Building Local Evidence The Right Design for the Right Question

Abstract

In the March issue, I emphasized that evidence is only as relevant as the context in which it is generated. Building that local evidence, however, raises a further question: how should such work be designed and judged?

During training, we learned to rank studies by their place on the evidence pyramid and to regard anything below a randomized controlled trial as somehow second-rate. Levels of evidence are a useful shorthand, but they are not a scorecard. A randomized trial is the best design for comparing two treatments. It is the wrong one for describing a rare injury, testing whether a new technique is feasible, or measuring muscle activity under different conditions. The better question to ask of any study is whether its design fits its question, given the patients, resources, and timeframe available.