Synthesizing Findings Across Multiple Randomized Trials
Posted on November 23, 2011
C. Hendricks Brown (bio) explores how to integrate findings from multiple studies.
How do we synthesize findings across multiple randomized trials? So, what we learned over a period of time is that the randomized trial, if you're doing it right, really gives you very, very strong evidence of a beneficial effect within that. It has a lot of integrity. Internal validity is what people call it to compare one intervention versus another.
But it often is not sufficient to be able to extrapolate to other communities or other settings. So other people have done similar trials in different areas, but they're all separate individual trials. So the question is how you can integrate the findings from a number of those studies together? So we're now in this process of understanding methodology for synthesizing the findings from multiple trials together.
People have done work on that before and they call it meta-analysis, but what that has is very limited questions can be answered by meta-analysis. Mostly those questions are: what's the overall effect of an intervention? But in our setting what we're looking at is how do we find variations of impact, who benefits or who might potentially be harmed from an intervention? Those are questions of moderation and as we look at the questions of moderation we're generally underpowered in each individual study to be able to analyze that data.
So we have to find ways of combining that data. So we've been integrating data sets together at the individual level, which requires us to share data, have people share data with us, and put those together and analyze that data. So that's this next phase of work that we're doing in methodology, this next step.
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.