Improving Social Determinants of Health with Predictive Analytics

July 20, 2021 - With the COVID-19 pandemic highlighting gaps in healthcare and access to service, providers are searching for ways to improve the well-being of their members. By implementing predictive analytics, UnitedHealthcare is addressing the social determinants of health to enhance care for those they serve.

“Around 80 percent of your health is determined by things that are not your genetics. There are things more such as what’s going on in the rest of your life, what we call social determinants of health — social, economic, gender orientation, and other markers that sometimes can lead to inequality,” Rebecca Madsen, chief consumer officer, UnitedHealthcare told HealthITAnalytics.

The social determinants of health have come into the spotlight with the COVID-19 pandemic enhancing health disparities. According to Madsen, 41 percent of Americans avoided medical care at the height of the pandemic.

“We know a lot of that was due to access, or non-essential care being effectively closed, switches to telemedicine. But we also know that this adversely and disproportionately affected underserved groups. And so that even showed more of a spotlight on why this is something we need to accelerate and build on the strong foundation that we have to be able to drive change,” Madsen said.

To address the issues regarding health disparities brought to light from the pandemic, UnitedHealthcare has created an advocacy system using predictive analytics to support members who might be struggling due to their social environment.

READ MORE: Social Determinants of Health Linked to Congenital Heart Disease

UnitedHealthcare’s advocacy system identifies those who need support in three different ways. The first is through a predictive analytics model and the use of machine learning. By evaluating de-identified claims from members, the system can determine the need for social services.

Data is then loading into an agent dashboard used by UnitedHealthcare advocates. When a member calls in, advocates can connect the caller to community resources at low or no cost. The second method the UnitedHealthcare system uses to identify the need for support is active listening by the advocates.

“Our advocates are trained to help the member and to listen for keywords,” Madsen explained.

Advocates are trained to listen to how callers sound on the phone. If a caller sounds stressed, advocates will ask the caller what’s going on. They will then listen for keywords or phrases such as “I’m hungry” or “I’m stressed” to determine the caller's needs.

“Then that enables us to connect again to consumers leveraging this comprehensive database to be able to plug them into support mechanisms,” Madsen said.

READ MORE: Why Interoperability is Key to Social Determinants of Health Efforts

Lastly, Madsen explained that advocates are equipped with questionnaires to ensure they are asking the right questions.

“We look at it holistically, at the data, the predictive modeling, the coding that the provider sent to us, the claims mining coupled with the human interaction to be able to identify the need that somebody has and get them the support,” Madsen said.

To provide comprehensive health coverage, every aspect of health should be examined. The approach means going beyond just physical health and developing an understanding of mental and social determinants that impact an individual.

“We know that so much of an individual’s health is determined by what happens outside of the doctor’s office. And we look at that as an organization, a chance to redefine a traditional insurance model, especially for the most vulnerable,” Madsen said.

Often when individuals face health challenges in their lives, Madsen said it creates a kind of snowball effect. Patients’ healthcare struggles can impact their mental health, environment, and their choices within that environment. Madsen explained that people need a source of support concerning the social determinants of health.

READ MORE: Deep Learning, Predictive Analytics Helps Identify Chronic Diseases

“I’ve learned just how appreciative people are of the support, and how important it is to continue to serve this mission and to be able to support people holistically and look at them as the whole person and all the needs that they have, is never more important than it is now,” Madsen said.

Since implementing the predictive analytics advocacy program at UnitedHealthcare, Madsen said the organization has seen a great response. Eligible individuals have accepted over 50 percent of the offers UnitedHealthcare has provided for support, and 100 percent of UnitedHealthcare’s Medicare population was referred into programs.

Moving forward, Madsen said UnitedHealthcare hopes to create a more digital environment within their predictive analytics advocacy model.

“Right now, it is a lot of predictive analytics in our call center, but we also know that people use multiple different touchpoints to be able to interact with both our health plan and to express their need,” Madsen explained.

Among the digital improvements to the program in the future, Madsen discussed creating a digital health survey. Through the survey, UnitedHealthcare could see how the program has been effective and understand why an individual declined an offer. With this information, the organization can improve training for advocates on cultural relevancy, providing better care for diverse populations.

Additionally, Madsen said the organization plans to continue improving its predictive models and basic data, such as bringing in more data and using machine learning and predictive analytics to ensure they are capturing the full breadth of the population.

While the predictive analytic program works to provide community-based social services to its members at little to no cost, Madsen stated that driving down medical costs is not exclusively the focus of this program. Instead, the goal is providing support and resources for those in need as well as improving upon health disparities and patient engagement.  

“It’s really about supporting their consumers, driving a different kind of experience, being sensitive to the full continuum of needs, and that every person is different, and they interact with the health system differently,” Madsen continued.

“Oftentimes, being able to have that conversation with somebody is life-changing. And being able to have that sensitivity and support and plug them into resources, a lot of that stuff, people would think that they have access to, but they really don’t. So, meeting this unmet need is a tremendous opportunity for the entire healthcare system.”