As health plan leaders move forward to embrace social determinants of health data and attempt to integrate that data into care management, they are tackling a host of challenges connected to and analogous to the ones that provider leaders are tackling.
That set of challenges was top of mind for a panel of four presenters from varied backgrounds, who co-presented a session entitled “The Next Frontier: Combining Clinical and Social Determinants of Health Data to Improve Outcomes and Reduce Disparities,” during the first day of the AHIP Institute & Expo 2019, this year’s annual conference of AHIP (America’s Health Insurance Plans), the Washington, D.C.-based national health plan association, which is being held at the Music City Center convention center in downtown Nashville.
This session around social determinants of health data shared the perspectives of a range of different leaders, from very different organizations. The speakers were Joyce Chan, vice president, Medicare, at HealthFirst, a New York City-based Medicaid health plan; David Lima, director, Medicare, at CareOregon, a Portland-based health plan that is primarily a Medicaid plan but which also serves a number of Medicare recipients and Medicare-Medicaid dual-eligible members; Christine Teigland, Ph.D., principal, health economics and advanced analytics, at Avalere Health, an Inovalon Company, a Washington, D.C.-based consulting firm; and Matthew Pickering, PharmD, RPh, Senior Director, Research and Quality Strategies, at the Pharmacy Quality Alliance, an Alexandria, Va.-based public-private collaborative focused on designing quality measures for the Medicare Part D program.
After a presentation by Teigland of Avalere in which she shared the results of an extended study of social determinants of health data across a database of 1.7 million Medicare Advantage beneficiaires, and in which she and her colleagues determined the considerable extent to which incorporating social determinants data in risk-adjusting health populations can provide new understandings of the care management quality of various plans, HealthFirst’s Chan shared with the audience her health plan’s unfolding journey around incorporating social determinants data in her and her colleague’s care management work.
Chan noted the reach of her health plan: with 1.4 million members in New York City, one in eight New York City residents is a HealthFirst member. HealthFirst is the largest Medicaid plan downstate, and the second largest in New York state. It is sponsored by the hospital systems in and around New York City. And it includes 155,000 Medicare beneficiaries.
Speaking of the SDOH challenges, Chan told the audience, “We really wanted member-specific social determinants of health data. There are Z codes, ICD-10 codes that allow providers to record SDH data. Codes appear to be good indicators of utilization.” But, she said, more need to be created and used. “So we turned to analytics and tried to develop a model to predict member-level social determinants. We started by using Z codes as well as member demographics and census data. The model we developed, the risk model, identified four times as many members at risk of social determinants (6.5 percent,) as did observed prevalence from claims diagnosis codes (1.7 percent).” What’s more, she said, in order to validate what they’d found, “We worked with a provider and found that our model was highly predictive of health risk.”
Next, Chan noted, “We integrated our model into our readmissions model, and found that, second only to a previous readmission, the social determinants of health risk model score was second most relevant predictor in our readmissions model. And predictive modeling is a viable way to identify members who likely will have social risk factors. This is especially important given the limited SODH data that exists today. And we’re hosting a hackathon this summer to develop data models.”
What’s more, Chan said, “We want to work with our hospital members on this, to make it even better. We’re going to refine our models with new data sources and additional survey data. And we’re going to create personas. We’ve identified around 30 personas. You can synthesizes needs and inform care managers about needs and actions needed.”
Further, Chan reported, “We’re deploying our data into our customer relationship management system, to give this information to our member-facing staff. There’s also a training component to it. We want to connect members to appropriate community resources and supports,” with the desired results being “better health outcomes, reduced medical cost, increased member engagement and retention.”
Importantly, Chan noted, with the increased flexibility created earlier this year by CMS (the federal Centers for Medicare and Medicaid Services) for Medicare Advantage plan leaders to reshape their plan benefits to meet social determinants needs, including for food, transportation and housing support, “Having this data can help you guide the benefits that will most help your population,” including over-the-counter medicine benefits, transportation benefits (including for Uber and Lyft use), and grocery allowances.
CareOregon’s breakthrough with dual-eligibles
At the opposite end of the country, leaders at the Portland-based CareOregon have been working hard to meet the needs of the 11,400 dual-eligible members enrolled in CareOregon Advantage, the plan’s program for dual-eligibles. All are at or below the federal poverty level, and about 35 percent are disabled individuals under the age of 35. “We have the same challenges as Joyce and her colleagues do,” David Lima said, referencing HealthFirst. Like them, he said, “We don’t have a standardized way of collecting this data, but it’s the right thing to do.”
In that context, Lima noted, “We saw that the AMA [American Medical Association] and UnitedHealth are partnering on a set of ICD-10 codes to expand” available SDOH-based coding. “We need some consistent, standardized way to collect this,” he said. “From the plan perspective, how do we think about this? What are the screening tools? How do our consumers think about social determinants?” Importantly, he noted, “Of our 1,130 dual-eligibles, 97 percent have had at least one social determinant or more. And we need to figure out how to get organized around it.”
Following the recent guidance from CMS giving health plans more flexibility in addressing the social determinants of health through potential new benefits, Lima noted that “We’re now providing our members with over-the-counter medication cards, with gym and wellness credits, with transportation support, with meals post-discharge, with coverage for routine comprehensive eye exams, and with VBID-based [Value-Based Insurance Design] medication adherence” programs.
We’re adding this year gym/wellness, transportation, and the transportation benefit will take them to using the gym and to pharmacies. And we’re adding a small meals benefit post-discharge. Also, routine physicals. We had vision and are expanding the network. For VBID, we’re adding medication adherence.” All of those added benefits, he said, speak to applying the Triple Aim concept in healthcare through product design, with the aims of reducing costs, improving quality, and satisfying members, using benefit flexibility mechanisms to facilitate those aims. And, as at HealthFirst, the leaders at CareOregon believe that incorporating social determinants of health data in their analysis of their populations, will ultimately help to fulfill those objectives.
