Predictability is Actionable: the SDOH Impact on Health and Financial Outcomes

At the end of October, we shared a look into a research project we conducted with the Virginia Tech’s Computational Modeling and Data Analytics (CMDA) Capstone Program to uncover the independently predictive value of social health insights. In short, we worked to find out if our social risk dataset and platform can predict disease prevalence at the community level and consider the implications of that that predictive power for better understanding health conditions affecting populations in any given area.

We go into the science in the prior post (linked above), but the ability to predict community prevalence of 4 conditions associated with poor outcomes and higher costs of care based on social data alone was impressive.

The percentages above represent an element of data science known as R² values, the level of variation of predictability from the independent variables, (in layperson’s terms, how close was the prediction when validated).

This is groundbreaking (if we do say so ourselves). These scores are meeting or beating other predictive modeling based on clinical data, and they’re doing it exclusively via social risk factors with zero access to patient records or claims.

Even depression stands out. Current predictive models return higher R2 values, but with much more significant barriers to access, requiring MRIs with no account for limited patient access and cost, or the voluntary sharing of smartphone use in adolescents.

The potential here to make a real difference in member or patient lives is enormous, and it leaves us ready to ask two exciting questions:

• What could you do if you were able to predict chronic conditions in member communities based just on our social risk data? Even in cases where claims and clinical data are incomplete or absent?
• And if you can do that with community SDOH data alone, what happens when you then match our individual level risk scores with the medical data you do have?

Taking Action on SDOH-Based Condition Prediction

Understanding that predictability is the first step in implementing focused, scalable interventions to improve quality of life and saving millions of dollars on the cost of care. Fortunately, that’s why we’re here.

As a refresher: SDOH (also known as HRSN) data and strategy are our entire area of expertise. At Socially Determined, we maintain an ongoing analysis of seven specific risk domains at the community and five at the individual level (above). We do this to allow for multiple use-cases across healthcare, from community and public health work to more granular, individual population health insights.

Now consider that data at scale, and what you could do with it.

More than 10% of the US population has diabetes, with 1.2 million Americans diagnosed every year. Additionally, 33% of adults in the US are overweight or obese, increasing the risk of hypertension, type 2 diabetes, coronary heart disease, and stroke. Furthermore, 33% of cardiovascular deaths are due to stroke, with stroke also as the top cause of adult disability. But how much of that is lacking from a clinical record? Or with identified community prevalence, how much could be addressed earlier rather than later through informed opportunities for intervention and care?

How are Diabetes, Coronary Heart Disease, and Stroke addressed with Focused Social Risk Interventions?

Beyond the predictability, the study determined that diabetes, obesity, and stroke, were all highly impacted by health literacy, housing environment, social connectedness, and education. These are exactly the sorts of challenges Socially Determined can address at a far more granular level with individual risk scoring.

For example, our payer and provider customers apply our person-level data to claims data, directly matching our findings with individual health status. This allows for any number of effective interventions they then use, significantly and positively impacting the lives of the identified patients, while also dramatically lowering costs through reduced ED utilization and hospital admission, as well as improved medication adherence. The now improved health status of these patients leads to an actuarially validated 5:1 ROI in avoided costs, one where each dollar saved directly maps to a better, healthier life.

It Doesn’t Stop There: Learn What’s Missing

A lot of discussion around social determinants of health tends to revolve around the known health factors and addressing their social causes. But there is so much more in SDOH data waiting to be used. Our work offers otherwise unavailable insight to client leadership on what is causing apparent unrelated costs in traditionally “healthy” populations.

For example, a partner of ours recently discovered a whopping 70% of their commercial members aged 20-29 were at risk for food insecurity, and those same members had a marked increase of cost of care without identifiable clinical attribution by $85+ PMPM. This is resolvable with food vouchers, produce prescriptions, or where applicable, medically tailored meal programs that are, in and of themselves, cost neutral at a typical $40-60 PMPM and have an immediate $60 PMPM savings. That’s still $30 less than that initial $90, but it’s also an immediate 66% reduction in outsized spending at scale.

Now, without a crystal ball, it’s impossible to determine what conditions those members would develop later in life because of social risk factors in their 20s. However, it’s not difficult for any physician or public health professional to recognize how poor nutrition in youth maps perfectly to... you guessed it: obesity, diabetes, coronary heart disease, and stroke.

Social determinants of health have a complex relationship with outcomes and the cost of care. They require a sophisticated solution, one capable of having and using the right data to deliver predictable outcomes to finally break through one of the most significant elements of healthcare that has been historically out of reach.

If your organization is ready to save lives and money, we’re here and we’re ready for you.

Let’s talk.