The Advantages of Adaptive Design
Posted on November 23, 2011
C. Hendricks Brown (bio) looks at the use of adaptive design as a response to some of the limitations imposed by traditional clinical trial design.
We've been interested in looking at different kinds of designs for evaluating programs, prevention programs as well as treatment programs. The classic model has been, as I had indicated, that people start with a model much like the FDA does a rigorous clinical trial. That doesn't work in many cases in here. It's good for the early stages of research on a prevention program or a treatment program, for example. Even a very complicated behavioral and social intervention you can still do rigorous randomized trials in this one.
But what we learn in here is that as you go into more complex research questions in here that you really need to adapt the kinds of studies. One of the real reasons is that because communities don't want to sit still and be control groups. They don't want half of their community to not get something that is really potentially very beneficial for us. So we've had this idea of trying to work very closely in partnership with communities around how do we take and design rigorous studies that meet not only their needs, but also the needs of the scientists in here in terms of evaluations?
So we talked about those in terms of general topics around adaptive designs and one way of thinking about this is the following. If you take two communities -- one at a time -- and you want to try and see whether an intervention, a complex intervention, works or not, you can select one of those communities and do an intervention and you use that group to compare with the other community that doesn't.
Well, that has a couple of problems with it. One of them is that it's only a sample of two. So even though you may have large numbers of people in those communities, you just still have a sample of two groups in here to be able to compare with. So, what we want to do is to try and find more ways of getting beyond just two communities at a time.
But the other problem is that there's obviously differences between communities and you often do research by selecting the community to get the intervention that you're more familiar with, you're more comfortable with, that they are more comfortable with you, and it's really not a fair comparison. It may be that there's measured and unmeasured characteristics of readiness in that community that are different and that's the reason why you might get a difference between the two communities.
So, what we've been looking at is ways of trying to do what we call rollout designs. So this idea of rollout designs, and we have a technical word for it, a dynamic weightless design, is a way that we've been working with communities as they want to implement a new prevention, a new program in their communities. So, as an example of that, we conducted one in Georgia -- a school district in Georgia -- with 150,000 children in it and they were interested in a suicide prevention program.
The suicide prevention program had never been evaluated before, but they were committed to putting this into all of the 32 schools that they had available in that area. So what we did was we said we're very interested in working with you and we recognize that you want it in all of the schools in here. What we'll do is work with you in terms of a design, so we randomly assigned which schools get it first.
So, it is a true randomized design and because of that we can meet the requirements of the science side of it, but it's also something that the school district was interested in it, because over time, every school gets the intervention and it wasn't adding any extra time, because the school system couldn't train everybody all at once and all we did was, we simply randomly assigned which school got it first.
So the schools that got it first were happy, because what they were willing to do is get something that potentially could benefit their children. The schools that got it later were also happy because they were willing to do that because they potentially got a little bit better intervention as we learned about what would be beneficial for how do you train people to deliver this intervention.
So, that kind of a rollout design is what we mean in one sense by adaptive designs. It's an example of that and I think that as we go to this phase of moving evidence-based programs into communities, taking into account the needs of those specific communities in here, we're going to need those kinds of designs as they get rolled out in large studies.
Excerpted from an interview with the researcher conducted at the 2011 NHSN Conference held in Miami, FL.
Please note that the feedback is viewed only by 4researchers staff and is not intended for communication with individual contributors.
Use the form below to submit feedback about this article. If you would like a response, please be sure to include your e-mail address.