Effect of Unblinding on Participants' Perceptions of Risk and Confidence in a Large Double-Blind Clinical Trial of Chemotherapy for Breast Cancer

Importance  Blinding patients to treatment regimen is an important component of high-quality randomized clinical trials, although concern exists about how receipt of a placebo will affect participants’ views, particularly among patients with cancer.

Objective  To determine whether unblinding of random assignment to placebo vs experimental agent in a large adjuvant breast cancer chemotherapy randomized clinical trial was associated with perception of greater chance of cancer recurrence and lower confidence in the decision to participate in the trial in participants who were randomized to placebo compared with those randomized to experimental therapy.

Design, Setting, and Participants  Serial telephone-based prospective survey substudy of all Eastern Cooperative Oncology Group Protocol 5103 (ECOG5103) participants enrolling between January 5 and June 8, 2010. In ECOG5103, patients were randomized to receive standard adjuvant chemotherapy for breast cancer with either placebo or bevacizumab for either approximately 6 or approximately 12 months. Treatment assignment was unblinded by 24 weeks, and then patients were surveyed.

Main Outcomes and Measures  Trial participants’ self-reported perceived risk of recurrence and confidence in study participation.

Results  Of 571 patients in this substudy who started protocol therapy, 550 were still in the study at unblinding and 492 (89%) responded to the unblinding survey. At unblinding, 336 of 477 (70%) believed that they had at most a small risk of breast cancer recurrence, and 342 of 480 (71%) reported perceiving at most a small risk of serious problem with therapy; most reported feeling very informed (421 of 483 [87%]) and having high levels of confidence in their study participation (420 of 483 [87%]). The 102 participants who learned that they had been randomized to placebo did not have greater perception of chance of recurrence (P = .48) or fear of recurrence (P = .69), feel less informed (P = .86), or have lower confidence in trial participation (P = .31) compared with the 390 participants who had been randomized to experimental therapy. Patients who learned that they had been randomized to bevacizumab perceived higher risk of having a serious problem caused by treatment (P = .01).

Conclusions and Relevance  In a placebo-controlled, double-blind randomized clinical trial of chemotherapy for breast cancer, unblinding to randomization allocation did not significantly affect most participants’ views except for chance of a serious problem from experimental treatment, regardless of receipt of placebo or experimental anticancer therapy along with standard chemotherapy.

There has been considerable controversy over the ethics of using placebos when treating patients with cancer and the clinical feasibility of placebo use in this setting.¹ Discussions often revolve around the tension between the ethics of research, which emphasize generating new knowledge, and the ethics of therapy, which emphasize caring for individuals.^2,3 In cancer care, there has been little prior research on the impact of use of placebos in the curative setting, where effective, standard treatments exist. To address this, researchers have developed an “add-on” design, according to which an experimental therapy or placebo is added to standard treatment.⁴

There is evidence that participants and clinicians have misperceptions about clinical trials, and patients’ feelings about randomization are a major predictor of refusal to join trials.^5-8 In particular, participants in a placebo-controlled trial who are unblinded to their receipt of placebo might experience postdecision regret about having joined the trial.⁹ However, there are only limited previous data in this area.^10-14

We sought to evaluate participant views and reactions to unblinding in a large randomized clinical trial (RCT) for patients with early breast cancer testing the addition of a new biologic therapy, bevacizumab, to standard chemotherapy. On the basis of promising results in patients with metastatic cancer and pilot safety data for adjuvant use,^15-18 the Eastern Cooperative Oncology Group Protocol 5103 (ECOG5103) was designed as a double-blind RCT to test the efficacy and safety of adding bevacizumab therapy.¹⁸ Patients were randomized in 1:2:2 proportions to add placebo (group A) or bevacizumab (groups B and C). The standard chemotherapy schedule varied on the basis of treating physician and patient choice, and treatment assignment was unblinded with the last placebo or bevacizumab infusion (week 18-22), at which point only group C participants continued study therapy, receiving open-label bevacizumab for 30 additional weeks.

Box Section Ref ID

Institutional review board approval for the clinical trial was received through participating sites for study participation, and written informed consent was obtained from participants prior to study enrollment. This substudy surveyed by telephone all patients enrolling in ECOG5103 between January 5 and June 8, 2010, at 5 time points including at baseline (pretreatment) and 18 to 22 weeks (shortly after unblinding).

Participants were asked their perceptions about the chance of cancer coming back in 5 years and chance of a serious problem from the study treatments (response options: almost zero; very small; small; moderate; large; very large; almost certain). The perceived risk magnitude with this 7-point response format has been shown to be a significant predictor of behavior.¹⁹

This consisted of 5 validated items measuring patients’ beliefs and anxieties concerning disease recurrence.^20-24

Two items assessed participants’ views on how informed they felt about the trial treatments and how confident they felt about joining the trial on an 11-point scale.^25,26

Testing between groups was performed by means of the Fisher exact test for categorical variables and by means of the Wilcoxon rank sum test for continuous or ordinal-level data. The primary end point was whether unblinding of random assignment led to differences between placebo (group A) and experimental agent (groups B and C) groups on survey measures. For chance of recurrence and serious problem, and the 2 decision-making questions, analyses were primarily performed preserving the original ordering of the survey scales. For ease of interpretation here, these questions were collapsed into 3 response categories and corresponding P values reported. Fear of recurrence survey data were scored accordingly.

Exploratory analyses were also conducted to explore differences between the 2 experimental treatment arms. Given the ordinal nature of both the survey questions, as well as the treatment arms, Jonckheere-Terpstra testing was used to compare across the 3 arms and Wilcoxon rank sum testing for subsequent pairwise comparisons.

Two-sided P values less than .05 were considered statistically significant. Although this substudy was mandatory for all patients enrolling in ECOG5103 between January 5 and June 8, 2010, participants were allowed to skip or refuse to answer any individual survey item and/or a survey itself; therefore, results shown reflect those who responded to each individual question.

During the substudy enrollment period, 571 participants started protocol therapy, and the 550 who had not withdrawn prior to unblinding were contacted, with 492 of 550 (89%) answering at least 1 survey question at unblinding (Figure). Groups were well balanced with respect to baseline characteristics (Table 1). There were also no differences with respect to these same characteristics between participants in this substudy and the parent trial.

At unblinding, most respondents, 336 of 477 (70%), perceived at most a small risk of recurrence, and 342 of 480 (71%) perceived at most a small risk of serious problem with therapy; most reported feeling very informed (421 of 483 [87%]) and having high levels of confidence in their study participation (420 of 483 [87%]) (Table 2). There were no differences between treatment groups in their perceptions of chance of recurrence (P = .48), fear of recurrence (P = .69), feeling informed (P = .86), or feeling confident about the decision to join the trial (P = .31). Patients who learned that they had been randomized to bevacizumab perceived a higher risk of having a serious problem caused by treatment (P = .01 when analyzing on original 7-point scale; P = .07 using 3 categories).

In an exploratory analysis conducted on the basis of concerns about longer duration of therapy in group C (bevacizumab therapy continued after unblinding, in contrast to the other groups), there were also no significant differences across the 3 groups with regard to chance of recurrence, fear of recurrence, and feeling informed or confident about joining the trial. Although there was a statistically significant difference across the 3 groups with respect to chance of serious problem (P = .04), there was no difference between group C and group B (P = .32).

There is controversy, concern, and sometimes confusion surrounding the use of placebos and blinding in clinical trials. Although often scientifically necessary to elucidate the full effects of new treatments, the use of placebos in the curative setting of cancer clinical trials is particularly controversial.^1,27 There has been little prior research assessing study participant reactions to receipt of placebo in a cancer trial and potential harms of unblinding. Thus, our findings from a large RCT are novel and reassuring. They are particularly compelling given the high-stakes trial studied, including a patient population documented to have generally high levels of anxiety and inaccurate heightened risk perceptions.^28,29 Therefore, use of placebo in this setting not only is scientifically sound and ethically justified but is acceptable to patients who participate in such studies.

Results should be interpreted in the context of potential limitations including concerns about generalizability to other study designs or populations, and nonresponse bias. Furthermore, participants’ views may change over time and participants may have greater regret when study efficacy data are revealed.³⁰ Finally, confidence in study participation may have been affected by publicity surrounding the experimental therapy in this study, bevacizumab, because US Food and Drug Administration approval of its use in patients with metastatic cancer was rescinded during the study, prior to unblinding for some, although patients continued to be accrued to this adjuvant study.³¹

These findings have important implications for future research. Given the low rates of cancer clinical trial accrual in adults in the United States, research to understand and improve the perceptions of patients and the public regarding trials is critical.¹⁴ Sharing results in general, and placebo allocation specifically, may improve communication surrounding clinical trials, augment public perception of trials, and lead to greater clinical trial accrual.^14,32-34

Accepted for Publication: February 3, 2015.

Corresponding Author: Ann H. Partridge, MD, MPH, Dana-Farber Cancer Institute, 450 Brookline Ave, Boston, MA 02215 (ahpartridge@partners.org).

Published Online: April 2, 2015. doi:10.1001/jamaoncol.2015.0246.

Author Contributions: Drs Partridge and O’Neill had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Partridge, Sepucha, Miller, Swaby, Sledge.

Acquisition, analysis, or interpretation of data: Partridge, Sepucha, O'Neill, Miller, Motley, Schneider, Dang, Northfelt, Sledge.

Drafting of the manuscript: Partridge, Sepucha, O'Neill, Motley.

Critical revision of the manuscript for important intellectual content: Partridge, Sepucha, O'Neill, Miller, Swaby, Schneider, Dang, Northfelt, Sledge.

Statistical analysis: O'Neill, Dang.

Obtained funding: Partridge, Miller, Schneider.

Administrative, technical, or material support: Partridge, Sepucha, Miller, Swaby, Schneider, Sledge.

Study supervision: Partridge, Miller.

Conflict of Interest Disclosures: Dr Schneider has acted as a consultant or in an advisory role for Genentech. Dr Dang has acted as an unpaid consultant or in an unpaid advisory role for Genentech and received research funding from Genentech. Dr Northfelt has received research funding from Genentech. No other disclosures are reported.

Funding/Support: This study was coordinated by the ECOG-ACRIN Cancer Research Group (Robert L. Comis, MD, and Mitchell D. Schnall, MD, PhD, group cochairs) and supported in part by Public Health Service grants CA23318, CA66636, CA21115, CA180820, CA180794, CA27525, CA49883, CA180795, CA77651, CA25224, CA33229, CA180816 and by grants from the National Cancer Institute, National Institutes of Health, and the Department of Health and Human Services. This work has also been supported by a Susan G Komen Promise Award (primary investigator: Schneider).

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: The content of this article is solely the responsibility of the authors and does not necessarily represent the official views of the National Cancer Institute.