Do we need randomised trials to evaluate diagnostic procedures?

Excerpt

The randomized controlled trial (RCT) was developed for comparing the effectiveness of two or more therapeutic modalities, and provides the only reliable means of performing such a comparison. Because one patient cannot be treated by two different methods at the same time, two patient populations are required for comparison of a new treatment and a reference method. As a result, it becomes necessary to accrue large patient numbers in an attempt to even out the variations in patient characteristics and disease severity that occur in a random fashion. The number of patients required depends on the expected magnitude of the difference between the two methods under investigation, compared with the magnitude of random population differences. …