When the referral isn't the finish line: Making SDOH programs actually close care gaps

A health plan asks a member about social needs, and the member says yes, they've struggled to get enough food this month. That question, rare a decade ago, has become common practice across hospitals and health plans as part of a broader push to identify unmet social needs before they become costlier medical problems.

An unresolved social need rarely stays limited to a social problem for long: food-insecure adults incur roughly $1,834 more a year in health care costs than food-secure adults, largely through higher rates of diabetes and heart disease, while transportation barriers alone lead an estimated 5% of U.S. adults, and as many as 14% of those with low incomes and 17% of those with a disability, to forgo care altogether. Left unresolved, these needs resurface later as emergency visits, hospitalizations, and complications that cost far more to treat than the original need would have cost to meet.

Preventing these undesirable outcomes depends on whether a need actually gets met, not just identified, and that's where the field still has real ground to cover. When a referral goes out to a local food bank or community organization, the encounter gets logged, and confirming whether that member actually received help remains the exception rather than the norm.

Closed-loop referral platforms exist, and have for years, precisely to solve this problem. Technically, a health plan can track a referral from hand-off to confirmed outcome. But having that capability and having it used consistently are different things. A community organization can be connected to a closed-loop system and still have no obligation to log what happened, because the relationship underneath the technology is typically voluntary, not contractual, and often unfunded. On paper, a program can appear successful because a health plan screened thousands of members and sent referrals. But whether those referrals actually resulted in food delivered, transportation arranged, or another need being met is a separate question – and the one that determines whether the investment made a difference.

California’s closed-loop example

The clearest evidence comes from North Carolina's NCCARE360 network. Duke researchers compared two five-month periods using the same closed-loop platform and community organizations. When a temporary COVID-19 program reimbursed organizations for food assistance and case management, 88% of referrals were confirmed resolved. After that funding ended, the resolution rate dropped to 30%. The platform hadn't changed — the reimbursement mechanism had. Lower resolution rates don't necessarily mean fewer people were helped. They may just indicate fewer organizations had the staff time to log it once they were no longer being paid for the work. Either way, the result is the same: a system that can no longer show what it accomplished.

Closing this gap doesn't require new technology, and California's Medicaid program (CalAIM) is an example of what workable governance can look like. Its Closed-Loop Referral requirement defines what constitutes a resolved referral and what information contracted providers must report back. But an independent review of the program’s early implementation found many community organizations struggled with the billing process, with some waiting six months or more for payment and others leaving the program entirely. The lesson isn't that funding and contracts don’t work – it’s that they must be paired with reporting and payment processes that community organizations can realistically navigate.

For health plans and other organizations looking to improve referral completion, three operational practices stand out:

• Fund the administrative work as its own line item, not an assumed byproduct of paying for the underlying service. Documentation and reporting require staff time and should be reimbursed accordingly.
• Simplify and speed up the payment processbefore asking organizations to take on more reporting. Faster, simpler reimbursement processes help community organizations maintain the capacity needed to document outcomes.
• Prioritize the highest-impact partnerships. Start by building outcome-based relationships with the community organizations serving the largest member populations, rather than trying to formalize every referral relationship at once.

None of this requires a new platform or a new mandate. It requires treating the organizations delivering social care with the same operational seriousness as the ones delivering clinical care: pay them on time, pay them for the paperwork, and start with the partners who touch the most members.

The member who said “Yes, they'd struggled to get enough food that month” is the reason any of this matters. Whether they actually got groceries that week is a question most plans still can't answer with confidence, not because there's no way to find out, but because no one funded the follow-through it takes to know. That gap is much more than a data problem – it’s a barrier to better care.

Organizations that prioritize referral completion as much as referral generation will be better equipped to measure results, improve health outcomes, and show that investments in social care are making a meaningful difference.

Rakshith Yashvanth is a senior product manager in ZeOmega’s Advanced Analytics & AI organization.