abstract
- Studies in which clinicians' therapeutic plans are determined before and after application of a new test offer an attractive alternative to randomized trials as a strategy for establishing the usefulness of diagnostic technologies. To maximize scientific credibility such studies should include a number of design features: specific clinical questions should be addressed; consecutive patients presenting with a clinical problem should be entered; the test's accuracy should be assessed; and changes in therapy should be described in sufficient detail to allow an estimate of the impact on patient outcome. Even with attention to these issues, the validity of before-after studies of therapeutic impact is limited by possible discrepancies between clinicians' stated plans and actual behavior, and by the often tenuous link between change in therapy and patient benefit. Before-after designs will be most useful in primary investigations which, if positive, should be followed by more rigorous studies to establish patient benefit.