Methodology

A priority of the Institute is the development of methods which have a direct impact on the design, conduct or analysis of our or other people’s studies. The MRC CTU at UCL is conducting the majority of the research into trials methodology, and their work is presented in three themes:

Design of trials, meta-analyses and observational studies

Multi-arm, multi-stage (MAMS) platform trials
Designing phase II (and III trials) based on an enhanced decision process at the end of phase II
Improving the design of stratified medicine trials and biomarker validation studies
Designing trials in uncommon diseases
Cluster randomised and stepped wedge trials
A flexible framework for complex time-to-event outcome trials
Planning and accounting for missing data
Improving the analysis and design of trials with longitudinal data or clusters of varying size
Designing trials with recurrent events as the primary outcome measure
Re-randomising patients into trials
Design, development and validation of prognostic models

Effective and efficient conduct of trials and meta-analyses

Trial conduct methodology:

Providing practical examples of how novel designs can be implemented
Evaluating and implementing strategies to ensure that data on randomised patients is not lost through patient withdrawal
Efficient trial monitoring
Getting trials started more quickly, and facilitating prompt reporting of outcome data

Meta-analysis conduct methodology:

Speeding up the evaluation of individual therapies in meta-analysis
Providing tools and guidance to promote greater awareness, understanding and use of IPD meta-analysis
Resolving outstanding issues in systematic review conduct

Analysis of trials, meta-analyses and observational studies

Analysing multi-arm multi-stage (MAMS) trials
Analysing time-to-event outcomes
Multivariable prognostic models and treatment-covariate interactions (including validation)
Appropriate analysis of longitudinal and clustered data
Causal models for answering questions not addressed by randomisation
Missing data and improved sensitivity analysis for missing outcome data
Design, development and validation of prognostic models