SPECIAL ARTICLE
Oncologists' Reluctance to Accrue Patients Onto Clinical Trials: An Illinois Cancer Center Study By Al B. Benson III, Janet Prokop Pregler, Judy A. Bean, Alfred W. Rademaker, Bruce Eshler, and Karen Anderson Clinical trials are recognized as the standard of care for the cancer patient, and the randomized, controlled trial represents the most definitive method to determine the effectiveness or ineffectiveness of a cancer treatment. However, less than 3% of all eligible patients enter a clinical trial. Of the 437 physician members of the Illinois Cancer Center (ICC), 244 responded to a survey designed to determine factors that present a significant barrier to entering patients on clinical trials. Rigid protocol design was the primary deterrent to accrual, especially for medical oncologists. Surgeons, radiation oncologists, and medical oncologists differed with respect to several factors, including willingness to seek a clinical trial, tendency to treat patients off study, quality-of-life issues, and the belief that trials were too excessive in time commitment (P < .05). Compared with hospital-based physicians, community oncologists had fewer patients on trial,
IN
RECENT YEARS, the prospective, randomized, controlled clinical trial has become the proving ground for new medical therapies. To be accepted as the standard of care by the medical community, treatment efficacy must be shown in a well-conducted, randomized trial. An important factor limiting the availability of promising new therapies has been inadequate accrual of patients onto clinical studies. Surveys of the general public have shown widespread support for the concept of clinical trials1 '2; however, fewer than 20% of cancer patients in hospitals with available research protocols actually participate. 3' 4 Nationwide, only about 3% of cancer patients receive treatment in a clinical trial.5 Many investigators, including the leadership of the National Cancer Institute (NCI), have publicly emphasized that cancer clinical trials represent a standard of care for the cancer patient and should be viewed as the treatment of choice for most stages of the various types of cancer.6 2 In addition, the current Director of the NCI has stated: "It is the strongly held view of the NCI that in many cases experimental therapy is the patient's best hope for survival."' 3 Of the 7 million patients with a diagnosis of cancer each year, only 25,000 are entered onto an NCI-approved study." There-
were more likely to enter patients on the basis of age, and were more concerned about aspects of informed consent and the financial burden of a trial (P < .01). One third of the physicians never pursued a clinical trial because of conflict with the priorities of individual care and excessive follow-up time. Fourteen percent indicated that they discouraged patients from participating in a clinical trial due to the risk of a patient receiving a placebo and patient follow-up requirements (P < .05). Subgroups of physicians differ in their reluctance to accrue patients, and there are clusters of beliefs expressed by physicians concerning their clinical trial activity. Current conduct of clinical trials needs to be reassessed, and intervention studies are required to determine the best methodology to alter physician reluctance to pursue clinical trials. J Clin Oncol 9:2067-2075. © 1991 by American Society of Clinical Oncology. fore, eligible patients may not be obtaining appropriate adjuvant chemotherapy, particularly for breast and colon cancer, and patients may not be receiving chemotherapy at proper dosages or time intervals. In an attempt to increase the total number of patients entered onto clinical trials, the NCI has initiated new programs directed towards physicians already involved in clinical cancer research. 7' 9,12'14 "High-priority clinical trials" have
been designated by the NCI and involve intergroup collaborations. The cooperative groups have
also been encouraged to improve quality control,
From the Division of Hematology/Oncology,Department of Medicine, the CancerCenter Biometry Section, and the Medical School, Northwestern University Medical School, Chicago; and the Illinois CancerCenter, Chicago,IL. Submitted December 13, 1990; accepted May 24, 1991. Supported by a grant from the National Cancer Institute, NationalInstitutes of Health (N02 P30-CA21742). Presented in part at the 25th Annual NationalMeeting of the American Society of Clinical Oncology, San Francisco, CA, May 21-23, 1989. Address reprint requests to Al B. Benson III, MD, Illinois Cancer Center. 200 S Michigan Ave, Suite 1700, Chicago, IL 60604-2404. c 1991 bvAmerican Society of Clinical Oncology. 0732-183X/91/0911-0007$3.00/0
Journal of Clinical Oncology, Vol 9, No 11 (November), 1991: pp 2067-2075
Downloaded from ascopubs.org by Boston University on April 7, 2019 from 128.197.229.194 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
2067
2068
BENSON ET AL
minimize ineligible patients, and reduce complexities of clinical trials by limiting data collection and patient testing. Physicians are being urged to use the NCI's Physician Data Query (PDQ) as a primary information source listing available clinical trials. There are efforts to increase public awareness of trials and to discuss insurancerelated economic issues. Previous investigators have identified specific obstacles to accrual of patients to clinical trials. The Southwest Oncology Group (SWOG) surveyed 54 participants in the Cooperative Group Outreach Program (CGOP). The results of this survey" indicated that a primary obstacle to community physician participation in clinical trials is the lack of studies that meet the requirements of the physician's patient population. Eligibility requirements, particularly for elderly patients, are often too rigid. The Toronto Leukemia Study Group examined the effect of rigid exclusion criteria on the calculation of remission duration and survival rates after combination chemotherapy for acute myelogenous leukemia.' 6 The overall remission rate varied 40%, depending on exclusion criteria applied to determine which cases would be analyzed. Investigations have shown that, apart from protocol exclusion criteria, physician refusal to enroll patients onto a trial is the single most important factor in low protocol participation.4'I"7
In one
report, physician refusal accounted for 52% of all failures to enroll eligible patients. Physician refusals outnumbered patient refusals by a margin of two to one.5 6
Only one published study directly examines physician reasons for excluding eligible patients from a randomized clinical trial. The survey of 94 National Surgical Adjuvant Breast and Bowel Project (NSABP) principal investigators used a self-administered, primarily open-ended questionnaire that asked physicians why they were not entering eligible patients onto an NSABP trial. Responses indicated that issues such as the doctorpatient relationship in a randomized trial, concerns about informed consent, and physicians' reluctance to discuss openly their uncertainty about the correct treatment all contributed to physician refusals to enroll eligible patients. A follow-up survey discussed breast cancer specialists' attitudes toward informed consent."' There are special issues inherent in the treatment
of breast cancer and in clinical trials of breast cancer therapies, including the historically emotional debate over the best treatment and the specific legal requirements in some states for full disclosure of all treatment options to all patients. Thus, attitudes toward clinical trials expressed by physicians who treat breast cancer primarily cannot be assumed to represent those physicians who treat other malignancies. To investigate the reasons physicians do not place patients on clinical trials further, 437 physician members of the Illinois Cancer Center (ICC) were surveyed. The purpose of the survey was to identify physician opinion regarding factors affecting accrual to clinical trials and to assess the extent to which these factors were deterrents to patient accrual. METHODS Subjects The ICC. the NCI Designated Consortium Research Center for the State of Illinois, has extensive experience in conducting and coordinating multiinstitutional research in clinical trials and cancer control. Membership includes 437 Illinois physicians representing medical oncology, hematology, radiation oncology, surgical oncology including otolaryngology, thoracic, and colorectal surgery, and gynecologic oncology. Physicians who were members as of February 1988 are included in the survey. Each of the 437 physicians was sent a 50-question survey. After 1 month, telephone follow-up of nonresponders was attempted, and a second questionnaire was sent.
Questionnaire The survey instrument was a five-page self-administered questionnaire revised after pilot testing among physicians at Northwestern University Medical School. Fifty questions were presented in four parts. The first section contained nine closed- and open-ended questions to determine the physician-respondent's educational background, patient population, self-appraised level of participation in randomized clinical trials, and type of practice. Physicians also were asked whether they were on the faculty of a medical school. Eleven true-false statements followed, eliciting opinions regarding informed consent, adjuvant trials, the treatment of patients according to one arm of an ongoing protocol without formally entering the patient, patient age, and the efficacy of standard therapies as barriers to the enrollment of otherwise eligible patients. Physicians were asked whether they actively sought trials for their patients and whether they ever encouraged (discouraged) their patients to enroll in trials. The third section consisted of 24 closed-ended questions inquiring whether a particular statement is true for nearly all, some, or none of the patients eligible for clinical trials. Physicians also were asked to indicate if any of the issues mentioned had been the major reason they did
Downloaded from ascopubs.org by Boston University on April 7, 2019 from 128.197.229.194 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
2069
CLINICAL TRIALS PATIENT ACCRUAL not enter a patient onto a trial. Factors addressed included informed consent, protocol design, and the impact of trials on patients and physicians. Physicians were encouraged to provide written comments about topics mentioned in the questionnaire and about clinical trials in general.
StatisticalAnalyses Frequency of responses and percentages were calculated using the Statistical Application System (SAS) statistical 2 2 package. ' The X test with correction for continuity when appropriate was used to assess associations between variables. Statistical significance was indicated by a P value < .05.
RESULTS
The response rate after two mailings and a telephone reminder was 58% (252 of 437). Nonresponding physicians were similar in proportion to responding physicians in terms of expertise (medical, surgical, or radiation oncologist; P = .47) and type of practice (community, hospital; P = .75). There was a lower percentage of universityaffiliated physicians in nonresponders compared with responders (49% v 78%; P < .0001). Eight of the 252 responding physicians were retired. As the forms were largely blank for these retired physicians, their responses were not included in the data set. Results are presented for the remaining 244 practicing oncologists. The number of responses across questions varied because not everyone answered each item. The typical respondent had completed training at a university teaching hospital, had fellowship training, and had some degree of contact with patients on a clinical trial during postgraduate clinical rotations. Respondents included 128 hematologists and medical oncologists (52%), 60 surgeons (25%), and 47 radiation oncologists (19%). Nine physicians (4%) physicians did not indicate a specialty. Sixty-six percent (160 of 244) of the physicians were members of a group directly involved in trials, whereas 11% (26 of 244) said they must make an independent effort to seek out a trial for their patients. Although not affiliated with a clinical trial group, 19% (47 of 244) had regular contact with physicians directly involved in trials. Ninety-three physicians (38%) were community-based, 144 (59%) were hospital-based, and seven (3%) did not indicate type of practice. Seventy-seven percent (189 of 244) were affiliated with a university, and 22% (53 of 244) were not. Sixty-four percent (156 of 244) indicated that 1% to 25% of their cancer patients were on trials.
Only 18% (44 of 244) responded that over 26% of their patients were participating. Sixty-three percent of respondents stated they were actively involved in finding protocols for their cancer patients. Although 77% usually encourage patients who are eligible to enter a randomized trial, 15% indicated that they at times discourage eligible patients from participating. Eight of the 11 true-false statements and all of the 24 closed-ended items were analyzed by determining the frequency in each response category. Individual statements were grouped by content area into five categories (informed consent, study design, physician-patient relationship, effect of trials on patients, and specific difficulties; Tables 1 to 5). One issue that may deter physicians from entering patients onto clinical trials is the necessity to obtain informed consent (Table 1). Nearly all respondents (98%) believe that informed consent should be required; however, 57% prefer a method of consent adapted to each individual patient. Forty percent believe that informed consent is obtained not to protect the patient but for legal, societal, or other reasons. Eighty-seven percent of respondents (200 of 230) stated that informed consent is difficult to obtain, although only 23% (49 of 209) indicated this had been a reason for not entering patients onto clinical trials. A small percentage indicated that informed consent was Table 1. Frequency of Responses to Statements Regarding Informed Consent for RCCT (n = 231-240) Always or Sometimes
Statement Requirement of written informed consent before entry onto an RCCT is justified. Is difficult to obtain. Gives more information than patients desire. Depresses patients. Pressures patients into making decisions they prefer their doctor made. Gives information that causes patients not to choose the best therapy. Prefer to adapt consent for each patient. Obtained for legal, societal, or other reasons Designed to control the actions of physicians.
Always True
True %)
(%)
98 87
84 6
83 81
19 13
78
14
77 57*
10
40* 10*
Abbreviation: RCCT, randomized cancer clinical trials. *The percentage responding "true" on true-false question.
Downloaded from ascopubs.org by Boston University on April 7, 2019 from 128.197.229.194 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
2070
BENSON ET AL
designed to control the actions of physicians. Remaining statements, including "informed consent may give patients more information than they desire" and "informed consent depresses patients," were true for at least some patients in the opinion of most physicians. For the six statements in Table 1, the percentage of physicians who indicated that these were major deterrents to patient accrual ranged from 8% to 23% (average, 14%). Eight statements concerning clinical trial design and physician responses are presented in Table 2. Eighty-two percent of respondents believe some or all protocols are too rigidly designed; 36% (73 of 203) stated rigid design had been a reason for excluding patients from trials. Fifty-five percent treat patients eligible for a trial in accordance with one arm of the protocol without formally entering the patient onto the trial. Half of the physicians (51%) reported excluding patients on the basis of age. Twenty-nine percent stated that standard treatment for some malignancies resulted in such high cure rates that they would not enter patients in a trial regardless of its design. Most physicians are not at ease when talking with patients about the choice of a treatment (Table 3). They also believe that research protocols undercut patients' beliefs in physician knowl-
Table 2. Frequency of Responses to Statements Regarding the Design of RCCT (n = 228-239)
Statement
When a protocol is available, physician finds the design scientifically sound. Difficult to get patients onto RCCT because protocols are too rigidly designed. Dislike entering patients on RCCT when there is a chance they will receive a placebo. Will treat patients eligible for a trial in accordance with one arm of protocol without entering the person Age of patient alone sometimes reason for exclusion. Cure rate for standard treatment is so high physician would not enter patients regardless of the design Age of patient alone sometimes reason for inclusion into RCCT Would not enter patients onto an adjuvant study regardless of its design.
Always or Sometimes True (%)
Always True
99
40
82
12
56
(%)
12
55* 51 *
29* 28*
*The percentage responding "true" on true-false question
Table 3. Frequency of Responses to Statements Regarding Physician and Patient Relationship (n = 232-233)
Statement
Always or Sometimes True (%)
Always True
55
6
51
2
35
1
28
4
Discussions concerning which treatment is the right one are uncomfortable. Patient's belief in physicians' knowledge and autonomous decisionmaking power is undermined. Patients feel that physician wants to "get rid of a hopeless case" if (s)he suggests entry onto RCCT Physician feels uncomfortable when patient is a research subject.
(%)
edge and decision-making power. Nonetheless, these issues were not detriments to accrual. Seven percent of respondents identified uneasiness in discussing treatment options as a reason for not entering patients. Only 5% indicated that research undermining patients' beliefs in their knowledge and decision-making power had prevented them from entering a patient on trial. Table 4 summarizes the negative aspects that most physicians believe affect patients involved in clinical trials. Ninety-one percent indicated that trials are inconvenient for patients, and 79% perceive that research is costly to participants. Inconvenience and high financial burden for patients involved in clinical trials were cited by 29% and 23% of respondents, respectively, as factors that had caused them not to place patients on trial. Most physicians (78%), however, believed that Table 4. Frequency of Responses to Statements Regarding Physicians' Perception of the Effect of Trials on Patients (n = 229-235)
Statement
Always True or Sometimes True (%)
Always True
91
10
79 78
7 27
64
3
62
3
61
3
33
1
RCCT are inconvenient for patients. RCCT impose high financial burden on the patient. Patients receive better care in RCCT Goals of RCCT conflict with priorities of individual patient care. Participation in RCCT reduces the amount of quality time a patient has before death. Patients become depressed when they are on RCCT Patients receive inferior care on RCCT
Downloaded from ascopubs.org by Boston University on April 7, 2019 from 128.197.229.194 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
(%)
2071
CLINICAL TRIALS PATIENT ACCRUAL
patients always or sometimes received better care on a trial. One third stated that trials always or sometimes resulted in inferior care. Nearly three fourths of the physicians (73%) stated that excessive time is required for patient follow-up on study (Table 5). Twenty-six percent (54 of 205) cited this as a reason they had not entered patients onto trials. More than half (54%) believe that once a patient is participating in a protocol, there is sometimes or always insufficient support for adequate follow-up. In further analysis, questions were crosstabulated by expertise of physicians to ascertain differences among the three physician groups (medical, surgical, and radiation oncologists). Seventy-seven percent of the medical oncologists reported attempting to find clinical trials for their patients, compared with 53% of surgeons and 38% of radiation oncologists (P < .001). Surgeons and radiation oncologists had greater preference for individual informed consent rather than one standard form (P = .003). The radiation oncologists were most likely to state that, for some tumors, the standard treatment has such a high cure rate that they would not enter patients onto a trial (P = .009). Sixty-seven percent of the radiation oncologists, 59% of medical oncologists, and 37% of surgeons indicated they sometimes use protocol guidelines to treat patients without formally enrolling them (P = .004). Only 2% of the medical oncologists would not enter patients onto an adjuvant study, compared with 15% and 13% for surgeons and radiation oncologists, respectively (P = .016). Significant differences (P < .05) between radiation oncologists, surgeons and medical oncologists
Table 5. Frequency of Responses to Statements Regarding Difficulties in Trial Procedures (n = 230-235)
Statement
Increased time required to observe patients on RCCT is excessive. Feels responsible to offer some form of treatment. Has difficulty getting support to observe treatment plan when patient is on RCCT. Feels responsible if a patient is not on the treatment arm shown to be superior
Always or Sometimes True (%)
Always True
73
18
59
16
54
7
25
3
(%)
were found in responses (always, sometimes, never true) to eight statements (Table 6). A higher percentage of radiation oncologists always or sometimes feels uncomfortable when their patients are on a clinical trial (51%) when compared with surgeons (29%) and medical oncologists (20%). Similarly, more radiation oncologists have at least some difficulty following through with treatment plans for patients on trials (71% v 47% and 51%). Radiation oncologists also feel that there is at least some conflict between clinical trials and individual patient care (79% v 51% and 62%). More surgeons felt that clinical trials were used to "get rid of a hopeless case" (52%) than did radiation oncologists (36%) or medical oncologists (27%). Fewer surgeons said that follow-up time was excessive (59% v 76%). When medical oncologists were compared with radiation oncologists and surgeons, fewer medical oncologists indicated that the following statements were true in part: discussions about treatment uncertainty were uncomfortable (47% v 65% and 64%), clinical trials reduce the patient's quality time (53% v 71% and 70%), and patients on trials become depressed (53% v 66% and 75%). The groups demonstrated significant differences (P < .05) in the frequency with which they endorsed the following five reasons for not entering patients onto trials (Table 7): (1) trial designs are too rigid, (2) trials place a high financial burden on patients, (3) trials require an excessive amount of time to observe a patient, (4) placebocontrolled trials are unacceptable, and (5) clinical trials are inconvenient for patients. A higher percentage of medical oncologists compared with the other two groups of physicians felt that these five factors were important obstacles and hindered their entering of patients. Community physicians were compared with hospital-based physicians. Twenty-four percent of hospital-based doctors had more than 25% of their patients on protocol, as compared with only 9% of community physicians (P = .002). Community physicians were more likely to include patients on trials on the basis of age by a margin of 39% to 21% (P = .004). More community physicians entered patients on trials when the standard cure rate was high (39% v 23%; P = .013). Significant differences (P < .01) between community and hospital physicians were found in responses (always, sometimes, never true) to six
Downloaded from ascopubs.org by Boston University on April 7, 2019 from 128.197.229.194 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
2072
BENSON ET AL Table 6.
Differences Between Physician Expertise Groups (P < .05) Radiation Oncologists (n =45-47)
Statement
Surgeons (n = 53-56)
Medical Oncologists (n= 122-125)
AorS (%)
A (%)
AorS (%)
A (%)
AorS (%)
A (%)
51 71 79 36 76
2 11 2 0 23
29 47 51 52 59
7 4 5 2 9
20 51 62 27 76
3 6 2 0 20
65 71 66
4 4 6
64 70 75
11 4 2
47 53 53
4 1 1
Physician feels uncomfortable when patient is on a trial. Support for follow-up is difficult. Clinical trials conflict with patient care. Clinical trials are for "hopeless cases." Follow-up time is excessive. Discussions about uncertain treatments are uncomfortable. Clinical trials reduce patients' quality time Patients on trials become depressed.
Abbreviations: A or S, always or sometimes true; A, always true.
statements (Table 8). More hospital physicians stated that patients always receive better care in clinical trials (35%
v 14%). More community
sions that they would rather the physician made
(P = .01); (2) discussions concerning uncertainty as to which treatment is better are uncomfortable
physicians stated that excessive time always is
(P = .014); and (3) trials impose a high financial
required to observe patients on trials (29% v 11%), that clinical trials are always too rigidly
burden on patients (P = .015).
designed (19% v 7%), that informed consent always is difficult to obtain (12% v 1%), that they always or sometimes feel responsible if the patient is randomized to an arm that eventually is not shown to be superior (31% v 21%), and that they always dislike placing a patient on a placebo arm
younger patients (11% v 38% had at least 65% of their patients aged 65 years or more), were members of an organization directly involved in trials (75% v 42%), and tended to place more than 25%
(19% v 6%). However, the groups did not differ when asked if these factors presented obstacles to
putting patients on trials. Finally, a higher percentage of community physicians indicated that they did not enter patients onto research protocols (Table 9) because (1) informed consent pressures patients to make deciTable 7. Significant Differences Between Physician Expertise Groups on Reasons for Not Entering Patients Onto Clinical Trials (P < .05) Radiation Medical Oncologists Oncologists Surgeons (n - 39-41) (n= 48-50) (n = 109-112)
Reason Clinical trials are too rigidly designed. Clinical trials impose a high financial burden on patients. Follow-up time is excessive. Physician dislikes trials with placebo arm. Clinical trials are inconvenient for the patient
23
20
47
15
10
31
20
13
33
13
17
30
27
16
35
NOTE. Data are percentages of physician expertise group indicating this reason.
University-affiliated physicians tended to treat
of their patients on trials (23% v 2%; all P < .001).
They also were more likely to seek out trials actively for their patients (68% v 47%; P = .009) and they were less likely to exclude patients from
trials because of high standard cure rates (25% v 42%; P = .03). University-affiliated physicians also included more surgical oncologists (28% v 16%) and fewer radiation oncologists (16% v 32%; P = .03). Significant differences (P < .05) between university and nonuniversity physicians were found in response (always, sometimes, never true) to nine statements (Table 10). For all but one of
Table 8. Differences Between Physician Practice Groups (P < .01) Community Hospital (n = 87-90) (n = 138-141) Statement
AorS
A
AorS
A
Patients receive better care on trials. Follow-up time is excessive. Clinical trials are too rigidly designed. Informed consent is difficult to obtain. Dislike patient not being randomized to superior arm. Physician dislikes trials with placebo arm.
68 78 88 90
14 29 19 12
83 69 78 86
35 11 7 1
31
7
21
0
60
19
53
6
NOTE. No statements had 01 < P < .05.
Downloaded from ascopubs.org by Boston University on April 7, 2019 from 128.197.229.194 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
2073
CLINICAL TRIALS PATIENT ACCRUAL Table 9. Significant Differences Between Physician Practice Groups on Reasons for Not Entering Patients Onto Clinical Trials (P < .05) Community (n = 80-82)
Reason Patients are pressured into decisions they want their physican to make. Discussions about uncertain treatments are uncomfortable. Clinical trials impose a high financial burden on patients.
Hospital (n = 119-122)
21
7
14
3
33
17
NOTE. Data are percentages of physician practice group indicating this reason.
these statements, university physicians indicated lower rates of these being always true. Fewer nonuniversity physicians indicated that clinical trials always or sometimes reduced patients' quality time. For each of the 24 closed-ended questions analyzed, university and nonuniversity physicians were similarly deterred from entering patients onto trials. Selected items dealing with overt recruitment or nonrecruitment were correlated with other items to obtain a better understanding of physician action. One third of the physicians in the survey do not pursue trials for their patients. Conflict between the goals of a trial and individual care and excessive time required for follow-up were reasons why these physicians did not actively enroll patients. Thirty-six respondents actively discourage paTable 10. Differences by Physician University Affiliation (P < .05) University (n = 179-184)
DISCUSSION
Nonuniversity (n = 48-51)
Statement
Aor S
A
Aor S
A
Follow-up time is excessive. Patients on trials become depressed. Clinical trials are inconvenient for the patient. Clinical trials impose a high financial burden on patients. Clinical trials reduce patients' quality time. Support for follow-up is difficult. Patients receive inferior care on trials. Informed consent gives patient more information than needed. Clinical trials are too rigidly designed.
70
13
80
37
62
1
57
10
89
6
92
24
80
4
77
19
57 51
2 4
78 66
8 16
32
0
39
4
84
16
82
31
83
9
80
22
NOTE Data are percentages.
tients from entering a trial. This group of physicians dislikes the chance that a patient may receive a placebo, believes protocols are rigidly designed, and believes that there is a lack of support for patient follow-up (P < .05). Physicians who do not enter patients onto clinical trials because there is a treatment having a high cure rate have objections to clinical research in general. Significant associations (P < .05) are found with the belief that trials result in inferior patient care, that patients became depressed when entered on a trial, and that trials reduce quality time (P < .05). These physicians again cite lack of support for follow-up. Doctors who do not enter their patients on adjuvant studies had other complaints. They stated more often that the goals of trials conflict with individualized care, and that they are uneasy with having patients in research studies. These physicians also were more likely to feel responsible if a patient is not randomized to the arm of a given study that is eventually shown to be superior. Open-ended comments at the end of the questionnaire were summarized in a descriptive manner to identify physician opinion. In general, informed consent was viewed positively. Negative opinions tended to dominate comments on study design and the effect of enrollment in trials on physicians. Comments were almost evenly divided between positive and negative on the doctorpatient relationship and the effect of trial enrollment on the patient.
Our study has identified several factors related to low patient accrual onto clinical trials. The majority of responding physicians identified negative aspects of clinical trials. These aspects pertain to informed consent (difficult to obtain, not a good vehicle for information transfer), study design (rigid exclusion criteria, placebo arms), physicianpatient relationship (physician autonomy), patients (inconvenience, high financial burden), and trial procedures (excessive follow-up time). Fewer positive aspects were identified (eg, design generally sound, better care on clinical trial). Although ethical and philosophical questions about clinical trials were of concern, practical matters such as rigid protocol design, inconvenience for patients, and extra work for physicians were critical reasons cited by the ICC physicians for not entering eligible patients onto clinical trials.
Downloaded from ascopubs.org by Boston University on April 7, 2019 from 128.197.229.194 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
2074
BENSON ET AL
Rigid design of protocols was mentioned most often as a rationale for not enrolling patients and has been confirmed by numerous other studies.3-5"1017'18"22 One survey evaluating multiinstitutional trials in breast cancer noted an average of 23 eligibility criteria per study, with considerable variation in the criteria among the studies."' 22 In the Toronto Leukemia Study Group Study,16 150 patients (55%) were excluded from calculations when all standard exclusion criteria were applied. Clearly, rigid protocol exclusion criteria not only decrease accrual, they also limit applicability of results, which become less useful to physicians who treat the entire spectrum of patients. In the ICC survey, other inflexible aspects of protocols such as frequent serial laboratory tests or radiologic testing on a specific day of a protocol were cited as cumbersome, especially in the community setting. Inconvenience for patients is a barrier to accrual that has received very little attention previously but was the second leading reason preventing physicians from entering their patients onto clinical trials. Patient inconvenience potentially covers all aspects of the clinical trial. Additional time required for protocol screening, registration, testing, treatment, and follow-up provide a cumulative deterrent for the patient, as well as for the physician. The combination of additional time commitment and costs provided the third significant reason preventing ICC physicians from entering patients on trials. These reasons have been noted anecdotally by several authors.'" Increasing accrual will require a decrease in physician reluctance to participate in clinical investigations. Our study confirms and augments the conclusions of previous investigations. While Taylor et al"2' observed higher rates for concern about the physician-patient relationship and the dislike of discussions regarding treatment uncertainty, our study showed similar levels of concern regarding randomization, clinician/scientist conflict, informed consent, and physician feeling of responsi5 bility. When compared with the SWOG survey," more ICC physicians felt that informed consent was difficult to obtain, while both surveys identified long-term follow-up as a significant problem. Representatives from the NCI have cited the following obstacles as hindering patient accrual to cancer clinical trials7 "'8 2 4: (1) unacceptability of
randomization for patients requiring individualized care; (2) large time commitment for data management and informed consent; (3) denial of insurance; (4) public misunderstanding of risks and benefits; (5) limited access for minority patients; and (6) cooperative groups tend to be political, ask irrelevant questions, and have less than ideal participation by cancer centers. Our study confirmed that issues of randomization, excessive time commitment, and financial implications were significant deterrents to accrual. The ICC's physician questionnaire suggests that focusing on concerns of particular physician groups (eg, community v hospital physicians) could contribute to the success of a clinical trials program. Excessive time commitment for clinical trials was a significant accrual factor in our sample as a whole, but especially for medical oncologists, communitybased physicians, and nonuniversity physicians. These subgroups are likely not to have the infrastructure of personnel and facilities in place to conduct clinical trials. Furthermore, physicians tend to have clusters of beliefs which may reinforce their reluctance to pursue clinical trials. The survey confirmed that community physicians when compared with hospital-based physicians were more likely to believe that trials were too rigidly designed and did not provide better care for patients. In addition, the community physicians were more likely to find a placebo arm an unacceptable alternative in a randomized study. When compared with their radiation oncology or surgical colleagues, the ICC medical oncologists were more likely to state that trials are too expensive and too rigid and that studies with a placebo arm are unacceptable. Positive opinions were identified in the survey. Greater than 75% of all physicians surveyed felt that patients received better care on clinical trials, that available protocols are usually scientifically sound, and that informed consent is justified. Favorable aspects of clinical trials such as these should be emphasized in encouraging change in accrual patterns. Based on the results of this study, we make the following recommendations to increase patient accrual to clinical trials: 1. Over 50% of physicians in this survey exclude patients from clinical trials on the basis of age. As a high percentage of cancer patients are elderly and with the growth of the aged population,
Downloaded from ascopubs.org by Boston University on April 7, 2019 from 128.197.229.194 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
2075
CLINICAL TRIALS PATIENT ACCRUAL physicians must be encouraged to consider eligible older patients for trials. 2. Excessive eligibility criteria need to be reassessed by scientists designing clinical trials. Physicians should invest their time only in trials with eligibility criteria consistent with their patient population. 3. Patient convenience when on a clinical trial is a priority. The process of scheduling, testing, treating, and follow-up needs to be conducted in a patient-friendly manner. 4. Follow-up assessments need to be simplified to increase both physician and patient compliance. The frequency of follow-up visits and the time commitment per visit need to be minimized. 5. Well-defined interventions to increase pa-
tient accrual to clinical trials should be designed and assessed. These interventions need to account for differences between physician groups. They must enhance aspects of the physician's belief system that are positive toward clinical trials, so that these beliefs may be translated into proactive accrual. If up to 78% of physicians truly believe that patients receive better care when participating in a clinical trial, as suggested from this survey, then eliminating barriers to patient entry should become an important cancer treatment priority. ACKNOWLEDGMENT We are indebted to Dr Nancy Cairns for review of the manuscript, Patti Jelen and Carole Johnson for their technical assistance, and Sharon Talarek and Sandra GulaGleason for assistance in the preparation of the manuscript.
REFERENCES 1. Kemp N, Skinner E, Toms J: Randomized clinical trials of cancer treatment: A public opinion survey. Clin Oncol 10:155-161, 1984 2. Cassileth BR, Lusk EJ, Miller DS, et al: Attitudes toward clinical trials among patients and the public. JAMA 8:968-970, 1982 3. Yee JY, Breaux SR: Accrual of radiotherapy patients to clinical trials. Cancer 52:1014-1016, 1983 4. Begg CB, Zelen M, Carbone PP, et al: Cooperative groups and community hospitals: Measurement of impact in the community hospitals. Cancer 52:1760-1767, 1983 5. Martin JF, Henderson WG, Zacharski LR: Accrual of patients into a multihospital cancer clinical trial and its implications on planning future studies. Am J Clin Oncol 7:173-182, 1984 6. Taylor KM, Margolese RG, Soskolne CL: Physicians' reasons for not entering patients in a randomized clinical trial of surgery for breast cancer. N Engl J Med 310:13631367,1984 7. Wittes RE, Friedman MA: Accrual to clinical trials. J Natl Cancer Inst 12:884-885, 1988 8. Accrual picking up, still nothing to write home about: unfunded MDs "subsidize entire program." Cancer Lett 14:1-5, 1988 9. Gelber RD, Goldhirsch A: Can a clinical trial be the treatment of choice for patients with cancer? J Natl Cancer Inst 80:886-887, 1988 10. Begg CB: Selection of patients for clinical trials. Semin Oncol 15:434-440, 1988 11. Friedman MA: Patient accrual to clinical trials. Cancer Treat Rep 71:557-558, 1987 (editorial) 12. Proceedings of The National Cancer Advisory Board 65th Meeting, Dept of Health and Human Services, Public Health Service, National Institutes of Health, National Cancer Institute, Bethesda, MD, February 1-3, 1988
13. Cancer and AIDS patient advocates list concerns on drug development. Cancer Lett 15:8, 1989 14. NCI mounts campaign to increase number of patients in clinical trials. Oncology 2:50-57, 1988 15. Carbone PP, Zelen M, Fisher B, et al: The present status and future outlook of clinical research in the community. Oncology Issues 4:8-12, 1989 16. The Toronto Leukemia Study Group: Results of chemotherapy for unselected patients with acute myeloblastic leukemia: Effect of exclusions on interpretation of results. Lancet 1:786-788, 1986 17. McCusker J, Wax A, Bennett JM: Cancer patient accessions into clinical trials. Am J Clin Oncol 5:227-236, 1982 18. Hunter CP, Frelick RW, Feldman AR, et al: Selection factors in clinical trials: Results from the community clinical oncology program physician's patient log. Cancer Treat Rep 71:559-565, 1987 19. Lee JY, Marks JE, Simpson JR: Recruitment of patients to cooperative group clinical trials. Cancer Clin Trials 3:381-384, 1980 20. Taylor KM, Shapiro M, Soskolne CL: Physician response to informed consent regulations for randomized clinical trials. N Engl J Med 60:1415-1420, 1987 21. SAS/STAT User's Guide, Version 6, fourth edition. Volume 1. Carey, NC, SAS Institute Inc, 1990 22. Begg CB, Engstrom PF: Eligibility and extrapolation in cancer clinical trials. J Clin Oncol 5:962-968, 1987 23. NCI, ASCO launch campaign to increase clinical trials participation. Oncol Commen 1988, pp 2-10 24. Antman K, Schnipper L, Frei E: The crisis in clinical research: Third-party insurance and investigational therapy. N Engl J Med 1:46-48, 1988
Downloaded from ascopubs.org by Boston University on April 7, 2019 from 128.197.229.194 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
Oncologists' Reluctance to Accrue Patients Onto Clinical Trials: An Illinois Cancer Center Study By Al B. Benson III, Janet Prokop Pregler, Judy A. Bean, Alfred W. Rademaker, Bruce Eshler, and Karen Anderson Clinical trials are recognized as the standard of care for the cancer patient, and the randomized, controlled trial represents the most definitive method to determine the effectiveness or ineffectiveness of a cancer treatment. However, less than 3% of all eligible patients enter a clinical trial. Of the 437 physician members of the Illinois Cancer Center (ICC), 244 responded to a survey designed to determine factors that present a significant barrier to entering patients on clinical trials. Rigid protocol design was the primary deterrent to accrual, especially for medical oncologists. Surgeons, radiation oncologists, and medical oncologists differed with respect to several factors, including willingness to seek a clinical trial, tendency to treat patients off study, quality-of-life issues, and the belief that trials were too excessive in time commitment (P < .05). Compared with hospital-based physicians, community oncologists had fewer patients on trial,
IN
RECENT YEARS, the prospective, randomized, controlled clinical trial has become the proving ground for new medical therapies. To be accepted as the standard of care by the medical community, treatment efficacy must be shown in a well-conducted, randomized trial. An important factor limiting the availability of promising new therapies has been inadequate accrual of patients onto clinical studies. Surveys of the general public have shown widespread support for the concept of clinical trials1 '2; however, fewer than 20% of cancer patients in hospitals with available research protocols actually participate. 3' 4 Nationwide, only about 3% of cancer patients receive treatment in a clinical trial.5 Many investigators, including the leadership of the National Cancer Institute (NCI), have publicly emphasized that cancer clinical trials represent a standard of care for the cancer patient and should be viewed as the treatment of choice for most stages of the various types of cancer.6 2 In addition, the current Director of the NCI has stated: "It is the strongly held view of the NCI that in many cases experimental therapy is the patient's best hope for survival."' 3 Of the 7 million patients with a diagnosis of cancer each year, only 25,000 are entered onto an NCI-approved study." There-
were more likely to enter patients on the basis of age, and were more concerned about aspects of informed consent and the financial burden of a trial (P < .01). One third of the physicians never pursued a clinical trial because of conflict with the priorities of individual care and excessive follow-up time. Fourteen percent indicated that they discouraged patients from participating in a clinical trial due to the risk of a patient receiving a placebo and patient follow-up requirements (P < .05). Subgroups of physicians differ in their reluctance to accrue patients, and there are clusters of beliefs expressed by physicians concerning their clinical trial activity. Current conduct of clinical trials needs to be reassessed, and intervention studies are required to determine the best methodology to alter physician reluctance to pursue clinical trials. J Clin Oncol 9:2067-2075. © 1991 by American Society of Clinical Oncology. fore, eligible patients may not be obtaining appropriate adjuvant chemotherapy, particularly for breast and colon cancer, and patients may not be receiving chemotherapy at proper dosages or time intervals. In an attempt to increase the total number of patients entered onto clinical trials, the NCI has initiated new programs directed towards physicians already involved in clinical cancer research. 7' 9,12'14 "High-priority clinical trials" have
been designated by the NCI and involve intergroup collaborations. The cooperative groups have
also been encouraged to improve quality control,
From the Division of Hematology/Oncology,Department of Medicine, the CancerCenter Biometry Section, and the Medical School, Northwestern University Medical School, Chicago; and the Illinois CancerCenter, Chicago,IL. Submitted December 13, 1990; accepted May 24, 1991. Supported by a grant from the National Cancer Institute, NationalInstitutes of Health (N02 P30-CA21742). Presented in part at the 25th Annual NationalMeeting of the American Society of Clinical Oncology, San Francisco, CA, May 21-23, 1989. Address reprint requests to Al B. Benson III, MD, Illinois Cancer Center. 200 S Michigan Ave, Suite 1700, Chicago, IL 60604-2404. c 1991 bvAmerican Society of Clinical Oncology. 0732-183X/91/0911-0007$3.00/0
Journal of Clinical Oncology, Vol 9, No 11 (November), 1991: pp 2067-2075
Downloaded from ascopubs.org by Boston University on April 7, 2019 from 128.197.229.194 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
2067
2068
BENSON ET AL
minimize ineligible patients, and reduce complexities of clinical trials by limiting data collection and patient testing. Physicians are being urged to use the NCI's Physician Data Query (PDQ) as a primary information source listing available clinical trials. There are efforts to increase public awareness of trials and to discuss insurancerelated economic issues. Previous investigators have identified specific obstacles to accrual of patients to clinical trials. The Southwest Oncology Group (SWOG) surveyed 54 participants in the Cooperative Group Outreach Program (CGOP). The results of this survey" indicated that a primary obstacle to community physician participation in clinical trials is the lack of studies that meet the requirements of the physician's patient population. Eligibility requirements, particularly for elderly patients, are often too rigid. The Toronto Leukemia Study Group examined the effect of rigid exclusion criteria on the calculation of remission duration and survival rates after combination chemotherapy for acute myelogenous leukemia.' 6 The overall remission rate varied 40%, depending on exclusion criteria applied to determine which cases would be analyzed. Investigations have shown that, apart from protocol exclusion criteria, physician refusal to enroll patients onto a trial is the single most important factor in low protocol participation.4'I"7
In one
report, physician refusal accounted for 52% of all failures to enroll eligible patients. Physician refusals outnumbered patient refusals by a margin of two to one.5 6
Only one published study directly examines physician reasons for excluding eligible patients from a randomized clinical trial. The survey of 94 National Surgical Adjuvant Breast and Bowel Project (NSABP) principal investigators used a self-administered, primarily open-ended questionnaire that asked physicians why they were not entering eligible patients onto an NSABP trial. Responses indicated that issues such as the doctorpatient relationship in a randomized trial, concerns about informed consent, and physicians' reluctance to discuss openly their uncertainty about the correct treatment all contributed to physician refusals to enroll eligible patients. A follow-up survey discussed breast cancer specialists' attitudes toward informed consent."' There are special issues inherent in the treatment
of breast cancer and in clinical trials of breast cancer therapies, including the historically emotional debate over the best treatment and the specific legal requirements in some states for full disclosure of all treatment options to all patients. Thus, attitudes toward clinical trials expressed by physicians who treat breast cancer primarily cannot be assumed to represent those physicians who treat other malignancies. To investigate the reasons physicians do not place patients on clinical trials further, 437 physician members of the Illinois Cancer Center (ICC) were surveyed. The purpose of the survey was to identify physician opinion regarding factors affecting accrual to clinical trials and to assess the extent to which these factors were deterrents to patient accrual. METHODS Subjects The ICC. the NCI Designated Consortium Research Center for the State of Illinois, has extensive experience in conducting and coordinating multiinstitutional research in clinical trials and cancer control. Membership includes 437 Illinois physicians representing medical oncology, hematology, radiation oncology, surgical oncology including otolaryngology, thoracic, and colorectal surgery, and gynecologic oncology. Physicians who were members as of February 1988 are included in the survey. Each of the 437 physicians was sent a 50-question survey. After 1 month, telephone follow-up of nonresponders was attempted, and a second questionnaire was sent.
Questionnaire The survey instrument was a five-page self-administered questionnaire revised after pilot testing among physicians at Northwestern University Medical School. Fifty questions were presented in four parts. The first section contained nine closed- and open-ended questions to determine the physician-respondent's educational background, patient population, self-appraised level of participation in randomized clinical trials, and type of practice. Physicians also were asked whether they were on the faculty of a medical school. Eleven true-false statements followed, eliciting opinions regarding informed consent, adjuvant trials, the treatment of patients according to one arm of an ongoing protocol without formally entering the patient, patient age, and the efficacy of standard therapies as barriers to the enrollment of otherwise eligible patients. Physicians were asked whether they actively sought trials for their patients and whether they ever encouraged (discouraged) their patients to enroll in trials. The third section consisted of 24 closed-ended questions inquiring whether a particular statement is true for nearly all, some, or none of the patients eligible for clinical trials. Physicians also were asked to indicate if any of the issues mentioned had been the major reason they did
Downloaded from ascopubs.org by Boston University on April 7, 2019 from 128.197.229.194 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
2069
CLINICAL TRIALS PATIENT ACCRUAL not enter a patient onto a trial. Factors addressed included informed consent, protocol design, and the impact of trials on patients and physicians. Physicians were encouraged to provide written comments about topics mentioned in the questionnaire and about clinical trials in general.
StatisticalAnalyses Frequency of responses and percentages were calculated using the Statistical Application System (SAS) statistical 2 2 package. ' The X test with correction for continuity when appropriate was used to assess associations between variables. Statistical significance was indicated by a P value < .05.
RESULTS
The response rate after two mailings and a telephone reminder was 58% (252 of 437). Nonresponding physicians were similar in proportion to responding physicians in terms of expertise (medical, surgical, or radiation oncologist; P = .47) and type of practice (community, hospital; P = .75). There was a lower percentage of universityaffiliated physicians in nonresponders compared with responders (49% v 78%; P < .0001). Eight of the 252 responding physicians were retired. As the forms were largely blank for these retired physicians, their responses were not included in the data set. Results are presented for the remaining 244 practicing oncologists. The number of responses across questions varied because not everyone answered each item. The typical respondent had completed training at a university teaching hospital, had fellowship training, and had some degree of contact with patients on a clinical trial during postgraduate clinical rotations. Respondents included 128 hematologists and medical oncologists (52%), 60 surgeons (25%), and 47 radiation oncologists (19%). Nine physicians (4%) physicians did not indicate a specialty. Sixty-six percent (160 of 244) of the physicians were members of a group directly involved in trials, whereas 11% (26 of 244) said they must make an independent effort to seek out a trial for their patients. Although not affiliated with a clinical trial group, 19% (47 of 244) had regular contact with physicians directly involved in trials. Ninety-three physicians (38%) were community-based, 144 (59%) were hospital-based, and seven (3%) did not indicate type of practice. Seventy-seven percent (189 of 244) were affiliated with a university, and 22% (53 of 244) were not. Sixty-four percent (156 of 244) indicated that 1% to 25% of their cancer patients were on trials.
Only 18% (44 of 244) responded that over 26% of their patients were participating. Sixty-three percent of respondents stated they were actively involved in finding protocols for their cancer patients. Although 77% usually encourage patients who are eligible to enter a randomized trial, 15% indicated that they at times discourage eligible patients from participating. Eight of the 11 true-false statements and all of the 24 closed-ended items were analyzed by determining the frequency in each response category. Individual statements were grouped by content area into five categories (informed consent, study design, physician-patient relationship, effect of trials on patients, and specific difficulties; Tables 1 to 5). One issue that may deter physicians from entering patients onto clinical trials is the necessity to obtain informed consent (Table 1). Nearly all respondents (98%) believe that informed consent should be required; however, 57% prefer a method of consent adapted to each individual patient. Forty percent believe that informed consent is obtained not to protect the patient but for legal, societal, or other reasons. Eighty-seven percent of respondents (200 of 230) stated that informed consent is difficult to obtain, although only 23% (49 of 209) indicated this had been a reason for not entering patients onto clinical trials. A small percentage indicated that informed consent was Table 1. Frequency of Responses to Statements Regarding Informed Consent for RCCT (n = 231-240) Always or Sometimes
Statement Requirement of written informed consent before entry onto an RCCT is justified. Is difficult to obtain. Gives more information than patients desire. Depresses patients. Pressures patients into making decisions they prefer their doctor made. Gives information that causes patients not to choose the best therapy. Prefer to adapt consent for each patient. Obtained for legal, societal, or other reasons Designed to control the actions of physicians.
Always True
True %)
(%)
98 87
84 6
83 81
19 13
78
14
77 57*
10
40* 10*
Abbreviation: RCCT, randomized cancer clinical trials. *The percentage responding "true" on true-false question.
Downloaded from ascopubs.org by Boston University on April 7, 2019 from 128.197.229.194 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
2070
BENSON ET AL
designed to control the actions of physicians. Remaining statements, including "informed consent may give patients more information than they desire" and "informed consent depresses patients," were true for at least some patients in the opinion of most physicians. For the six statements in Table 1, the percentage of physicians who indicated that these were major deterrents to patient accrual ranged from 8% to 23% (average, 14%). Eight statements concerning clinical trial design and physician responses are presented in Table 2. Eighty-two percent of respondents believe some or all protocols are too rigidly designed; 36% (73 of 203) stated rigid design had been a reason for excluding patients from trials. Fifty-five percent treat patients eligible for a trial in accordance with one arm of the protocol without formally entering the patient onto the trial. Half of the physicians (51%) reported excluding patients on the basis of age. Twenty-nine percent stated that standard treatment for some malignancies resulted in such high cure rates that they would not enter patients in a trial regardless of its design. Most physicians are not at ease when talking with patients about the choice of a treatment (Table 3). They also believe that research protocols undercut patients' beliefs in physician knowl-
Table 2. Frequency of Responses to Statements Regarding the Design of RCCT (n = 228-239)
Statement
When a protocol is available, physician finds the design scientifically sound. Difficult to get patients onto RCCT because protocols are too rigidly designed. Dislike entering patients on RCCT when there is a chance they will receive a placebo. Will treat patients eligible for a trial in accordance with one arm of protocol without entering the person Age of patient alone sometimes reason for exclusion. Cure rate for standard treatment is so high physician would not enter patients regardless of the design Age of patient alone sometimes reason for inclusion into RCCT Would not enter patients onto an adjuvant study regardless of its design.
Always or Sometimes True (%)
Always True
99
40
82
12
56
(%)
12
55* 51 *
29* 28*
*The percentage responding "true" on true-false question
Table 3. Frequency of Responses to Statements Regarding Physician and Patient Relationship (n = 232-233)
Statement
Always or Sometimes True (%)
Always True
55
6
51
2
35
1
28
4
Discussions concerning which treatment is the right one are uncomfortable. Patient's belief in physicians' knowledge and autonomous decisionmaking power is undermined. Patients feel that physician wants to "get rid of a hopeless case" if (s)he suggests entry onto RCCT Physician feels uncomfortable when patient is a research subject.
(%)
edge and decision-making power. Nonetheless, these issues were not detriments to accrual. Seven percent of respondents identified uneasiness in discussing treatment options as a reason for not entering patients. Only 5% indicated that research undermining patients' beliefs in their knowledge and decision-making power had prevented them from entering a patient on trial. Table 4 summarizes the negative aspects that most physicians believe affect patients involved in clinical trials. Ninety-one percent indicated that trials are inconvenient for patients, and 79% perceive that research is costly to participants. Inconvenience and high financial burden for patients involved in clinical trials were cited by 29% and 23% of respondents, respectively, as factors that had caused them not to place patients on trial. Most physicians (78%), however, believed that Table 4. Frequency of Responses to Statements Regarding Physicians' Perception of the Effect of Trials on Patients (n = 229-235)
Statement
Always True or Sometimes True (%)
Always True
91
10
79 78
7 27
64
3
62
3
61
3
33
1
RCCT are inconvenient for patients. RCCT impose high financial burden on the patient. Patients receive better care in RCCT Goals of RCCT conflict with priorities of individual patient care. Participation in RCCT reduces the amount of quality time a patient has before death. Patients become depressed when they are on RCCT Patients receive inferior care on RCCT
Downloaded from ascopubs.org by Boston University on April 7, 2019 from 128.197.229.194 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
(%)
2071
CLINICAL TRIALS PATIENT ACCRUAL
patients always or sometimes received better care on a trial. One third stated that trials always or sometimes resulted in inferior care. Nearly three fourths of the physicians (73%) stated that excessive time is required for patient follow-up on study (Table 5). Twenty-six percent (54 of 205) cited this as a reason they had not entered patients onto trials. More than half (54%) believe that once a patient is participating in a protocol, there is sometimes or always insufficient support for adequate follow-up. In further analysis, questions were crosstabulated by expertise of physicians to ascertain differences among the three physician groups (medical, surgical, and radiation oncologists). Seventy-seven percent of the medical oncologists reported attempting to find clinical trials for their patients, compared with 53% of surgeons and 38% of radiation oncologists (P < .001). Surgeons and radiation oncologists had greater preference for individual informed consent rather than one standard form (P = .003). The radiation oncologists were most likely to state that, for some tumors, the standard treatment has such a high cure rate that they would not enter patients onto a trial (P = .009). Sixty-seven percent of the radiation oncologists, 59% of medical oncologists, and 37% of surgeons indicated they sometimes use protocol guidelines to treat patients without formally enrolling them (P = .004). Only 2% of the medical oncologists would not enter patients onto an adjuvant study, compared with 15% and 13% for surgeons and radiation oncologists, respectively (P = .016). Significant differences (P < .05) between radiation oncologists, surgeons and medical oncologists
Table 5. Frequency of Responses to Statements Regarding Difficulties in Trial Procedures (n = 230-235)
Statement
Increased time required to observe patients on RCCT is excessive. Feels responsible to offer some form of treatment. Has difficulty getting support to observe treatment plan when patient is on RCCT. Feels responsible if a patient is not on the treatment arm shown to be superior
Always or Sometimes True (%)
Always True
73
18
59
16
54
7
25
3
(%)
were found in responses (always, sometimes, never true) to eight statements (Table 6). A higher percentage of radiation oncologists always or sometimes feels uncomfortable when their patients are on a clinical trial (51%) when compared with surgeons (29%) and medical oncologists (20%). Similarly, more radiation oncologists have at least some difficulty following through with treatment plans for patients on trials (71% v 47% and 51%). Radiation oncologists also feel that there is at least some conflict between clinical trials and individual patient care (79% v 51% and 62%). More surgeons felt that clinical trials were used to "get rid of a hopeless case" (52%) than did radiation oncologists (36%) or medical oncologists (27%). Fewer surgeons said that follow-up time was excessive (59% v 76%). When medical oncologists were compared with radiation oncologists and surgeons, fewer medical oncologists indicated that the following statements were true in part: discussions about treatment uncertainty were uncomfortable (47% v 65% and 64%), clinical trials reduce the patient's quality time (53% v 71% and 70%), and patients on trials become depressed (53% v 66% and 75%). The groups demonstrated significant differences (P < .05) in the frequency with which they endorsed the following five reasons for not entering patients onto trials (Table 7): (1) trial designs are too rigid, (2) trials place a high financial burden on patients, (3) trials require an excessive amount of time to observe a patient, (4) placebocontrolled trials are unacceptable, and (5) clinical trials are inconvenient for patients. A higher percentage of medical oncologists compared with the other two groups of physicians felt that these five factors were important obstacles and hindered their entering of patients. Community physicians were compared with hospital-based physicians. Twenty-four percent of hospital-based doctors had more than 25% of their patients on protocol, as compared with only 9% of community physicians (P = .002). Community physicians were more likely to include patients on trials on the basis of age by a margin of 39% to 21% (P = .004). More community physicians entered patients on trials when the standard cure rate was high (39% v 23%; P = .013). Significant differences (P < .01) between community and hospital physicians were found in responses (always, sometimes, never true) to six
Downloaded from ascopubs.org by Boston University on April 7, 2019 from 128.197.229.194 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
2072
BENSON ET AL Table 6.
Differences Between Physician Expertise Groups (P < .05) Radiation Oncologists (n =45-47)
Statement
Surgeons (n = 53-56)
Medical Oncologists (n= 122-125)
AorS (%)
A (%)
AorS (%)
A (%)
AorS (%)
A (%)
51 71 79 36 76
2 11 2 0 23
29 47 51 52 59
7 4 5 2 9
20 51 62 27 76
3 6 2 0 20
65 71 66
4 4 6
64 70 75
11 4 2
47 53 53
4 1 1
Physician feels uncomfortable when patient is on a trial. Support for follow-up is difficult. Clinical trials conflict with patient care. Clinical trials are for "hopeless cases." Follow-up time is excessive. Discussions about uncertain treatments are uncomfortable. Clinical trials reduce patients' quality time Patients on trials become depressed.
Abbreviations: A or S, always or sometimes true; A, always true.
statements (Table 8). More hospital physicians stated that patients always receive better care in clinical trials (35%
v 14%). More community
sions that they would rather the physician made
(P = .01); (2) discussions concerning uncertainty as to which treatment is better are uncomfortable
physicians stated that excessive time always is
(P = .014); and (3) trials impose a high financial
required to observe patients on trials (29% v 11%), that clinical trials are always too rigidly
burden on patients (P = .015).
designed (19% v 7%), that informed consent always is difficult to obtain (12% v 1%), that they always or sometimes feel responsible if the patient is randomized to an arm that eventually is not shown to be superior (31% v 21%), and that they always dislike placing a patient on a placebo arm
younger patients (11% v 38% had at least 65% of their patients aged 65 years or more), were members of an organization directly involved in trials (75% v 42%), and tended to place more than 25%
(19% v 6%). However, the groups did not differ when asked if these factors presented obstacles to
putting patients on trials. Finally, a higher percentage of community physicians indicated that they did not enter patients onto research protocols (Table 9) because (1) informed consent pressures patients to make deciTable 7. Significant Differences Between Physician Expertise Groups on Reasons for Not Entering Patients Onto Clinical Trials (P < .05) Radiation Medical Oncologists Oncologists Surgeons (n - 39-41) (n= 48-50) (n = 109-112)
Reason Clinical trials are too rigidly designed. Clinical trials impose a high financial burden on patients. Follow-up time is excessive. Physician dislikes trials with placebo arm. Clinical trials are inconvenient for the patient
23
20
47
15
10
31
20
13
33
13
17
30
27
16
35
NOTE. Data are percentages of physician expertise group indicating this reason.
University-affiliated physicians tended to treat
of their patients on trials (23% v 2%; all P < .001).
They also were more likely to seek out trials actively for their patients (68% v 47%; P = .009) and they were less likely to exclude patients from
trials because of high standard cure rates (25% v 42%; P = .03). University-affiliated physicians also included more surgical oncologists (28% v 16%) and fewer radiation oncologists (16% v 32%; P = .03). Significant differences (P < .05) between university and nonuniversity physicians were found in response (always, sometimes, never true) to nine statements (Table 10). For all but one of
Table 8. Differences Between Physician Practice Groups (P < .01) Community Hospital (n = 87-90) (n = 138-141) Statement
AorS
A
AorS
A
Patients receive better care on trials. Follow-up time is excessive. Clinical trials are too rigidly designed. Informed consent is difficult to obtain. Dislike patient not being randomized to superior arm. Physician dislikes trials with placebo arm.
68 78 88 90
14 29 19 12
83 69 78 86
35 11 7 1
31
7
21
0
60
19
53
6
NOTE. No statements had 01 < P < .05.
Downloaded from ascopubs.org by Boston University on April 7, 2019 from 128.197.229.194 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
2073
CLINICAL TRIALS PATIENT ACCRUAL Table 9. Significant Differences Between Physician Practice Groups on Reasons for Not Entering Patients Onto Clinical Trials (P < .05) Community (n = 80-82)
Reason Patients are pressured into decisions they want their physican to make. Discussions about uncertain treatments are uncomfortable. Clinical trials impose a high financial burden on patients.
Hospital (n = 119-122)
21
7
14
3
33
17
NOTE. Data are percentages of physician practice group indicating this reason.
these statements, university physicians indicated lower rates of these being always true. Fewer nonuniversity physicians indicated that clinical trials always or sometimes reduced patients' quality time. For each of the 24 closed-ended questions analyzed, university and nonuniversity physicians were similarly deterred from entering patients onto trials. Selected items dealing with overt recruitment or nonrecruitment were correlated with other items to obtain a better understanding of physician action. One third of the physicians in the survey do not pursue trials for their patients. Conflict between the goals of a trial and individual care and excessive time required for follow-up were reasons why these physicians did not actively enroll patients. Thirty-six respondents actively discourage paTable 10. Differences by Physician University Affiliation (P < .05) University (n = 179-184)
DISCUSSION
Nonuniversity (n = 48-51)
Statement
Aor S
A
Aor S
A
Follow-up time is excessive. Patients on trials become depressed. Clinical trials are inconvenient for the patient. Clinical trials impose a high financial burden on patients. Clinical trials reduce patients' quality time. Support for follow-up is difficult. Patients receive inferior care on trials. Informed consent gives patient more information than needed. Clinical trials are too rigidly designed.
70
13
80
37
62
1
57
10
89
6
92
24
80
4
77
19
57 51
2 4
78 66
8 16
32
0
39
4
84
16
82
31
83
9
80
22
NOTE Data are percentages.
tients from entering a trial. This group of physicians dislikes the chance that a patient may receive a placebo, believes protocols are rigidly designed, and believes that there is a lack of support for patient follow-up (P < .05). Physicians who do not enter patients onto clinical trials because there is a treatment having a high cure rate have objections to clinical research in general. Significant associations (P < .05) are found with the belief that trials result in inferior patient care, that patients became depressed when entered on a trial, and that trials reduce quality time (P < .05). These physicians again cite lack of support for follow-up. Doctors who do not enter their patients on adjuvant studies had other complaints. They stated more often that the goals of trials conflict with individualized care, and that they are uneasy with having patients in research studies. These physicians also were more likely to feel responsible if a patient is not randomized to the arm of a given study that is eventually shown to be superior. Open-ended comments at the end of the questionnaire were summarized in a descriptive manner to identify physician opinion. In general, informed consent was viewed positively. Negative opinions tended to dominate comments on study design and the effect of enrollment in trials on physicians. Comments were almost evenly divided between positive and negative on the doctorpatient relationship and the effect of trial enrollment on the patient.
Our study has identified several factors related to low patient accrual onto clinical trials. The majority of responding physicians identified negative aspects of clinical trials. These aspects pertain to informed consent (difficult to obtain, not a good vehicle for information transfer), study design (rigid exclusion criteria, placebo arms), physicianpatient relationship (physician autonomy), patients (inconvenience, high financial burden), and trial procedures (excessive follow-up time). Fewer positive aspects were identified (eg, design generally sound, better care on clinical trial). Although ethical and philosophical questions about clinical trials were of concern, practical matters such as rigid protocol design, inconvenience for patients, and extra work for physicians were critical reasons cited by the ICC physicians for not entering eligible patients onto clinical trials.
Downloaded from ascopubs.org by Boston University on April 7, 2019 from 128.197.229.194 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
2074
BENSON ET AL
Rigid design of protocols was mentioned most often as a rationale for not enrolling patients and has been confirmed by numerous other studies.3-5"1017'18"22 One survey evaluating multiinstitutional trials in breast cancer noted an average of 23 eligibility criteria per study, with considerable variation in the criteria among the studies."' 22 In the Toronto Leukemia Study Group Study,16 150 patients (55%) were excluded from calculations when all standard exclusion criteria were applied. Clearly, rigid protocol exclusion criteria not only decrease accrual, they also limit applicability of results, which become less useful to physicians who treat the entire spectrum of patients. In the ICC survey, other inflexible aspects of protocols such as frequent serial laboratory tests or radiologic testing on a specific day of a protocol were cited as cumbersome, especially in the community setting. Inconvenience for patients is a barrier to accrual that has received very little attention previously but was the second leading reason preventing physicians from entering their patients onto clinical trials. Patient inconvenience potentially covers all aspects of the clinical trial. Additional time required for protocol screening, registration, testing, treatment, and follow-up provide a cumulative deterrent for the patient, as well as for the physician. The combination of additional time commitment and costs provided the third significant reason preventing ICC physicians from entering patients on trials. These reasons have been noted anecdotally by several authors.'" Increasing accrual will require a decrease in physician reluctance to participate in clinical investigations. Our study confirms and augments the conclusions of previous investigations. While Taylor et al"2' observed higher rates for concern about the physician-patient relationship and the dislike of discussions regarding treatment uncertainty, our study showed similar levels of concern regarding randomization, clinician/scientist conflict, informed consent, and physician feeling of responsi5 bility. When compared with the SWOG survey," more ICC physicians felt that informed consent was difficult to obtain, while both surveys identified long-term follow-up as a significant problem. Representatives from the NCI have cited the following obstacles as hindering patient accrual to cancer clinical trials7 "'8 2 4: (1) unacceptability of
randomization for patients requiring individualized care; (2) large time commitment for data management and informed consent; (3) denial of insurance; (4) public misunderstanding of risks and benefits; (5) limited access for minority patients; and (6) cooperative groups tend to be political, ask irrelevant questions, and have less than ideal participation by cancer centers. Our study confirmed that issues of randomization, excessive time commitment, and financial implications were significant deterrents to accrual. The ICC's physician questionnaire suggests that focusing on concerns of particular physician groups (eg, community v hospital physicians) could contribute to the success of a clinical trials program. Excessive time commitment for clinical trials was a significant accrual factor in our sample as a whole, but especially for medical oncologists, communitybased physicians, and nonuniversity physicians. These subgroups are likely not to have the infrastructure of personnel and facilities in place to conduct clinical trials. Furthermore, physicians tend to have clusters of beliefs which may reinforce their reluctance to pursue clinical trials. The survey confirmed that community physicians when compared with hospital-based physicians were more likely to believe that trials were too rigidly designed and did not provide better care for patients. In addition, the community physicians were more likely to find a placebo arm an unacceptable alternative in a randomized study. When compared with their radiation oncology or surgical colleagues, the ICC medical oncologists were more likely to state that trials are too expensive and too rigid and that studies with a placebo arm are unacceptable. Positive opinions were identified in the survey. Greater than 75% of all physicians surveyed felt that patients received better care on clinical trials, that available protocols are usually scientifically sound, and that informed consent is justified. Favorable aspects of clinical trials such as these should be emphasized in encouraging change in accrual patterns. Based on the results of this study, we make the following recommendations to increase patient accrual to clinical trials: 1. Over 50% of physicians in this survey exclude patients from clinical trials on the basis of age. As a high percentage of cancer patients are elderly and with the growth of the aged population,
Downloaded from ascopubs.org by Boston University on April 7, 2019 from 128.197.229.194 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
2075
CLINICAL TRIALS PATIENT ACCRUAL physicians must be encouraged to consider eligible older patients for trials. 2. Excessive eligibility criteria need to be reassessed by scientists designing clinical trials. Physicians should invest their time only in trials with eligibility criteria consistent with their patient population. 3. Patient convenience when on a clinical trial is a priority. The process of scheduling, testing, treating, and follow-up needs to be conducted in a patient-friendly manner. 4. Follow-up assessments need to be simplified to increase both physician and patient compliance. The frequency of follow-up visits and the time commitment per visit need to be minimized. 5. Well-defined interventions to increase pa-
tient accrual to clinical trials should be designed and assessed. These interventions need to account for differences between physician groups. They must enhance aspects of the physician's belief system that are positive toward clinical trials, so that these beliefs may be translated into proactive accrual. If up to 78% of physicians truly believe that patients receive better care when participating in a clinical trial, as suggested from this survey, then eliminating barriers to patient entry should become an important cancer treatment priority. ACKNOWLEDGMENT We are indebted to Dr Nancy Cairns for review of the manuscript, Patti Jelen and Carole Johnson for their technical assistance, and Sharon Talarek and Sandra GulaGleason for assistance in the preparation of the manuscript.
REFERENCES 1. Kemp N, Skinner E, Toms J: Randomized clinical trials of cancer treatment: A public opinion survey. Clin Oncol 10:155-161, 1984 2. Cassileth BR, Lusk EJ, Miller DS, et al: Attitudes toward clinical trials among patients and the public. JAMA 8:968-970, 1982 3. Yee JY, Breaux SR: Accrual of radiotherapy patients to clinical trials. Cancer 52:1014-1016, 1983 4. Begg CB, Zelen M, Carbone PP, et al: Cooperative groups and community hospitals: Measurement of impact in the community hospitals. Cancer 52:1760-1767, 1983 5. Martin JF, Henderson WG, Zacharski LR: Accrual of patients into a multihospital cancer clinical trial and its implications on planning future studies. Am J Clin Oncol 7:173-182, 1984 6. Taylor KM, Margolese RG, Soskolne CL: Physicians' reasons for not entering patients in a randomized clinical trial of surgery for breast cancer. N Engl J Med 310:13631367,1984 7. Wittes RE, Friedman MA: Accrual to clinical trials. J Natl Cancer Inst 12:884-885, 1988 8. Accrual picking up, still nothing to write home about: unfunded MDs "subsidize entire program." Cancer Lett 14:1-5, 1988 9. Gelber RD, Goldhirsch A: Can a clinical trial be the treatment of choice for patients with cancer? J Natl Cancer Inst 80:886-887, 1988 10. Begg CB: Selection of patients for clinical trials. Semin Oncol 15:434-440, 1988 11. Friedman MA: Patient accrual to clinical trials. Cancer Treat Rep 71:557-558, 1987 (editorial) 12. Proceedings of The National Cancer Advisory Board 65th Meeting, Dept of Health and Human Services, Public Health Service, National Institutes of Health, National Cancer Institute, Bethesda, MD, February 1-3, 1988
13. Cancer and AIDS patient advocates list concerns on drug development. Cancer Lett 15:8, 1989 14. NCI mounts campaign to increase number of patients in clinical trials. Oncology 2:50-57, 1988 15. Carbone PP, Zelen M, Fisher B, et al: The present status and future outlook of clinical research in the community. Oncology Issues 4:8-12, 1989 16. The Toronto Leukemia Study Group: Results of chemotherapy for unselected patients with acute myeloblastic leukemia: Effect of exclusions on interpretation of results. Lancet 1:786-788, 1986 17. McCusker J, Wax A, Bennett JM: Cancer patient accessions into clinical trials. Am J Clin Oncol 5:227-236, 1982 18. Hunter CP, Frelick RW, Feldman AR, et al: Selection factors in clinical trials: Results from the community clinical oncology program physician's patient log. Cancer Treat Rep 71:559-565, 1987 19. Lee JY, Marks JE, Simpson JR: Recruitment of patients to cooperative group clinical trials. Cancer Clin Trials 3:381-384, 1980 20. Taylor KM, Shapiro M, Soskolne CL: Physician response to informed consent regulations for randomized clinical trials. N Engl J Med 60:1415-1420, 1987 21. SAS/STAT User's Guide, Version 6, fourth edition. Volume 1. Carey, NC, SAS Institute Inc, 1990 22. Begg CB, Engstrom PF: Eligibility and extrapolation in cancer clinical trials. J Clin Oncol 5:962-968, 1987 23. NCI, ASCO launch campaign to increase clinical trials participation. Oncol Commen 1988, pp 2-10 24. Antman K, Schnipper L, Frei E: The crisis in clinical research: Third-party insurance and investigational therapy. N Engl J Med 1:46-48, 1988
Downloaded from ascopubs.org by Boston University on April 7, 2019 from 128.197.229.194 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
