Background
Randomized controlled trials (RCT) remain the gold standard for evaluating the effectiveness of interventions in healthcare research. Critical to the success of these trials is recruitment of study participants in order to have sufficient statistical power to detect the effects of interventions [
1]. Recruitment is increasingly challenging for clinical trials across all domains of health research, particularly for studies recruiting from the community. Potential barriers include characteristics and stigma associated with the condition being studied, differences in cultural expectations and experience with research across study sites, and challenges engaging community healthcare providers and systems. However, because translational trials evaluate the effectiveness of interventions within the community [
2], these studies are crucial to evaluating the implementation of these interventions in real-world contexts.
Mental health studies can be particularly challenging for recruitment. For those with major depressive disorder (MDD), withdrawal from regular activities is common. Eligible people may be less likely to engage and more difficult to retain in clinical trials, even if leaving the house to attend study appointments is not required [
3]. A review of recruitment to depression treatment trials offering cognitive behavioral therapy (CBT) by computer reported enrollment rates of 2%–60% and uptake rates (actually participating in the treatment) of 3%–25%, representing the wide range of variability in acceptance of and participation in depression treatment trials [
4]. The stigma of mental health conditions is an additional barrier for many [
5]. Participating in mental health studies requires both acceptance of the condition and willingness to address it. This may create a significant struggle for the participant if shame or stigma are involved [
6].
An individual’s culture, social background, and community also powerfully influence their perception of research and likelihood of participation [
7,
8]. In many rural, medically underserved areas, patients commonly have a low opinion of the quality of their locally available healthcare and limited direct experience with participation in clinical research [
9]. Research teams seen as an extension of this establishment may be viewed with skepticism [
10]. In this setting, investigators need to build trust and perform recruitment activities simultaneously. This trust takes time to build and establish and relies on relationships among researchers, patients, and care providers. For example, the Appalachian region is well-known for its overrepresentation of a number of serious medical problems [
11]. Despite the need, there are significant misgivings of the medical system among many people in this region [
9]. This can lead to delays in care and resistance to enrolling in trials [
12,
13].
Multicenter trials necessarily deal with these cultural values at each of the different study sites, with each community having its own sets of values. Efforts to build trust and creative solutions to study-related problems that work in one area may not work in another. This represents an added barrier to successfully recruiting and demands even more effort and flexibility among members of the research team.
The purpose of this paper is to present a case study of recruitment and retention for Program ACTIVE II, a successful, NIDDK-funded behavioral health trial [
14]. We present our recruitment strategy, barriers, and costs associated with this trial.
Methods
Study design
Program ACTIVE II was a multisite RCT for adults with type 2 diabetes and MDD [
14]. The purpose of the study was to test the comparative effectiveness of manualized (i.e. conducting an intervention according to a manual) exercise (EX) and cognitive behavioral (“talk”) therapies (CBT) for the treatment of depression, individually or in combination (EX+CBT), against usual care (UC). The study protocol was approved by the Institutional Review Boards at each of the individual sites (Indiana University 1105005684, 1308973934; Ohio University 11F031; West Virginia University H-23246).
Potential study participants were identified using a variety of methods, described below, from the communities surrounding the three participating study sites. Research staff trained in the study protocol and approved by the local site Institutional Review Boards (IRB) conducted all contacts with potential participants to assess their interest and eligibility. During screening calls, information was provided about the study using an approved script. If interested, participants gave verbal consent to continue with eligibility screening. Self-reported medical information was collected (see Table
1 for medical exclusion criteria). Those not excluded based on medical information were then screened for psychiatric eligibility. Potential participants who were eligible at medical and psychiatric screening by phone were invited to an in-person baseline eligibility assessment (baseline) where each potential participant provides written informed consent before completing any study-related activities. These baseline assessments were held at fitness centers within the three areas where recruitment took place. Each person then completed a psychiatric interview by phone. Following all baseline activities, an enrollment committee formally evaluated baseline results to determine eligibility and appropriateness for the study for each potential participant. Those not eligible were referred to other services. Eligible participants were randomized, notified of their randomization group (EX, CBT, EX+CBT, or UC), and assigned to an intervention provider, if relevant. Details of the study design and eligibility criteria have been detailed elsewhere [
15].
Table 1
Study exclusion criteria
• Age 18 years or older • Not able to walk • Currently pregnant or planning on becoming pregnant in the next year • History of diabetic ketoacidosis • Continuous insulin therapy since type 2 diabetes diagnosis • History of stroke or transient ischemic attack • Recent cardiac event (diagnosed angina, PTCA, any cardiac intervention for CAD or tachydysrhythmias in the past 6 months) • Laser surgery for proliferative retinopathy in the past 6 months • Aortic stenosis or other sever valvular heart disease or atrial fibrillation • Lower limb amputation • Asensory proliferative retinopathy • Uncontrolled hypertension • Current use of daytime oxygen (severe COPD) • Class III or IV heart failure • Diagnosis of type 2 diabetes < 1 year | • No diagnosis of current MDD • Active suicidal ideation or history of suicide attempt • Bipolar depression • Evidence of psychotic symptoms or history of psychotic disorder • Current substance abuse or dependence • Co-morbid anxiety or eating disorders where MDD is not the primary presentation • If on antidepressant medication, stable dose and usage for at least 6 weeks • Current therapy for MDD by a mental health professional |
Recruitment approach
Participants were recruited from rural southeastern Ohio (OH), north-central West Virginia (WV), and central Indiana (IN) communities. To ensure that recruitment goals were met, the study utilized multiple recruitments approaches, continually assessing the effectiveness of each, and adjusting strategies as necessary [
16]. We classified recruitment identification strategies by who made the initial contact. Initial contacts initiated by the
potential participant were classified as Inbound and those the
study team initiated as Outbound.
Inbound recruitment strategies
Inbound recruitment used a flexible, multifaceted approach, involving outreach to physicians, health departments, libraries, grocery stores, pharmacies, advertisements in newspapers, and flyers posted in public areas. As recruitment methods were implemented, each was reviewed for effectiveness, altering or expanding as necessary to meet recruitment goals.
The main initial recruitment strategy for the study was via physician practices. This included providers affiliated with the study and those seeing patients with diabetes (e.g. endocrinologists, primary care providers, nurse practitioners). Providers were presented study information and asked if contact with patients could occur (e.g. letters to patients, in-office contact, phone calls to patients). Patients who received study information were instructed to call the study team. This outreach was later expanded to include pharmacists, diabetes educators, and diabetes education classes. Study staff also attended health fairs hosted by healthcare providers, community organizations, social justice groups, employer-hosted employee health fairs, and diabetes education groups, providing information about diabetes and depression, study promotional materials, and study flyers. Additional flyers and posters were distributed throughout the communities at libraries, churches, community centers, pharmacies, grocery stores, and to health departments to distribute at diabetes classes.
Advertisements were placed in local community newspapers. Limited advertising was also instituted on Facebook, targeting adult users with an interest in diabetes. Further outreach occurred via radio public service announcements (at the OH site), television segments (OH and WV), and radio segments (IN). These media spots aired locally on public radio and television stations as well as during local evening news broadcasts.
Additional outreach was conducted by direct mail or email to fitness facility members (at the IN and OH sites) and patient lists (WV and IN). Study emails were also sent to registrants of the Indiana Clinical and Translational Sciences Institute’s (CTSI) clinical trials registry (INResearch.org). At the IN site, advertisements were placed on the university’s electronic classifieds board, accessible by students, staff, and faculty.
Outbound recruitment strategies
Electronic medical records from physician practices and volunteer research registries were used to identify prospective participants, who were then directly contacted by the study team. This strategy was utilized at the IN and WV sites. WV conducted outreach to local physician offices. Interested practices allowed the study team contact patients first by mail and then by phone. The IN site partnered with ResNet, a service of the Indiana CTSI, to access patient data in the Eskenazi and IU Health systems. In addition, the IN site utilized the Indiana CTSI’s INresearch.org participant registry. After sending initial emails introducing the study, registrants with type 2 diabetes were contacted by a research assistant. The Ohio University IRB did not approve Outbound calling at the OH site. Unlike the other two sites, OH did not have institutional agreements with healthcare systems to share protected health information with researchers to facilitate Outbound outreach.
Analyses
All contacts with potential participants were logged at the time of contact into a study database. Contact information included the date, time, and nature of the contact. Logged contacts included phone calls (such as study introduction calls, screening, scheduling, psychiatric interviews, etc.), mailings (such as appointment scheduling letters), and in-person appointments (such as eligibility assessments). In addition to logging the contact, the time needed to complete the contact was also recorded, allowing us to track the effort spent on various recruitment tasks. These records were reviewed to determine time spent on recruitment activities, broken down into phone screening and baseline eligibility assessment contacts. Descriptive statistics were used to characterize count, cost, and proportions of individuals in each outcome category. Qualitative data were collected from process notes recorded from team meetings over the study period.
Discussion
Community recruitment of participants is critical to the advancement of all forms of science, particularly behavioral translational trials. Increasingly, clinical trials must compete for participant attention, interest, and time against a backdrop of considerable life demands and opportunities to easily obtain reliable health information. Funders and investigators recognize the central role that recruitment plays in the success of funded research. Underestimation of the effort and costs needed for successful recruitment is a modifiable variable that investigators can and should address during the study planning phase of the research. Here, the Program ACTIVE II study [
14] experience is shared as a case example of the breadth and depth of recruitment strategies that were required to overcome barriers to study success as well as the monetary costs incurred (Table
3).
Endemic to the Program ACTIVE II study design, we encountered the following challenges: the specific eligibility criteria of the study (duration of type 2 diabetes for at least one year while meeting full DSM-IV-TR criteria for current MDD; medically appropriate for community-based exercise) [
21], requirements for study participation (attending 2–3 appointments required for the baseline assessment, 10 weekly appointments for the CBT intervention, six exercise classes over eight weeks for the exercise intervention, and two appointments for each of three follow-up assessments) [
21‐
23], stigma associated with both type 2 diabetes and depression [
24], participant beliefs about study participation (e.g. lack of personal gain) [
23,
24], diversification of communication channels to reach potential participants [
22], health literacy issues [
22], and co-morbid illnesses that may impede participation in studies [
22,
23]. Each of these characteristics may have been relevant during study recruitment and reflected in the effort necessary for successful study enrollment.
These findings from the Program ACTIVE II recruitment experience demonstrate the considerable time, effort, and costs associated with recruiting participants for this federally funded behavioral RCT. These expenditures far exceeded investigator expectations and demonstrated important lessons for investigators and sponsors of future clinical trials. The costs incurred specifically for recruitment activities were nearly $1400 per participant, with $836 of this amount attributable to the direct costs required for recruitment of each enrolled participant, in line with direct recruitment costs reported by other diabetes studies [
25,
26]. While grant application budgets tend to be focused on the costs associated with interventions, recruitment costs are substantial and should not be underestimated.
A second lesson was the amount of time needed to meet our recruitment target. Our team anticipated that we would complete recruitment activities within the first 24 months of the funding period. Instead, we required 4.5 years to achieve our goal. The extension of this period was a function of the large number of people we needed to contact to compete enrollment. The recruitment rates of this study (0.6%–1%) were significantly lower than those observed in our previous pilot study (8%) that used identical eligibility criteria [
17], though similar to the 1.3% recruitment rate experienced by another depression treatment trial [
27]. In the Program ACTIVE pilot work, Inbound recruitment approaches were successfully used to achieve a smaller target sample of 50 [
17]. However, replication of those same methods did not yield similar recruitment rates in the larger trial at the same site (Table
4).
Continual reassessment of strategies and adaptations to the recruitment plan were needed throughout the recruitment period. The Outbound calling method was particularly important for Program ACTIVE II as people with depression may not seek mental health treatment because of stigma [
5,
28,
29] or take the initiative to make the initial contact with a study [
30]. This was especially relevant for our study considering that we recruited people with current MDD, who are relatively unlikely to initiate treatment for their depression [
31].
There are limitations to the data presented in this paper. First, tracking of staff activities was not the primary goal of study data collection so some estimation was required in our calculations. As a result, the total effort of study staff may be underestimated for some activities (we erred on the side of underestimation). Additionally, figures reported are based on actual costs over the period of recruitment (2012–2016) and have not been adjusted for inflation. These numbers should be adjusted by investigators for use in future research budgets.
A critical element in the ultimate success of this trial was the flexibility and continued support by the funding agency as the investigators adapted to barriers to recruitment. The agency permitted additional time to complete recruitment and demonstrated confidence in the investigators, who were the most knowledgeable about their local barriers and needs. Had restrictions been placed on study funding based original assumptions about recruitment rates and strategies, the investigators would not have been able to demonstrate the effectiveness of the interventions [
32].
Acknowledgements
The Program ACTIVE Research Team wishes to thank the following individuals for their many contributions to this study: Rachel Clift, RN; Susan Eason, M.A.; David Donley, M.S.; Alex Tylka, M.S.; Lindsey Sams, B.S.; Jaclyn Babe, B.S.; Christiaan Abildso, Ph.D; Daniel Bonner, M.S., Melinda Ruberg, B.S.; Frank Schwartz, M.D., Bernadette Heckman, Ph.D.; Cammie Starner; Lynn Petrik; Tracey Garrett; Molly Long, M.S.; Michael Craven, M.S.; Allysia Ramirez, M.S.; Amber Massa, M.S.; Chelsea Holbert, M.A.; Debby Wimer; Kelly Chaudoin, M.A.; Sarah Mielens, M.A.; Ellen Knapp, M.A.; Kent Crick, M.S.; Danielle Epler, M.S.; Michelle Weinstein, M.S.; Kasey Goodpaster, Ph.D.; Brett McKinney; and Kisha Wilkerson.
The authors also thank all of the participants of Program ACTIVE II who gave generously of their time and energy to engage in a new research activity. We are grateful for their generosity of spirit and their willingness to try something new.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.