The meSH study used the InCharge Health mobile app in an implementation study with three aims: (1) improve patient adherence to hydroxyurea, (2) improve provider hydroxyurea prescribing behaviors, and (3) evaluate barriers and facilitators to implementation of mobile health interventions [6,7]. The meSH study design used a staggered site initiation across 7 sites, with the first 2 sites beginning the enrollment process in September 2019 for providers and in November 2019 for patients. The SCDIC consortium includes clinical sites in academic and nonacademic settings and in both rural and urban settings. Development of this cohort used implementation science principles previously described in detail (ClinicalTrials.Gov NCT03380351) [12].

For our evaluation of meSH recruitment strategies, the eligibility criterion was functioning as the lead study coordinator for a site (all lead study coordinators agreed to participate). For the meSH study itself, the eligibility criteria for providers included being a physician (including a physician in training) or advanced practice provider (ie, nurse practitioner or physician assistant) who cared for at least 1 patient with SCD for an anticipated minimum of 12 months from study enrollment and had access to a smartphone (either Android or iOS) or a computer with internet connectivity (the HU Toolbox app can be accessed via the internet on any device) [6,7]. Eligibility criteria for patients included being aged 15 to 45 years; receiving treatment at or being affiliated with one of the SCDIC sites; speaking English; having a confirmed SCD diagnosis by a hemoglobin fractionation test; owning a smartphone (either Android or iOS); and being treated with hydroxyurea therapy, defined as already receiving hydroxyurea therapy (at least 1 previous prescription for hydroxyurea in the past 3 months) or initiating hydroxyurea therapy (the first prescription must have been written at study enrollment) [6,7].

Enrollment of patients was initially planned to occur in person during routine clinic visits. Enrollment of providers was planned in person, to occur during staff meetings, at clinics, or in the providers’ offices. The method of enrollment, modification of the enrollment method, and refusal reasons were tracked prospectively. Ineligibility was documented during the screening process and captured via CONSORT forms, and reasons for declining were collected by site study coordinators. REDCap (Research Electronic Data Capture; Vanderbilt University), a secure electronic database, was used to administer and store the surveys, survey data, and consent forms.

To better understand the barriers and adaptations to study recruitment, we surveyed the primary research coordinator responsible for recruitment at each of the 7 participating sites with 9 open-ended questions regarding recruitment strategies, adaptations to enrollment methods, and challenges for enrolling patients and providers (Multimedia Appendix 1). The interview guides included were open-ended questions developed to elicit (1) methods used for recruitment, specifically the details of modifications to the original recruitment process listed in the protocol, such as adaptations or deviations from the in-person recruitment process originally planned; (2) temporal variations of each recruitment modality (ie, how long each recruitment strategy was used and when it was changed), (3) reasons for modification of recruitment strategies, and (4) the perception of success for each recruitment strategy. Responses were deidentified and tabulated, and a thematic content analysis was conducted. Responses were first organized inductively by the main domains they addressed and then categorized based on the identified thematic group, allowing for the organic emergence of themes.

We summarized provider and patient participant characteristics (as frequencies and percentages) overall and by site to illustrate site-level differences. Statistical comparisons were made using chi-square tests with an α level of .05. All analyses were conducted using SAS (version 9.4; SAS Institute).

For deidentified participant data access, the original data for this study will be made available at BioLINCC [13]. The meSH study attained institutional review board (IRB) approval at all 7 participating sites (St. Jude Children’s Research Hospital, 19-0159; Augusta University IRB, 1528751; UIC IRB, 2020-0087; Duke University IRB, Pro00073506; Mount Sinai IRB, STUDY-20-00303; Medical University of South Carolina [MUSC] IRB, Pro00097832; Washington University at St. Louis IRB, 202003150). All patient participants (or their legal guardians) signed informed consent, and all provider participants gave verbal assent before any study procedure.

All eligible providers in all 7 sites were approached to participate. Of the 104 approached, 89 enrolled, an 86% study acceptance rate. Of the 89 providers enrolled, most were female (65/89, 73%), aged 25 to 44 years (53/89, 60%), White (47/89, 53%), and not Hispanic or Latino (84/89, 94%); most considered themselves SCD specialists [12] (54/89, 61%) (Table 1). The provider recruitment and enrollment duration ranged from 2 to 7 months across all 7 sites. Initially, all sites planned to use an in-person recruitment approach. Some opted to introduce the study via a presentation at regularly scheduled staff meetings, followed by in-person individual provider enrollment. With the onset of the COVID-19 pandemic in March 2020, most sites suspended in-person research activities, which posed recruitment challenges [3]. At that time, recruitment to meSH pivoted from an in-person to a virtual approach. Virtual enrollment of providers was conducted by email. An introductory email was sent by the site study coordinators that detailed the meSH study and inquired about interest in participation. This email was followed by reminder emails every 1 to 2 weeks until recruitment ended in March of 2021. In total, 41 of 54 (76%) providers approached virtually accepted, compared to 48 of 50 (96%) providers who were approached in-person and enrolled (P<.001).

The most common challenges site coordinators faced in recruiting providers were a lack of response, a lack of interest due to an impersonal experience or perceived lack of need for intervention, a lack of time in provider schedules, and technical issues with the provider’s phone, such as outdated software that inhibited them from downloading the intervention app.

The total patient population treated with hydroxyurea across all 7 sites was 2208 patients. Of those, 2157 (97.6%) patients were screened for eligibility. Of the 2157 individuals screened, 611 (28.3%) patients were eligible for the study, and 343 (56%) were approached for enrollment (Figure 1).

Of the 343 eligible patients approached, 293 accepted participation and were successfully enrolled across all 7 sites, an 85.4% participation rate. Among those enrolled, most patients were aged between 26 to 45 years (164/293, 56%), were Black or African American (283/293, 97%), were not Hispanic or Latino (279/293, 95%), were never married (242/293, 83%), had a high school degree or some college or vocational training (176/293, 60%), were not employed (115/293, 39%), and had a household income of $25,000 or less (166/293, 57%) (Table 2). In total, 126 of 143 (88%) patients virtually approached were enrolled, compared to 167 of 200 (83.5%) patients who were approached in person (P=.23) (Figure 1).

Although slightly lower than the projected 46 patients per site (322 total patients), our total patient participant sample size of 293 met the requirements of the a priori statistical design [5].

A total of 1546 patients were screened and deemed ineligible across all 7 sites, with the most common reason for ineligibility being “not treated with hydroxyurea” for 49.8% of patients. A total of 50 patients who were approached declined participation. Reasons for declining participation included not being interested in research, not having time to participate in research, and not being interested in the app because of already having a reminder system or thinking the app would not help them.

Patient recruitment challenges fell into 3 themes: lack of interest in participating in research, lack of resources (staff and space), and lack of proper documentation to verify study eligibility. Specifically, challenges included difficulty verifying hydroxyurea status, technical issues such as nonworking cell phones or phone storage issues, inability to reach patients due to reduced clinic flow as patients opted for telehealth or telemedicine visits, wrong or disconnected phone numbers, and lack of interest in research participation.

Patient recruitment was initially designed to be conducted fully in person. In-person recruitment involved screening and approaching eligible patients during a regularly scheduled, nonacute clinic visit. Due to the COVID-19 pandemic restrictions for in-person research activities, virtual enrollment was initiated. The sites most affected by clinic shutdowns were the first 2 sites, as previously reported [9]. Although clinic lockdowns were no longer in effect when the remaining 5 sites opened enrollment, these sites were also affected by long-term changes related to the pandemic (eg, staff resignation, remote working). The subsequent sites, however, benefited from institutional adjustments to the ongoing pandemic and amended their protocols to include remote recruitment before initiating enrollment. The approach for virtual enrollment included telephone and Zoom (Zoom Video Communications, Inc) calls. Research coordinators at some sites attempted to join clinicians during telehealth patient visits. However, leveraging telehealth clinical visits to conduct study enrollment was unsuccessful. It was not feasible for some sites to add extra time to telehealth clinical visits for research purposes because of already strained logistic issues with establishing connections and providers feeling overwhelmed with managing a new system. One site used the electronic health record (EHR) to send patients secure messages through the patient portal. This approach was intended to create study awareness among eligible participants and to prime them to expect a recruitment phone call. Other sites that transitioned to virtual enrollment contacted patients via phone calls and text messages.

While most of the challenges coordinators faced were during the enrollment phase, coordinators also expressed challenges during the screening portion of the recruitment phase. One challenge was verifying eligibility due to poor documentation of hydroxyurea use in the patients’ charts. In some cases, coordinators found it difficult to search the EHR to find a definitive answer to whether participants were taking hydroxyurea and, if they were, the start and stop dates. Multiple study coordinators reported having a difficult time finding patients that were interested in participating in the study. When asked, most participants gave vague answers such as “just being uninterested in research.” At the same time, some gave more specific answers such as “not having time for study-related activities” or “participating in too many studies (research fatigue).” To overcome these challenges, some sites modified their screening process to optimize identifying eligible patients. For instance, one site sought the recommendations of the hematologists for participants to enroll and for permission to approach them. However, this strategy of seeking provider input ultimately was not sustainable as it consumed too much time for the research coordinators and providers. To save time, the research coordinators sent courtesy messages to the SCD providers to let them know which patients they planned to approach for enrollment. Providers accepted this approach as they believed it led to the least interference with clinic flow and impact on their time.

During the enrollment phase, some coordinators also mentioned a lack of physical and human resources, such as a shortage of research staff (as some resigned during the COVID-19 pandemic) and a lack of space to conduct research, as areas of their clinics and buildings reduced space for research activities due to restrictions on in-person contact. Multiple coordinators mentioned that the biggest hindrance was a need for more staff to help support study implementation. The reasons for staff shortage varied. Still, one common theme was the presence of other studies competing for the time and focus of a limited number of study coordinators. Coordinators adapted their recruitment strategies to overcome challenges. Coordinators with working space restrictions began to rely heavily on remote methods (eg, phone, text message, telehealth) for recruitment instead of in-person methods and successfully sustained recruitment activities. Those with technology issues worked closely with their IT departments to reach feasible solutions, including a tip sheet provided by the IT department to the coordinators for reference for future problems. For sites experiencing staffing shortages and competing studies, coordinators prioritized the enrollment goals of the meSH study while more staff were being hired.