Vist et al (2008) have attempted to answer this question by comparing people who participate in a randomised with their counterparts who did not take part in a particular trial. Whilst longitudinal follow-up of participants was possible for trial volunteers, such data was not readily available for people who did not participate.
This systematic review identified only those studies where comparative data was available, that is there was longitudinal follow-up data available on participants in the trial and also similar data on people who decided not to participate. The review identified a total of 30 new non-randomized cohort studies (45 comparisons) and includes five RCTs (yielding six comparisons; where being invited to participate or not was randomised) and 80 non-randomized cohort studies (130 comparisons), with 86,640 patients treated in RCTs and 57,205 patients treated outside RCTs.
The main analysis showed that participating in a randomised trial is neither beneficial nor harmful compared to counterparts who receive conventional care outside of a trial. This is reassuring and questions the assumption that taking part in a randomised trial is regarded as being risky. It also highlights the generalisability of the results from trials.
You can read the full review here.