Improving the efficiency of clinical trials with adaptive designs


Clinical trials have been very successful in evaluating the effectiveness of new treatments. However, over recent years the cost of trials has been steadily increasing, putting pressure on limited healthcare and research resources. There is therefore a great need for novel approaches to improve the efficiency of clinical trials, avoiding research waste and reducing the time taken to find effective new treatments.

Adaptive designs are an exciting innovation in clinical trials that have the potential to improve both the efficiency of trials and the outcomes of patients who volunteer for trials. In essence, adaptive designs allow information collected during the trial to be used to change (adapt) the design to optimise it. A large number of adaptations are possible, with different ones being useful in different situations.

Types of adaptive design

Multi-arm multi-stage is a type of design that allows a trial to test multiple interventions simultaneously and to use information collected to drop treatments that are found to be ineffective so that the trial can focus on the most promising ones. This is good for the patients recruited to the trial, who are more likely to get better treatment as ineffective treatments are more likely to be discarded during the trial; it is also beneficial in terms of the efficiency of the trial, as fewer patients are needed to achieve the same level of statistical power to detect whether the treatment is effective.

Adaptive designs can also be used when there is limited information available about the treatment’s likely effect on the primary outcome (the main outcome that researchers are interested in measuring during the trial). Normally this makes it difficult to choose an appropriate sample size (number of participants needed to take part in the study) and could mean that there is not enough information for researchers to draw conclusions about the effectiveness of a new treatment or mean  that more patients have participated in the trial than needed to. The sample-size reassessment design is an adaptive approach that allows the sample size to be reassessed partway through the trial as information available. This allows an increase in sample size to be implemented where justified by promising results mid-trial, or a decrease if the number recruited so far is sufficient.

There is a growing interest in identifying which patients a treatment works well for and which it does not. The hope of precision medicine is to be able to quantify these individual differences using biomarkers. Biomarkers are biological measurements that may be associated with different effectiveness of a treatment, for example cancer treatments are often targeted at certain types of tumour mutations. Adaptive designs can be used to guide allocation of the patient to the treatment that has been most successful for similar patients. This leads to better treatment for the patient and to improved chances of identifying specific patient groups that respond well to new treatments.

Want to find out more?

Adaptive designs can be very valuable for improving the power, efficiency and ethics of a trial. The Biostatistics Research Group in Newcastle has considerable expertise in developing adaptive designs for a variety of situations. Please contact James Wason, if you would like to find out more about adaptive designs. If you would like support designing an adaptive trial please contact the Research Design Service North East.

This post was written by James Wason, Professor of Biostatistics, Biostatistics Research Group, Institute of Health & Society, Newcastle University.


Leave a Reply

Please log in using one of these methods to post your comment: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s