The impact of sharing results of a randomized breast cancer clinical trial with study participants
Background There has been growing interest in providing clinical trial participants with study results yet only limited information exists regarding the process and impact of sharing results. We sought to evaluate patient perceptions of how results had been shared from a large randomized cooperative group trial, and the impact of learning results. Patients and methods A subset of women who participated in NCCTG 9831 (A Phase III Trial of Adjuvant Chemotherapy with or without Trastuzumab for Women with HER2-positive Breast Cancer) were mailed surveys after the preliminary study results were released to the public and mailed to participants. Results One hundred and 67 of 228 surveys sent (73%) were returned; 61% reported receiving trastuzumab on study; 4% reported recurrent disease. Ninety-five percent of participants were glad they received results; 81% were satisfied with how results were shared; 23% were more anxious after learning the results. Sixty-nine percent correctly interpreted the results. Logistic regression revealed that satisfaction with the process of receiving results was associated with satisfaction with treatment (P = 0.04), and increased anxiety was associated with dissatisfaction with treatment (0.02), incorrect interpretation of results (0.04), and not having received trastuzumab (P < 0.0001). Conclusion Sharing results directly with study participants is met with overwhelmingly favorable responses from patients, although some may not initially understand the findings. The potential for increased anxiety should be considered, and psychosocial support may be required by some. A plan to share results should be routinely and prospectively considered in the design of cancer clinical trials.
KeywordsBreast cancer Clinical trial results Cancer communication Research ethics Adjuvant trastuzumab
An estimated 1.3 million individuals are diagnosed with cancer each year in the United States alone . Thousands of patients participate in clinical trials each year in the hope of furthering medical research and obtaining state of the art medical care [2, 3]. There has been growing interest in providing clinical trial participants with study results [4, 5, 6, 7]. Sharing results has been considered to show respect and appreciation for study participants, and to engage patients and the public more fully in clinical research. Following clinical trial participation, patients are rarely informed directly of the study results unless the information would influence their future care. A national consensus conference, the Summit Series on Clinical Trials, suggested that providing the results to trial participants be considered the “ethical norm.”  Recently, several NCI-funded cooperative groups have approved the concept of offering results to study participants and have considered guidelines for sharing results [8, 9]. However, no routine mechanism is in place at present to share study results and there are only limited data available regarding how to share results or how patients react to the information [10, 11, 12, 13, 14].
In previous work, we found most patients would be interested in learning results of trials in which they have participated [13, 15]. A survey of Cancer and Leukemia Group B (CALGB) clinicians found that most oncology clinicians do not share results routinely, but would be willing to consider this approach . However, there have been concerns expressed by patients, advocates, and clinicians that include: (1) the potential negative emotional impact on patients; (2) the possibility of poor understanding of patients; and (3) the increased demand for resources [5, 16, 17].
We sought to evaluate the process of sharing results information and the impact of results information among participants in a large multicenter randomized clinical trial on satisfaction with the process of receiving results and breast cancer-related anxiety.
Setting, study subjects and study procedures
North Central Cancer Treatment Group Study 9831 was a randomized phase III trial of doxorubicin and cyclophosphamide (AC) followed by weekly paclitaxel (T) with or without trastuzumab (H) as adjuvant treatment for women with HER-2 overexpressing node-positive breast cancer. The primary endpoint was disease free survival (DFS). Patients were randomized to one of three post-operative chemotherapy regimens: Arm A, AC every 21 days for four cycles followed by T alone every week for 12 weeks as control arm (AC-T); Arm B, AC followed by T followed by H weekly for 52 weeks (AC-T-H); and Arm C, AC followed by T with H weekly for 12 weeks followed by H weekly alone for 40 weeks (AC-TH-H). The study was open to accrual from May, 2000 through April 2005, when a planned interim analysis lead to release of the results to the public . Based on the interim analysis, the National Cancer Institute issued on April 26th, 2005 a press release which announced a statistically significant improvement in disease-free survival among women who had been randomized to chemotherapy plus trastuzumab (AC-TH-H) when compared to chemotherapy alone. These findings were rapidly disseminated as the media coverage was extensive. The trial’s Data Safety Monitoring Committee (DSMC) recommended that these results be communicated in writing to all study investigators and patients, recommended that women on the control arm be offered trastuzumab if they had completed chemotherapy within 6 months. Institutions that had enrolled patients were provided by NCCTG with a patient letter to submit to institutional review boards (IRBs) and to send to study participants detailing the interim trial results and recommendations. This letter was written at a fairly high literacy level, having a Flesch-Kincaid grade level of 13.1 (using Microsoft Word).
Eight high-accruing, geographically diverse institutions in the United States collaborated to recruit all living patients that they had enrolled through their centers on the clinical trial to participate in this survey study which was independent of the NCCTG and of the N9831 study team. Upon IRB approval of this research at each of these institutions, patients who participated in N9831 who were (a) alive at the last follow-up and (b) had been sent the letter of results were sent the survey along with a letter explaining the purpose of this study. Informed consent was implied by response to survey. The investigators of this survey study did not have access to the N9831 data files, and could not verify answers from the patients to factual questions.
We developed a brief questionnaire to assess the patient’s view of the process and impact of sharing results. The survey assessed treatment received and disease status, interpretation of results, impact of the results on anxiety and satisfaction, views of their medical care in light of the trial results, and reaction to the manner in which results were communicated, and a series of questions previously used to evaluate participant views of the quality of care on a clinical trial .
Because this survey was designed for a specific patient population in a specific circumstance, we were unable to pilot it in its entirety. However, most of the items had been used in other study populations and refined through focus groups, and by preliminary review of the Cancer and Leukemia Group B (CALGB) Quality of Life and Patient Advocacy Committees.
The endpoints of this study were to evaluate patient satisfaction with how results were shared and the proportion of participants who had increased disease-related anxiety after receiving the study results. Satisfaction with how study results were provided was categorized into five categories, and dichotomized for analysis as satisfied (“very satisfied”, “satisfied”) and not satisfied (“neither satisfied nor dissatisfied”, “dissatisfied”, and “very dissatisfied”). Anxiety (as measured by the question “Has learning the study results had an influence on your anxiety about your breast cancer?”) was dichotomized as more anxious (“yes I am a lot more anxious”, “yes I am a little more anxious”) and not more anxious (“no I am neither more nor less anxious”, “yes I am a little less anxious”, “yes I am a lot less anxious”, “not sure”). These ordered variables were dichotomized for three reasons: first, because the main interest was in satisfaction or not and anxiety or not; second, because there were too few patients who were “very dissatisfied” or “a lot more anxious” to model these results; and third, because although the five categories were ordered they could not be assigned numerical values (as there is no way to tell, for example, if the distance between “dissatisfied” and “very dissatisfied” is the same as the distance between “neither satisfied nor dissatisfied” and “dissatisfied”). Step-up logistic regression modeling was used to evaluate factors associated with satisfaction with the process and anxiety . All covariates were grouped as binary variables. The significance level used for stepping up came from the likelihood ratio test, and a significance level <0.05 was used as the cutoff point. The significance levels in tables and text were the levels from when a covariate was added. A Fisher’s exact test was applied to assess the association between treatment satisfaction and treatment received. Sample size was based on the number of patients identified at participating institutions.
Characteristics of survey respondents (N = 167)
51 (range 26–76)
Highest level of education
Less than high school graduate
High school/vocational school graduate
Native American or Alaskan native
Treatment on protocol, self-reported
Adriamycin, cyclophosphamide, then paclitaxel (AC-T)
Adriamycin, cyclophosphamide, then paclitaxel, then trastuzumab alone (AC-T-H)
Adriamycin, cyclophosphamide, then paclitaxel and trastuzumab, then trastuzumab alone (AC-TH-H)
Recurrent disease (excluding new cancer in opposite breast)
Dana-Farber/Partners Cancer Care
Memorial Sloan Kettering Cancer Center
Dartmouth Hitchcock Medical Center
Johns Hopkins University
University of California at San Francisco
University of Chicago
Southeastern Cancer Community Oncology Practice
How women heard results
Understanding and interpretation of results
Fifty-six percent of women reported that they found the information in the results letter easy to understand, with 63% indicating that there was “just the right amount” of information. Two percent of women felt there was too much information, and 17% believed there was not enough (14% were not sure, and 4% did not respond to this survey item). The women were asked to respond to the following question to determine their understanding of the overall results:
A similar number of women had their breast cancer return following chemotherapy alone compared to chemotherapy plus herceptin
Fewer women had their breast cancer return following chemotherapy alone compared to chemotherapy plus herceptin
More women had their breast cancer return following chemotherapy alone compared to chemotherapy plus herceptin
Sixty-eight percent of respondents chose answer C, the correct answer. Incorrect answers were selected by 14% of respondents (answers A and B), and 16% of women indicated that they were not sure of the answer.
Perceptions of care on trial after receiving results
In an effort to evaluate whether learning results had a detrimental effect on participant views of clinical trials, women were asked about their perceptions of the care they received on trial, in light of the results. Six percent of women indicated that prior to receiving results they regretted participating in the clinical trial, and 4% of women regretted participation after receiving results. Eighty-seven percent of women were satisfied with treatment. Overall, women rated the quality of their care highly with only 4% of women rating their care as fair or poor. Ninety-three percent of women felt they had been treated with dignity and respect during the trial, and just 5% of women felt they had been subjected to more tests and procedures than they thought necessary. Two percent felt they had been “treated like a guinea pig” in the study. Ninety-one percent of women described their overall experience with the clinical trial up until the time of the survey as positive, and 89% of women indicated that they would recommend participating in a clinical trial to someone else who had been diagnosed with cancer.
Reactions to the manner in which results were communicated
Study participants’ perceptions of the process of learning results
% Study participants
Comfortable receiving results by mail
Blank/did not receive results by mail
All right to have received results by mail only, with the option of calling your doctor or a member of the research team with any questions
Would have preferred to have been offered results first, in a two-step process
Would have declined the offer to receive results
Glad to have received results
Yes, quite a bit
Yes, a little
No, not at all
Would want to be informed of study results again if participated in other clinical studies
Satisfied with the process of receiving results
Neither satisfied nor dissatisfied
Effect of learning results on anxiety
Emotional impact of learning results
Logistic regression model of increased anxiety
Increased anxiety (%)
Odds ratio (95% CI)
P-value when added
0.09 (0.04, 0.21)
Satisfied with treatment
0.18 (0.05, 0.69)
Interpretation of study results
0.29 (0.09, 0.95)
Individuals with cancer who are considering treatment on a clinical trial are provided with detailed information about both the standard treatment approach and the experimental therapy. Following trial participation, however, study participants are not routinely offered trial results. Some clinicians may provide updates to their own patients who participated, though this is done infrequently . The decision to provide results to an individual patient is likely dependent on a range of patient, physician, and trial variables. There are very limited empiric data to guide investigators and physicians who chose to share study results.
When the preliminary results of N9831 were disclosed, and the benefit of trastuzumab became clear, the DSMC and investigators felt obliged to share results with participants, particularly those for whom treatment might change based on the findings. The present study evaluates how a subset of those patients reacted to the receipt of results from the trial. The observation that most patients were comfortable receiving results is consistent with limited previous research in this area among women with breast cancer [13, 15, 20].
For the clinical trial considered in the present study (N9831), the NCCTG DSMC did not give women the option of declining results, but some reported that they would have preferred to have been offered this opportunity. It is recognized that not all patients want trial results [21, 22]. Ten percent of participants in our previous study declined to receive results and were pleased to have this option . Contacting study participants with results of their trials may cause significant anxiety or an increase in fear of recurrence among some participants. Reactions to results may be heavily influenced, not only by the nature of the results, but by an individual’s coping style, how well (s)he dealt with the illness, the extent to which (s)he was involved in treatment decisions, time since therapy, and a range of factors related to the patient’s current mental and physical health, many of which were not evaluated in this study [23, 24, 25, 26, 27]. One unanticipated finding was the association of satisfaction with the process of learning results and satisfaction with treatment. Because satisfaction with treatment was highly correlated with treatment received, this likely reflects greater dissatisfaction and increased anxiety observed among the women who did not receive trastuzumab. It would seem prudent to consider these findings when providing results to study participants in future studies and to anticipate the need for additional support, particularly for patients who received the less effective treatment and/or experienced a recurrence.
We specifically chose N9831 for this survey study as it was a large multicenter randomized clinical trial for a common disease, and many participants were still available for follow-up. However, the findings may not be generalizable to other patient populations, other types of clinical trials, and other study outcomes. The generalizability may be limited by the lack of a control group in our study (i.e., everyone surveyed was sent the results), and by non-responder bias. Women who did not respond to our survey (27%) may have had different reactions than those who answered our survey: more distressed individuals may have been less likely to respond to a survey requesting their input. Furthermore, the small number of participants including few dissatisfied participants decreased the ability to assess predictors of low satisfaction. In addition, use of non-validated measures for breast cancer-related anxiety as well as satisfaction with care are potential limitations, and some patients may not have understood the questions.
We believe a plan to share results should be considered in the design of phase III clinical trials, and possibly some phase II trials. In general, this plan should involve a two-step process: participants should be offered results first, with the option of declining. Some have suggested that participants should be offered the possibility of receiving results during the informed consent process at trial initiation . In light of our findings and reports from other investigators, the potential for both confusion and disappointment should be considered . It is also important to consider that levels of anxiety may have been greater (or less) had there been fewer patients who mis-understood the results information. Educational and psychosocial support may be needed for some. Efforts to optimize the clarity and readability of results information for study participants would likely improve understanding of study findings.
This study fills a significant gap in the available literature regarding communication surrounding clinical trials. Dissemination of trial results clearly transcends oncology and impacts the entire medical research enterprise. Sharing trial results may facilitate communication between clinicians and patients, improve the quality of care delivered to patients, increase patient satisfaction with study participation, and ultimately, lead to greater clinical trial participation.
- 3.Comis RL, Aldige CR, Stovall EL et al (2000) A quantitative survey of public attitudes towards cancer clinical trials—a Harris Interactive poll. pp 1–10. www.cancersummit.org
- 4.Cancer Leadership Counsel Cancer Research Foundation of America, Coalition of National Cancer Cooperative Groups et al. (2000) The summit series on clinical trials website—recommendations and action steps. www.cancersummit.org
- 8.Shurin S (2001) Bioethics Committee: report and invitation. Child Oncol News 2:14Google Scholar
- 9.Coalition of National Cancer Cooperative Groups Inc (2002) Guidelines for notifying patients about early closure of cancer clinical trials. www.cancertrialshelp.org [Last accessed on 12/1/02]
- 15.Partridge AH, Burstein HJ, Bluman LG et al (2002) Should patients with cancer be offered aggregate results of clinical trials in which they have participated? Proc Am Soc Clin Oncol 21:259a. Abstract 1034Google Scholar
- 19.Agresti A (2002) Chapter 6, categorical data analysis, 2nd edn. New Jersey, Wiley, pp 211–267Google Scholar