UT Researchers Develop Methodology for Understanding Social Determinants of Health with Greater Precision
Among health practitioners and policymakers, there’s a growing consensus that the social determinants of health — factors like access to healthy food, education, reliable transportation, a safe neighborhood, friendships — contribute to an individual’s health outcomes.
But which social determinants matter the most? For example, is having a strong social network a bigger predictor of good health than having access to fresh food? And is that true for all people? Or does the relative weighting of factors change across different populations?
“Not all social determinants of health are equal, and we know that certain populations are more vulnerable to specific social determinants than others,” explained S. Craig Watkins, executive director of the IC² Institute. “But there’s not a good understanding of how social determinants matter with any degree of precision.”
IC² researchers have set out to address this knowledge gap by developing a precision-oriented methodology for evaluating social determinants of health (SDoH). Their findings were recently published in PLOS Mental Health.
A Data Sandbox: All of Us
For useable data, the researchers tapped All of Us, an ambitious health data collection program sponsored by the National Institutes of Health (NIH). All of Us includes data from a variety of sources: participant surveys, genomic analyses, electronic health records and wearables. With over 800,000 participants to date, and a goal of reaching a million or more, All of Us represents a giant “sandbox” for advancing health and disease-related research.
Watkins said, “It’s one of the few data sets that gives you the ability to understand social determinants of health in relationship to other kinds of health-oriented data. This multi-dimensional data gives you the opportunity to develop models for understanding what’s happening with a given health outcome.”
Project Scope
The IC² researchers narrowed their study to a single diagnostic health code: mental depressive disorder. As noted in their paper, mental depressive disorder (MDD) is one of the most common forms of mental health disorders in the U.S., with 8.4% of all adults having had at least one major depressive episode in 2020.
The researchers laid out three primary questions:
- To what degree, if at all, do demographic characteristics predict risk of depression?
- To what degree, if at all, do social determinants of health factors predict risk of depression?
- How do specific demographic factors moderate the role of these social determinants of health on the risk of depression?
Matt Kammer-Kerwick, the project’s lead principal investigator, described the parameters of the project: “What we were setting out to do is take different demographic features — in this case, race, ethnicity, gender and sexual identity — and try to understand them in relationship to selected social determinants — food insecurity, discrimination, neighborhood social cohesion and loneliness.”
Using correlational moderation analysis, the researchers were able to identify specific contexts in which the selected SDoHs have a more pronounced relationship with MDD.
Kammer-Kerwick added that the “heart of the work” was Question 3 — the “moderation layer” — where researchers looked at potential degrees of moderation, or interaction, between different demographic characteristics and different social determinants of health.
Nuanced Findings
The most basic takeaway from the research? “It’s complicated,” said Kammer-Kerwick.
“We were able to answer the ultimate research question, which was, ‘Are these SDoH factors differentially impacting people’? The answer is yes. Our findings illustrate that the complexity of people’s living conditions can have significant differential impact on MDD. But it’s nuanced.” – Matt Kammer-Kerwick
Loneliness and food insecurity, for example, contributed to depression in expected ways: the higher the loneliness, the higher the food insecurity, the more likely a person will be diagnosed with MDD. But for the other social determinants — social cohesion and discrimination — the results were more nuanced.
For White, Hispanic and Black populations, researchers determined that as social cohesion goes up, the likelihood of being diagnosed with MDD goes down. But for the Asian population, the opposite is true. (More research is needed to determine causality.)
The apparent impact of discrimination on the risk of MDD was much weaker than researchers expected. “But,” Kammer-Kerwick added, “that doesn’t mean that there’s a weak relationship between discrimination and MDD. It means that we’re probably collecting data in ways that don’t provide as much insight on the moderating effect for that particular variable.”
Kammer-Kerwick summed up the results: “This research has surfaced additional questions for us. But I think we demonstrated that the extra precision we’re looking for is possible to obtain.”
Creating Decision-Making Tools for Community Partners
This exploration of the interplay of demographic characteristics and social determinants of health wasn’t a one-and-done: The researchers are currently working with additional combinations of health conditions, social determinants, and demographic factors to refine their modeling techniques and expand their understanding of social determinants and the nuanced interplay of SDoH factors among different segments of society.
The ultimate goal for the IC² Institute is to use these modeling techniques to develop decision-making tools for community partners. Kammer-Kerwick explains: “If we were to partner with a healthcare provider, for example, we could use this kind of modeling to develop a patient risk stratification tool. What we have here is a methodology for introducing more precision into decision-making in the context of a complexity of factors.”
MEDIA CONTACT
Matt Kammer-Kerwick
Director, Bureau of Business Research at the IC² Institute
512-422-8943