Humana Calls for Social Determinants of Health in Risk Adjustment

In a recent joint article with the National Quality Forum (NQF), Humana called for a new risk adjustment model that accounts for both clinical factors and social determinants of health.

“We believe it is time for a meaningful reconsideration of the goal of, and approach to, risk adjustment writ large in quality measurement and payment systems,” the writers argued in the article, which HealthPayerIntelligence received by email. “The discussion should no longer be about the dichotomy of clinical risk versus social risk.”

The writers did not recommend a specific risk adjustment model, but rather outlined some steps to initiating a fruitful conversation on the subject of a clinical-social risk adjustment model.

The conversation needs to engage providers about what resources are available to them and how to improve access to resources. The results of this engagement could reduce the number of providers penalized for working with high-cost populations with a solely clinical risk adjustment model. This approach could also prevent justifying low quality care for these populations with a clinical-social risk adjustment model.

Another topic for discussion would be how to present data to stakeholders and the public in an actionable and accurate manner.

The new risk adjustment model should also be able to encompass any social factors that impact health, as well as the ways that these factors are interconnected.

Evidence linking social determinants of health and health outcomes grows each year. But these factors, which account for 80 percent of health outcomes according to a commonly cited statistic, are largely absent from the models intended to predict costs and reimbursement.

While many providers, regulators, and researchers have been vocal about the demand for greater inclusion of social factors in risk adjustment, some consumer and patient activists and employers have voiced an opposing view.

To opponents of social risk adjustment models, the potential models pose both practical and ethical problems.

On the practical side, it will require new data types to input into the model, which will be a challenge to establish and enact.

And ethically, opponents fear that including the social risk factors in risk adjustment could lead to lower quality standards. If an area is expected to have high cost and potentially low outcomes based on its demographics and social factors, a clinical-social risk adjustment model may not provide the financial incentive to enforce needed changes.

There is also resistance to the implications that certain demographic influences determine health outcomes.

"Some institutions, including the Centers for Medicare and Medicaid Services (CMS), have struggled with the optics of a payment model that implicitly acknowledges any inequity in expected outcomes based on social characteristics,” the writers explained.

CMS is currently midway through the process of phasing in its Medicare Advantage risk adjustment model. The model stratifies dual enrollees who are enrolled in both Medicare and Medicaid, allowing for some adjustment based on social determinants of health associated with dual enrollees.

By explicitly including social determinants of health in the calculations, the risk adjustment model could become an unwanted thermometer for persistent racial care disparities, opponents of social risk adjustment argued.

However, the writers found these objections to be insufficient reasons to further delay adding social factors to risk adjustment.

Age is a social determinant of health which is already successfully incorporated into risk adjustment models, the article pointed out. In fact, out of NQF’s over 190 quality measures, 89 percent took age into account.

“We believe the consternation that surrounds accounting for social factors in risk-adjustment models is too myopic; it should make us uncomfortable with the integrity of the concept of clinical risk and should highlight the need to better characterize differences between these ideas,” the writers stated.

In the press release, William Shrank, MD, chief medical officer of Humana and one of the article’s co-authors, added that coronavirus has played a particular role in revealing disparities and demonstrating the need for an improved risk-adjustment model.

“The COVID-19 crisis has underscored the inter-relatedness of social context and physical health,” Shrank said.

“Our proactive outreach to our COVID-19-positive and highest risk members has uncovered high rates of social isolation and considerable barriers to accessing healthy food, which put them at higher risk for bad outcomes. Given the associations between social risk and clinical risk, it stands to reason that risk adjustment for payment models should incorporate the breadth of patient characteristics that predict the need for health care services.”

One way that payers have incorporated social factors in the past is by increasing capped provider payments based on social determinants of health and costs, as Vermont’s OneCare Medicaid ACO did in 2017.