- Email: [email protected]
Sociodemographic Analysis of Patients in Radiation Therapy Oncology Group Clinical Trials
Int. J. Radiation Oncology Biol. Phys., Vol. 40, No. 1, pp. 9 –15, 1998 Copyright © 1998 Elsevier Science Inc. Printed in the USA. All rights reserved 0360-3016/98 $19.00 1 .00
PII S0360-3016(97)00833-X
●
Clinical Investigation SOCIODEMOGRAPHIC ANALYSIS OF PATIENTS IN RADIATION THERAPY ONCOLOGY GROUP CLINICAL TRIALS ROBERT M. CHAMBERLAIN, PH.D.,* KATHRYN A. WINTER, M.S.,‡ SRINIVASAN VIJAYAKUMAR, M.D.,§ ARTHUR T. PORTER, M.D.,\ M. ROACH, III, M.D.,¶ OSCAR STREETER, M.D.,# JAMES D. COX, M.D.† AND MELISSA L. BONDY, PH.D.* Departments of *Epidemiology and †Radiation Oncology, The University of Texas M. D. Anderson Cancer Center, Houston, TX; ‡ Radiation Therapy Oncology Group, Philadelphia, PA; §Department of Radiation Oncology, University of Chicago, Chicago, IL; \ Department of Radiation Oncology, Wayne State University, Detroit, MI; ¶Department of Radiation Oncology, University of California, San Francisco, CA; #Department of Radiation Oncology, University of Southern California, Los Angeles, CA Purpose: To assess the degree to which the sociodemographic characteristics of patients enrolled in Radiation Therapy Oncology Group (RTOG) clinical trails are representative of the general population. Methods and Materials: Sociodemographic data were collected on 4016 patients entered in 33 open RTOG studies between July 1991 and June 1994. The data analyzed included educational attainment, age, gender, and race. For comparison, we obtained similar data from the U.S. Department of Census. We also compared our RTOG data with Surveillance Epidemiology and End Results (SEER) data for patients who received radiation therapy, to determine how RTOG patients compared with cancer patients in general, and with patients with cancers at sites typically treated with radiotherapy. Results: Overall, the sociodemographic characteristics of patients entered in RTOG trials were similar to those of the Census data. We found that, in every age group of African-American men and at nearly every level of educational attainment, the proportion of RTOG trial participants mirrored the proportion in the census data. Significant differences were noted only in the youngest category of African-American men, where the RTOG accrues more in the lower educational categories and fewer with college experience. For African-American women, we found a similar pattern in every age group and at each level of educational attainment. As with men, RTOG trials accrued a considerably larger proportion of younger, less educated African-American women than the census reported. Using SEER for comparison, the RTOG enrolled proportionately more African-American men to trials all cancer sites combined, and for prostate and head and neck cancer. In head and neck trials, the RTOG enrolled nearly twice as many African-American men than would be predicted by SEER data. In lung cancer trials, RTOG underrepresented African-American men significantly; however, there was no difference for brain cancer trials. There were no racial differences in RTOG accrual and SEER incidence data for women on trials in brain, lung, and head and neck cancer. However, the RTOG trials accrued nearly twice the proportion of African-American women in cervical cancer trials and in all sites combined, compared to the SEER data. Conclusions: Comparisons with the U.S. Census and SEER show that African-Americans are proportionally well represented in cancer clinical trials conducted by the Radiation Therapy Oncology Group. The comparative analysis indicates that all educational levels in each age group of African-Americans generally mirror the U.S. Census, with one exception. The exception is a significant overrepresentation of less-educated African-Americans in the youngest age category. This exception is counter to the expectation that better-educated patients are more likely to enroll in trials. When compared with SEER data, the RTOG trials either parallel or overrepresent African-American men and women, with the only exception being in lung cancer, where men are underrepresented. These results show that, in comparison to the Census and SEER data, the RTOG has fulfilled its commitment to enroll African-American patients in its clinical trials. © 1998 Elsevier Science Inc. Cancer clinical trials, Minority representation, African-Americans, Cooperative groups, Recruitment, Accrual, Enrollment.
INTRODUCTION
America. RTOG conducts multicenter clinical trials that evaluate the effects of ionizing radiation in the treatment of cancer. Over the last 15 years, it has enrolled over 20,000 patients in clinical studies that have resulted in significant advances in the understanding and treatment of cancer.
The Radiation Therapy Oncology Group (RTOG), formed in 1971, has approximately 260 full and affiliate members and includes most of the major cancer centers in North
R.N., R.T.T., at RTOG Headquarters and to Margaret R. Spitz, M.D., Virginia L. Ernster, Ph.D., Katherine A. McGlynn, Ph.D. and M. Elizabeth Hammond, Ph.D. of the RTOG Epidemiology Committee for their comments on the data tables and figures. Accepted for publication 30 September 1997.
Reprint requests to: Melissa L. Bondy, Ph.D., Department of Epidemiology, Box 189, The University of Texas M. D. Anderson Cancer Center, 1515 Holcombe Blvd., Houston, TX 77030. Acknowledgements—This work was supported in part by Public Health Service grants CA21661, CA37422 and CA32115. The authors are indebted to Thomas F. Pajak, Ph.D. and JoAnne Stetz, 9
10
I. J. Radiation Oncology
●
Biology
●
Physics
RTOG operates with the principle that cancer therapy can be significantly improved by subjecting promising interventions to carefully designed clinical trials, culminating in phase III trials in which new treatments can be tested against standard therapies. A premise of this concept is that trial results should be applicable to the population served and so transferable to clinical practice. To ensure that patients enrolled in a clinical trial should represent the general population to which the trial results will be applied, the sociodemographic profile of those patients should be similar (1–3). Findings from trials with insufficient numbers of minority and women participants cannot be applied with confidence to the general population (4 –7). These issues have been a concern to the National Cancer Institute (NCI), and cooperative groups, such as the RTOG, have been mandated to assess their compliance with this goal. The National Institute of Health (NIH) recently established guidelines enforcing the inclusion of minority and underserved populations in NIH-funded clinical trials. These guidelines evolved from a series of NIH policies and guidelines, beginning in 1985 with the report of the PHS Task Force on Women’s Health, followed by published policies in the NIH Guide to Grants and Contracts in subsequent years. The general purpose of these policies and guidelines was to encourage researchers to ensure that research results and benefits would be applicable to women and minority persons. In 1993, Congress passed into law the NIH Revitalization Act (Public Law 103-43), these existing NIH policies and guidelines. In response to these guidelines and to NCI special initiatives, investigators have directed considerable effort to increase the proportions of women and minority participants in cancer treatment trials and have generally achieved a racial/ethnic composition that parallels the composition of the general population (8, 9). The RTOG has initiated 12 prostate cancer trials, involving more than 5500 men. The treatment results of 8 trials have been published. Another one closed recently, and the remaining 3 are currently accruing patients. However, information about race has been difficult to glean from the resulting data. Even when trials enroll proportions of minority participants similar to those in the general population, the small numbers of these participants often preclude a rigorous statistical comparison. To overcome this problem, Roach et al. (10) used a modified meta-analysis approach, combining data from several RTOG prostate cancer trials, and showed that stagefor-stage and equivalent therapy, race was not an independent prognostic indicator of survival. The RTOG also analyzed their non-small cell lung and brain cancer studies for racial differences, and found no differences (11, 12). Recently, the NCI modified its policy to provide for ethnicbased subset analysis to be taken into account in calculating sample size in Phase III trials in which differences are thought to exist. To determine if there has been an underrepresentation of African-Americans or other bias, we evaluated the educational attainment, gender, age, and race of patients prospectively enrolled in the RTOG trials by a comparative analysis to the U.S. Census and the SEER.
Volume 40, Number 1, 1998
Table 1. Distribution of 56 RTOG studies by type
Type of protocol
Number of open protocols
Number of RTOG patients
Open only to RTOG members (New disease) (Recurrent disease) RTOG-run intergroup protocols NonRTOG-run intergroup protocols Combined protocols
33 (23) (10) 8 15 56
4016 (3375) (641) 764 752 5,532
MATERIALS AND METHODS Patient population Between July 1991 and June 1994, 5532 patients were entered into 56 protocols available to the RTOG members. Table 1 shows that 33 protocols were only open to RTOG members. Twenty-three of those protocols were for newly diagnosed cancers, whereas the other 10 protocols were for recurrent tumors. During this period, the RTOG members had access to 23 protocols in which more than one national cancer cooperative group participated. These were designated intergroup protocols and are not included in this analysis. In RTOG-run nonintergroup studies, each subject was required to complete or have completed a questionnaire soliciting information on sociodemographic information, such as age, race, highest level of educational attainment, marital status, and data regarding income before and after treatment (Fig. 1). Method of payment for treatment was also collected. For each category, patients were given the option of ‘‘not answering.’’ From the U.S. Department of Census, we obtained a data sample from the 1990 census on educational attainment by age, gender, and race (13). We compared the census data on educational attainment with the education levels of the RTOG trial patients to determine how well they represented the general population. To determine how RTOG patients compared with cancer patients in general, and with patients with cancer at sites definitively treated with radiotherapy (14), we compared our sociodemographic data on RTOG trial patients to the Surveillance Epidemiology and End Results (SEER) data for patients who received radiation therapy. Breast cancer in women was excluded because RTOG had not conducted any breast trials during the study period.
Statistical analysis We calculated descriptive statistics to compare the RTOG demographic data with both SEER and census data. Chisquared values were calculated to compare the observed RTOG accrual to the expected accrual based on the U.S. population from the census and SEER data. These chisquared values were then evaluated by using multiple comparisons, which requires adjustment of the significance level. The Hommel procedure was used to make this adjustment (15).
Analysis of patients in radiation therapy
●
R. M. CHAMBERLAIN et al.
Fig. 1. Patient questionnaire.
11
12
I. J. Radiation Oncology
●
Biology
●
Physics
Table 2. Distribution of demographic characteristics in the study population Number
Percent
848 908 1009
30.7% 32.8% 36.5%
249 300 660 1174 382
9.0% 10.8% 23.9% 42.5% 13.8%
569 2196
20.6% 79.4%
339 2426 2765
12.3% 87.7% 100.0%
Education Years of School Completed 12th Grade or Less High School Graduate Any College Age Years ,45 45–54 55–64 65–74 751 Gender Female Male Race African-American White Total
RESULTS Of the 4016 patients entered into 33 RTOG trials, 3375 patients had newly diagnosed cancer; of these, 170 (5.0%) did not complete the questionaire (Table 1). These data include adequate numbers of white and African-American patients for a comparative analysis. Other racial groups had insufficient numbers to make comparisons, 170 (5.3%) were Hispanic, 73 (2.3%) were of other races, and four (0.1%) did not answer the question. Included in this study are 2426 (87.7%) white and 339 (12.3%) African-American patients. Data for 2765 white and African-American patients with complete demographic data on all of the pertinant study variables are shown in Table 2. More than a third (36.5%) of the patients had attended college or technical school, and another third (32.8%) had graduated from high school only. Only 18 patients (0.5%) reported never attending any school. More than 99% of the individuals answered questions regarding their race and marital status. Questions related to education and birthdate were also usually answered, with 93% compliance. However, we omitted income data because questions about income level before and after cancer diagnosis were not answered by 27% and 29% of the individuals, respectively. Our figures show only data for African-Americans because, for every age group and at each level of educational attainment, the proportions of white RTOG patients were comparable to the proportions in the census and SEER data. The proportion of African-American male subjects 65 years old or older was comparable with the census data in the age and educational categories (Fig. 2). In the younger AfricanAmerican males, RTOG accrued significantly fewer with college experience and more high school graduates. In the age group 20 –54 years, 45.3% of the African-American male RTOG trial participants did not complete high school in comparison with 18.9% in the U.S. census (p 5 0.0001).
Volume 40, Number 1, 1998
Similar but not statistically significant, comparisons were observed in the 55– 64 year group (p 5 0.09). In the 20 –54 year age group, 52.3% of the AfricanAmerican female trial participants had attained less than 12th grade education compared with 17.5% in the census (p 5 0.0001) (Fig. 3). For the other age groups, the results were parallel with the census. Figure 4 shows comparisons between the patients accrued in RTOG trials and SEER incidence data by cancer sites treated with radiation as the main therapeutic modality. We compared only sites where RTOG focuses its clinical treatment efforts. For all cancer sites combined, the percentages of African-American men enrolled in RTOG trials (11.9%) was significantly higher (p , 0.015) than the SEER data (10.4%). The percentages of African-American men (12.0%) in RTOG prostate cancer trials significantly (p , 0.0002) exceeded those tabulated in the SEER data (9.2%). A similar significantly (p , 0.0001) higher representation of African-American men enrolled in RTOG head and neck cancer trials (23.1%) was observed in comparison to those tabulated by SEER (12.6%). RTOG brain cancer trials enrolled the same percentage (3.0%) of African-American men as observed in the SEER data. Only in RTOG lung cancer trials were African-American men significantly (p , 0.041) underrepresented.
Fig. 2. Percentage of African-American males enrolled in RTOG trials compared with the percentage of African-American males enumerated in the U.S. Census by age and educational attainment.
Analysis of patients in radiation therapy
●
R. M. CHAMBERLAIN et al.
13
jective was guided by the premise that clinical trial samples should reflect the population to which the results will eventually be applied. We conducted a prospective study of patients entered into the RTOG treatment protocols during a 3-year period. Overall, the patient population was 12.3% African-American and 5.0% Hispanic. Because the analysis stratified the minority subsamples by age and education, statistical requirements precluded a detailed analysis of the Hispanic group. Compared with population data from SEER and the census, our findings suggest that African-Americans and women are well represented in RTOG trials. Overall, there were more educated whites and less-educated AfricanAmericans in RTOG trials than in the general population. A further notable feature of our study is that half of the RTOG subjects were older than 65, reflecting the age distribution of cancer patients, and 30% were over 70, which may reflect the effect prostate trials have on the distribution. The results of studies of patient entry into trials have been inconsistent, but suggest that patients in clinical trials tend to be of higher performance status and higher socioeconomic status and to be younger (17–19). A Danish study
Fig. 3. Percentage of African-American females enrolled in RTOG trials compared with the percentage of African-American females enumerated in the U.S. Census by age and educational attainment.
Figure 4 shows notable racial differences among women in the RTOG accrual compared with SEER incidence data, in cancer at all sites combined (p , 0.0001) and cervical cancer (p , 0.0001). In these two categories, the RTOG trials accrued nearly twice the proportion of African-American women as reported in the SEER data. For trials in brain, lung, and head and neck cancer, RTOG accruals of African-American women did not significantly differ from SEER data. In summary, the RTOG population proportionally parallels or disproportionally favors African-American patients for a majority of the age groups and by educational attainment when compared with SEER and the census. In fact, RTOG was even more successful in recruiting AfricanAmerican patients to prostate, head and neck, and cervical cancer protocols compared with SEER statistics. DISCUSSION The initial objective of the authors was to assess the degree to which RTOG trial participants represented the minority population in general, and minority cancer patients specifically. This interest parallels the work by Tejeda and colleagues (16), in their study of minority representation in all trials funded by the National Cancer Institute. Our ob-
Fig. 4. Percentage of African-American males and females enrolled in RTOG trials compared with the percentage of AfricanAmerican males and females in SEER (1991) who received radiation therapy as primary treatment.
14
I. J. Radiation Oncology
●
Biology
●
Physics
using structured interviews of outpatients demonstrated that 98% approved of study participation and that there were no differences related to socioeconomic status (20). Interestingly, a British opinion study that surveyed 1022 people about four hypothetical protocols suggested that younger persons and those with higher socioeconomic status were less likely to participate in clinical research (21). The converse, that educational status correlates with a greater desire and ability to participate in clinical decision-making, was demonstrated in a series of 75 women with advanced breast cancer (22); youth and educational status were positively related to a willingness to challenge physician authority. Although the foregoing might imply that young AfricanAmerican patients with limited education or lower socioeconomic status would more readily enroll in clinical trials, this may not be the case in the United States. Fear of experimentation, especially within the African-American community, is particularly still prevalent and is partly based on real abuses, the most frequently cited being the Tuskegee Study of Untreated Syphilis in the Negro Male conducted by the U.S. Public Health Service between 1936 and 1972 (23). There is also evidence that the currently used informed-consent forms are too complex for some patients. In one study, 79% of 68 oncologists reviewing current consent forms judged that the forms were ‘‘much’’ or ‘‘somewhat’’ too technical (24). The question ‘‘who volunteers for clinical trials?’’ is being more frequently asked as we move toward the end of the third decade of NCI-supported clinical trials (25). In general, there are three major variables associated with enrollment into clinical trials: ● ● ●
The clinical trial characteristics; the patient characteristics; the physician characteristics.
A number of the features of clinical trials may limit patient participation. For example, protocols may not be available or offered to patients (17). Gender-specific trials will obviously limit the participation in clinical trials. For example, the RTOG has strong initiatives in prostate cancer, which affects the overall gender ratios and age distribution of RTOG subjects. The trial design itself may act as a barrier to participation, if eligibility clauses require that the subject live within a given distance from the institution or require performance of many nonreimbursed tests. Furthermore, randomized trials with notreatment arms or with very different treatment arms, such as
Volume 40, Number 1, 1998
radiotherapy vs. surgery, may be difficult to understand and, therefore, deter participation (26–29). Patients have many reasons for choosing or not choosing to participate in trials. A recent review of the literature indicates that subject perception of personal benefit must outweigh two major categories of barriers: (a) Time and hassle involved in participation; and (b) negative personal and family attitudes about trials (30). Previous studies indicate that the first category of barriers is by far the most important. Finally, physician attitudes and motivation are extremely important in the enrollment of patients into clinical trials. A survey mailed to investigators comparing segmental mastectomy to a radical mastectomy found the most frequently cited reason for patients not entering a study was perceived disruption of the doctor-patient relationship by substitution of study-imposed interventions based on randomization rather that physician judgement (9, 27, 31). A report from the NCI Clinical Cancer Oncology Program shows evidence that an investigators’ knowledge about trials and about enrolling patients in trials is a primary determinant in successful accrual of subjects (32). Investigators conducting prevention trials find the guidelines especially difficult to achieve, because few successful recruitment methods for disease-free minority populations have been developed and rigorously tested. In some of the largest prevention trials for breast and prostate cancer, fewer than 5% of the participants have been from minority populations. Despite efforts to recruit participants that accurately represent the target population, prevention study populations often do not represent the general population with respect to demographic variables such as race/ethnicity, educational attainment, gender, and age. The high proportion of minority populations accrued in RTOG treatment trials strengthens the applicability of study results to these populations. RTOG trials aim to answer questions not only about the entire study population, but also about important subgroups of patients. This commitment of the RTOG led to the initiation of the Special Populations and Epidemiology Committees, with goals to evaluate participation, study design, and relevance to women and minorities, and its efforts are effectively reducing the disparities between the makeup of treatment populations. We have shown, in this study, that the RTOG has successfully recruited minorities and women to treatment protocols and that results of these trials can be extended to the general population.
REFERENCES 1. Begg, C. B.; Engstrom, P. F. Eligibility and extrapolation in cancer clinical trials. J. Clin. Oncol. 5:962–968; 1987. 2. Mohiuddin, S. M.; Hilleman, D. E. Gender and racial bias in clinical pharmacology trials. Ann. Pharmacother. 27:972–973; 1993. 3. Svensson, C. K. Representation of American blacks in clinical trials of new drugs. JAMA 261:263–265; 1989. 4. Antman, K.; Amato, D.; Wood, W.; Corson, J.; Suit, H.; Proppe, K.; Carey, R.; Greenberger, J.; Wilson, R.; Frei, E., III. Selection bias in clinical trials. J. Clin. Oncol. 3:1142– 1147; 1985.
5. Kaluzny, A.; Brawley, O.; Garson-Angert, D.; Shaw, J.; Godley, P.; Warnecke, R.; Ford, L. Assuring access to state-ofthe-art care for U.S. minority populations: The first 2 years of the minority-based Community Clinical Oncology Program. J. Natl. Cancer Inst. 85:1945–1950; 1993. 6. King, A. C.; Harris, R. B.; Haskell, W. L. Effect of recruitment strategy on types of subjects entered into a primary prevention clinical trial. Ann. Epidemiol. 4:312–320; 1994. 7. Roberson, N. L. Clinical trial participation: Viewpoints from racial/ethnic groups. Cancer 74:2687–2691; 1994. 8. McCabe, M. S.; Varricchio, C. G.; Padberg, R. M. Efforts to
Analysis of patients in radiation therapy
9. 10.
11.
12.
13.
14.
15. 16.
17. 18.
recruit the economically disadvantaged to national clinical trials. Semin. Oncol. Nurs. 10:123–129; 1994. Taylor, K. M.; Shapiro, M.; Soskolne, C. L. Physician response to informed consent regulations for randomized clinical trials. Cancer 60:1415–1422; 1987. Roach, M.; Krall, J.; Keller, J. W.; Perez, C. A.; Sause, W. T. The prognostic significance of race and survival from prostate cancer based on patients irradiated on Radiation Therapy Oncology Group protocols (1976 –1985). Int. J. Radiat. Oncol. Biol. Phys. 24:441– 449; 1992. Graham, M. L.; Geitz, et al. Comparison of prognostic factors and survival among black patients and white patients treated with irradiation for non-small– cell lung cancer. J. Natl. Cancer Inst. 84:1731–1735; 1992. Simpson, J.; Horton J., et al. Influence of location and extent of surgical resection on survival of patients with glioblastoma multiforme: Results of three consecutive Radiation Therapy Oncology Group (RTOG) clinical trials. Int. J. Radiat. Oncol. Biol. Phys. 26:239 –244; 1993. U.S. Bureau of the Census, Current Population Survey. Educational attainment of persons 15-years-old and over, by age, sex, race and Hispanic origin. Department of Commerce, Bureau of the Census, 20-489; March 1995. Kosary, C. L.; Ries, L. A. G.; Miller, B. A.; Hankey, B. F.; Harras, A.; Edwards, B. K. (eds). SEER Cancer Statistics Review, 1973–1992: Tables and Graphs, National Cancer Institute. NIH Pub. No. 96-2789. Bethesda, MD, 1995. Hommel, G. A stagewise rejective multiple test procedure based on a modified Bonferroni test. Biometrika 75:383–386; 1988. Tejeda, H.; Green, S.; Trimble, E.; Ungerleider, R.; Friedman, M.; Brawley, O. The representation of African-Americans, Hispanics, and whites in National Cancer Institute sponsored cancer treatment trials. J. Natl. Cancer Inst. 88:812– 816; 1996. Hunter, C. P.; Frelick, R. W.; Feldman, A. R. Selection factors in clinical trials. Cancer Treat. Rep. 71:559 –565; 1987. Martin, J. F.; Henderson, W. G.; Zacharski, L. R. Accrual of patients into a multihospital cancer clinical trial and its implications on planning future studies. J. Clin. Oncol. 7:173–182; 1984.
●
R. M. CHAMBERLAIN et al.
15
19. McCusker, J.; Wax, A.; Bennett, J. M. Cancer patient accessions into clinical trials. J. Clin. Oncol. 5:227–236; 1982. 20. Saubrey, N.; Jensen, J.; Rasmussen, P. E. Danish patients attitudes to scientific and ethical questions. Acta Med. Scand. 215:99 –104; 1984. 21. Kemp, N.; Skinner, E.; Toms, J. Randomized clinical trials in cancer treatment—a public opinion survey. Clin. Oncol. 10: 155–161; 1984. 22. Llewellyn-Thomas, H. A.; McGreal, M. J.; Thiel, E. C. Patients willingness to enter clinical trials: Measuring the association with perceived benefits and preference for decision participation. Soc. Psy. Med. 32:35– 42; 1991. 23. Jones, J. H. Bad blood: The Tuskegee syphilis experiment. New York: The Free Press; 1993. 24. Penman, D. T.; Holland, J. C.; Bahna, G. F. Informed consent for investigational chemotherapy. Patients’ and physicians’ perceptions. J. Clin. Oncol. 2:849 – 855; 1984. 25. Gehan, E. A.; Schneiderman, M. A. Historical and methodological developments in clinical trials of the National Cancer Institute. Stats. Med. 9:871– 880; 1990. 26. Cassileth, B. R.; Knuiman, M. W.; Abeloff, M. D. Anxiety levels in patients randomized to adjuvant therapy vs. observation for early breast cancer. J. Clin. Oncol. 4:972–974; 1986. 27. Ganz, P. A. Clinical trials: Concerns of the patient and the public. Cancer 65:2394 –2399; 1990. 28. MacKillop, W. J.; Ward, G. K.; O’Sullivan, B. The use of expert surrogates to evaluate clinical trials in non-small cell lung cancer. Br. J. Cancer 54:661– 667; 1986. 29. MacKillop, W. J.; Ward, G. K.; O’Sullivan, B. Non-small cell lung cancer. How oncologists want to be treated. Int. J. Rad. Oncol. Biol. Phys. 13:929 – 4; 1987. 30. Swanson, G. M.; Ward, A. J. Recruiting minorities into clinical trials: toward a participant-friendly system. JNCI 87: 1747–1759; 1995. 31. Taylor, K. M.; Kelner, M. Informed consent: The physician’s perspective. Soc. Psy. Med. 24:135–143; 1987. 32. Feigl, P.; Diehr, P.; Allison, C. Community cancer care evaluation. Final report of NCI contract NO1-CN-350009; 1987.
PII S0360-3016(97)00833-X
●
Clinical Investigation SOCIODEMOGRAPHIC ANALYSIS OF PATIENTS IN RADIATION THERAPY ONCOLOGY GROUP CLINICAL TRIALS ROBERT M. CHAMBERLAIN, PH.D.,* KATHRYN A. WINTER, M.S.,‡ SRINIVASAN VIJAYAKUMAR, M.D.,§ ARTHUR T. PORTER, M.D.,\ M. ROACH, III, M.D.,¶ OSCAR STREETER, M.D.,# JAMES D. COX, M.D.† AND MELISSA L. BONDY, PH.D.* Departments of *Epidemiology and †Radiation Oncology, The University of Texas M. D. Anderson Cancer Center, Houston, TX; ‡ Radiation Therapy Oncology Group, Philadelphia, PA; §Department of Radiation Oncology, University of Chicago, Chicago, IL; \ Department of Radiation Oncology, Wayne State University, Detroit, MI; ¶Department of Radiation Oncology, University of California, San Francisco, CA; #Department of Radiation Oncology, University of Southern California, Los Angeles, CA Purpose: To assess the degree to which the sociodemographic characteristics of patients enrolled in Radiation Therapy Oncology Group (RTOG) clinical trails are representative of the general population. Methods and Materials: Sociodemographic data were collected on 4016 patients entered in 33 open RTOG studies between July 1991 and June 1994. The data analyzed included educational attainment, age, gender, and race. For comparison, we obtained similar data from the U.S. Department of Census. We also compared our RTOG data with Surveillance Epidemiology and End Results (SEER) data for patients who received radiation therapy, to determine how RTOG patients compared with cancer patients in general, and with patients with cancers at sites typically treated with radiotherapy. Results: Overall, the sociodemographic characteristics of patients entered in RTOG trials were similar to those of the Census data. We found that, in every age group of African-American men and at nearly every level of educational attainment, the proportion of RTOG trial participants mirrored the proportion in the census data. Significant differences were noted only in the youngest category of African-American men, where the RTOG accrues more in the lower educational categories and fewer with college experience. For African-American women, we found a similar pattern in every age group and at each level of educational attainment. As with men, RTOG trials accrued a considerably larger proportion of younger, less educated African-American women than the census reported. Using SEER for comparison, the RTOG enrolled proportionately more African-American men to trials all cancer sites combined, and for prostate and head and neck cancer. In head and neck trials, the RTOG enrolled nearly twice as many African-American men than would be predicted by SEER data. In lung cancer trials, RTOG underrepresented African-American men significantly; however, there was no difference for brain cancer trials. There were no racial differences in RTOG accrual and SEER incidence data for women on trials in brain, lung, and head and neck cancer. However, the RTOG trials accrued nearly twice the proportion of African-American women in cervical cancer trials and in all sites combined, compared to the SEER data. Conclusions: Comparisons with the U.S. Census and SEER show that African-Americans are proportionally well represented in cancer clinical trials conducted by the Radiation Therapy Oncology Group. The comparative analysis indicates that all educational levels in each age group of African-Americans generally mirror the U.S. Census, with one exception. The exception is a significant overrepresentation of less-educated African-Americans in the youngest age category. This exception is counter to the expectation that better-educated patients are more likely to enroll in trials. When compared with SEER data, the RTOG trials either parallel or overrepresent African-American men and women, with the only exception being in lung cancer, where men are underrepresented. These results show that, in comparison to the Census and SEER data, the RTOG has fulfilled its commitment to enroll African-American patients in its clinical trials. © 1998 Elsevier Science Inc. Cancer clinical trials, Minority representation, African-Americans, Cooperative groups, Recruitment, Accrual, Enrollment.
INTRODUCTION
America. RTOG conducts multicenter clinical trials that evaluate the effects of ionizing radiation in the treatment of cancer. Over the last 15 years, it has enrolled over 20,000 patients in clinical studies that have resulted in significant advances in the understanding and treatment of cancer.
The Radiation Therapy Oncology Group (RTOG), formed in 1971, has approximately 260 full and affiliate members and includes most of the major cancer centers in North
R.N., R.T.T., at RTOG Headquarters and to Margaret R. Spitz, M.D., Virginia L. Ernster, Ph.D., Katherine A. McGlynn, Ph.D. and M. Elizabeth Hammond, Ph.D. of the RTOG Epidemiology Committee for their comments on the data tables and figures. Accepted for publication 30 September 1997.
Reprint requests to: Melissa L. Bondy, Ph.D., Department of Epidemiology, Box 189, The University of Texas M. D. Anderson Cancer Center, 1515 Holcombe Blvd., Houston, TX 77030. Acknowledgements—This work was supported in part by Public Health Service grants CA21661, CA37422 and CA32115. The authors are indebted to Thomas F. Pajak, Ph.D. and JoAnne Stetz, 9
10
I. J. Radiation Oncology
●
Biology
●
Physics
RTOG operates with the principle that cancer therapy can be significantly improved by subjecting promising interventions to carefully designed clinical trials, culminating in phase III trials in which new treatments can be tested against standard therapies. A premise of this concept is that trial results should be applicable to the population served and so transferable to clinical practice. To ensure that patients enrolled in a clinical trial should represent the general population to which the trial results will be applied, the sociodemographic profile of those patients should be similar (1–3). Findings from trials with insufficient numbers of minority and women participants cannot be applied with confidence to the general population (4 –7). These issues have been a concern to the National Cancer Institute (NCI), and cooperative groups, such as the RTOG, have been mandated to assess their compliance with this goal. The National Institute of Health (NIH) recently established guidelines enforcing the inclusion of minority and underserved populations in NIH-funded clinical trials. These guidelines evolved from a series of NIH policies and guidelines, beginning in 1985 with the report of the PHS Task Force on Women’s Health, followed by published policies in the NIH Guide to Grants and Contracts in subsequent years. The general purpose of these policies and guidelines was to encourage researchers to ensure that research results and benefits would be applicable to women and minority persons. In 1993, Congress passed into law the NIH Revitalization Act (Public Law 103-43), these existing NIH policies and guidelines. In response to these guidelines and to NCI special initiatives, investigators have directed considerable effort to increase the proportions of women and minority participants in cancer treatment trials and have generally achieved a racial/ethnic composition that parallels the composition of the general population (8, 9). The RTOG has initiated 12 prostate cancer trials, involving more than 5500 men. The treatment results of 8 trials have been published. Another one closed recently, and the remaining 3 are currently accruing patients. However, information about race has been difficult to glean from the resulting data. Even when trials enroll proportions of minority participants similar to those in the general population, the small numbers of these participants often preclude a rigorous statistical comparison. To overcome this problem, Roach et al. (10) used a modified meta-analysis approach, combining data from several RTOG prostate cancer trials, and showed that stagefor-stage and equivalent therapy, race was not an independent prognostic indicator of survival. The RTOG also analyzed their non-small cell lung and brain cancer studies for racial differences, and found no differences (11, 12). Recently, the NCI modified its policy to provide for ethnicbased subset analysis to be taken into account in calculating sample size in Phase III trials in which differences are thought to exist. To determine if there has been an underrepresentation of African-Americans or other bias, we evaluated the educational attainment, gender, age, and race of patients prospectively enrolled in the RTOG trials by a comparative analysis to the U.S. Census and the SEER.
Volume 40, Number 1, 1998
Table 1. Distribution of 56 RTOG studies by type
Type of protocol
Number of open protocols
Number of RTOG patients
Open only to RTOG members (New disease) (Recurrent disease) RTOG-run intergroup protocols NonRTOG-run intergroup protocols Combined protocols
33 (23) (10) 8 15 56
4016 (3375) (641) 764 752 5,532
MATERIALS AND METHODS Patient population Between July 1991 and June 1994, 5532 patients were entered into 56 protocols available to the RTOG members. Table 1 shows that 33 protocols were only open to RTOG members. Twenty-three of those protocols were for newly diagnosed cancers, whereas the other 10 protocols were for recurrent tumors. During this period, the RTOG members had access to 23 protocols in which more than one national cancer cooperative group participated. These were designated intergroup protocols and are not included in this analysis. In RTOG-run nonintergroup studies, each subject was required to complete or have completed a questionnaire soliciting information on sociodemographic information, such as age, race, highest level of educational attainment, marital status, and data regarding income before and after treatment (Fig. 1). Method of payment for treatment was also collected. For each category, patients were given the option of ‘‘not answering.’’ From the U.S. Department of Census, we obtained a data sample from the 1990 census on educational attainment by age, gender, and race (13). We compared the census data on educational attainment with the education levels of the RTOG trial patients to determine how well they represented the general population. To determine how RTOG patients compared with cancer patients in general, and with patients with cancer at sites definitively treated with radiotherapy (14), we compared our sociodemographic data on RTOG trial patients to the Surveillance Epidemiology and End Results (SEER) data for patients who received radiation therapy. Breast cancer in women was excluded because RTOG had not conducted any breast trials during the study period.
Statistical analysis We calculated descriptive statistics to compare the RTOG demographic data with both SEER and census data. Chisquared values were calculated to compare the observed RTOG accrual to the expected accrual based on the U.S. population from the census and SEER data. These chisquared values were then evaluated by using multiple comparisons, which requires adjustment of the significance level. The Hommel procedure was used to make this adjustment (15).
Analysis of patients in radiation therapy
●
R. M. CHAMBERLAIN et al.
Fig. 1. Patient questionnaire.
11
12
I. J. Radiation Oncology
●
Biology
●
Physics
Table 2. Distribution of demographic characteristics in the study population Number
Percent
848 908 1009
30.7% 32.8% 36.5%
249 300 660 1174 382
9.0% 10.8% 23.9% 42.5% 13.8%
569 2196
20.6% 79.4%
339 2426 2765
12.3% 87.7% 100.0%
Education Years of School Completed 12th Grade or Less High School Graduate Any College Age Years ,45 45–54 55–64 65–74 751 Gender Female Male Race African-American White Total
RESULTS Of the 4016 patients entered into 33 RTOG trials, 3375 patients had newly diagnosed cancer; of these, 170 (5.0%) did not complete the questionaire (Table 1). These data include adequate numbers of white and African-American patients for a comparative analysis. Other racial groups had insufficient numbers to make comparisons, 170 (5.3%) were Hispanic, 73 (2.3%) were of other races, and four (0.1%) did not answer the question. Included in this study are 2426 (87.7%) white and 339 (12.3%) African-American patients. Data for 2765 white and African-American patients with complete demographic data on all of the pertinant study variables are shown in Table 2. More than a third (36.5%) of the patients had attended college or technical school, and another third (32.8%) had graduated from high school only. Only 18 patients (0.5%) reported never attending any school. More than 99% of the individuals answered questions regarding their race and marital status. Questions related to education and birthdate were also usually answered, with 93% compliance. However, we omitted income data because questions about income level before and after cancer diagnosis were not answered by 27% and 29% of the individuals, respectively. Our figures show only data for African-Americans because, for every age group and at each level of educational attainment, the proportions of white RTOG patients were comparable to the proportions in the census and SEER data. The proportion of African-American male subjects 65 years old or older was comparable with the census data in the age and educational categories (Fig. 2). In the younger AfricanAmerican males, RTOG accrued significantly fewer with college experience and more high school graduates. In the age group 20 –54 years, 45.3% of the African-American male RTOG trial participants did not complete high school in comparison with 18.9% in the U.S. census (p 5 0.0001).
Volume 40, Number 1, 1998
Similar but not statistically significant, comparisons were observed in the 55– 64 year group (p 5 0.09). In the 20 –54 year age group, 52.3% of the AfricanAmerican female trial participants had attained less than 12th grade education compared with 17.5% in the census (p 5 0.0001) (Fig. 3). For the other age groups, the results were parallel with the census. Figure 4 shows comparisons between the patients accrued in RTOG trials and SEER incidence data by cancer sites treated with radiation as the main therapeutic modality. We compared only sites where RTOG focuses its clinical treatment efforts. For all cancer sites combined, the percentages of African-American men enrolled in RTOG trials (11.9%) was significantly higher (p , 0.015) than the SEER data (10.4%). The percentages of African-American men (12.0%) in RTOG prostate cancer trials significantly (p , 0.0002) exceeded those tabulated in the SEER data (9.2%). A similar significantly (p , 0.0001) higher representation of African-American men enrolled in RTOG head and neck cancer trials (23.1%) was observed in comparison to those tabulated by SEER (12.6%). RTOG brain cancer trials enrolled the same percentage (3.0%) of African-American men as observed in the SEER data. Only in RTOG lung cancer trials were African-American men significantly (p , 0.041) underrepresented.
Fig. 2. Percentage of African-American males enrolled in RTOG trials compared with the percentage of African-American males enumerated in the U.S. Census by age and educational attainment.
Analysis of patients in radiation therapy
●
R. M. CHAMBERLAIN et al.
13
jective was guided by the premise that clinical trial samples should reflect the population to which the results will eventually be applied. We conducted a prospective study of patients entered into the RTOG treatment protocols during a 3-year period. Overall, the patient population was 12.3% African-American and 5.0% Hispanic. Because the analysis stratified the minority subsamples by age and education, statistical requirements precluded a detailed analysis of the Hispanic group. Compared with population data from SEER and the census, our findings suggest that African-Americans and women are well represented in RTOG trials. Overall, there were more educated whites and less-educated AfricanAmericans in RTOG trials than in the general population. A further notable feature of our study is that half of the RTOG subjects were older than 65, reflecting the age distribution of cancer patients, and 30% were over 70, which may reflect the effect prostate trials have on the distribution. The results of studies of patient entry into trials have been inconsistent, but suggest that patients in clinical trials tend to be of higher performance status and higher socioeconomic status and to be younger (17–19). A Danish study
Fig. 3. Percentage of African-American females enrolled in RTOG trials compared with the percentage of African-American females enumerated in the U.S. Census by age and educational attainment.
Figure 4 shows notable racial differences among women in the RTOG accrual compared with SEER incidence data, in cancer at all sites combined (p , 0.0001) and cervical cancer (p , 0.0001). In these two categories, the RTOG trials accrued nearly twice the proportion of African-American women as reported in the SEER data. For trials in brain, lung, and head and neck cancer, RTOG accruals of African-American women did not significantly differ from SEER data. In summary, the RTOG population proportionally parallels or disproportionally favors African-American patients for a majority of the age groups and by educational attainment when compared with SEER and the census. In fact, RTOG was even more successful in recruiting AfricanAmerican patients to prostate, head and neck, and cervical cancer protocols compared with SEER statistics. DISCUSSION The initial objective of the authors was to assess the degree to which RTOG trial participants represented the minority population in general, and minority cancer patients specifically. This interest parallels the work by Tejeda and colleagues (16), in their study of minority representation in all trials funded by the National Cancer Institute. Our ob-
Fig. 4. Percentage of African-American males and females enrolled in RTOG trials compared with the percentage of AfricanAmerican males and females in SEER (1991) who received radiation therapy as primary treatment.
14
I. J. Radiation Oncology
●
Biology
●
Physics
using structured interviews of outpatients demonstrated that 98% approved of study participation and that there were no differences related to socioeconomic status (20). Interestingly, a British opinion study that surveyed 1022 people about four hypothetical protocols suggested that younger persons and those with higher socioeconomic status were less likely to participate in clinical research (21). The converse, that educational status correlates with a greater desire and ability to participate in clinical decision-making, was demonstrated in a series of 75 women with advanced breast cancer (22); youth and educational status were positively related to a willingness to challenge physician authority. Although the foregoing might imply that young AfricanAmerican patients with limited education or lower socioeconomic status would more readily enroll in clinical trials, this may not be the case in the United States. Fear of experimentation, especially within the African-American community, is particularly still prevalent and is partly based on real abuses, the most frequently cited being the Tuskegee Study of Untreated Syphilis in the Negro Male conducted by the U.S. Public Health Service between 1936 and 1972 (23). There is also evidence that the currently used informed-consent forms are too complex for some patients. In one study, 79% of 68 oncologists reviewing current consent forms judged that the forms were ‘‘much’’ or ‘‘somewhat’’ too technical (24). The question ‘‘who volunteers for clinical trials?’’ is being more frequently asked as we move toward the end of the third decade of NCI-supported clinical trials (25). In general, there are three major variables associated with enrollment into clinical trials: ● ● ●
The clinical trial characteristics; the patient characteristics; the physician characteristics.
A number of the features of clinical trials may limit patient participation. For example, protocols may not be available or offered to patients (17). Gender-specific trials will obviously limit the participation in clinical trials. For example, the RTOG has strong initiatives in prostate cancer, which affects the overall gender ratios and age distribution of RTOG subjects. The trial design itself may act as a barrier to participation, if eligibility clauses require that the subject live within a given distance from the institution or require performance of many nonreimbursed tests. Furthermore, randomized trials with notreatment arms or with very different treatment arms, such as
Volume 40, Number 1, 1998
radiotherapy vs. surgery, may be difficult to understand and, therefore, deter participation (26–29). Patients have many reasons for choosing or not choosing to participate in trials. A recent review of the literature indicates that subject perception of personal benefit must outweigh two major categories of barriers: (a) Time and hassle involved in participation; and (b) negative personal and family attitudes about trials (30). Previous studies indicate that the first category of barriers is by far the most important. Finally, physician attitudes and motivation are extremely important in the enrollment of patients into clinical trials. A survey mailed to investigators comparing segmental mastectomy to a radical mastectomy found the most frequently cited reason for patients not entering a study was perceived disruption of the doctor-patient relationship by substitution of study-imposed interventions based on randomization rather that physician judgement (9, 27, 31). A report from the NCI Clinical Cancer Oncology Program shows evidence that an investigators’ knowledge about trials and about enrolling patients in trials is a primary determinant in successful accrual of subjects (32). Investigators conducting prevention trials find the guidelines especially difficult to achieve, because few successful recruitment methods for disease-free minority populations have been developed and rigorously tested. In some of the largest prevention trials for breast and prostate cancer, fewer than 5% of the participants have been from minority populations. Despite efforts to recruit participants that accurately represent the target population, prevention study populations often do not represent the general population with respect to demographic variables such as race/ethnicity, educational attainment, gender, and age. The high proportion of minority populations accrued in RTOG treatment trials strengthens the applicability of study results to these populations. RTOG trials aim to answer questions not only about the entire study population, but also about important subgroups of patients. This commitment of the RTOG led to the initiation of the Special Populations and Epidemiology Committees, with goals to evaluate participation, study design, and relevance to women and minorities, and its efforts are effectively reducing the disparities between the makeup of treatment populations. We have shown, in this study, that the RTOG has successfully recruited minorities and women to treatment protocols and that results of these trials can be extended to the general population.
REFERENCES 1. Begg, C. B.; Engstrom, P. F. Eligibility and extrapolation in cancer clinical trials. J. Clin. Oncol. 5:962–968; 1987. 2. Mohiuddin, S. M.; Hilleman, D. E. Gender and racial bias in clinical pharmacology trials. Ann. Pharmacother. 27:972–973; 1993. 3. Svensson, C. K. Representation of American blacks in clinical trials of new drugs. JAMA 261:263–265; 1989. 4. Antman, K.; Amato, D.; Wood, W.; Corson, J.; Suit, H.; Proppe, K.; Carey, R.; Greenberger, J.; Wilson, R.; Frei, E., III. Selection bias in clinical trials. J. Clin. Oncol. 3:1142– 1147; 1985.
5. Kaluzny, A.; Brawley, O.; Garson-Angert, D.; Shaw, J.; Godley, P.; Warnecke, R.; Ford, L. Assuring access to state-ofthe-art care for U.S. minority populations: The first 2 years of the minority-based Community Clinical Oncology Program. J. Natl. Cancer Inst. 85:1945–1950; 1993. 6. King, A. C.; Harris, R. B.; Haskell, W. L. Effect of recruitment strategy on types of subjects entered into a primary prevention clinical trial. Ann. Epidemiol. 4:312–320; 1994. 7. Roberson, N. L. Clinical trial participation: Viewpoints from racial/ethnic groups. Cancer 74:2687–2691; 1994. 8. McCabe, M. S.; Varricchio, C. G.; Padberg, R. M. Efforts to
Analysis of patients in radiation therapy
9. 10.
11.
12.
13.
14.
15. 16.
17. 18.
recruit the economically disadvantaged to national clinical trials. Semin. Oncol. Nurs. 10:123–129; 1994. Taylor, K. M.; Shapiro, M.; Soskolne, C. L. Physician response to informed consent regulations for randomized clinical trials. Cancer 60:1415–1422; 1987. Roach, M.; Krall, J.; Keller, J. W.; Perez, C. A.; Sause, W. T. The prognostic significance of race and survival from prostate cancer based on patients irradiated on Radiation Therapy Oncology Group protocols (1976 –1985). Int. J. Radiat. Oncol. Biol. Phys. 24:441– 449; 1992. Graham, M. L.; Geitz, et al. Comparison of prognostic factors and survival among black patients and white patients treated with irradiation for non-small– cell lung cancer. J. Natl. Cancer Inst. 84:1731–1735; 1992. Simpson, J.; Horton J., et al. Influence of location and extent of surgical resection on survival of patients with glioblastoma multiforme: Results of three consecutive Radiation Therapy Oncology Group (RTOG) clinical trials. Int. J. Radiat. Oncol. Biol. Phys. 26:239 –244; 1993. U.S. Bureau of the Census, Current Population Survey. Educational attainment of persons 15-years-old and over, by age, sex, race and Hispanic origin. Department of Commerce, Bureau of the Census, 20-489; March 1995. Kosary, C. L.; Ries, L. A. G.; Miller, B. A.; Hankey, B. F.; Harras, A.; Edwards, B. K. (eds). SEER Cancer Statistics Review, 1973–1992: Tables and Graphs, National Cancer Institute. NIH Pub. No. 96-2789. Bethesda, MD, 1995. Hommel, G. A stagewise rejective multiple test procedure based on a modified Bonferroni test. Biometrika 75:383–386; 1988. Tejeda, H.; Green, S.; Trimble, E.; Ungerleider, R.; Friedman, M.; Brawley, O. The representation of African-Americans, Hispanics, and whites in National Cancer Institute sponsored cancer treatment trials. J. Natl. Cancer Inst. 88:812– 816; 1996. Hunter, C. P.; Frelick, R. W.; Feldman, A. R. Selection factors in clinical trials. Cancer Treat. Rep. 71:559 –565; 1987. Martin, J. F.; Henderson, W. G.; Zacharski, L. R. Accrual of patients into a multihospital cancer clinical trial and its implications on planning future studies. J. Clin. Oncol. 7:173–182; 1984.
●
R. M. CHAMBERLAIN et al.
15
19. McCusker, J.; Wax, A.; Bennett, J. M. Cancer patient accessions into clinical trials. J. Clin. Oncol. 5:227–236; 1982. 20. Saubrey, N.; Jensen, J.; Rasmussen, P. E. Danish patients attitudes to scientific and ethical questions. Acta Med. Scand. 215:99 –104; 1984. 21. Kemp, N.; Skinner, E.; Toms, J. Randomized clinical trials in cancer treatment—a public opinion survey. Clin. Oncol. 10: 155–161; 1984. 22. Llewellyn-Thomas, H. A.; McGreal, M. J.; Thiel, E. C. Patients willingness to enter clinical trials: Measuring the association with perceived benefits and preference for decision participation. Soc. Psy. Med. 32:35– 42; 1991. 23. Jones, J. H. Bad blood: The Tuskegee syphilis experiment. New York: The Free Press; 1993. 24. Penman, D. T.; Holland, J. C.; Bahna, G. F. Informed consent for investigational chemotherapy. Patients’ and physicians’ perceptions. J. Clin. Oncol. 2:849 – 855; 1984. 25. Gehan, E. A.; Schneiderman, M. A. Historical and methodological developments in clinical trials of the National Cancer Institute. Stats. Med. 9:871– 880; 1990. 26. Cassileth, B. R.; Knuiman, M. W.; Abeloff, M. D. Anxiety levels in patients randomized to adjuvant therapy vs. observation for early breast cancer. J. Clin. Oncol. 4:972–974; 1986. 27. Ganz, P. A. Clinical trials: Concerns of the patient and the public. Cancer 65:2394 –2399; 1990. 28. MacKillop, W. J.; Ward, G. K.; O’Sullivan, B. The use of expert surrogates to evaluate clinical trials in non-small cell lung cancer. Br. J. Cancer 54:661– 667; 1986. 29. MacKillop, W. J.; Ward, G. K.; O’Sullivan, B. Non-small cell lung cancer. How oncologists want to be treated. Int. J. Rad. Oncol. Biol. Phys. 13:929 – 4; 1987. 30. Swanson, G. M.; Ward, A. J. Recruiting minorities into clinical trials: toward a participant-friendly system. JNCI 87: 1747–1759; 1995. 31. Taylor, K. M.; Kelner, M. Informed consent: The physician’s perspective. Soc. Psy. Med. 24:135–143; 1987. 32. Feigl, P.; Diehr, P.; Allison, C. Community cancer care evaluation. Final report of NCI contract NO1-CN-350009; 1987.