Clinical trials give the best evidence about the benefits and harms of drugs and other health interventions. This research theme aims to improve how clinical trials are run and then, once the study has completed, how the data collected are analysed and interpreted.
Traditionally, many clinical trials specify a rigid protocol with preset numbers of patients on each treatment or dose, even though there is potentially limited information about the treatment or intervention in question. The techniques developed at the BSU will allow clinical trialists to review data as it is collected and improve the design of the study as it progresses – for example, stopping the use of one treatment or increasing the number of patients assigned to another treatment – whilst maintaining the integrity of the study. These methods can also use biomarker measurements on individual trial participants to inform trial adaptations, maximise patient benefit and increase the number of successful trials.
Despite careful design and conduct, many clinical trials suffer complications which make the analysis difficult. When individuals allocated to a placebo start taking the active treatment, as happens in late-stage cancer trials, the benefit of treatment is underestimated, and we are developing statistical methods to compare treatment with no treatment. When some trial outcomes are incomplete, results may be biased, and we are developing statistical methods which can estimate the benefit of treatment while acknowledging the uncertainty about what these missing data might really be. Other issues we are working on include how to define outcome measures in Systemic Lupus Erythematosus and in phase II cancer trials.
Other Research Themes:
- ESH: Evidence Synthesis to inform Health
- SGX: Statistical Genomics
- COLD: Methods for the Analysis of Complex Observational and Longitudinal Data