One reason I have been skeptical of the "trial effect" is that trials do not enroll a random sample of patients. Ethical research requires informed consent, and if patients who consent to trials have different characteristics than those who decline, it seems plausible that they will have different medical courses. UK researchers led by Andrew Clark recently put this thesis to the test (Eur J Heart Failure; also reported in the December issue of Nature Medicine). In their study, they asked a large sample of patients whether they were willing to enter a clinical trial. They then followed the clinical course of patients who declined, and compared them with patients who consented to participation but were never enrolled in a clinical trial. They found that patients who accepted enrollment had better clinical outcomes- even when factors like age, other sicknesses, or drug use.
The finding raises a number of interesting questions about tensions between study validity and informed consent. It does not suggest that we should relax consent standards to reduce bias- though some may be tempted to view the study in this way. It does, however, raise questions about how findings in clinical trials should be interpreted when applying them in real clinical settings. And it provides another problem for those who are attached to the position that trial participation is, in itself, therapeutic. (photo credit: funkandjazz, Skew, 2007)