Hospital recruitment for a pragmatic cluster-randomized clinical trial: Lessons learned from the COMPASS study

Abstract

BackgroundPragmatic randomized clinical trials are essential to determine the effectiveness of interventions in “real-world” clinical practice. These trials frequently use a cluster-randomized methodology, with randomization at the site level. Despite policymakers’ increased interest in supporting pragmatic randomized clinical trials, no studies to date have reported on the unique recruitment challenges faced by cluster-randomized pragmatic trials. We investigated key challenges and successful strategies for hospital recruitment in the Comprehensive Post-Acute Stroke Services (COMPASS) study.MethodsThe COMPASS study is designed to compare the effectiveness of the COMPASS model versus usual care in improving functional outcomes, reducing the numbers of hospital readmissions, and reducing caregiver strain for patients discharged home after stroke or transient ischemic attack. This model integrates early supported discharge planning with transitional care management, including nurse-led follow-up phone calls after 2, 30, and 60 days and an in-person clinic visit at 7–14 days involving a functional assessment and neurological examination. We present descriptive statistics of the characteristics of successfully recruited hospitals compared with all eligible hospitals, reasons for non-participation, and effective recruitment strategies.ResultsWe successfully recruited 41 (43%) of 95 eligible North Carolina hospitals. Leading, non-exclusive reasons for non-participation included: insufficient staff or financial resources (n = 33, 61%), lack of health system support (n = 16, 30%), and lack of support of individual decision-makers (n = 11, 20%). Successful recruitment strategies included: building and nurturing relationships, engaging team members and community partners with a diverse skill mix, identifying gatekeepers, finding mutually beneficial solutions, having a central institutional review board, sharing published pilot data, and integrating contracts and review board administrators.ConclusionsAlthough we incorporated strategies based on the best available evidence at the outset of the study, hospital recruitment required three times as much time and considerably more staff than anticipated. To reach our goal, we tailored strategies to individuals, hospitals, and health systems. Successful recruitment of a sufficient number and representative mix of hospitals requires considerable preparation, planning, and flexibility. Strategies presented here may assist future trial organizers in implementing cluster-randomized pragmatic trials.Trial registrationClinicaltrials.gov, NCT02588664. Registered on 23 October 2015.Electronic supplementary materialThe online version of this article (doi:10.1186/s13063-017-2434-1) contains supplementary material, which is available to authorized users.

Challenges of conducting research in long-term care facilities: a systematic review

“…Many of the challenges and barriers to research described above would be similar to studying a frail elderly patient population in acute care or community samples. For example, in terms of recruiting sites from an acute hospital setting, in two recent larger scale clinical trials on subjects of variable age groups, 31% and 43% of hospitals contacted eventually participated [ 64 , 65 ], and while higher than the LTC site recruitment figures quoted above, these are still low. In terms of under-recruitment and attrition of elderly subjects in clinical studies, this factor is certainly not unique to LTC and is evident in many health care settings [ 10 , 66 , 67 ].…”

“…Staff time constraint or staff non-cooperation as a barrier to research participation is also not exclusive to LTC. For example, they are common reasons for non-participation in research in hospitals or palliative care facilities as well [ 65 , 73 ]. Methodological challenges are not limited to research in LTC.…”

Implementation of a billable transitional care model for stroke patients: the COMPASS study

“…Ninety-five hospitals were eligible and invited to participate, of which 41 (43%) signed a letter of agreement to participate, received IRB approval, and were randomized in 40 units. Details of hospital recruitment have been published [11]. Participating hospitals, compared to non-participating hospitals, were more likely to be Primary Stroke Centers (59% vs. 41%), located in metropolitan areas (54% vs. 45%), and have high stroke volume (29% vs. 17% having ≥300 stroke discharges per year) [11].…”

“…The COMPASS study design and selection of hospitals are described in detail elsewhere [10, 11]. Briefly, hospitals were randomized to either the intervention (COMPASS-TC) or control (usual care) arm in Phase 1.…”

A Person‐Centered Approach to Poststroke Care: The COMprehensive Post‐Acute Stroke Services Model

“…As the COMPASS coordinating site, the Wake Forest Baptist Medical Center Comprehensive Stroke Center was designated the vanguard site, where the COMPASS model was developed, piloted, and refined 6 weeks before launching the COMPASS trial. Based on this experience, we continuously refine implementation of the intervention to address challenges identified in clinical practice, training, and workflow …”