Rare Disease Trials:Beyond the Randomised Controlled Trial

The development of new treatments to aid patients who suffer from rare diseases is a challenging area of medicine, particularly since the patient populations are limited. Therefore, traditional clinical trial designs and their sample size calculations often require a large proportion of the total patient population to be recruited into the clinical trial. Due to this, many novel designs of clinical trials seek to increase the benefit to the patients recruited into the trials. This is a motivation for response adaptive randomisation designs and their extension, covariate adjusted response adaptive (CARA) randomisation designs. These designs use previous patients' outcomes (and the CARA design also uses the previous patients' covariates) from within the trial to predict which treatment will be superior for future patients, and prioritise the allocation of said predicted superior treatment. In this thesis, two methods to maximise the benefit to patients are explored. The first method focuses on increasing the benefit to patients within the trial. A CARA trial design, which can be used for several different types of covariates and patient outcomes, is explored using two simulation studies; one includes a continuous covariate and outcome, the other includes two binary covariates and a survival outcome. The design is then extended to incorporate historical trial data. This extension is evaluated using two simulations studies that incorporate a continuous covariate and outcome. Different versions of both trial designs are evaluated in simulations across a wide range of scenarios. The second method is an alternative sample size calculation for a randomised controlled trial, which optimises the trial sample size such that the benefit to the whole patient population is maximised. Two different versions of the approach are investigated and compared using a continuous patient outcome trial, for a range of scenarios