ABSTRACT
In randomized clinical trials a new therapy is usually compared with the standard treatment or with placebo. Within the intention-to-treat analysis, all patients are used for the overall comparison. But independent of the overall result, the clinicians want to know whether there are subsets of patients with different outcomes. If the new therapy is superior to the standard, the question is whether there are patients who will not benefit from or may be harmed by the new treatment. The same holds if there is no advantage to the new therapy. There may well be patients who will benefit or may be harmed by the new drug. The background of these concerns is the tendency in recent years to select the appropriate treatment for a specific patient based on his or her prognostic or predicitive factors.
