Comparison of Characteristics and Outcomes of Trial Participants and Nonparticipants: Example of Blood and Marrow Transplant Clinical Trials Network 0201 Trial

Abstract

BACKGROUND Controversy surrounds the question of whether clinical trial participants have better outcomes than comparable patients who are not treated on a trial. We explored this question using a recent large, randomized, multi-center study comparing peripheral blood (PB) with bone marrow (BM) transplantation from unrelated donors (URD), conducted by the Blood and Marrow Transplant Clinical Trials Network (BMT CTN). METHODS AND FINDINGS We compared characteristics and outcomes of study participants (n=494) and non-participants (n=1384) who appeared eligible and received similar treatment without enrolling on the BMT CTN trial at participating centers during the study time-period. Data were obtained from the Center for International Blood and Marrow Transplant Research. Outcomes were compared between the two groups using Cox proportional hazards regression models. No significant differences in age, sex and disease distribution, race/ ethnicity, HLA matching, comorbidities and interval from diagnosis to HCT were seen between the participants and non-participants. Non-participants were more likely to have lower performance status, lower-risk disease, and older donors, and to receive myeloablative conditioning and anti-thymocyte globulin. Non-participants were also more likely to receive PB grafts, the intervention tested in the trial (66% vs. 50% p<0.001). Overall survival, transplant-related mortality, and incidences of acute or chronic GVHD were comparable between the two groups though relapse was higher (HR 1.22, 95% CI 1.02–1.46, p=0.028) in non-participants. CONCLUSION Despite differences in certain baseline characteristics, survival was comparable between study participants and non-participants. The results of the BMT CTN trial appear generalizable to the population of trial-eligible patients.

Allogeneic Hematopoietic Cell Transplantation for Adult Chronic Myelomonocytic Leukemia

“…This is contrary to what has been reported in other studies. 33–36 It is also interesting to note that no deaths in the bone marrow graft group were due to graft failure. It is unclear why those patients with BM had poorer survival; however our study is limited in that only a small number of subjects received bone marrow grafts (16%).…”

Alternative consent models for comparative effectiveness studies: Views of patients from two institutions

“…Unfortunately, most such data currently come from trials testing unapproved treatments. Although few studies compare outcomes of patients enrolled in CER studies to those of patients not enrolled, findings from one highly visible CER study (SUPPORT 2010) and other retrospective analyses (Foglia et al 2015) seem to suggest that participating in a CER randomized study may be associated with better, if not equivalent, outcomes (Khera et al 2015;Lantos and Feudtner 2015).…”

Are cancer patients better off if they participate in clinical trials? A mixed methods study

“…Identification and review of research literature We identified nine eligible studies that included twelve comparisons of trial and non-trial patients: Arrieta et al [28], Bertelli et al [29], De Placido et al [30], Field et al [31], Goldman et al [32], Khera et al [33], Led Du et al [34], Svensson et al [35] and Templeton et al [36]. Figure 2 contains a PRISMA flow diagram showing the selection of studies.…”

“…Characteristics of the studies All nine studies included comparisons concerning overall survival [28][29][30][31][32][33][34][35][36], some also included progression-free survival [29,30,32,34]. Eight of the nine studies were retrospective cohort The Kaplan-Meier analyse showed a trial effect (p = 0.007), but the univariate Cox proportional hazard ratio did not (p = 0.089) studies.…”

“…Non-trial patients restricted to those meeting trial eligibility criteria In the five other studies, study design or trial phase could not be identified [28,31,32,34,36] e Three studies reported no baseline difference [28,30,34], and two studies adjusted using multivariate or multivariable analyses [31,33] f Two studies adjusted using multivariate analysis [28,32], one through restriction of cohort [34], and another did not identify any baseline difference [33]. For three studies, control of the confounder was not applicable [29,35,36] g One study explicitly adjusted for stage in the multivariate analysis [28], one study according to high vs low risk in multivariate analysis [33]. Four other studies in which the patients had metastatic disease at treatment start adjusted for, e.g.…”