Study Participants, Future Patients, and Outcomes That Matter in Cancer Clinical Trials

Abstract

Randomized clinical trials (RCTs) define the efficacy of new cancer treatments and drive changes in clinical practice. Although the role of RCTs in generating knowledge is universally accepted, the often-cited benefits of trial participation for individual patients is open for debate. The trial effect is composed of potential gains offered by the new treatment (ie, the treatment effect) and potential gains offered by virtue of participating in the trial itself (ie, the participation effect). 1 Gains due to treatment effect can only be realized by patients in an experimental group that proves to be superior to standard care, while participation effect can be experienced by participants in both groups.In this issue of JAMA, Iskander et al 2 explore whether a participation effect exists in cancer RCTs. The meta-analysis included 85 comparisons of outcomes between trial participants and non-trial participants treated with the same intervention. Across all studies, participation in a clinical trial was associated with a lower mortality risk (unadjusted hazard ratio, 0.76 [95% CI, 0.69-0.82]). However, when the authors restricted the analysis to high-quality studies that adjusted for confounders, the participation effect was lost (adjusted hazard ratio, 0.91 [95% CI, 0.80-1.05]); this survival difference also disappeared when accounting for publication bias. This finding is consistent with previous research on this topic. 1,3 A participation effect might arise if differences in quality of care related to structure (ie, treatment facilities) and process (ie, increased monitoring/follow-up, clinical expertise) inherent to an RCT translate to improved patient outcomes. Although the unadjusted results of the study by Iskander et al could be interpreted to support a participation effect, the fact that outcome differences disappear after adjusting for confounders suggests this observation is largely driven by what we already know-patients enrolled in clinical trials are different from those in routine practice. Patients in trials are generally younger, fitter, have fewer comorbidities, and come from higher socioeconomic groups 4,5 ; this enrollment bias largely explains the participation effect. The implications of this finding are important for understanding how trials are often viewed in clinical practice. The participation effect is often used to promote the view that "a clinical trial is the best treatment option," 6 but this may be a false narrative.Differences between patients in trials and those in clinical practice leads to differences in outcome between RCTs and routine care, know as the efficacy-effectiveness gap. 7 This gap can profoundly influence prognostic discussions between oncologists and patients. For example, contemporary RCTs in