Community-Level Social Risk Adjustment Did Not Address Payment Disparities

Despite the notion that social risk adjustment may improve health equity in value-based payment models, incorporating community-level social risk factors into Medicare risk adjustment did little to address payment disparities, a study published in JAMA Health Forum found.

Health equity is a critical priority when shifting from fee-for-service to value-based payment models. Failing to account for social factors in risk adjustment may disincentivize care delivery for vulnerable patient populations, lead to inaccurate performance assessments, and further health disparities.

Using community-level social risk measures to calculate payments and spending targets in value-based payment models is a potential approach to social risk adjustment. The data is widely available and easy to implement, according to researchers. However, studies have found that community-level social risk may not accurately represent individual social needs.

Researchers used data from a Medicare Advantage plan and survey data on self-reported social risk to determine the association between community-level social risk and spending and to assess how incorporating community-level social risk influences risk adjustment.

The community-level social risk measure considered was the 2019 Area Deprivation Index (ADI) of a beneficiary’s census block group, which measures neighborhood socioeconomic disadvantage.

The study population included 61,469 beneficiaries living in 42,078 census block groups. Beneficiaries in the highest ADI quintile, or most disadvantaged areas, were more likely to be younger, Black, dually eligible for Medicare and Medicaid, low income, and have more self-reported social needs compared to those in less disadvantaged areas.

Over half of the beneficiaries (53.8 percent) reported at least one social need, with financial strain and food insecurity being the most common.

ADI explained just 0.02 percent of the observed variation in spending. When grouping beneficiaries by geography, ADI explained 0.61 percent of the spending variation at the census block group level and did not account for any variation at the county level.

When considered alone, ADI was associated with higher spending; a one-point increase in ADI was tied to an $8.77 increase in annual spending. When taken with demographic and clinical characteristics, a one-point increase in ADI was associated with an $11.08 decrease in spending.

“One explanation of this pattern is that, as observed in our study, high-cost beneficiaries disproportionately reside in disadvantaged areas, such that ADI serves as a proxy for higher spending,” researchers wrote. “But when conditioned on demographic and clinical characteristics, ADI serves more as a proxy for structural access barriers and is therefore associated with lower spending, as has been shown for other markers of structural access barriers.”

Incorporating ADI into a risk adjustment model using demographics and clinical characteristics led to payment inequities for White beneficiaries, suburban and rural beneficiaries, those in low ADI quintiles, and those with a high burden of self-reported social needs.

ADI incorporation decreased predicted spending for Black beneficiaries and individuals in high ADI quintiles.

Although directly incorporating ADI into risk adjustment models could reduce predicted spending for certain populations, it explained little variation in healthcare spending and was weakly correlated with self-reported social risk factors.

“This prompts caution and nuance when using the ADI for social risk adjustment and raises concern that adjusting for community-level social risk may not address payment disparities for many beneficiaries with high levels of social risk,” researchers concluded.