Compass Rose for Generating Community-based Referrals via Epic: Best Practices and Lessons Learned

<h3>Objective:</h3> The purpose of this study was to study the implementation of the Compass Rose toolset. <h3>Study Design and Analysis:</h3> We used information gathered from site visits and team meetings to understand how Compass Rose was being utilized across sites. We also assessed patterns of social risks and need for assistance among our patient populations. <h3>Setting or Dataset:</h3> We used data from Epic on social needs screening and Compass Rose utilization. <h3>Population Studied:</h3> Three sites, representing 13 urban and rural primary care practices in Minnesota, Wisconsin, and Florida participated in the Compass Rose pilot. <h3>Intervention/Instrument:</h3> The screening tool used in the pilot sites included five questions to assess needs related to transportation, food insecurity, utility payments, housing, and a final question to assess patients’ willingness to receive help. A Compass Rose episode was automatically generated in the patient’s medical record in Epic if patients indicated that they had at least one of the four social risk factors on the screening tool. The workflows for use of the tool, including enrollment criteria, patient outreach, and use of the Findhelp platform for community-based referrals were standardized; however, each pilot site had different staff members responsible for using the tool. <h3>Outcome Measures:</h3> We assessed patterns of social risk factors across our patient population as well as the extent to which patients wanted assistance from the clinic. <h3>Results:</h3> Between January 2024 and March 2024, the pilot sites generated 861 Compass Rose episodes. Of those patients with an episode (i.e., at least one social risk factor), 486 (56%) declined assistance and 375 (44%) requested support to help address their need(s). The most common type of support requested related to transportation needs (42%), followed by assistance related to food insecurity (25%), utilities (12%), and housing (9%). <h3>Conclusions:</h3> Findings indicate that Compass Rose was adaptable across a variety of clinical contexts, easy to use, and essential for the critical closing the referral feedback loop between staff at community-based organizations and clinical teams. We had lower referral completions than expected because of the difficulty connecting with patients asynchronously. As a next step, we will explore solutions to the challenges identified during implementation, including developing and refining workflows that support both synchronous and asynchronous interventions.