The SDOH Gap in Level Funded Plan Design: What Claims Data Shows and What Plan Sponsors Ignore

Social determinants of health drive healthcare utilization in ways that claims data captures indirectly but plan design ignores entirely. Members missing appointments because they lack transportation. Diabetics whose glucose control deteriorates because they cannot afford the diet their condition requires. Rising emergency department utilization driven by housing instability rather than acute illness. The SDOH gap in level funded plan design is both a cost management failure and a harm to members that the plan architecture can address. The evidence base from Medicaid and Medicare programs is more developed than employer plan evidence, and where this article extrapolates, it says so.

What Claims Data Shows Without Saying

#

Claims data does not include a field for food insecurity or housing instability. But the patterns are visible to anyone who knows how to look.

Repeated emergency department visits for conditions that should be managed in primary care signal something other than acute illness. A diabetic member presenting to the ED for hyperglycemia multiple times per year is not receiving effective primary care management. The claims data shows the ED visits and the diagnosis codes. It does not show why the member is not managing their condition between visits.

Medication nonadherence is visible through prescription fill gaps. A member prescribed a maintenance medication who fills it once and never refills has a medication adherence problem. The claims data shows the gap. It does not show whether the gap reflects cost barriers, transportation barriers to pharmacy access, health literacy gaps, or something else.

Research published in Population Health Management estimated that approximately 50 percent of patients do not take their medications as prescribed, with cost cited as one of the most common reasons for nonadherence. The same study found that individuals with low income who face difficulties meeting basic needs such as food, clothing, housing, and transportation show lower medication adherence rates than those without such difficulties. One survey of primarily low income adults found that spending less on basic needs to pay for medication was significantly more likely among individuals with fair or poor health status and a greater number of chronic conditions.

The CMS Accountable Health Communities model screened nearly 483,000 Medicare and Medicaid beneficiaries and found that 15 percent were eligible for navigation services based on unmet social needs. More than half of those navigation eligible beneficiaries reported more than one core health related social need across five domains: housing instability, food insecurity, transportation problems, utility difficulties, and interpersonal violence.

For commercially insured populations, research from Cigna examined 5.1 million members and found that 27 percent lived in a zip code where the median income was at or below 200 percent of the federal poverty line. The researchers identified populations with social needs who had common conditions for which employers often provide no cost or low cost benefit programs, including diabetes, behavioral health conditions, high risk pregnancy, and overweight or obesity.

The analytical challenge is distinguishing SDOH driven patterns from purely clinical or behavioral patterns in claims data without individual level SDOH screening. A member who misses appointments may have a transportation barrier or may have scheduling conflicts or may have disengaged from care for reasons unrelated to social needs. A TPA that can screen for SDOH can differentiate. A TPA that cannot screen can only observe the downstream patterns in claims and make inferences.

The Benefit Design Responses That Exist

#

The responses to SDOH needs in employer plans are developing but not yet standard. Medicaid programs have more experience, and much of the employer plan approach extrapolates from Medicaid evidence.

Transportation assistance through non emergency medical transportation benefits covers rides to appointments. Medicaid programs have extensive data on NEMT utilization and cost impact. Employer plan adoption is minimal but growing through vendors like Lyft Health, Uber Health, and specialized NEMT coordinators. Research published in Healthcare found that unreliable transportation leads to missed appointments, with rescheduled appointments often delayed for months. These disruptions compromise continuity and adherence to care. The University of Michigan estimates that one in four adults in the United States experience transportation insecurity, making it difficult to get to medical appointments or pick up prescriptions.

Food and nutrition programs include food pharmacy models, produce prescription programs, and medically tailored meals. The Geisinger Fresh Food Farmacy provides food assistance to diabetic patients and has published data showing A1c improvement in participants. Research from JAMA Internal Medicine found that addressing unmet basic resource needs as part of chronic cardiometabolic disease management can improve clinical outcomes. Healthy People 2030 notes that food insecure adults are at higher risk for chronic conditions including coronary heart disease, diabetes, obesity, and cancer.

Community resource navigation connects members with local social services through screening and referral. Platforms like Findhelp and Unite Us maintain databases of community resources covering housing assistance, utility assistance, food banks, legal aid, and other services. The navigation function can be integrated into the TPA’s member services or outsourced to these platforms. The CMS Accountable Health Communities model evaluation found that referral to community services was feasible at scale and that beneficiaries who received navigation services showed increased awareness of and connection to available resources.

The evidence base for each intervention is not equivalent. Medication adherence support programs have the strongest evidence for claims cost reduction across multiple published studies. Transportation to appointments reduces missed visits and ED utilization in Medicaid populations; the employer plan evidence is extrapolated but structurally sound because the mechanism is the same regardless of payer. Food and nutrition programs show clinical improvement in Medicaid and health system populations, with the translation to employer plan claims cost reduction inferred from clinical improvement rather than directly measured in most cases.

Which Interventions Produce Measurable Claims Cost Reduction

#

The evidence hierarchy for SDOH interventions in employer plans requires distinguishing what is proven from what is plausible.

Medication adherence support produces measurable claims reduction across multiple published studies. This is an SDOH intervention because cost is the primary barrier to adherence for many members. A 2021 analysis found that food insecure individuals who do not take medications as prescribed and do not engage regularly with care spend 45 percent more on healthcare than their food secure peers, according to the Center on Budget and Policy Priorities. The analysis found that food insecure people spent an average of $6,100 on medical care per year compared to $4,200 in food secure households after controlling for other SDOH and demographic variables.

Transportation to appointments has moderate evidence. The employer plan evidence is extrapolated from Medicaid, but the mechanism is identical: the member cannot get to the appointment, misses care, the condition worsens, higher cost care follows. A cost estimate from Health Affairs estimated that food insecure families have 20 percent greater healthcare expenditures than food secure families, an annual difference of $2,456. The transportation mechanism operates similarly.

Food and nutrition programs show developing evidence. Clinical improvement is documented in multiple programs. Geisinger Fresh Food Farmacy data, Feeding America partnership evaluations, and medically tailored meal programs show glycemic improvement and weight management outcomes. The translation to employer plan claims cost is inferred from clinical improvement. Improved A1c means fewer diabetes complications, fewer emergency department visits for hyperglycemia, fewer cardiovascular events downstream.

Housing stability interventions have strong Medicaid evidence but minimal employer plan data. Research from the CDC notes that housing cost burden is associated with overall poor health and increased risk of disease including hypertension and cardiovascular disease. But the cost and complexity of housing interventions exceed what most small group plans can support.

The honest statement is that employer plan SDOH intervention evidence is less developed than Medicaid evidence. The extrapolation is reasonable where the mechanisms are the same. A level funded plan with a transportation benefit is not conducting original research; it is applying a mechanism that has been tested elsewhere to a population that faces the same barrier.

The Plan Design Opportunity for Level Funded

#

A level funded plan has a structural advantage in addressing SDOH: the employer and TPA can see their own claims data. A fully insured employer cannot. The claims patterns that signal SDOH needs are visible to a TPA that knows how to look. The plan design response can be built into the benefit.

NEMT as a covered benefit costs $3 to $5 PMPM in most implementations. The transportation benefit prevents missed appointments for members who would otherwise have no way to reach care. One avoided emergency department visit per year among the transportation constrained population produces a positive return on the benefit cost.

Medication cost reduction programs can include copay assistance programs, 90 day fill incentives, and transparent PBM arrangements that reduce out of pocket costs. These are plan design decisions, not add on programs. A plan that designs around medication affordability is addressing the cost related nonadherence that drives downstream claims.

Community resource navigation through the TPA’s member services or through outsourced platforms can be integrated into chronic disease management programs. A care manager working with a diabetic member can ask about food access, transportation to appointments, and medication affordability as part of standard outreach. The screening does not require a separate program; it requires training the existing staff to ask the questions and connecting them to resources that can respond.

SDOH screening can be integrated into wellness visits or annual health assessments. The Hunger Vital Sign screening tool is two questions. The Transportation Security Index is a validated 16 question survey. The screening does not need to be elaborate to be useful; it needs to be systematic enough to identify members who would benefit from intervention.

The cost of these interventions is modest relative to the claims cost they address. The gap is not budget; it is design. Most level funded plans do not incorporate SDOH response because no one in the design process thought to include it. The claims data that would reveal the need is not analyzed for SDOH signals. The benefit design that would address the need is not built into the plan.

Closing

#

The SDOH gap in level funded plan design is a gap between what the data shows and what the plan does about it. Closing it does not require transforming the health plan into a social services agency. It requires identifying the specific SDOH signals in claims data, building the specific benefit design responses that address the highest cost signals, and integrating those responses into the level funded core.

The employers who do this will produce better member outcomes and lower claims costs than those who treat SDOH as someone else’s problem. The evidence is strongest for medication adherence and developing for transportation, food, and other interventions. The extrapolation from Medicaid evidence is reasonable where the mechanisms are the same. The plan design opportunity is structural, not programmatic.