Reasons for Ineligibility for a Randomized Clinical Trial

324 Brief communication REASONS FOR INELIGIBILITY FOR A RANDOMIZED CLINICAL TRIAL This is a post hoc examination of smokers not eligible for a study of smoking cessation among African Americans. The parent study, Quit for Life, was a randomized controlled trial for smoking cessation among African Americans.1 Here, we describe study selection criteria and the reasons for ineligibility among nonparticipants. We then consider how selection criteria might be modified to conduct more generalizable research among the poor and underserved. For many health issues, minority groups bear a disproportionately high burden of morbidity and mortality.2 Clinical trials in the United States are conducted predominantly among middle-class, white men.3 Historical, economic , and cultural barriers have contributed to the underrepresentation of minority and low-income populations in clinical research.3,4 Clinical trial eligibility criteria may constitute another reason for low rates of participation. Overly strict enrollment criteria can limit successful accrual in clinical trials5 and systematically exclude subpopulations from participating . Abetter understanding of the characteristics of ineligible participants and the effects of selection criteria on enrollment should enhance the design and implementation of future research among underrepresented populations. Method Quit for Life (QFL) used a randomized, single-blind design to examine the effects of a culturally tailored brochure and videotape on smoking cessation compared with a nontailored, commonly used brochure and videotape. Both groups of participants received eight weeks of transdermal nicotine patches. Patients were recruited through referral and active recruitment from outpatient clinics at a large, urban academic hospital predominantly serving African Americans. A total of 1,902 persons underwent eligibility screening for Quit for Life, 787 of whom were invited to enroll in the study. The remaining 1,115 persons were not enrolled in the study. Eight hundred and ninety six persons completed screening and were found ineligible by at least one exclusion criterion. These are the participants included in the present analyses. The Emory University School of Medicine Human Investigation Committee approved the study protocol. Received October 18, 2001; revised July 18,2002; accepted August 2, 2002. Journal of Health Care for the Poor and Underserved · Vol. 14, No. 3 · 2003 DOI: 10.1177/1049208903255441 Jean et al. 325 QFL had 12 eligibility criteria. Five criteria were designed to screen out persons who might compromise the goals or design of the study. These criteria excluded from the study non-African Americans, those in treatment for alcohol or other drug use, persons using other forms of tobacco, persons having a friend in the study, or those in "precontemplation" (not considering making a change) regarding quitting smoking. Three criteria were designed to screen out persons for whom nicotine patch use would be contraindicated: pregnant smokers, those younger than 18 years of age and those weighing less than 100 pounds. These criteria protected the health and safety of participants. Three criteria were designed to screen out smokers who did not have the minimal resources required to adhere to study protocols. A VCR was required for participants to view the videotape component of the intervention. A telephone was required because booster phone calls were part of the intervention. A permanent address was required for future contacts. One criterion was designed to screen out those smoking fewer than eight cigarettes a day, since little data exist on the safety and efficacy of the nicotine patch for such smokers. Post hoc analyses of the reasons for ineligibility were conduced using SAS.6 Means are given to summarize the age, weight, and number of cigarettes smoked per day for those with complete and incomplete information. The two-sample f-test was used to compare these characteristics between the two groups. Frequencies and percentages were calculated for persons meeting each study exclusion criterion. The frequency and percentage of persons meeting one versus multiple exclusion criteria were also calculated. Chisquare tests were performed to determine if there were any significant (p < 0.05) positive or negative bivariate associations between pairs of exclusion criteria. Results The 896 persons included in the present analysis had a mean age of 41, weighed an average of 164 pounds, and smoked an average of 12 cigarettes per day. Two hundred and nineteen persons were excluded from further analysis because of incomplete or inconsistent data. They were similar to the remaining participants with respect to age and weight but differed significantly in their daily cigarette consumption, which averaged 14 per day (f = 2.69, ϕ = 0.0076). Table 1 displays the number and percentage of participants ineligible for the study due to each exclusion criterion. Almost half of inéligibles smoked fewer than eight cigarettes per day. For 14 percent of persons, this was the sole cause for ineligibility. Had the exclusion criteria been fewer than five cigarettes per day, we would have excluded only 167 persons, instead of 392, thus increasing the number of eligible individuals by 225 (not shown). One-third of inéligibles had no access to a VCR, but this was the sole cause of ineligibility 326 Reasons for Ineligibility TABLE 1 PROPORTIONS OF PERSONS MEETING STUDY EXCLUSION CRITERIA3 EXCLUSION CRITERIA ALL PERSONS MEETING THIS CRITERION η PERCENTAGE PERSONS MEETING ONLY THIS CRITERION η PERCENTAGE Fewer than eight cigarettes/day 392 43.7 121 13.5 No access to a VCR 340 38.0 57 6.4 Precontemplation 284 31.7 62 6.9 No access to a phone 258 28.8 52 5.8 In treatment 135 15.1 33 3.7 Other tobacco use 115 12.8 29 3.2 No permanent home 90 10.0 4 0.5 Not African American or black 45 5.0 17 1.9 Friend in study 21 2.3 7 0.8 Pregnant 19 2.1 3 0.3 Less than 100 pounds 7 0.8 1 0.1 Younger than 18 years of age 0 0.0 0 0.0 a Percentage column totals exceed 100 percent as patients could be ineligible for multiple reasons. TABLE 2 RANKED PROPORTIONS OF PERSONS MEETING ONE OR MORE EXCLUSION CRITERIA NUMBER OF EXCLUSION CRITERIA PERCENTAGE CUMULATIVE PERCENTAGE 386 308 123 64 11 4 43.1 34.4 13.7 7.1 1.2 0.5 43.1 77.5 91.2 98.3 99.5 100 for only 6 percent of potential participants. Nearly one-third of inéligibles were in "precontemplation" regarding quitting smoking; for 7 percent of persons , this was the sole cause for ineligibility. One-quarter of inéligibles had no access to a phone. This was the sole cause of ineligibility for 6 percent of potential participants. Many participants (111) were excluded solely for economic reasons (no access to a phone, no access to a VCR, no permanent home). Table 2 displays the number of potential participants excluded from QFL due to multiple exclusion criteria. Forty-three percent were ineligible due to only one criterion. The remainder, a majority of potential participants (56 percent ), were excluded because they failed to meet two or more criteria. Jean et al. 327 TABLE 3 BIVARIATE ASSOCIATIONS BETWEEN PAIRS OF EXCLUSION CRITERIA3 1 2 3 4 5 6 7 8 9 10 11 12 1. Fewer than eight cigarettes per day 392 108* 138* 62* 29* 45 9* 10* 3* 6 3 0 (43.8) (12.1) (15.4) (6.9) (3.2) (5.0) (1.0) (1.1) (0.3) (0.7) (0.3) (0) 2.NoVCR 340 87* 149* 54 34* 69* 16 4 8 4 0 (38.0) (9.7) (16.6) (6.0) (3.8) (7.7) (1.8) (0.5) (0.9) (0.5) (0) 3. Precontemplation 284 67* 26* 25* 28 9 0* 0* 2 0 (31.7) (7.5) (2.9) (2.8) (3.1) (1.0) (0) (0) (0.2) (0) 4. No phone 258 49* 26 60* 8 3 3 1 0 (28.8) (5.5) (2.9) (6.7) (0.9) (0.3) (0.3) (0.1) (0) 5. In treatment 135 12 40* 6 3 1 0 0 (15.1) (1.3) (4.5) (0.7) (0.3) (0.1) (0) (0) 6. Other tobacco use 155 5* 6 3 2 0 0 (12.8) (0.6) (0.7) (0.3) (0.2) (0) (0) 7. No permanent home 90 1 2 1 0 0 (10.0) (0.1) (0.2) (0.1) (0) (0) 8. Not African American or black 45 1 2 0 0 (5.0) (0.1) (0.2) (0) (0) 9. Friend in study 21 1 0 0 (2.1) (0.1) (0) (0) 10. Pregnant 19 0 0 (2.1) (0) (0) 11. Less than 100 pounds 7 0 12. Younger than 18 years of age 0 (0.8) (0) 0 (0) a η (percentage). Percentages refer to the proportion of all 896 excluded persons to whom the pair of exclusion criteria applied. *p < 0.05: denotes a significant association between exclusion criteria in corresponding row and column Table 3 is a bivariate representation of the frequency and percentage of people who failed each exclusion criterion. This table allows for visual examination of the degree of overlap or disjunction between pairs of exclusion criteria. Data on the diagonal midline are the frequency and percentage of potential participants who failed a single criterion. The off-diagonal data are the frequency and percentage of people who failed the two criteria on the vertical and horizontal axes. Several of the most prevalent causes of ineligibility were associated with each other. Not surprisingly, 17 percent of the group had no access to either a VCR or to a telephone. Fifteen percent of persons smoked fewer than eight cigarettes per day and were in precontemplation. Twelve percent of persons smoked fewer than eight cigarettes per day and did not have 328 Reasons for Ineligibility access to a VCR. Eight percent of inéligibles were in precontemplation and did not have access to a telephone. Conversely, some causes of ineligibility had little or no overlap. For example, no pregnant women were in precontemplation regarding quitting smoking (i.e., all pregnant women were contemplating quitting smoking). Discussion The most common cause of ineligibility was smoking fewer than eight cigarettes per day, followed by having no access to a VCR and being in precontemplation regarding quitting smoking. More than half of study candidates were ineligible for multiple reasons. In numerous instances, pairs of exclusion criteria were significantly associated. These findings highlight several fundamental issues in conducting research among the poor and underserved. African Americans are disproportionately represented among the poor and underserved, and smoking patterns among African Americans are poorly understood. Most studies using pharmacotherapy for smoking cessation require participants to smoke more than 10 or sometimes 20 cigarettes per day to be eligible for inclusion.7"9 It has been found that compared with the general population, African Americans smoke many fewer cigarettes per day.10 Future studies tailored to African Americans may have to drop the required smoking level even further, perhaps as low as 5 cigarettes per day, to ensure the study includes a broad and representative spectrum of African American smokers. Being in precontemplation with regard to smoking cessation also proved to be a significant barrier to eligibility. Again, many smoking cessation investigators use screening criteria designed to cull out precontemplaters. This is done to maximize the likelihood that participants will comply with protocols and complete all measures. However, a recent study found that when given access to pharmacotherapy, participants in a lower stage of change achieved quit rates comparable with those in a more advanced stage.11 To reduce barriers to participation and improve the representativeness of study populations, motivational factors might best be used as process measures and not eligibility criteria. It is important to avoid picking the most affluent persons to participate in studies, so that results are generalizable to the poor and underserved. In many ethnic and cultural subgroups, eligibility requirements involving resources, material goods, or transportation can create barriers to participation. In QFL, lacking access to a VCR and phone, and lacking a permanent address, proved to be common and frequently overlapping causes of ineligibility. Future studies should scrutinize eligibility requirements for hidden economic barriers and remove as many as possible. Sometimes this might require changes in the proposed intervention. For example, videotapes might be viewed at an on-site VCR, which eliminates a large barrier for some individuals. Researchers might also choose to send postcards or make home visits instead of phone Jean et al. 329 calls. However, investigators should be aware that removing one requirement might not increase eligibility, especially if another criterion excludes virtually the same potential participants. Some of these changes might increase the cost of research projects, but they would improve participation and result in more truly generalizable findings. These considerations also affect the potential for disseminating successful interventions. If participants do not have the resources for the parent study, will most of the target population be in the same situation? If the study has the potential for creating policy change to make more resources available, these requirements may be justified. If not, protocols should be adjusted to match the context of the end user. The present study has several limitations. It is an observational study: no intervention or comparison group is included to assess whether changes in eligibility criteria would actually result in a more representative study population . Second, because this analysis was not planned for in the original study, the screening form collected minimal descriptive data, which limited our analyses. Also, we are aware that exclusion data are necessary in order to control for potential confounding or extraneous variables that can substantially reduce the internal validity of a study, whereas certain exclusion criteria are needed for practical study-design issues. Attention to relatively small changes such as those discussed here can result in more representative populations. Investigators should resist the temptation to adopt, whole cloth, protocols and criteria used in trials conducted among affluent populations. Specifically, in selecting screening criteria for tailored interventions, target population health behavior patterns and economic constraints should strongly influence screening criteria. This will ensure that the study population is representative and that the trial is a truer test of the effectiveness of the intervention. REFERENCES 1. Ahluwalia JS, Richter KP, Mayo MS, et al. Quit for Life: A randomized trial of culturally sensitive materials for smoking cessation in African Americans. J Gen Intern Med 1999 Apr;14(Suppl 2):6. 2. United States Department of Health and Human Services. Healthy People 2000: Midcourse review and 1995 revisions. Washington, DC: Government Printing Office, 1995.(DHHS Pub. No. (PHS) 91-50213.) 3. Swanson GM, Ward AJ. Recruiting minorities into clinical trials: Toward a participant-friendly system. J Natl Cancer Inst 1995 Dec 6;87(23):1747-59. 4. McCarthy CR. Historical background of clinical trials involving women and minorities. Acad Med 1994 Sep;69(9):695-98. 5. LaRue LJ, Alter M, Traven ND, et al. Acute stroke therapy trials: Problems in patient accrual. Stroke 1988 Aug;19(8):950-54. 6. SAS Institute Inc. SAS user's guide: Statistics. 6th ed. Cary, NC: SAS Institute, 1990. 7. Silagy C, Mant D, Fowler G, et al. Meta-analysis on efficacy of nicotine replacement therapies in smoking cessation. Lancet 1994 Jan 15;343(8890): 139-42. 330 Reasons for Ineligibility 8. Daughton D, Susman J, Sitorius M, et al. Transdermal nicotine therapy and primary care. Importance of counseling, demographic, and participant selection factors on 1-year quit rates. The Nebraska Primary Practice Smoking Cessation Trial Group. Arch Fam Med. 1998 SepOct ;7(5):425-30. 9. Westman EC, Levin ED, Rose JE. The nicotine patch in smoking cessation. A randomized trial with telephone counseling. Arch Intern Med 1993 Aug 23;153(16):1917-23. 10. Caraballo RS, Giovino GA, Pechacek TF, et al. Racial and ethnic differences in serum continine levels of adult cigarette smokers, Third National Health and Nutrition Examination Survey, 1988-1991. JAMA 1998 Jul 8;280(2):135-59. 11. Jolicoeur DG, Ahluwalia JS, Richter KP, et al. The use of nicotine patches with minimal intervention . Prev Med 2000 Jun;30(6):504-12. Samuel Jean, MPH Medical Student Department of Preventive Medicine University of Kansas Medical Center KiMBER P. Richter, PhD, mph Co-Director of Health of the Public and Assistant Professor Department of Preventive Medicine University of Kansas Medical Center Jasjit S. Ahluwalia, md, mph, ms Chair and Professor Department of Preventive Medicine University of Kansas Medical Center Kristin H. Schmelzle, MA Biostatistician PRA International, Kansas City Matthew S. Mayo, PhD Assistant Professor Department of Preventive Medicine University of Kansas Medical Center ...