Wednesday, 27 April 2011

Adaptive..yawn..trial design

My name is Paul and I am interested in clinical trial design.
OK, it is hardly the chat-up line of the year. I also don't think I am going to win any awards for most interesting hobby; certainly not combined with my other interest in all things related to gut bacteria and gut hyperpermeability. I do, however, stand by my curious amateur/semi-professional interest in how clinical trials are designed and run and how such design can impact on the results obtained.
My interest was piqued recently by this communication from the National Institutes of Health (NIH) and their proposal to use adaptive trial design in HIV research.
There is a good description of what an adaptive trial could contain here in the curiously titled Orphanet Journal of Rare Diseases. The article by Chow and Chang is a keeper for me because they go through, in quite some detail, the various manifestations of adaptive design as well as providing a point-by-point checklist of things to watch out for when using such a trial design.
The basic premise of an adaptive design is that during a clinical trial, adjustments are made according to the data that is produced. So for example suppose you want to look at a specific intervention - dietary change for example. You have your randomly assigned groups (dietary change vs. diet as usual) and commence with the study. Suppose you set specific 'goals' for your dietary change group reflective of some change you are expecting, which at some point, you need to measure. You bring in an independent person to break the allocation codes and look at the data produced at that point. If you have reached / exceeded those goals, you then alter the trial to bring the diet as usual group onto the dietary change and continue with both groups in the experimental group. Its called a 'drop-the-loser' design and just happens to be the same design that we used on our ScanBrit trial.
This is just one example of an adaptive trial design. There are lots of other combinations. One of the main strengths of such a design is that it is flexible and responsive to what happens during your trial, particularly for very early clinical trials where a new drug, diet, instrument, etc is under preliminary evaluation.
Kudos to the NIH for bringing adaptive design centre stage.