Controlled assessment of diagnostic techniques: methodological problems.

Abstract

The increasing requests for diagnostic tests is one of the many causes of rising costs in health care systems. Many health care professionals think today that this increase could be reduced if diagnostic tests were accurately assessed, as are drugs. If there are similarities between a controlled treatment trial and a diagnostic method trial, there are also many differences relating to the sample size, the choice of criteria for final judgment, the study design and the clinician's behaviour when faced with the problem. These differences make it difficult to organise controlled trials of diagnostic techniques. A different approach using simulation methods and a decision tree appears more appropriate in the evaluation of diagnostic techniques.