Extensions to the fragility index, for non-binary data

Background

The fragility index [Walsh 2014] calculates how many events would be required to change a statistically significant trial result into a non-significant result for binary data. A smaller fragility index indicates a more “fragile” trial. The index has recently gained prominence as a practical way of demonstrating how vulnerable a trial result is. In a recent study [Ridgeon 2016], the median fragility index for multicentre randomised controlled trials in critical care medicine reporting mortality was 2, and 40% (of 56 trials) had a fragility index of less than or equal to 1. This result is startling, and shows how weak much of the evidence we rely on is, which is perhaps why reproducibility is a problem in many studies. The fragility index is currently not considered at the planning and design stage of a clinical trial. Furthermore it is currently only applicable to trials with a binary outcome looking to detect superiority. Developing the fragility index to allow its use within non-inferiority and equivalence trials as well as trials with ordinal or continuous outcomes would greatly help our understanding of the robustness of these trial results. Additionally allowing for the fragility index to be incorporated at the design stage may lead to a stronger evidence base.

Aims

This PhD will develop methodology to allow the use of the fragility index to be extended. The student will consider how the fragility index could be incorporated at the design phase of a clinical trial and allow for its use in alternative frameworks. They will propose new indices, based on the ideas underlying the fragility index, which can be applied to ordinal scales, continuous scales and survival data, including a review of the relevant literature. They will investigate which of the new indices are more easily understood by clinicians, policy makers and patients, and make recommendations on where they should be used in the future, e.g. in Data Monitoring Committee reports. They will investigate how vulnerable a selection of recently published trials are, across a range of outcome data types. The primary aim of this PhD is to create indices that can be easily understood by anyone, and they could become very widely used in the clinical trial community. It is expected that they will be used by statisticians in sample size discussions at the design stage of clinical trials as well as in interpreting final results. The results will be useful to those involved in trial design, Data Monitoring Committees, trial analysis, and disseminating results.

Training Outcomes

The student will have the opportunity to gain training at three different institutions, both at the level of a standard PhD student, and to join training undertaken within the various trials units. They will have opportunities to present work, run workshops, and gain feedback through the UK-CRC Statistics Operational Group, which meets twice a year.

This MRC programme is joint between the Universities of Edinburgh and Glasgow. You will be registered at the host institution of the primary supervisor detailed in your project selection.

All applications should be made via the University of Edinburgh, irrespective of project location:

http://www.ed.ac.uk/studying/postgraduate/degrees/index.php?r=site/view&id=919

Please note you must apply to one of the projects and you are encouraged to contact the primary supervisor prior to making your application. Additional information on the application process if available from the link above.

For more information about Precision Medicine visit:

http://www.ed.ac.uk/usher/precision-medicine

http://www.gla.ac.uk/colleges/mvls/graduateschool/precisionmedicine/