J Clin Oncol 25: by American Society of Clinical Oncology INTRODUCTION

Transcription

1 VOLUME 25 NUMBER 32 NOVEMBER JOURNAL OF CLINICAL ONCOLOGY R E V I E W A R T I C L E Compliance With Patient-Reported Outcomes in Multicenter Clinical Trials: Methodologic and Practical Approaches Stephanie R. Land, Marcie W. Ritter, Joseph P. Costantino, Thomas B. Julian, Walter M. Cronin, Sarah R. Haile, Norman Wolmark, and Patricia A. Ganz From the National Surgical Adjuvant Breast and Bowel Project Operations and Biostatistical Centers; Department of Biostatistics, Graduate School of Public Health, University of Pittsburgh; Allegheny General Hospital; Pittsburgh, PA; and University of California Los Angeles, Schools of Public Health and Medicine, Jonsson Comprehensive Cancer Center, Los Angeles, CA. Submitted April 13, 2007; accepted June 21, Supported by Public Health Service Grants No. U10-CA-37377, U10-CA , U10-CA-12027, and U10-CA from the National Cancer Institute, National Institutes of Health, Department of Health and Human Services. Authors disclosures of potential conflicts of interest and author contributions are found at the end of this article. Address reprint requests to Stephanie R. Land, PhD, National Surgical Adjuvant Breast and Bowel Project Biostatistical Center, 201 N Craig St, Ste 350, Pittsburgh, PA 15213; land@ pitt.edu by American Society of Clinical Oncology X/07/ /$20.00 DOI: /JCO A B S T R A C T Purpose This report describes interventions undertaken by the National Surgical Adjuvant Breast and Bowel Project (NSABP) to improve compliance with patient-reported outcome (PRO) assessments in the setting of multicenter cancer clinical trials. We describe the effectiveness of several interventions and of observational factors. Methods PRO submission rates were analyzed for the following three NSABP protocols: the Study of Raloxifene and Tamoxifen (STAR), B-32, and B-35. Institutions participating in protocol B-35 were randomly assigned to receive automated reminders of upcoming assessments or not. Compliance was analyzed with a logistic repeated measures mixed modeling. Results Compliance was high in the three protocols, with rates greater than 80% for nearly all time points. Institutions were a significant source of variability (P.01). The largest institutions had the highest compliance in STAR (odds ratio [OR] 0.68 for 50 participants enrolled and OR 0.82 for 50 to 99 participants enrolled v larger institutions; P.001). Midsized institutions had highest compliance in B-32 (OR 4.63 for 31 to 50 patients enrolled and OR 3.12 for 50 patients enrolled v small institutions; P.007). Compliance increased with participant age in STAR (OR 0.57, 0.89, and 1.01 for ages 50, 50 to 60, and 60 to 70 years, respectively, v 70 years; P.001). Race was significant in B-32 (OR 2.63 for white v nonwhite; P.001) and in STAR (OR 1.41 for white v nonwhite; P.001). Treatment group was significant in B-32 (OR 0.74; P.006). The B-35 prospective reminder did not improve compliance significantly (P.30), but in B-32, delinquency sanctions were significant (OR 1.56; P.007). Conclusion Compliance in NSABP PRO studies is higher now than a decade ago. Results for compliance initiatives were mixed. Age and race are important factors, but institutional variation remains significant and largely unexplained. J Clin Oncol 25: by American Society of Clinical Oncology INTRODUCTION The National Surgical Adjuvant Breast and Bowel Project (NSABP) is a multicenter cancer clinical trials cooperative group funded by the National Cancer Institute. It comprises a network of several hundred medical facilities in the United States, Canada, and Puerto Rico. Studies are designed primarily for the prevention and treatment of early-stage breast and colorectal cancer. The first NSABP trial was opened in 1958; the first trial including a patient-reported outcome (PRO) was opened in By the end of the 1990s, six protocols included PROs. During that time, the NSABP experienced the difficulties with questionnaire form compliance that were also reported by other groups. 1-7 The timely completion and submission of questionnaires is a critical factor in the success of PRO (eg, quality of life [QOL]) studies in clinical trials. Frequently, QOL data cannot be reliably obtained retrospectively, 2 and this same issue also applies to many other types of PRO data. Furthermore, a substantial rate of missing data threatens the credibility of the results because the available data might not be representative of the experience of all the patients. 8 A wide literature exists for addressing such informative missingness by statistical means However, these 5113

2 Land et al methods generally require untestable assumptions about the missing data mechanism. The only completely satisfactory solution is to ensure timely form completion during a study. 2,8 There are many possible factors related to questionnaire compliance. Participant factors might include demographics, such as age, 2,13-16 education, 2,5,14,17 and physical functioning, 2,14-16,18 disease stage and treatment, 5,13-15,19,20 and insurance coverage. Institutional factors 2,7,18,20-23 include the size of the institution, the structure of the institution (eg, centralized study coordination v coordination at many satellite clinics), and the attitude, experience, and time commitment of the research staff. Study design, including the number and timing of the assessments and the design of the questionnaire, may also impact compliance. Finally, efforts by the headquarters of the cooperative group to improve compliance may have an impact. In 2000, the NSABP Biostatistical Center began a series of initiatives (some modeled on work performed in other cooperative groups) to improve form compliance. This report describes the initiatives undertaken to improve compliance, the evaluation of those initiatives to date, and the methodology for the prospective randomized evaluation of the initiatives. Our purpose is to summarize what has been learned and accomplished in the context of prevention and adjuvant breast cancer trials and to describe the ongoing scientific study of form compliance in the NSABP PRO program. METHODS Setting NSABP clinical trials are run centrally from the Operations Center and Biostatistical Center in Pittsburgh, PA. A typical study is led by two medical officers (protocol officer and chair), a lead statistician, a Behavioral and Health Outcomes (BAHO) protocol officer, a BAHO statistician, and a pathologist. Several hundred institutions (networks of health care providers) enroll participants and administer the studies. Clinical data forms and questionnaires are (historically) sent by mail or facsimile from institutions to the Biostatistical Center, where the data are stored and analyzed. NSABP PRO studies have primarily been nested within randomized phase III clinical trials of treatments or preventive agents. This report will describe PRO form compliance in three phase III NSABP protocols (Table 1). Participants in these trials provided consent forms, and the studies themselves were approved by institutional review boards. The Study of Raloxifene and Tamoxifen (STAR) compared raloxifene versus tamoxifen for the prevention of breast cancer in high-risk postmenopausal women and included two PROs, a symptom checklist (SCL) and QOL. 24,25 The present analysis will consider only the SCL. Protocol B-32 was designed to compare the following two methods of lymph node resection for clinically node-negative breast cancer patients: sentinel node resection followed by conventional axillary dissection (SNAD) versus sentinel node resection alone (SN). B-32 is still in active follow-up and has not yet been reported. Protocol B-35 is an ongoing study to compare anastrozole versus tamoxifen in the treatment of ductal carcinoma in situ. B-32 and B-35 include QOL substudies. Whenever possible, the PRO assessment schedule is linked to the treatment and clinical follow-up schedule so that questionnaires may be completed by the participant during an office visit. When that is not possible, institutions are encouraged to collect the questionnaires by mail or to administer the questionnaire over the phone. Compliance Interventions The NSABP has made operational changes to facilitate correct institutional implementation of BAHO studies and timely form submission, including the creation of a new position entitled BAHO Compliance Officer, held by one of the authors (M.W.R.). Broadly, the initiatives are designed for use by the institutions, by our internal staff, by the participant, or by the medical officers of the protocol (Table 2). Communication with the institution begins with the protocol documents. BAHO studies are now routinely integrated into the protocol and all other documents pertaining to the study. One example is the guidelines for nurses and clinical research associates (CRAs) regarding the administration of studies. For each protocol and PRO, the guidelines include detailed answers to frequently asked questions (eg, appropriate methods for administering questionnaires). Access to protocols, BAHO guidelines, forms, and other materials appearing via the NSABP members-only Internet site was restructured and simplified. This integration makes the information more accessible to clinical staff at institutions and communicates that PRO outcomes are given the same priority as other outcomes. In recent years, clinical centers have been reimbursed for participation in PRO substudies, either with credit via the National Cancer Institute Community Clinical Oncology Program (CCOP) or by direct monetary reimbursement from industrial sponsors. For some protocols, the reimbursement has included a credit for entering a participant and additional credits for each submitted questionnaire. Dialogue with institutions has also been an area of effort. The importance of the PRO research effort has been highlighted in numerous presentations at meetings for NSABP member investigators and nurses/cras and in periodic newsletters sent to institutions. A one-time telephone survey of institutions was conducted with the following aims: to remind institutions that the NSABP is committed to collecting accurate and timely PROs, to educate institutions on the proper use of PRO forms, and to identify both barriers to compliance and solutions. Expectation and receipt of forms is now communicated in several ways. In all NSABP trials, confirmation faxes are sent to institutions on randomization to verify the participant s treatment assignment. These faxes have been augmented with the anticipated PRO assessment schedule. The list of assessment dates is printed on two pages; one is for the participant s chart, and a second is given to the participant. Monthly status reports for some protocols have been modified to indicate, for each scheduled PRO assessment (from baseline forward) and every participant, whether the corresponding form has been received by the Biostatistical Center. In this way, the institutions were reminded of forms that were missed in the past, some of which may have been completed but never submitted. For several protocols, the NSABP developed a prospective monthly reminder listing the next required assessment date for participants who are scheduled to complete PRO forms in the upcoming 3 months. In current protocols, the same information is available in real-time via the Internet. The NSABP has developed an institutional performance rating system that includes compliance with PROs. The evaluation serves as a tool with which institutions can monitor their performance. In addition, the rating system allows the NSABP to assist institutions with an unfavorable rating and gain knowledge from institutions with an excellent rating. Such a performance report was developed for protocol B-32 in June 2002 to after the trial was underway. In addition, beginning in October 2003, the Quality Assurance (QA) Committee of the NSABP began reviewing the institutional performance reports and assigning sanctions for poor performance. These ranged from a request that the institution submit a corrective action plan to loss of privileges to accrue to the protocol. For internal use at the Biostatistical Center, software has been developed for routine compliance monitoring. As problems become apparent at particular institutions, the BAHO Compliance Officer provides targeted interventions. Missing Data forms were developed for most PROs, including B-32 and B-35 QOL assessments. The QOL Missing Data (QMD) form is to be completed by a nurse/cra as a substitute for a missed QOL questionnaire. The nurse/cra indicates the major reasons that the BAHO assessment was not performed. The receipt of the QMD form provides closure with respect to compliance follow-up. In addition, the form indicates whether the missing questionnaire should be considered missing at random (when, for example, staff oversight is the cause of the missing questionnaire) or informative (eg, 5114 JOURNAL OF CLINICAL ONCOLOGY

3 Compliance With Patient-Reported Outcomes Table 1. NSABP PRO Studies Included in Compliance Analysis Protocol STAR B-32 B-35 Treatment modality Daily pill Surgery Daily pill Blinding Double-blind Not blinded Double-blind Trial enrollment, No. of participants 19,747 5,611 3,104 PRO substudies SCL QOL QOL Substudy enrollment dates 7/1/ /4/2004 5/1/2001-2/27/2004 2/5/ /28/2004 Substudy enrollment, No. of 19, ,275 participants Assessment schedule Baseline; 6, 12,... 72months Baseline; 1 week post-op, Baseline; 6, 12,... 72months 2-3 weeks post-op; 6,12,...36months No. of institutions Institution size, No. of participants Minimum th percentile Median th percentile Maximum Participant age, years 49 No. 1, % No. 9, % No. 6, % No. 1, % Participant race White No. 18, ,118 % Nonwhite No. 1, % Abbreviations: NSABP, National Surgical Adjuvant Breast and Bowel Project; PRO, patient-reported outcome; STAR, Study of Raloxifene and Tamoxifen; SCL, symptom checklist; QOL, quality of life; post-op, postoperative. Cancer Trials Support Unit is considered one institution. PRO study participants. participant refusal or illness). Institutions are not reimbursed for the submission of a QMD. In recent studies, the NSABP has improved participant communication with the use of trifold study brochures designed in a collaboration of the NSABP Patient Advocate and NSABP BAHO Committees. These are designed to encourage recruitment and compliance. The Biostatistical Center also engaged the medical officers by circulating monthly reports of PRO compliance by institution, which were integrated into reports of other study variables such as adverse events. Statistical Methods Form compliance was analyzed with logistic mixed-effects modeling, in which the outcome variable was a binary indicator of the receipt of an expected form for a given institution, participant, and assessment time point. A form was considered expected if the participant had survived past the scheduled time point and had not withdrawn consent for the clinical protocol. In B-32 and B-35, participants were not expected to continue PROs after a recurrence or second primary cancer. The model included participant factors (age, race, and treatment group), institutional factors, calendar time from the protocol initiation, and the assessment time point. Several institutional size factors were considered, including total number of participants enrolled onto parent protocol, number of participants in PRO substudy, proportion PRO of total enrollment (PRO enrollment divided by total parent protocol enrollment), and number of satellites. The type of institution (CCOP, Cancer Trials Support Unit, or regular member) was also considered. Factors were also included to account for the presence of compliance-enhancement initiatives, discussed earlier. The institutions were included as random effects. The primary analysis considered receipt of a questionnaire at any time; in secondary analyses, submission was required to be timely (within 3 weeks of expected date). Results are presented in terms of odds ratios (OR). Computation was performed using SAS versions 8.2 and 9.1 (SAS Institute, Cary, NC). For protocol B-35, institutions were randomly assigned to receive or not receive the prospective monthly reminder of upcoming assessments. The unit of randomization is the institution to avoid contamination between participants being managed by the same clinical staff. The random assignment was stratified by the institution s compliance record in protocols B-30 and B

4 Land et al Table 2. Compliance Enhancement Initiatives Initiative (composite compliance 88% v 88%). New institutions were randomly assigned without stratification. Using simulation, the power for the comparison of intervention and control groups in B-35 was estimated to be 80% for an anticipated improvement in mean compliance from 80% to 85%. In evaluations of compliance initiatives, the power depends on the following three sample sizes: the number of forms required per participant, the number of institutions, and the total number of participants. Statistical power also depends on the distribution of the institution sizes, the variation of compliance across institutions, the expected effect size, and the variation across assessment time points. For the present sample size calculation, the sample sizes were based on including all B-35 QOL substudy participants. Other parameters were estimated from past NSABP trials. RESULTS Recommendation Initiatives directed at the institution Integration of PRO studies into protocol documents, including nursing guidelines; improved access to PRO materials and information via Internet Reimbursement to institutions for PRO participation Targeted communications with institutions, either individually or through mass communication such as newsletters (contacts via postal mail, facsimile, , and telephone) Presentations at national NSABP group meetings Telephone survey to assess barriers and identify solutions List of anticipated PRO assessments given to institution? upon random assignment Monthly status reports indicating receipt, expectancy,? or delinquency for all PRO assessments Prospective reminder of upcoming assessments Institutional performance rating? Quality Assurance Committee sanctions Initiatives directed at Biostatistical Center Biostatistical Center internal monthly reports of compliance by institution and time point QMD forms Initiatives directed at the participant List of anticipated PRO assessments given to? participant at random assignment Participant brochure? Initiatives directed at medical officers Monthly report circulated to protocol chairs and? officers presenting PRO compliance by institution Abbreviations: PRO, patient-reported outcome; NSABP, National Surgical Adjuvant Breast and Bowel Project; QOL, quality of life; QMD, QOL missing data. recommended;? value has not been determined. The compliance in all three protocols was high. STAR SCL form compliance was 99% at baseline, with submission rates ranging from 83% to 95% for the other time points. In B-32, compliance ranged from 99% at baseline to 81% at the 3-year time point. Compliance in B-35 was 100% at baseline (by definition of substudy participation). Currently, compliance with later time points ranges from 94% to 72% for the most recent time point, but B-35 is an ongoing study, and submission rates will likely increase over time. The institution was a significant (P.01) source of variability in compliance in all trials examined. In STAR, large institutions had the highest compliance (OR 0.68 for institutions with fewer than 50 participants enrolled and OR 0.82 for institutions with 50 to 99 participants enrolled v larger institutions; P.001). In B-32, compliance was greatest for midsized institutions (OR 4.63 for institutions with 31 to 50 participants enrolled and OR 3.12 for institutions with 51 or more participants enrolled v small institutions; P.007), but this finding did not occur in B-35. The other institution-level variables we considered (number of participants in PRO substudy, proportion of enrolled participants who were also in PRO substudy, number of satellite institutions, and type of institution [regular, CCOP, or Cancer Trials Support Unit]) did not impact compliance. Compliance decreased significantly (P.001) by assessment time point in all protocols, with ORs of approximately 0.8 for each successive time point in both B-32 and B-35. In STAR, the effect of time point was of negligible magnitude (OR 0.996). Compliance was further diminished over calendar time in all three studies (P.001), with varying effect sizes (for each successive year after the opening of the protocol, OR 0.72 for B-32, OR 0.51 for B-35, and OR 0.86 for STAR). Participant age did not predict compliance in B-32 and B-35. In STAR, however, compliance increased with age (OR 0.57, 0.89, and 1.01 for ages 50, 50 to 60, and 60 to 70, respectively, compared with participants aged 70 years; P.001). Race was significant in B-32 (OR 2.63 for white v nonwhite; P.001) and in STAR (OR 1.41 for white v nonwhite; P.001). Treatment group was significant in B-32 (OR 0.74 for SNAD v SN; P.006). Treatment group was also statistically significant in STAR (P.019), but the effect was negligible (OR 0.95 for tamoxifen v raloxifene), and no difference between treatment arms was greater than 1% at any time point. Treatment group was not significant in B-35. The institution of a performance report in B-32 did not significantly impact compliance (P.21), but the implementation of QA sanctions did (OR 1.56; P.007). The prospective reminder was not a significant predictor of compliance in B-35 (P.30). Compliance did appear to be improved in the reminder group for some time points, but this effect has reversed for the most recent assessment (Fig 1). The introduction of a status report, which listed past forms submission and missing assessments, was not associated with a change in compliance in B-35 (P.10). Results for timely submission were largely consistent with the results for overall compliance, with a few exceptions. There was no difference in timely submission by institution size in either B-32 or B-35. Timely submission in STAR decreased significantly over time (OR 0.87 for each calendar year; P.001). Calendar time was not significant for timely submission in B-32 (P.70). Treatment group did not predict timely compliance in B-32 (P.18). The initiation of the B-32 performance report indicating PRO form submission with a qualitative score was actually associated with a decrease in timely submission (OR 0.67; P.002), whereas the QA sanction was not significant (P.56). The QMD forms, provided for 654 (97%) of 675 missed assessments in B-32 and for 485 (66%) of 733 missed assessments in B-35, also provide insights into the cause of missing assessments. Additional 5116 JOURNAL OF CLINICAL ONCOLOGY

5 Compliance With Patient-Reported Outcomes Compliance (%) Prospective reminder Control Assessment Time Point (months) Fig 1. Quality of life compliance by intervention group across assessment time points in National Surgical Adjuvant Breast and Bowel Project B-35. QMD forms for B-35 will be submitted in the future because this is an ongoing trial. The reasons that the questionnaire was missed at the clinic visit are listed by treatment for B-32 (Table 3) and overall for B-35 (Table 4). The proportion of missed assessments attributed to staff oversight or understaffing was close to one half in both protocols. Participants not appearing for appointments accounted for approximately one fifth. Attributions to the participant s medical or emotional condition were rare in both protocols. Participant refusal was low in B-32 ( 1% refused attributed to burden) but much higher in B-35 (20%). DISCUSSION Despite the mixed analytic results for compliance initiatives, there has been an improvement in compliance in recent studies compared with studies undertaken in the 1990s. For example, NSABP protocol B-23, which opened in May of 1997, reported QOL form compliance of 70% to 82% for early time points and Table 4. Reasons for Missed QOL Assessments in NSABP B-35 Reason QOL Was Not Done in Clinic No. of Assessments % Staff oversight or understaffing Patient unavailable (eg, scheduling or transportation difficulties) Patient refused to complete questionnaire Clinical staff concerned for patient s medical or emotional condition Patient felt too ill or emotionally upset Abbreviations: QOL, quality of life; NSABP, National Surgical Adjuvant Breast and Bowel Project. 50% to 59% for the latest time points, which is in stark contrast with the high compliance achieved in the protocols reported here. It has been widely observed that PROs are receiving greater attention in the scientific community, but the improvement might also be attributed to changes at the NSABP. Given this general observation, we recommend integration of PRO studies into the parent protocol documents and procedures, involvement by the protocol medical officers, and the appointment of a dedicated person to facilitate compliance. Anecdotally, reimbursement of NCI credits to CCOP institutions improves motivation; it is less clear whether this is true for non-ccop institutions that receive monetary reimbursement. Although this report has focused on systematic evaluation of compliance, the success of some interventions is best evaluated in an ad hoc fashion. For example, in July of 2001, 223 patients were identified as being enrolled onto the QOL study of NSABP B-30 (a randomized chemotherapy trial) and having a missing baseline QOL form. A tailored letter was sent to the study coordinators at each of the relevant institutions. By November, the number of patients with a missing baseline form was reduced to 45 (2% of enrollment). This is one of the experiences that leads us to believe in the benefit of targeted communication. And because targeted communication requires central monitoring, we also recom- Table 3. Reasons for Missed QOL Assessments in NSABP B-32 Treatment SNAD SN Total Reason QOL Was Not Done in Clinic No. of Assessments % No. of Assessments % No. of Assessments % Staff oversight or understaffing MD felt patient was too ill MD felt patient was too emotionally upset Patient failed to appear for appointment Patient refused because she felt too ill Patient refused because she had too little time Patient refused because she dislikes completing form or complains of burden Other Total Abbreviations: QOL, quality of life; NSABP, National Surgical Adjuvant Breast and Bowel Project; SNAD, sentinel node resection followed by conventional axillary dissection; SN, sentinel node resection alone; MD, doctor

6 Land et al mend that effort be given to programming for internal reports. Presentations at cooperative group meetings and our survey of institutions have also led to productive discussions. Our recommendations with respect to providing detailed routine information to institutions are more equivocal. We have not evaluated the benefit of providing, at the time of enrollment, a list of anticipated assessment dates. The prospective reminder in B-35 did not produce consistently favorable results. It seems that the improvement ( 2% to 3%) was smaller than the study was powered to address. That benefit might not justify the cost of developing this intervention in another data center. The most recent time points showed lower compliance for the intervention group, although this result might be spurious as a result of small sample sizes. Because this study is ongoing, the present analysis should be viewed as preliminary. The submission status report is standard for other forms at the NSABP, so this will be continued without further evaluation. Although it was encouraging that the QA sanctions were associated with an increase in compliance, the performance report was associated with a decrease. We cannot recommend implementing an institutional performance report without further evaluation. It could be, in part, that this report seeks mainly to address forms already missed, rather than to improve compliance prospectively, so that the present analysis was not suited to reveal the benefit. It might also be that the decrease in compliance was coincident with the initiation of this initiative but not caused by it. It is also possible that simple positive reinforcement would be more effective. The United Kingdom Cancer Research Campaign (UK CRC) Clinical Trials Center recommends sending a letter of thanks to the institution coordinators when the first questionnaires are submitted early in a study. 30 The QMD is useful for detecting barriers to participation, for bookkeeping, and for the analysis of study outcomes in the presence of missing data. Therefore, we recommend the QMD, although we have not tested its effect on compliance. We have not formally evaluated the benefit of patient brochures. The UK CRC has found that compliance is improved when the patient is provided with clear information about the purpose of the study. 30 On that basis, we assume our brochures are helpful if they are given to patients. Other cooperative groups have reported success with more intensive patient-directed compliance interventions. The Interdisciplinary Group for Cancer Evaluation in Italy reported increasing submission rates by nearly 50% by mailing reminders of outstanding questionnaires directly to patients. 6 The UK CRC has used a strategy in which patients are first contacted by telephone before questionnaires are mailed, questionnaires are mailed directly to patients, and finally, patients are sent a reminder if the questionnaires are not returned within 2 weeks. 30 Although these approaches seem to be successful in assuring compliance, the NSABP has sought to avoid the cost and patient privacy issues associated with direct contact between the NSABP central offices and patients. In addition, it has been our view that the office visit allows the most controlled, standardized environment for questionnaire completion. Therefore, we have discouraged questionnaire completion in the participant s home except as a last resort. Treatment group had a significant effect on compliance in B-32, with compliance higher in the experimental (SN) group. This is the only nonblinded study reported here. It is possible that the assessments were seen as less important for patients who were not in the experimental group (explaining the 2% higher rate of staff oversight or understaffing for SNAD and the 2.5% higher rate of patient refusal, according to QMD results). The difference is not explained by a difference in compliance with clinic visits, which was in fact higher in the SNAD group. A detailed examination of this bias will be required for the primary PRO analysis. Our results indicated that compliance was higher for white participants than nonwhite, which is a result that we have not seen reported elsewhere. Barriers to clinical trial recruitment and retention among nonwhite racial groups have received much attention ; barriers to PRO submission also warrant examination. This examination should include an assessment of resources at institutions with large minority populations. There has been no consistent finding regarding participant age and compliance either in our experience or in reports from other groups. 5,17,18 In the NSABP Breast Cancer Prevention Trial, compliance was unrelated to all participant factors tested, including age, race/ethnicity, education, income, and baseline risk of breast cancer. 17 Deteriorating health status is one factor in patient compliance in advanced disease trials, 2 but our QMD results indicate that this is not a factor for adjuvant breast cancer trials. Participant attitude might also be expected to play a role, but both anecdotal and survey evidence suggest that most participants appreciate the opportunity to complete QOL questionnaires. 30 QMD results from B-32 indicate that few participants complained of questionnaire burden ( 1%). In B-35, the rate of participant refusal was higher (20.2%). The reason for this is unclear. The questionnaire was just seven items longer than in B-32. The high rate might have resulted from our removal of the other reasons option that accounted for approximately 30% of QMD responses in B-32. We also speculate that, because B-35 participants had very early-stage disease, they were less committed to the clinical trial than participants in B-32. Other groups have reported that institutional factors seem to be more important than patient factors. 2,7 For example, the German Breast Cancer Study Group found that compliance with PROs depended more on the institution and the attitude of the treating physician than on patient characteristics. 7 In the NSABP Breast Cancer Prevention Trial, as in our B-32 results, midsized institutions had the highest follow-up compliance. 17 The Eastern Cooperative Oncology Group has also reported that midsized institutions had highest baseline compliance. 18 One might speculate that midsized institutions take advantage of economies of scale, without becoming too large to function optimally. Future studies might also do more to examine the sources of variation between institutions, including staff-to-patient ratio, history of cooperative group membership, and staff turnover rate. Anecdotally, noncompliance has been attributed to a sense among clinical staff that the PRO outcomes were not as important relative to the clinical outcomes. The high rate of staff oversight or understaffing reported on QMD forms is consistent with that explanation. However, physician attitudes seem to have improved in recent years JOURNAL OF CLINICAL ONCOLOGY

7 Compliance With Patient-Reported Outcomes The primary end point of the present report was the submission of questionnaires. However, compliance interventions might reasonably be evaluated based on other criteria, such as the apparent correctness of the data (when applicable), submission error rates (eg, duplicate form submission or errors in the coding of the visit time point), rates of missing items on forms, or number of compliance-related contacts made by the cooperative group headquarters staff to the institutional staff. The issues are complex. The present report provides a systematic update of our recent experience in breast cancer prevention and adjuvant treatment studies as a foundation for future initiatives. AUTHORS DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST Although all authors completed the disclosure declaration, the following author(s) indicated a financial or other interest that is relevant to the subject matter under consideration in this article. Certain relationships marked with a U are those for which no compensation was received; those relationships marked with a C were compensated. For a detailed description of the disclosure categories, or for more information about ASCO s conflict of interest policy, please refer to the Author Disclosure Declaration and the Disclosures of Potential Conflicts of Interest section in Information for Contributors. Employment or Leadership Position: None Consultant or Advisory Role: Walter M. Cronin, AstraZeneca (C) Stock Ownership: None Honoraria: Walter M. Cronin, AstraZeneca Research Funding: Joseph P. Costantino, Eli Lilly & Co Expert Testimony: None Other Remuneration: None AUTHOR CONTRIBUTIONS Conception and design: Stephanie R. Land, Joseph P. Costantino, Norman Wolmark, Patricia A. Ganz Administrative support: Joseph P. Costantino, Thomas B. Julian, Walter M. Cronin Provision of study materials or patients: Thomas B. Julian Collection and assembly of data: Stephanie R. Land, Joseph P. Costantino, Sarah R. Haile Data analysis and interpretation: Stephanie R. Land, Marcie W. Ritter, Joseph P. Costantino, Walter M. Cronin, Sarah R. Haile, Patricia A. Ganz Manuscript writing: Stephanie R. Land, Marcie W. Ritter, Joseph P. Costantino, Walter M. Cronin Final approval of manuscript: Stephanie R. Land, Marcie W. Ritter, Joseph P. Costantino, Thomas B. Julian, Walter M. Cronin, Norman Wolmark, Patricia A. Ganz Other: Thomas B. Julian [Protocol oversight, B-32, B-35] REFERENCES 1. Bernhard J, Gusset H, Hurny C: Practical issues in quality of life assessment in multicentre trials conducted by the Swiss Group for Clinical Cancer Research. Stat Med 17: , Bernhard J, Cella DF, Coates AS, et al: Missing quality of life data in cancer clinical trials: Serious problems and challenges. Stat Med 17: , Hayden KA, Moinpour CM, Metch B, et al: Pitfalls in quality-of-life assessment: Lessons from a Southwest Oncology Group breast cancer clinical trial. Oncol Nurs Forum 20: , Hopwood P, Harvey A, Davies J, et al: Survey of the administration of quality of life (QL) questionnaires in three multicentre randomised trials in cancer: The Medical Research Council Lung Cancer Working Party and the CHART Steering Committee. Eur J Cancer 34:49-57, Kaasa S, Hjermstad MJ, Jordhoy MS, et al: Compliance in quality of life data: A Norwegian experience. Stat Med 17: , Mosconi P, Torri V, Cifani S, et al: The multi-centre assessment of quality of life: The Interdisciplinary Group for Cancer Care Evaluation (GIVIO) experience in Italy. Stat Med 17: , Olschewski M: Compliance with QOL assessment in multi-centre German breast cancer trials. Stat Med 17: , Coates A, Gebski V: Quality of life studies of the Australian New Zealand Breast Cancer Trials Group: Approaches to missing data. Stat Med 17: , Troxel AB, Fairclough DL, Curran D, et al: Statistical analysis of quality of life with missing data in cancer clinical trials. Stat Med 17: , Land S, Wieand HS, Day R, et al: Methodological issues in the analysis of quality of life data in clinical trials: Illustrations from the National Surgical Adjuvant Breast and Bowel Project (NSABP) Breast Cancer Prevention Trial, in Mesbah M, Cole B, Lee M (eds): Statistical Design, Measurement and Analysis of Health Related Quality of Life. Boston, MA, Kluwer Academic Publishers, 2002, pp Little R, Rubin D: Statistical Analysis With Missing Data. New York, NY, Wiley and Sons, Pauler DK, McCoy S, Moinpour C: Pattern mixture models for longitudinal quality of life studies in advanced stage disease. Stat Med 22: , Fairclough DL: Multiple imputation for nonrandom missing data in longitudinal studies of health-related quality of life, in Mesbah M, Cole BF, Lee M (eds): Statistical Design, Measurement and Analysis of Health Related Quality of Life: Design, Measurements and Analysis. Boston, MA, Kluwer Academic Publishers, 2002, pp Italian Group for Evaluation of Outcomes in Oncology: Patient compliance with quality of life questionnaires. Tumori 85:92-95, Kopp I, Lorenz W, Rothmund M, et al: Relation between severe illness and non-completion of quality-of-life questionnaires by patients with rectal cancer. J R Soc Med 96: , Walker M, Brown J, Brown K, et al: Practical problems with the collection and interpretation of serial quality of life assessments in patients with malignant glioma. J Neurooncol 63: , Ganz PA, Day R, Costantino J: Compliance with quality of life data collection in the National Surgical Adjuvant Breast and Bowel Project (NSABP) Breast Cancer Prevention Trial. Stat Med 17: , Hahn EA, Webster KA, Cella D, et al: Missing data in quality of life research in Eastern Cooperative Oncology Group (ECOG) clinical trials: Problems and solutions. Stat Med 17: , Hopwood P, Stephens RJ, Machin D: Approaches to the analysis of quality of life data: Experiences gained from a Medical Research Council Lung Cancer Working Party palliative chemotherapy trial. Qual Life Res 3: , Osoba D, Zee B: Completion rates in healthrelated quality-of-life assessment: Approach of the National Cancer Institute of Canada Clinical Trials Group. Stat Med 17: , Bernhard J, Gusset H, Hurny C: Quality-of-life assessment in cancer clinical trials: An intervention by itself? Support Care Cancer 3:66-71, Hurny C, Bernhard J, Joss R, et al: Feasibility of quality of life assessment in a randomized phase III trial of small cell lung cancer: A lesson from the real world The Swiss Group for Clinical Cancer Research (SAKK). Ann Oncol 3: , Young T, Maher J: Collecting quality of life data in EORTC clinical trials: What happens in practice? Psychooncology 8: , Vogel VG, Costantino JP, Wickerham DL, et al: Effects of tamoxifen vs raloxifene on the risk of developing invasive breast cancer and other disease outcomes: The NSABP Study of Tamoxifen and Raloxifene (STAR) P-2 trial. JAMA 295: , Land SR, Wickerham DL, Costantino JP, et al: Patient-reported symptoms and quality of life during treatment with tamoxifen or raloxifene for breast cancer prevention: The NSABP Study of Tamoxifen and Raloxifene (STAR) P-2 trial. JAMA 295: , Phillips RA, Faulkner M, Gassman J, et al: Recruitment of African Americans with chronic renal insufficiency into a multicenter clinical trial: The African American Study of Kidney Disease and Hypertension. J Clin Hypertens (Greenwich) 6: , Unson CG, Dunbar N, Curry L, et al: The effects of knowledge, attitudes, and significant others on decisions to enroll in a clinical trial on osteoporosis: Implications for recruitment of older African-American women. J Natl Med Assoc 93: ,

8 Land et al 28. Gorelick PB, Harris Y, Burnett B, et al: The recruitment triangle: Reasons why African Americans enroll, refuse to enroll, or voluntarily withdraw from a clinical trial An interim report from the African-American Antiplatelet Stroke Prevention Study (AAASPS). J Natl Med Assoc 90: , Gorelick PB, Richardson D, Hudson E, et al: Establishing a community network for recruitment of African Americans into a clinical trial. The African-American Antiplatelet Stroke Prevention Study (AAASPS) experience. J Natl Med Assoc 88: , Fallowfield L: Compliance issues in quality of life assessment: Experiences of two Cancer Research Campaign sponsored groups. Stat Med 17: , Batel-Copel LM, Kornblith AB, Batel PC, et al: Do oncologists have an increasing interest in the quality of life of their patients? A literature review of the last 15 years. Eur J Cancer 33:29-32, 1997 Acknowledgment We thank the courageous participants without whom this research could not have been accomplished. We thank Barbara C. Good, PhD, Director of Scientific Publications for the National Surgical Adjuvant Breast and Bowel Project, and Wendy L. Rea for editorial assistance, and Myoungkeun Lee, MS, for statistical programming JOURNAL OF CLINICAL ONCOLOGY