It’s always been pretty obvious that factors such as wealth, race, education, and the quality of food and water have enormous impacts on health. But only in the past few years have the medical professions tried to quantify and capture these factors. Given that the field is increasingly digitized and data-driven, health IT is responsible for collecting and analyzing social determinants of health (SDoH). Data can both call out SDoH and help to address its effects.
An example of calling out SDoH was cited by Dr. Sherri Onyiego, Medical Director for the Texas Market at Equality Health. They use claims data to track use of emergency rooms, medical equipment, and generic versus brand medications.
They also obtain public health data by ZIP code and data from a Health Information Exchange (HIE). That data can reveal important information that might not make it into the patient’s own record, such as an emergency room visit.
Onyiego mentioned the recent, scandalous revelation that maternal morbidity among non-Hispanic Black people is much higher than average (and are getting worse) as an example of feeding SDoH data into analytics.
An oft-cited statistic estimates that 80% of health outcomes can be attributed to SDoH. Dr. Geoff Rutledge, chief medical officer of HealthTap, says that variations in healthcare outcomes between adjacent counties can be huge. About 25% of the disparity can be attributed to difficulties getting access to healthcare. Even more disparities are caused by known social inequities: poverty, race, and so on.
Social determinants of health can hover over populations for decades and generations. For instance, researchers found higher rates of cardiovascular diseases and poor outcomes among people today whose neighborhoods were redlined in the 1930s.
Clinicians Taking Responsibility
Naturally, the medical profession can’t fix all SDoH. For instance, air pollution has numerous negative health effects, and clearly affects dense urban environments more than other neighborhoods. Similarly, lead levels in children are tied to both race and income. Fixing these problems is an international undertaking, although hospitals can do more to reduce their own waste and carbon emissions (video).
Policy-makers and political leaders often single out an institution that serves the public—schools, clinical settings, even libraries—and dump a welter of tasks on these institutions that they weren’t originally designed to handle. Simply because a school or clinic deals with people who have many needs, these institutions are thrust into compensating for the lack of money and support that policy-makers and political leaders refused to give needy people in the first place.
So before we go further we should ask: Why should a clinician be responsible for finding a food source or an apartment for a patient?
One of my interviewees said that in the U.S., at least, the medical profession should take responsibility for SDoH because we contribute to its problems. Medical care is rife with discrimination on the basis of race and gender, particularly toward LGBTQ+ people. Our industry has put 100 million Americans in debt and causes 40% of bankruptcies.
Eden Brownell, director of behavioral science at mPulse Mobile, introduced me to the term “techquity,” which means ensuring health equity in technology and tailoring interventions to the person.
So let’s face SDoH head on. Current uses for SDoH in medicine include connecting individuals to services such as Meals on Wheels, and using it to identify at-risk people. We’ll see many examples of these uses in this series. SDoH is important in improving climate resilience too.
A Mandate to Address SDoH
According to Hillit Meidar-Alfi, founder and chief executive officer of Spatially Health, value-based care models are pushing healthcare providers to use SDoH. After all, value-based care purportedly rewards a provider for discovering problems early and ameliorating them, whereas the traditional fee-for-service model rewards treating the condition when it is relatively advanced. Spatially Health applies geospatial analysis to expand a provider’s understanding of SDoH data. Like many SDoH tools, the analysis identifies vulnerable patients and does risk stratification.
The Inpatient Quality Reporting (IQR) program, used by Centers for Medicare & Medicaid Services (CMS) to rate hospitals, now contain two measures of SDoH. These measures are very crude: The first just records whether the hospital is screening patients 18 years of age or older, and the second records how many patients have problems related to SDoH. Hopefully, at least, the measures will stimulate hospitals to ask the right questions, and CMS will know how needy their populations are.
Maria Gil, Data-Tech-AI Partner, Consumer and Healthcare at Genpact, says that the incorporation of SDoH into Star ratings, along with the advent of more powerful analytics and AI, have made a noticeable impact on clinical sites, causing them to ramp up their collection and use of SDoH. More than 80% of payers, she says, plan to increase their spending on understanding SDoH. Still, less than 40% of these institutions have incorporated data into their work so far. (See a survey cosponsored by Genpact.)
The Healthy People project of the Department of Health and Human Services (HHS) now incorporates SDoH. An example goal is: “Reduce the proportion of children who have ever experienced a parent or guardian who has served time in jail.”
The U.S. Census Bureau’s Opportunity Project collects this data in order to improve the use of funds for transit, climate resilience, and other issues affecting under-resourced communities. The CDC has an Environmental Justice Index to find correlations between environmental problems and health.
Rutledge says that Medicare, which already penalizes hospitals when one of their patients is readmitted within 30 days, will increase penalties if the hospital has not addressed the patient’s SDoH.
Complementing this penalty, the Office of Health Policy within HHS released a report measuring the benefits of addressing SDoH in healthcare.
The nonprofit Patient-Centered Outcomes Research Institute (PCORI) funded a roundtable on sharing SDoH data, organized and led by the Center for Open Data Enterprise (CODE). One of the resulting papers lists a large number of public sources for SDoH data (pages 26-30). This paper, dated August 2020, pointed out that this data could help prioritize efforts at COVID-19 testing and contact tracing—today, of course, we would add vaccinations.
I talked to several researchers from CODE: president Joel Gurin, research and communications manager Matthew Rumsey, and roundtables program manager Paul Kuhne.They called COVID-19 a “wake-up call” regarding the crucial need to collect and apply SDoH data. As it turned out, unfortunately, geographic, economic, and racial disparities distorted the approach to COVID-19 and reduced the effectiveness of public health and clinical efforts.
The report also said that SDoH data could improve “health system resilience” generally (page 7), which meant addressing social problems such as low income, poor housing, and lack of Internet access. The importance of protecting privacy is also recognized.
The report reflects the importance of outreach to under-served communities through contacts who “share a cultural background with communities they work in and speak the languages that those communities speak.” (pages 21-22). “Researchers could build partnerships for data collection with churches, soup kitchens, CBOs, and other entities that have already built trust within the community.” (page 25)
A similar call for equity through the use of SDoH data, including community engagement, was made in an IBM paper in 2022.
Another report from CODE calls for better tagging of data with SDoH. We should know, for instance, not only how many people in a city are unhoused (homeless), but how many Asians, etc.
The CODE managers said that data sharing needs to be improved even within HHS itself. It’s still hard to determine how much social support a person has, or even to get data on income.
They called for more fine-grained data: not just at the level of the county or even the ZIP code, but at the level of the census tract. They recognize that when data is collected—especially at a fine-grained level—privacy must be protected.
According to Andrew Eye, CEO of ClosedLoop, governments publish directories showing where there are shortages of healthcare providers. He also says that researchers have developed an area deprivation index and a social vulnerability index.
Complications of SDoH
Gerry Miller, founder and CEO of Cloudticity, reminds us that 5% of patients create 50% of healthcare costs. SDoH data can help us identify and help these people—perhaps even before they have diagnosable diseases. He says that most patients would be perfectly happy sharing SDoH data with payers, and that we should increase the number who do so by making that option an opt-out choice instead of opt-in.
Sagility employs SDoH to help manage the total cost of care. SDoH can help predict who has a high risk of hospitalization and who will need interventions to prevent admissions to a long-term care facility such as a skilled nursing facility. An aging population also accentuates the effects of SDoH, according to SV Krishnan of Sagility. From the ages of 70 to 80, he says, frailty is a good predictor of healthcare usage—but the value of the measure decreases after 80.
Jessica Probst, senior specialist, real world evidence at OM1, points out that many people from minority or low-income groups are faced with barriers that inhibit equitable access to health care. Patients with limited access to routine or preventative care can experience higher disease burden and increased use of high-cost healthcare resources, such as inpatient or emergency care.
Miller says we need to provide deidentified PHI. Even more important, he would like laws that prevent discrimination based on lifestyle choices.
In fact, even communities are at risk from the release of data, because calling out their vulnerabilities could lead to red-lining by powerful institutions. Gil says that machine learning can produce insights for smaller, fine-grained populations. Miller says that fine-grained data would help with situations such as closing a single school instead of a whole school district after an outbreak of an infectious disease.
Aspects of Data Addressed in This Series
Whenever you incorporate new data into your modus operandi, you first have to define what data you need, which can be as precise as particulates in the air or as vague as cultural expectations. You have to know how you will share data with the people who can act on it, roping in questions of formats, conformance, and data interoperability that are familiar to professionals in health IT.
Now the data must be collected, with an eye to protecting individual privacy, and classified, which raises questions of terminology, units of measurements, and standards.
Finally, the clinicians and public health authorities must determine how to act, which can be difficult because so many of the determinants lie outside their control.
The rest of this series examines what companies and public health agencies are doing in these areas of SDoH.
