From Screening to Resolution: Governing Social-Needs Referrals and Accountable Community Partnerships

Consider a patient who leaves the hospital with a referral for food assistance. The electronic message is transmitted, the receiving organization acknowledges it, and the hospital records the task as complete. The patient may still face an ineligible program, an inaccessible location, an unfamiliar telephone number, or a service that has reached capacity. This illustrative situation exposes a governance problem: different participants can use the word “complete” to describe different events.

A mixed-methods study across three Chicago-area hospitals found gaps between discharge referrals and patients’ use of community resources. Patients and community organizations described barriers involving access, language, perceived benefit, staffing, funding, and the ability to track referrals. The small sample supports understanding implementation problems; it does not provide a population estimate of referral effectiveness. Its practical lesson is that transmission and utilization need separate attention. [1]

An evaluation framework developed at MetroHealth separated screening, identified need, consent, referral placement, referral acceptance, and referral outcome. Completion varied across these stages and across patient characteristics. This makes a useful starting point for executive oversight, provided that the organization also states what it means by a resolved outcome and incorporates the patient’s perspective. A service organization’s status code is not automatically equivalent to a meaningful improvement in someone’s circumstances. [2]

The executive decision is therefore broader than which screening tool or referral platform to buy. It includes what the health system promises, which needs it can help address, what it asks community partners to do, and how it responds when assistance is unavailable. A credible program can acknowledge limits while remaining accountable for a respectful, useful response.

Define the purpose of the program in terms that patients and staff can understand. The organization might offer help identifying relevant resources, assistance with an application, navigation through a complex eligibility process, or a defined service through a funded partnership. Those are different commitments. Avoid implying that a positive screen guarantees housing, financial assistance, or another scarce resource.

A randomized pilot comparing clinic-based resource specialists with a centralized call center found no significant difference in patient-reported social needs over follow-up. An exploratory analysis of referral receipt was more favorable, but it does not carry the same causal interpretation as the randomized comparison. The organizational question is not whether local or centralized navigation is universally superior. It is which model can provide reliable support for the intended population and which outcomes that model can reasonably influence. [3]

Interest in help also varies among people who report a need. A cross-sectional pediatric study found that a substantial proportion of caregivers with identified social needs were not interested in assistance from the clinical practice. The study examines participation, not whether declining help was beneficial or harmful. Its implication is to ask what support is wanted, rather than assume that every positive answer authorizes a referral. [4]

The service promise should describe how preferences are recorded, whether patients may return later, and what happens if a resource is unavailable. Staff need permission to explain uncertainty honestly. Promising a solution to encourage screening can damage trust if the available service is only a directory or an unanswered referral. The program should be designed around its actual capabilities and a plan for strengthening them.

Social-needs information can be sensitive. Patients may be deciding whether disclosure will help, whether they will be judged, who will see the information, and whether sharing it could have consequences they do not want. An efficient questionnaire cannot settle those questions by itself. The introduction, setting, language, and response to a declined question are part of the intervention.

Qualitative research on patient engagement found both positive experiences of feeling supported and negative experiences involving discomfort, suspicion, ineligibility, technology problems, or frustration with resources. These findings should inform how staff explain the purpose of screening and how they respond afterward. They do not support a single script as sufficient for every patient or setting. [5]

A scoping review of digital screening tools identified a developing evidence base with varied technologies, small or single-site studies, and integration challenges. Digital access can make screening more convenient, but convenience should be assessed from the patient’s perspective. Provide a feasible alternative when a portal, tablet, written form, or remote conversation is inaccessible or unsuitable. [6]

In a qualitative study involving Black and Latine primary-care patients, participants described weighing the perceived risks of disclosure against expectations of receiving useful support. Mistrust, privacy concerns, judgment, and structural barriers mattered. These findings support transparent communication and accountability throughout the pathway. They should not be turned into assumptions about what an individual patient will prefer because of their race, ethnicity, or language. [7]

Explain what information is requested, why it may help care, who may receive it, and how participation can be declined. Use qualified language support and accessible communication according to the organization’s requirements. Collect what the program can use responsibly. If a patient reports an immediate safety concern, the established clinical or emergency response pathway should take priority over routine resource navigation.

Navigation includes finding a suitable service, checking eligibility, arranging contact, helping with an application, resolving failed connections, and learning what happened. These activities take time and can require several forms of expertise. A referral platform can record the work, but it does not determine who has the capacity or authority to complete it.

Research comparing on-site and remote navigator availability in pediatric community health centers found that on-site availability was associated with more clinician referrals, while contact and resource receipt did not differ significantly between groups. This distinguishes clinician behavior from patient outcomes. A navigator’s visibility can change referral volume without resolving the later barriers that determine whether a family receives help. [8]

An emergency-department implementation study likewise found losses along the pathway from screening to outreach and community referral. Staff discomfort, assumptions about whom to screen, and patient concerns affected implementation. The study supports examining reach and workflow together, rather than treating an electronic referral as evidence that the intended population has been served. [9]

Give each active referral a named accountable role, a next action, and a review date. Define who handles unanswered outreach, incorrect contact details, eligibility questions, and changes in patient preference. These may be shared tasks, but shared work still needs an accountable owner. The hospital should not leave a patient to coordinate between organizations that each believe the other is following up.

An EHR-integrated assessment and navigation platform has been examined in a preliminary randomized study in Baltimore primary-care clinics. The study offers useful implementation information, but its early enrollment, small groups, and exploratory subgroup comparisons warrant caution. Digital integration should be evaluated as part of a service model, with attention to who is reached and what assistance is actually delivered. [10]

The system should distinguish at least six events: a need is identified, the patient wants help, sharing is authorized as applicable, a referral is accepted, a service is received, and progress is assessed. These events may not occur in a simple sequence. A patient can change priorities, a partner can decline a referral, or a service can provide temporary help while the underlying need persists.

Avoid a single closure button that erases those differences. Recommended outcome states include service received, patient reports improvement, patient reports no improvement, patient declined further assistance, ineligible, service unavailable, unable to contact, and outcome unknown. These are proposed operational categories. The local team should adapt them with patients and partners, define each clearly, and avoid collecting detail that serves no useful purpose.

Preserve the ability to record more than one outcome. A food referral may be successful while a housing referral remains unresolved. A family may obtain short-term support and still need sustained assistance. Patient-level summaries should make these distinctions visible without requiring staff to read a long chain of free-text notes or mistakenly infer that every need is resolved because one task closed.

Community organizations are service partners with their own missions, expertise, constraints, and accountability. They are not an unlimited destination for work generated by a health system. A referral program that increases demand without addressing capacity can place strain on the very organizations patients depend on.

A realist study involving third-sector organizations in Scotland described how referral quality, link-worker experience, follow-up, relationships, resources, and perceived power imbalances influenced collaboration. Its findings are explanations drawn from participants and contexts, rather than trial estimates of patient benefit. They nevertheless show why a directory of organizations is an inadequate partnership strategy. [11]

Before expanding referrals, agree on eligibility, service availability, communication routes, expected response intervals, the information genuinely needed, and who will handle a failed connection. Discuss funding for coordination and reporting as well as service delivery. If the health system requires a partner to use new software or complete additional documentation, those requirements have a cost even when no invoice is submitted.

Create a regular forum where community partners can identify inaccurate referrals, unmet needs, burdensome processes, and capacity changes. Give that forum a route to decisions about contracts, staffing, data requirements, and program scope. Consultation without the ability to change the arrangement can become another demand on already constrained organizations.

Some problems involve several systems at once. Housing instability may involve benefits, tenancy, disability accommodation, or another legal issue. A healthcare navigator should recognize when a specialized referral is appropriate without making promises about legal eligibility or outcomes. Similarly, a health system should not expect one community organization to resolve every type of social need.

A national survey on social-needs screening and medical-legal partnerships found varying experience with screening and willingness to use legal assistance. This supports asking about preferences and explaining what a specialized partnership offers. The survey cannot establish that such a partnership improves health or that its results generalize to every community. [12]

For any specialized pathway, define the handoff and confidentiality arrangements with the qualified receiving service. Tell patients which organization is offering assistance and what information will be shared. Distinguish a referral from the establishment of a professional relationship, an eligibility determination, or a guarantee of service. The health system’s responsibility includes making the pathway understandable and following up on access problems within its agreed role.

Maintain a route back when a specialist partner cannot help. A declined referral should produce useful information about the reason and the next feasible option, where one exists. It should not silently return to a queue that repeats the same unsuccessful referral. Repeated failure is a signal for partnership redesign or a candid conversation about a service gap.

The ACORN dashboard development study in the Veterans Health Administration illustrates the value of involving end users in defining variables, filters, patient-level information, and usability. It documents design and implementation work. Dashboard use is not itself evidence that social needs were resolved or that health outcomes improved. [13]

An executive dashboard should answer a small set of questions: whom are we reaching, who wants assistance, where does the pathway stall, what help is delivered, what do patients report afterward, and which service gaps require an organizational response? Each measure should have a defined numerator, denominator, source, time window, and owner.

Report both overall reach and the transitions between stages. Screening completion among all eligible patients answers a different question from service receipt among accepted referrals. Reporting only the latter can make a program appear effective while overlooking people who were never offered screening, declined because they distrusted the process, or could not be reached after discharge.

Separate the person, the need, and the referral as units of analysis. One person may have several needs and several referrals for the same need. Counting all those referrals as people served inflates reach. A measurement dictionary should explain how repeat contacts, reopened cases, duplicate referrals, and recurring needs are handled. These definitions belong in routine operating practice, not only in an analyst’s spreadsheet.

A large English observational study found differences in offers of social prescribing and subsequent referrals by patient and area characteristics. Such data can identify inequitable patterns, but they do not establish the cause of each difference or the quality of assistance received. The executive response should investigate the pathway rather than infer that a group simply lacks interest. [14]

Compare screening offers, participation, requested help, referral acceptance, service receipt, and known outcomes across relevant groups where data quality and privacy permit. Language, geography, disability-related access needs, age, insurance arrangements, and other contextual factors may affect different stages in different ways. Avoid interpreting a small subgroup percentage without its denominator and uncertainty.

Qualitative research on implementation across a multistate health system described unclear roles, disconnected workflows, limited communication, and distress when screening identified needs that available resources could not address. Community health workers described important relational and cultural roles that were not always fully integrated. These accounts suggest that equity requires attention to work design and authority as well as reporting categories. [15]

An observed disparity should lead to a specific investigation. Is the screen offered consistently? Is the language understandable? Can patients contact the partner during available hours? Are eligibility rules excluding the people the program intended to help? Does a digital workflow require a stable telephone number? The answers should guide changes with patients and community partners, not a generic message asking staff to improve their numbers.

A recent retrospective pediatric study found differences in some utilization measures after social-needs screening was introduced, while hospitalization and preventive-visit findings were not uniformly favorable or significant. Its before-and-after comparison cannot remove all effects of time, case mix, or concurrent changes. It supports further evaluation, not a guarantee that screening reduces acute-care use. [16]

Systematic reviews reinforce that caution. One review of screening and referral programs found promising outcomes but substantial risks of bias. Another review of EHR-integrated screening and interventions found more frequent reporting of process measures than clinical or economic outcomes, with mixed results for health, cost, and utilization. The intervention’s intensity and the outcome definition matter. [17, 18]

The program should therefore distinguish three levels of claims. It may demonstrate that a workflow operates as intended. It may show that people receive services or report improvement. It may eventually show effects on health or healthcare use through an appropriate evaluation. Progress at one level is valuable, but it does not establish the next level automatically.

Reduced utilization is not always the right immediate objective. A useful referral may help someone obtain previously inaccessible primary care, legal assistance, nutrition support, or another needed service. Some appropriate use may increase. Evaluate whether the change serves patient priorities and the program’s purpose before classifying more or fewer encounters as a success or failure.

Caregiver interviews within a pediatric weight-management navigation study found differing experiences of screening, data sharing, and available resources. Participants emphasized trust and the need for useful, tailored assistance. The small qualitative study does not establish comparative effectiveness, but it helps define what a meaningful follow-up conversation should explore. [19]

Ask whether the patient was able to use the resource, whether it helped with the need they prioritized, what burden the process created, and whether further assistance is wanted. A partner’s confirmation of service delivery and a patient’s assessment of usefulness are complementary. Neither should be discarded merely because they disagree.

Use language that allows partial progress. “Improved,” “temporarily addressed,” “still needs help,” and “no longer wants assistance” can be more accurate than a binary resolved/unresolved label. A need can recur because circumstances change, even after an appropriate referral and useful service. Recurrence should be visible without being treated automatically as a failure by the patient or the navigator.

Follow-up itself should be proportionate. Repeated requests to verify outcomes can become burdensome or intrusive. Agree on preferred contact methods and times, define a reasonable outreach approach, and record when further contact is not wanted. If the outcome remains unknown, retain that uncertainty rather than treating silence as success or noncompliance.

In one randomized trial in pediatric urgent care, adding longitudinal in-person navigation to individualized written resources did not produce a significant between-group difference in social risks or child and caregiver health. Both groups improved on some measures over time. Because both received an intervention, those within-group changes do not establish the effect of either approach compared with no assistance. [20]

A secondary analysis of a different randomized trial found a lower hospitalization risk for children whose families received navigation, without a corresponding difference in emergency-department visits. These are distinct outcomes, and the analysis had incomplete follow-up data. The finding supports the possibility of benefit in that context, while leaving questions about generalization, mechanism, and the contribution of specific program components. [21]

A randomized trial involving Medicaid members with type 2 diabetes found no significant differences in its measured health or utilization outcomes between navigation and usual care. Social needs persisted, and missing laboratory values limited interpretation of the primary clinical outcome. A well-intended referral intervention can encounter conditions it cannot overcome, including resource scarcity and incomplete engagement. [22]

These studies should help leaders formulate realistic questions: which patients, which needs, which intensity of support, which available services, and which outcome over what period? They should not be flattened into either an unconditional endorsement or a blanket rejection of social-care integration. A local evaluation should make the proposed pathway from assistance to benefit explicit and examine where that pathway succeeds or breaks down.

A community-resource assessment within the Illinois WISEWOMAN program documented partnerships over several years, including changes when organizations discontinued services. Its contribution is a structured way to identify partners, gaps, and development needs. Counting partnerships does not establish that each service has sufficient capacity or that every referred patient receives help. [23]

Maintain information about what a partner can currently deliver, rather than treating the directory as a permanent asset. Review service areas, eligibility, language access, contact methods, waiting times, and referral restrictions with the partner. Record who validated the information and when it will be checked again. An outdated resource list can create avoidable work for patients, navigators, and community staff.

When a major gap persists, route it to the level capable of responding. The response may involve funding, a new partnership, a simplified contract, shared staffing, advocacy, or a narrower service promise. It may also require acknowledging that the health system cannot create the missing resource on its own. Leaving the gap in an individual navigator’s queue obscures a structural problem.

Use aggregate unmet-need patterns to inform community investment, while protecting privacy and avoiding misleading inference. The people who reach the health system or agree to screening may not represent the whole community. Combine program data with community expertise and other appropriate population information before deciding that a particular need is absent or a neighborhood has adequate services.

A 2025 systematic review of social-needs data collection and use found mixed utilization and cost findings across varied programs. Some studies reported favorable changes, while others found no significant effect or increases in some types of use. The review supports context-sensitive evaluation and clearer reporting, rather than importing a published return-on-investment estimate into a different service model. [24]

Include the health system’s screening, navigation, technology, training, oversight, and evaluation costs. Include partner coordination and service costs, even when they are funded elsewhere. Distinguish program costs from clinical cost changes, avoided spending from shifted spending, and patient benefit from financial return to a particular organization.

A mixed-methods evaluation of a Massachusetts Medicaid nutrition and housing program described differences in enrollment across service types and challenges involving contracts, administrative work, and housing shortages. It also identified the importance of infrastructure funding, communication, and simpler workflows. These are implementation findings from a specific partnership, not proof that every nutrition or housing referral produces the same economic result. [25]

The investment case should state who receives the benefit and who bears the work. A community organization should not have to finance the hospital’s reporting requirements through unpaid labor. Where possible, use payment and governance arrangements that support both service delivery and the coordination needed to make it accessible. Review the arrangement as demand and community capacity change.

The following register is proposed for local implementation. Its purpose is to keep unresolved transitions visible and assign decisions to the right level. It is not a validated performance instrument.

Assign a backup for each accountable role. Referrals should not become unowned during leave, turnover, or a change in contract. Provide an escalation route when a partner cannot respond or the patient’s circumstances change. Urgent clinical and safety concerns must remain connected to the organization’s established response systems.

Use the register to review cases that teach the team something about the pathway. A small sample of failed, successful, declined, and unknown-outcome referrals can reveal definitions or workarounds that an aggregate dashboard misses. Review should support improvement and respectful accountability, with access limited to people who need the information for the agreed purpose.

Choose a defined population, a manageable set of needs, and partners willing and able to participate. Establish the service promise, patient communication, data definitions, and responsibilities before opening a large referral pipeline. The initial scope should reflect available capacity and the importance of the need, not merely the easiest field to add to the electronic record.

Test representative scenarios with staff and partners. Include a patient who wants help, one who declines, one who cannot use the digital route, one whose eligibility is unclear, and one whose referral is accepted but whose need remains unresolved. These are workflow tests using fictional or appropriately protected scenarios, not outcome evidence. They can reveal missing decisions before patients encounter the same problems.

During implementation, review reach, workload, service availability, and patient experience together. A rise in referral volume may be desirable, but it may also expose a capacity limit. Define who can temporarily narrow or pause new referrals to a saturated service while maintaining support for active cases and explaining the situation to patients.

Evaluate the pathway against a prespecified purpose and feasible comparison. If the goal is reliable connection, measure connection and its barriers. If the goal includes health outcomes or cost changes, use an evaluation design and follow-up period capable of addressing those claims. Account for incomplete follow-up, selection into the program, concurrent initiatives, and changes in service availability.

Before expansion, ask whether the next setting has comparable resources, language support, staffing, and relationships. Preserve the core definitions while adapting the workflow with local participants. The program should grow because its service pathway is understood and supported, with uncertainty visible in the decision to expand.

This article is a focused narrative synthesis of research from different populations, clinical settings, and social-service systems. It is not a systematic review, a pooled estimate of benefit, or a comparison of commercial platforms. Several cited studies are qualitative, cross-sectional, preliminary, or observational. Their findings help identify experiences, barriers, and associations but cannot establish every proposed causal relationship.

Reviews include heterogeneous interventions and may overlap in their underlying studies. Navigation intensity, usual care, eligibility, available resources, and outcome definitions differ. A result in pediatric urgent care, an integrated health system, or a national social-prescribing program should not be assumed to apply unchanged to another population. Some source assessments were limited to their indexed primary abstracts; detailed implementation claims were restricted accordingly.

The responsibility register, referral states, and implementation recommendations are proposed operating practices for local testing. They do not guarantee resolution of social needs or replace clinical, social-work, legal, privacy, or community expertise. The central accountability is to make a respectful offer, provide the assistance promised, identify what happened, and act on the gaps the program exposes.