EHR-Linked Social Needs Screening Enters a Diabetes Care Trial in an Urban Safety-Net System

Uncontrolled type 2 diabetes rarely reflects glycemia alone. Food insecurity, housing instability, transportation barriers, and other health-related social needs can interrupt medication access, clinic attendance, and self-management long before a laboratory value rises into dangerous territory. This BMJ Open protocol describes a pilot type 3 hybrid implementation-effectiveness randomised trial designed to test an electronic health record-integrated triage tool paired with a closed-loop referral system for patients with uncontrolled diabetes in a US urban safety-net setting.

The intervention is built around a practical clinical problem: social needs are often identified, documented, and then lost in the handoff to community support. By connecting screening, referral generation, care team documentation, and referral tracking inside one workflow, the study aims to measure not only whether patients are referred, but whether those referrals are completed and whether the process can be delivered reliably in routine care.

A single missed handoff can leave a patient with uncontrolled diabetes without the resources needed to take the next step.

Why Social Needs Function as a Diabetes Care Signal

The protocol frames social needs as the downstream, actionable expression of larger social drivers of health. In bedside terms, that distinction matters because the day-to-day barriers are often the ones a diabetes team can actually detect and address. A patient may understand insulin dosing perfectly and still fail therapy if there is no refrigeration, no stable address for supplies, no transportation to appointments, or no money for food that supports a prescribed regimen.

Inside primary care and specialty clinics, these problems are easy to recognize and difficult to operationalize. Traditional screening can uncover need, but screening alone does not guarantee service connection. The trial therefore uses a closed-loop model, in which referrals are not treated as a one-way administrative task. They become part of a feedback system that can show whether a community-based organization received the referral, whether contact occurred, and whether support was actually delivered.

That design reflects a central implementation challenge in safety-net medicine: identifying need is not the same as closing it.

Randomisation, Fidelity, and the Logic of a Hybrid Trial

Investigators plan a stratified randomised design with intention-to-treat analyses, strengthening internal validity and limiting selection and analytic bias. The protocol also adopts a type 3 hybrid implementation-effectiveness framework, a structure suited to interventions that are not being tested in a vacuum. The question is not only whether the tool can improve referral processes, but whether it can be integrated into clinic workflows with enough fidelity to matter at scale.

The study measures implementation from multiple angles through integration of the electronic health record, the referral platform, and care team documentation. That multidimensional setup should allow assessment of intervention uptake, fidelity, and referral outcomes rather than relying on a single endpoint. In practice, this is important because a social needs workflow can fail at several points: screening may not happen, referrals may not be accepted, patients may not be reachable, or documentation may never circle back to the care team.

The protocol’s structure anticipates those failures rather than ignoring them. It is built to show where the pipeline breaks.

Community Capacity Will Shape What the Data Mean

The safety-net setting gives the trial immediate clinical relevance, but it also imposes constraints that may shape interpretation. The single-system design may limit generalisability to other healthcare environments, especially those without similar staffing, referral infrastructure, or community partnerships. Heterogeneity among community-based organisations is another likely source of variation. Some groups may have robust staffing and intake systems; others may operate with fewer resources and less capacity to absorb referrals.

Those differences matter because referral completion can be influenced by the receiving organisation as much as by the referring clinic. A closed-loop system can document that a referral was sent, yet the downstream service connection still depends on organisational capacity, timing, and patient follow-through. In a real-world social needs intervention, the community partner is not a passive endpoint. It is part of the treatment pathway.

Referral completion may therefore reflect both the strength of the intervention and the strength of the community network around it.

Measuring Whether Care Actually Reaches the Patient

A particularly important feature of the protocol is its attention to ascertainment. Service connection may be incomplete because referral outcome data are not uniformly available across community-based organisations, which creates a clear measurement challenge. If a patient receives support outside the formal electronic record, the system may miss it unless supplemental follow-up captures the outcome.

For implementation research, that is not a minor technicality. Incomplete outcome visibility can blur the difference between a true failure of referral delivery and a failure of documentation. The trial acknowledges that gap and plans supplemental patient follow-up when needed. That approach should help distinguish between a referral that never reached a patient and one that happened outside the electronic footprint.

The distinction is clinically consequential. A workflow that appears weak on paper may still be functioning in practice, while a seemingly successful referral stream may conceal poor patient contact if tracking is incomplete.

What This Protocol Could Change at the Bedside

If the intervention performs as intended, clinicians could gain a more dependable way to move from social needs screening to action. For patients with uncontrolled diabetes, the bedside consequence would be concrete: fewer abandoned referrals, clearer accountability across the clinic-community interface, and better visibility into whether social barriers are being addressed rather than merely recorded.

That is especially relevant in urban safety-net care, where diabetes control is often shaped by instability outside the exam room. A tool embedded in the electronic health record may be able to reduce friction for clinicians who already face heavy documentation burdens. It may also make social needs workflows easier to sustain because the process is integrated into existing care-team documentation instead of layered on as an extra task.

The protocol remains a pilot, so it does not promise clinical outcome improvements yet. Its value lies in testing whether a closed-loop model can function with enough fidelity, reach, and measurable follow-through to justify a larger trial. The most important unanswered questions are operational ones: how consistently the system identifies need, how reliably it completes referrals, and how often the receiving community organizations can document service connection across a real-world safety-net population.


Reporting by Dr. A. Nikhil Krishnan, MBBS for Medicine World Foundation, with additional data provided by BMJ Open.
PRIMARY CLINICAL EVIDENCE & REGISTRY DATA