the most effective treatment and the power.
Tuesday 19 June 12:30 - 13:30
(Room A-011)
Louise Hartley (WMS)  Inter-continental differences in the clinical effectiveness of medical interventions: a study using panoramic meta-analysis
Spring term seminars 2012
Tuesday 10 January 12:30 - 13:30
(Room A-011)
Thomas Hamborg (WMS) Sequential and adaptive designs tutorial II
Group-sequential and combination test methods
This second talk of the series will be concerned with (adaptive) group sequential designs for clinical trials. In these designs one or more so called interim analyses are conducted during the course of a trial, for the purpose of monitoring efficacy and/or safety. Whilst interim analyses can serve a variety of purposes, a primary goal is to determine if the trial should be stopped before its planned termination time because superiority or inferiority of the intervention is clearly established or because the ultimate demonstration of a relevant treatment effect has become unlikely (futility). In this talk the statistical issues related to interim monitoring with the possibility of early stopping will be discussed and the three main statistical approaches to designing clinical trials allowing for such analyses, namely classical group sequential methods, combination test methods and conditional error functions, will be explained and compared. Emphasis is placed on the flexible alpha-spending approach and the construction of stopping boundaries when using it. Slides:
Tuesday 31 January 13:00 - 14:00
(Room A-150)
Nick Parsons (WMS) Sequential and adaptive designs tutorial III
Treatment and subgroup selection

Slides:
Wednesday 22 February 13:00 - 14:00
(Room A-030)
Peter Kimani (WMS) Sequential and adaptive designs tutorial IV

Estimation of treatment effects
This is the last of a series of four talks on adaptive and group-sequential methods for clinical trials. The first talk gave the motivation for adaptive clinical trials while the second and the third talks described how to analyse adaptive clinical trials without inflating the overall type I error rate. In the same way one needs to account for adaptation while analysing adaptive clinical trials, one needs to account for adaptation while estimating treatment effects after adaptive clinical trials. The fourth talk will give some of the methods available in the literature for point estimation after an adaptive clinical trial. Slides: