Social needs and social determinants of health (SDOH) significantly outrank medical care when considering the impact on a person’s length and quality of life [1]. Rising health care costs and worsening life expectancy in the United States are, in large part, the direct result of these unmet social needs and SDOH challenges [2-6]. Challenges related to social needs and SDOH disproportionately impact racial and ethnic minority populations and those with low income and are critical factors in explaining many health-related disparities, ultimately leading to higher mortality rates in these populations [7].

Currently, many sources of data on social needs and SDOH challenges can be found within a typical electronic health record (EHR) and some vendors have started adding specific data fields for collecting information on social needs and SDOH challenges. However, no universally accepted or standardized format exists for documenting social needs and SDOH information [8-13]. Furthermore, delivery systems can now take advantage of recent health IT advances in offering automated integration of new sources of relevant social needs and SDOH information [8,14,15]. The development of cross-provider regional health information exchanges, such as Maryland Health Information Exchange (known as CRISP), also offers the opportunity to connect social needs and SDOH data from multiple clinical organizations and to share information across medical and social sectors (eg, primary care practices and community-based organizations [CBOs]) [16].

The extensive range of available social needs and SDOH data will offer a better understanding of how individual and neighborhood challenges could impact a patient’s social needs and SDOH and increased risk of adverse health outcomes [17-22]. However, many technical and organizational challenges must be surmounted before social needs–integrated health IT solutions can be implemented on a wide scale. For example, the literature provides evidence of existing gaps in the development of best practices for integrating social needs and SDOH data into the clinical workflow and the clinicians’ core clinical decision support (CDS) tools [23] and data sharing with CBOs [24].

Also, multiple practical gaps exist in addressing social needs and SDOH in underlying care management tools. Over the past decade, the new generation of “value-based” health insurance plans and integrated delivery systems (eg, Patient-Centered Medical Homes, and Accountable Health Organizations) have fully embraced available Health IT systems to implement clinically oriented risk identification and stratification CDS tools addressing the needs of medically underserved patients [25]. These mechanisms are commonly used to identify and refer the subset of patients who are high-risk or high-cost (ie, with serious and multiple chronic conditions) who need care management services [25]. Today, more providers are placed at financial risk for a wider range of nontraditional performance-based target indicators (eg, neighborhood health indicators) and more diverse populations. Hence, there is a rapidly growing gap regarding the use of social needs and SDOH risk information, relative to clinical or morbidity risk factors, within current case finding and risk stratification methodologies. Such a gap is critical for the large proportion of patients, especially among minority populations, who, in addition to having high clinical needs, concomitantly face behavioral, socioeconomic, and community-resource challenges [26].

Some health plans and provider systems have started to address the social needs of their patients by implementing EHR-based survey assessment tools and navigation services in their clinical and digital workflows [27-31]. Moreover, a growing body of evidence such as the Accountable Health Communities model by the Centers for Medicare & Medicaid Services provides support for the implementation of such services in the health care settings and their impact on improving the care process and health outcomes [30,32,33]. However, automated and standardized social needs and SDOH risk identification metrics and data collection and curation processes are sorely lacking. To avoid burdensome and inefficient social needs assessment (data collection from all patients at each visit), it will be essential to develop automated screening tools using EHR or community databases to help identify the subset of patients who would most benefit from EHR-integrated social needs assessment and data collection [34]. Also, currently in most health systems, the referral to CBOs is based on clinic staff’s one-off attempt to identify available CBOs and to make referral arrangements. And, in cases where an automated directory of services is used, information is rarely exchanged that might benefit the patient [24,29,30]. The disconnect between health systems and CBOs and the lack of established data-sharing mechanisms have many negative implications for the patient and the system. For example, it is difficult for the providers to identify whether the social needs of their patients were addressed [8,28,35]. From a system perspective, such disconnect makes it difficult to evaluate the impact of identification, assessment, and referral of social needs on health outcomes and health care use, as well as to evaluate the efforts of health systems to address disparate access to medical and social services [28].

Thus, integrating data on social needs and SDOH challenges along with clinical risk information within operational CDS systems built into EHRs is an effective approach to addressing health-related social needs. To achieve this goal applied research is needed to (1) identify optimal solutions for the effective collection and application of social needs and SDOH information within EHRs; (2) develop EHR-integrated CDS tools and closed-loop referral systems; and (3) implement the CDS tools and closed-loop referral systems in the digital and clinical workflows at health care systems and collaborating CBOs.

The objective of this mixed methods study including a randomized controlled trial (RCT) and a qualitative phase is to address the current gaps in the EHR-based social needs assessment and navigation services. Thus, we plan to assess the feasibility, acceptability, and effectiveness of an EHR-integrated digital platform to identify patients with social needs and provide navigation services including CBO referrals. The digital platform includes a CDS tool and a closed-loop CBO referral. The CDS tool contains a social risk score and an overview of different risk factors and comorbidities contributing to the patient’s social risk score. The CDS tool also provides recommendations on assessment for social needs based on the social risk score and other available information (refer to Multimedia Appendix 1 for an explanation and overview of the CDS tool).

Using both patient- and population-level data, we developed a social predictive risk score based entirely on electronic information readily available within most health care delivery systems. We also validated the model using multiple denominators to ensure generalizability and retrained and tested the model for each subpopulation of interest (eg, individuals aged 65 years or older, racial and ethnic minority populations, and those living in the most and least disadvantaged neighborhoods) [34]. The predictive risk score helps providers systematically identify patients at risk of having social needs based on their demographic characteristics, clinical comorbidities, previous social needs, and clinical outcomes such as hospitalization and emergency department visits. Those patients represent likely targets for further assessment of their social needs and ultimately potential referral to CBOs to address such needs. Using this systematic electronic case-finding screening approach helps the health care system avoid burdensome and inefficient social needs assessment (eg, primary data collection from every patient at every visit).

Building on the current evidence on the experience of patients, providers, and health care systems with the EHR-based social needs assessment and navigation services [32] as well as the impact of such systems on care process and health outcomes [33], we plan to test the central hypothesis that implementing the digital platform for social needs identification and providing navigation services will (1) be feasible in digital and clinical workflows of primary care ambulatory clinics and CBOs; (2) be acceptable by patients, clinical providers, and CBO staff; and (3) demonstrate effectiveness in improving care process and ultimately the clinical outcomes.

We will apply an RCT design to compare selected outcomes between the intervention (digital platform) and control (standard of care) groups during their primary care ambulatory visits. The RCT will take place in 5 primary care ambulatory clinics at Johns Hopkins Health System (JHHS) and 5 CBOs in Baltimore City. The CBOs are specialized in providing services related to different categories of social needs including residential instability (ie, homelessness, housing insecurity, and poor-quality housing), food insecurity, and transportation. We will identify a clinical provider champion in each pilot primary care clinic who will serve as the liaison between the study team and the clinic. The provider champion will help the research team with the implementation of the platform in the digital and clinical workflow and will advocate for the study among other providers. Figure 1 describes the RCT’s procedural flow.

We will identify eligible patients via EHR in the JHHS pilot primary care clinics. A research project manager will contact patients after obtaining their permission via mail, email, and patient portal communications. After verifying eligibility and introducing the study, interested patients will complete consent processes. We will randomize the patients to the intervention or control arm for a 3-month study period. The patients in the intervention group will be contacted by staff at Hopkins Community Connection (HCC; a team of trained social workers and care managers, serving as a hub at JHHS providing standardized social needs assessment and navigation services) within 48 weekday hours of enrollment and receive social needs assessment and navigation services based on their social risk score and other information available in the CDS tool. The patients in the control group will receive the standard of care consisting of provider assessment, assessing, and addressing social needs on an as-needed basis based on the information provided during the visit. The patients in the intervention group will complete an exit interview via phone 3 months after the intervention, assessing their experience with the digital and clinical work process. For patients in the control group, we will retrieve any information on social needs assessment and referral services provided as a standard of care from their EHR.

To evaluate the feasibility and acceptability of the digital platform and its impact on the clinical and digital workflow and patient care we will conduct focus groups with the care teams at various pilot primary care clinics (eg, clinical providers), HCC staff, and CBO staff to determine satisfaction of users and identify any facilitators or barriers to using the digital platform.

We will assess whether the EHR-integrated digital platform is superior to the standard of care and more feasible and acceptable. Thus, we will track the outcomes of feasibility, acceptability, and effectiveness through the RCT and qualitative study components (Table 1). We will report feasibility outcome measures as weekly use rates of the digital platform (the CDS tool and closed-loop CBO referral). The acceptability outcome measure will be the provider’s responses to the truthfulness of a statement indicating a willingness to use the digital platform in the future. Effectiveness (ie, the primary outcomes) will be measured by tracking the 3-month process and clinical outcomes versus the baseline visits comparing the intervention and control arms.

We will adhere to the standards for trial design, analysis, and reporting per the CONSORT (Consolidated Standards of Reporting Trials) [43]. We will perform univariate analysis including demographic characteristics, clinical comorbidities, previous social needs, and clinical outcomes such as hospitalization and emergency department visits among all patients as well as in the intervention and control arms. We will use frequencies, means, and standard deviations related to feasibility and acceptability outcomes. For the effectiveness outcomes, we will use the change in the number of identified social needs and provided navigation services from baseline to 3-month follow-up and proportion with any hospitalization and emergency department visits comparing the 2 study arms using t test and Pearson chi-square tests. We will also use a logistic regression model to include other potential predictors of the outcome to improve the precision of the estimate. All trial procedures will also be in concordance with ClinicalTrials.gov regulations (NCT05574699).

For focus groups, we will deidentify the transcripts and upload them into MAXQDA [42] for data management and analyses. We will deductively develop an initial codebook from the interview guide and apply the deductive codes to the first few interview transcripts. In addition, we will inductively identify emergent subcodes for each parent code, resulting in several subcodes. This initial coding process will be conducted by the PI and coinvestigators with expertise in qualitative research. We will discuss the disagreements in coding and reach a consensus by mutual agreement. One coder will then code the remainder of the interview data and both reviewers will discuss emergent findings and modifications to the coding framework. We will engage in a process of constant comparison of emergent findings throughout the analytic process, and when no new codes can be identified, we will consider thematic saturation to have been achieved.

This study protocol was approved by the Johns Hopkins University, School of Medicine’s institutional review board (IRB; #IRB00354803).

We will disseminate the study results regardless of effect direction and size through publications in peer-reviewed journals and presentations at conferences. Moreover, we will develop guidelines and algorithms to support other health care systems wishing to apply our methods and tools. These will include logic flow for an EHR-integrated digital platform containing the risk assessment process, the navigation services, and the referral process. In addition, we will provide information on challenges surrounding data sharing between primary care practice sites and CBOs. Our recommendations will suggest how to address the variabilities in information technology infrastructure across different CBOs. We will also develop recommendations for how best to scale the implementation of the EHR-integrated digital platform, across the entire JHHS, regionally, and nationally. Finally, we will vigorously disseminate our findings and methods to major provider networks, technology vendors, policy makers, and other stakeholders.