The number of cancer survivors alive in the United States has increased fivefold from the 1970s to present. At present, there are over 10 million cancer survivors alive in the United States, 2.4 million are breast cancer survivors (12). It is estimated that 16% of the US population over age 65 yr are cancer survivors (8). With the increasing success of cancer therapies, there is increasing interest in developing effective interventions to improve health behaviors (e.g., diet, exercise, and appropriate screenings) and to address the psychosocial and rehabilitation needs of survivors (10).
One challenge to conducting such intervention research is finding sufficient eligible and interested survivors within a defined geographic area who are willing to travel to the research site for one or more intervention and/or measurement visits. Some interventions may address the participant burden of getting to a central research location by focusing on home, phone, or Web-based approaches (5,6). This is possible for some but not all types of interventions. For example, the quality of an exercise training intervention that can be delivered with a home or a phone-based intervention varies dramatically from what can be taught by regular meetings with a certified fitness professional in a fitness facility setting. Another challenge to intervention research in cancer survivors (and other populations) is sustainability once the research study is complete. Sustainability and dissemination of successful interventions would be facilitated by research that shows it is possible to intervene in the community, with community-based interventionists.
Within this report, we present an approach for addressing the recruitment of breast cancer survivors into community-based exercise interventions by describing methods used in conducting the Physical Activity and Lymphedema (PAL) trial. This novel recruitment approach was developed based on experiences from the pilot study for PAL, called the Weight Training for Breast Cancer Survivors Study (WTBS) (1). Women recruited for WTBS were required to come to a university-based fitness facility. This limited recruitment of potential participants to those who lived close to the intervention facility or those with personal resources available to overcome proximity challenges. Further, when the study was over, the facility and the intervention research staff were no longer available to participants. This created a barrier to sustaining and disseminating the intervention beyond the study, despite requests from the breast cancer community to do so. Therefore, in setting out to do the PAL trial, multiple approaches were applied to overcome these issues.
PAL trial overview.
The primary purpose of the PAL trial was to assess the safety and efficacy of a year-long twice-weekly strength training intervention among women with and at risk for breast cancer-related lymphedema. The recently completed PAL trial recruited and randomized 295breast cancer survivors 1 to 15 yr after breast cancer diagnosis as well as the efficacy of the same intervention to improve strength, function, and quality of life. Women were randomized to receive the intervention either during the study (treatment group) or at the end of the study (control group). All activities were reviewed and conducted in accordance with the University of Pennsylvania's human subjects protection programs. Written informed consent was obtained from all participants.
The PAL trial intervention took place at YMCA's and privately owned gyms in the Philadelphia Metropolitan Area. All of the fitness centers approached to collaborate on this project were eager to do so; in fact, several fitness centers expressed interest that were later not used due to lack of survivor residential density around the facility from which to recruit. Leadership at each YMCA association and the privately owned gym expressed interest in the potential for fostering a new market of members by having programming for breast cancer survivors. All participating fitness centers offered the "senior citizen" discounted membership for PAL trial participants (paid for by the grant). The intervention staff were drawn from the certified fitness staff of these facilities, so that the intervention could continue after the end of the study. The intervention consisted of twice-weekly strength training for 1 yr, starting with 3 months of twice-weekly supervised small group training (two to six participants per group) led by a certified fitness professional. Each session lasted 60 to 90 min. With few exceptions, participants lived within a 20-min drive of the facility at which they participated in the exercise intervention. Each of the interventionists went through 24h of specialized training to ensure that they understand the training protocol, the safety precautions, and the process evaluation measures.
Study measurements were taken at the University of Pennsylvania's campus at baseline, at 3 months, at 6 months, and at 1 yr. Participants were paid up to $145 over the course of the year for completing measurement visits. Free parking was provided for all measurement visits. All participants were required to attend a lymphedema education lecture that took place at multiple locations, including the University of Pennsylvania and each of the eight participating fitness centers. Two to three lymphedema education lecture sessions were offered to each of the 11 recruitment waves.
The inclusion/exclusion criteria for PAL were designed to balance three goals: recruitment feasibility, excluding those for whom the risks of the intervention outweighed the potential benefits, and the ability to test hypotheses. Eligible participants were female breast cancer survivors 1 to 15 yr postdiagnosis. Eligibility criteria are outlined in Table 1. A priori power calculations indicated a goal to recruit and to randomize 288 breast cancer survivors to address the primary safety outcomes regarding onset and worsening of lymphedema. Interested women were screened with questions regarding demographics, medical history (including breast cancer diagnosis and treatment), and physical activity.
Intervention locations were chosen based on density of survivors residing within one zip code of a fitness center (determined by cancer registry data) as well as interest and ability of the specific fitness center to assist with the trial (e.g., available staff with national fitness trainer certifications). Recruitment took place in waves, according to the geographic location of the participating fitness centers. By the end of recruitment, there were eight participating fitness centers (7-79 participants per site) and 11 recruitment waves (7-40 participants per wave), and 17 certified fitness professionals had been trained to deliver the protocol. Partnering facilities included fitness centers associated with the Philadelphia and Vicinity YMCA (n = 4), the Family YMCA of Burlington County in Southern New Jersey (n = 1), the Community YMCA of Eastern Delaware County, PA (n = 2), and Sisters in Shape Fitness (a privately owned fitness center in North Philadelphia). PAL targeted minority recruitment specifically to African American women, given this is the largest minority racial group in Philadelphia. To facilitate this goal, two of the facilities recruited for the intervention were located within the primarily African American neighborhoods of West Philadelphia and Logan.
Two major methods of recruitment were used: passive and active. Active recruitment was defined as approaching survivors directly through a letter. Passive recruitment was defined as exposure to media or study materials or an interaction with the study staff at a support group, conference, or community event. Active recruitment has been shown to be successful in other trials when the researcher makes direct contact versus hospital staff giving out brochures or displaying flyers (7).
Letters from state and hospital registries.
Letters were mailed in waves, corresponding to zip codes surrounding a specific intervention site, in an attempt to target women who lived closest to a particular facility, given prior observations that proximity to an exercise facility is associated with physical activity level (13). Letters mailed from the Pennsylvania and New Jersey State Cancer Registries accounted for age (<90 yr), date of diagnosis, sex, and race and excluded women for whom the primary cancer site was not breast. Mailings consisted of a letter inviting women to participate, a postage paid return postcard to indicate interest or to actively decline interest, and an e-mail address and phone number for those interested in contacting study staff to learn more. The Pennsylvania State Cancer Registry included 25,517 potential participants in the targeted zip codes. In New Jersey, the registry included 3424 potential participants in the targeted zip codes. Both state registries allowed survivors to be contacted without prior permission from a treating physician, given that study participation required a physician's signature. Letters from the New Jersey registry were sent from the registry office on registry letterhead; PAL study staff never saw the mailing list. Letters from the Pennsylvania registry were sent from the University of Pennsylvania on University of Pennsylvania letterhead. Any patients with privacy complaints regarding the letter sent from the University of Pennsylvania were given the appropriate contact at the Pennsylvania state registry. Out of 24,528 letters mailed from the Pennsylvania state registry, only 12 women (0.0004%) responded negatively when contacted. In addition, 1345 letters sent from the Pennsylvania registry mailing list were returned, and within these returns, 31 women were newly determined to be deceased beyond what the registry already excluded before providing the mailing list.
PAL also sent 969 letters through three hospital registries, including the Delaware County Memorial Hospital (DCMH), a community hospital in Lansdowne, PA; the Hospital of the University of Pennsylvania (HUP), a large urban tertiary care teaching hospital serving Philadelphia and the surrounding area; and the Virtua Health System in New Jersey, which serves Southern New Jersey. Each hospital identified a list of potential participants, followed by a process to obtain physicians approval before mailing letters. Each hospital required that PAL be approved by the institution's human subjects protection program (institutional review board [IRB]) to approach any patients.
Active recruitment targeted to minority racial groups.
Strategies for self-identified racial minority enrollment differed slightly from the recruitment of white participants. The mailings to the residents living within zip codes close to fitness facilities targeting African American women were specifically tailored to the population and therefore included more materials than other mailings. The mailings to the Sisters in Shape Fitness Center recruitment wave included a color postcard that displayed the owner (an African American woman) holding weights. The mailings to the West Philadelphia YMCA zip codes included a personalized letter from the executive director and fitness director with a photo of both men in front of strength training equipment (both are African American).
Passive recruitment for PAL included participation of PAL staff in 32 community events, 21 forms of print media, and 26 types of broadcast media, with proportionally more media interviews among the African American than mainstream press. Female members of the West Philadelphia YMCA were sent letters (964 total) informing them about the study and inviting them to share the opportunity with anyone they knew who might be eligible to participate. Information was also sent to local Baptist churches in West Philadelphia. The PAL research staff met with several prominent African American community leaders with access to the breast cancer survivor population and attended events specifically targeted to African American breast cancer survivors (two survivor conferences, three support groups, and an event at a local hospital). Finally, PAL offered a one-time line-dancing event at the West Philadelphia YMCA to African American women who did not respond to the initial mailing in an effort to boost enrollment at this site. Twenty-one of the 2044 women who were mailed invitations attended. Notably, no one from this event was recruited into the study.
In all, 28,921 letters were mailed to breast cancer survivors in the Philadelphia metropolitan region: 27,952 from one of two participating state registries and 969 from one of three participating hospital registries. There were 25,517 potential participants from the Pennsylvania registry. PAL recruitment was completed in 17 months, and over that time 3200 women contacted the study with interest or to decline. Of these women, 506 were eligible, 1257 did not meet eligibility requirements, 999 declined, 65 women self-reported that they lived too far from the intervention site, and 373 did not complete their initial screening (Table 2A). The most common reason for ineligibility was timing (not yet 1 yr out from diagnosis, more than 5 yr since diagnosis without lymphedema, and more than 15 yr since diagnosis). Other reasons for ineligibility were that women had not had lymph nodes removed, currently had cancer, had had bilateral cancer, were moving within the study period, had physical limitations precluding exercise participation, or already did strength training. There was no way to track race, age, or other descriptors of women who declined by postcard or phone message. Therefore, screening information is only available for those who were both eligible and interested. Although state registries provided PAL with the largest proportion of consented and randomized participants, the response to letters from hospital registries was slightly higher (Table 2B). Figure 1 shows the flow of participants from recruitment to randomization.
Three hundred and seventy-one women attended an orientation session and signed a consent form to participate in PAL (Table 2C). The majority of consented women (67%) reporting having heard about the study from a direct letter sent in cooperation with the state cancer registries. The method of recruitment for randomized women mirrored those for consented women and varied only slightly between white and nonwhite participants (Table 3).
Most of the women recruited into PAL were between the ages of 50 and 54 yr, and the age distribution of randomized study participants was similar to the distribution of age in the state registry (Table 4). Racial self-identification data were collected on 652 women who completed at least part of the phone screening for the study: 401 women were white and 223 self-identified as African American. Of the 371 consented women, 124 self-identified as minorities and 112 identified as African American. Of the total 295 randomized women, 103 self-identified as minorities. This exceeded the a priori goal of recruiting 98 self-identified minority women. Ninety-four of the 103 minority women randomized into the PAL trial were African American. Table 4 includes specific information on white, African American/black, and Asian races. All other races were included in the other category, and all unknowns were excluded. In the Pennsylvania and the New Jersey state cancer registries, 81% and 84% of breast cancer survivors identified as white, respectively. By comparison, 62% of women screened for PAL were white and 65% of randomized PAL participants were white. Further, the proportion of African American participants were 32.4%, 30.1%, and 31.8% among those screened, consented, and randomized, respectively.
Table 5 shows baseline characteristics of the 295 women randomized into the PAL trial within lymphedema and treatment group status. The groups are well balanced across treatment group status, within lymphedema status, for all demographic, cancer history, and lymphedema variables at baseline. There were no between-group differences (across treatment status, within lymphedema status) in any variables included in Table 5 (results not shown).
As the number of cancer survivors continues to increase, more behavioral intervention research will be needed to test approaches to preventing, controlling, or treating late and long-term effects of necessary life-saving cancer therapies. Further, not all of these interventions can be home, phone, or Web based. It is desirable that such interventions be disseminable within the community. Accordingly, it is important for researchers to be able to efficiently recruit cancer survivors into interventions that are community based. In the PAL trial, the majority of participants recruited into a community-based exercise intervention heard about the study through letters sent in cooperation with two state cancer registries. That said, although support groups and events and media attention were not the most successful means of recruitment, they provided staff with social and networking ties, which are an immeasurable asset to recruitment in a community-based intervention. It is not possible to know whether the recruitment letters sent through the registries would have been as successful without the social and networking ties from support groups and media attention. However, there are examples in the literature of other survivorship studies that have successfully recruited through registries that have not focused on media attention (9,11). These two prior examples differ from the PAL trial in two ways: one was larger than PAL but was survey based only (11), the other was much smaller and did not deal with the geographic recruitment issue addressed herein, as all exercise occurred at a single health club or in the home (9).
Letters from hospital registries only accounted for 3% of randomized participants. Despite this low percentage, it is still considered a vital part of the recruitment plan because hospital registries may allow access to women sooner after diagnosis than is possible from state cancer registry data. Cost and staff time are two components to consider when working with a hospital registry. It is essential to meet with hospital staff and physicians to get "buy in" for the study from key decision makers. Once IRB approval is granted, there is also a significant burden of hospital administrative staff time. It has been observed that when recruiting through hospital registries and using physician signed letters, it is helpful to be able to review the possible participants for eligibility before approaching physicians. This creates a smaller list for physicians to review, creating a quicker turnaround (2). By comparison to working with hospital registries, working with state registries is much less costly and seems to net a much larger potential pool of participants. The amount of time required to get nearly 25,000 addresses of recruitable survivors from the Pennsylvania state cancer registry was a matter of weeks, and little staff time was allocated to this effort. The amount of time required for getting the 969 letters mailed through the hospital registries was months, and the staff time required was manyfold greater. From the perspective of return on investment, state cancer registries are clearly a superior method of recruiting.
Another issue with hospital-based registries is that it may be necessary to recruit through several hospital registries to find a diverse enough sample. Pakilit et al. (11) observed that a cancer center registry had a higher response rate than a community hospital, but the community hospital registry was provided a more diverse sample. Compared with a state registry, hospitals may offer a much smaller and possible homogeneous sample. Another challenge of hospital registries for behavioral trials that require attendance to regular onsite intervention activities is geography. For example, we observed very few women within the targeted zip codes who had been seen at HUP (n = 94) despite a very large breast cancer program at the University of Pennsylvania (approximately 800 surgeries per year).
PAL spent a significant amount of time dedicated to the recruitment of minority participants. The majority of minority recruitment efforts were focused on African American women. PAL staff attended proportionally more support groups and conferences focused on African American than white women. However, the registry still proved to be the most successful recruitment tool for both racial groups, indicating that registries can provide a racially diverse sample. At the recommendation of African American research colleagues at the University of Pennsylvania, extra materials (including photographs of African American exercise leaders) were included in letters sent to African American survivors. It is not possible to know whether this positively affected the recruitment efforts. However, the success of minority recruitment in the PAL trial supports the use of this approach in future studies.
Registries are not uniform in their research and information policies. One report indicated that 71% of state cancer registries had a passive approach in telling patients about the registry, such as through a Web site or a brochure in a waiting room; therefore, most patients are unaware that their personal information is being collected this way (4). Policies are state specific; some may allow patients to be contacted for research purposes whereas others may not. For those that allow contact, many require a signed physician's approval before letters can be sent. PAL was successful in part because of the cooperation of two state registries that allowed survivors to be contacted without prior permission from a treating physician. It is not known exactly what effect a requirement of written physician approval before contacting patients would have had on recruitment for the PAL trial. It can be assumed that recruitment would have certainly been extended beyond 17 months and cost more.
Another valid concern when recruiting with registry data is the possibility of violating patient confidentiality. However, there is some evidence that most patients are open to contact. In a survey of 33 nationwide cancer registry directors, Beskow et al. (4) observed that 51.5% categorized less than 1% of their patients as flagged (flagged patients have declined to be contacted for research purposes). In our study, only 12 women were upset by being contacted through the registry (0.0004% of total letters mailed from the Pennsylvania cancer registry mailing list). If more patients were actively notified about the existence of a registry, patients who would not want to be contacted could opt out at the time of notification. As other studies have indicated, an awareness of the registry may make patients more likely to respond to invitations to participate in research and potentially allowing the registries to keep in better contact with survivors (3).
To summarize, state cancer registries were a useful tool in recruiting racially and geographically diverse participants into a community-based exercise intervention for breast cancer survivors. This supports the goal of conducting health behavior interventions within communities where the survivors live. The geographic recruitment and the community-based intervention approaches described herein may support the sustainability and the dissemination of health behavior interventions to cancer survivors beyond the end of research funding.
This study was supported by the National Cancer Institute (R01-CA106851 to Schmitz) and by the Public Health Services Research Grant RR024134 from the National Institutes of Health to the University of Pennsylvania.
The results of the present study do not constitute endorsement by the American College of Sports Medicine.
The authors thank Robin Otto and Craig Edelman, Pennsylvania Cancer Registries, Pennsylvania Department of Health, Harrisburg, PA; Constance Sumner, R.N., and Jazmine Alaverez, Black Women's Health Alliance; Elyse Caplan, M.A., and Amy B. Grillo, Living Beyond Breast Cancer; Jennifer Leith, Susan G. Komen Foundation Philadelphia Affiliate; Donna Dunkin, Linda Creed Foundation; Wilma Morgan, O.T.R./L., Janice Buhler, M.S., P.T., O.C.S., Vicki Moran, Cheri Doll, M.S.P.T., and Carolyn Weaver, R.N., B.S.N., A.O.C.N., Fox Chase Cancer Center Philadelphia, PA; Judy Neuman, Director, Oncology Data and Clinical Data, Phyllis Duda, Nurse Practitioner, Doug Marshall, Community Educator, Barbara Francks, R.N., Eric Miller, M.D., and Adrienne Kirby, Ph.D., F.A.C.H.E., R.N., Virtua Health System; Serena Harris, Family YMCA of Burlington County; Gregory Lyles, Executive Director, and Alonzo Holder, Fitness Director, West Philadelphia YMCA; Mary Francis Reilly and Claine Crew, Main Line YMCA; Kathy Perry and Kathy Disque, Ambler Area Branch YMCA; Duane Johnson, Executive Director, and Miriam Ranalli, Lansdowne YMCA; Debbie Mignogna, Branch Director, Ridley Area YMCA; Donna Kern, Northeast YMCA; Ann Stephanie, Nurse Practitioner, Cancer Institute of NJ Cooper University Hospital; Tracy Harper, D.P.T., Christine Hayes Physical Therapy; Carol Waties, Director, Special Populations and Advocacy of the American Cancer Society; J. Ehmann, M.P.T., C.W.S., C.L.T., and Amanda Zavodnick, M.S.W., L.S.W., Chestnut Hill Hospital; Anne Matthews, O.T.R., C.H.T., C.L.T.-L.A.N.A., Suzanne R. Smith, R.N., M.S.N., Oncology Clinical Specialist/IRB Coordinator, Deborah Markiewicz, M.D., Monica E. Smith, M.D., Rachelle M. Lanciano, M.D., Linda S. Callans, M.D., and Mark A. Rovito, M.D., Delaware County Memorial Hospital; Amy Davis; LympheDIVAs; Assistant Professor Patricia K. Bradley, Ph.D., R.N., C.S., Villanova University College of Nursing; Kay Stevens, Director, Community Outreach of Mercy Hospital of Philadelphia; Veronica Suber, Christian Stronghold Baptist Church, Philadelphia, PA; Stephanie Evans, Salvation Army; Melanie Marchand, Sisters in Shape Fitness; Karen Neyer, LSW Assistant Program Director, Wellness Community of Philadelphia; Lia Daniels and Kara Stephenson, Breast Cancer Resource Center at the Princeton YWCA; Loretta Mikulski, South Jersey Breast Cancer Coalition; Ernestine Delmoor, M.P.H., National Black Leadership Initiative on Cancer; Janet Cash, Sisters Health Initiative; Joanne Poorman, Church Secretary, Ardmore Presbyterian Church; Novella K. Lyons, Executive Director, Women of Faith and Hope, Inc; and Rena Rowan, Breast Center, Abramson Cancer Center of the University of Pennsylvania.