Sonia Sarkar was previously the Chief Policy and Engagement Officer for the Baltimore City Health Department.
Fifteen years ago, I volunteered at clinics in East and West Baltimore. There, the patients I met taught me that the best medicine in the world can only go so far if you have no food to take it with. This seems obvious—but it is not how we conduct business in our healthcare system. What I’ve learned is that so-called “social determinants of health” (SDOH)—food, housing, jobs, transportation, and more—have a far greater influence on health outcomes than medical care alone.
SDOH have not been particularly high-priority for the healthcare sector as a whole. But in 2016, that prioritization changed. In my first few weeks on the job as Chief Policy and Engagement Officer at the Baltimore City Health Department, I was stunned to read that the Centers for Medicaid and Medicare Services (CMS) were launching their Accountable Health Communities (AHC) model. AHC is a national pilot in 32 regions across the country to screen patients, refer them to community resources, and then align local stakeholders to address SDOH gaps, such as advocating for policies that make healthy food affordable and accessible.
AHC sent a definitive signal that addressing SDOH was no longer a risky bet but rather something that CMS could consider covering in the long run. It also meant that those of us engaged in this work had to stop viewing SDOH interventions as one-off activities and begin developing comprehensive approaches to integrate those activities within our healthcare strategies.
One key pillar we focused on in Baltimore was embedding SDOH in our approach to technology and data. Along with health department colleagues and partners across the city, we began to envision what a true technology infrastructure to support this work would look like.
Eager, new SDOH vendors presented us with slide decks and algorithms that they promised would fix everything. The more we listened to those technology vendors, the more questions we had for them, because we knew that any software would inevitably intersect with all of the complicated societal factors that surround social determinants. What happens when a patient in Baltimore is identified as food insecure, but there aren’t any grocery stores within a one mile radius of their house?
Three Realities To Consider in SDOH Technology
The challenges that come with SDOH were around long before the advent of integrated data systems, and they will exist long after —especially if we’re not intentional about the role of technology in this work. Three things to pay attention to:
1-Humans build systems, and systems are human.
Many of the vendors who’ve come my way have spoken at length about the sophisticated algorithms they had developed to “risk stratify” patients with differing social needs into categories that would allow providers to manage their health. While organizing patient data in this way can be extraordinarily helpful and predictive, it also set off red flags regarding biases that may get baked into those systems and further harm patients.
Virginia Eubanks, in her book Automating Inequality, points out that our approaches to poverty alleviation and social services programs—which are an inherent part of SDOH—are rooted in a history of viewing families who access those programs as a burden on society. As a result, algorithms used by state agencies or other entities often create systems where families are penalized further for volunteering information about their social needs and/or seeking assistance.
For technology vendors, public health agencies, and healthcare providers that interact with SDOH technologies, we should ask ourselves whether we are naming potential prejudices and addressing them across design and implementation.
2-The data can only address part of the story.
One of the refrains I heard constantly in planning for AHC—and that I worried about myself—was whether asking patients certain questions might open a Pandora’s box related to historical policies and gaps in the social services landscape. For example, in Baltimore, the waitlist for Section 8 housing is more than five years long—and yet, we wanted to make sure we were asking patients about their housing status since it’s so integral to health.
While technology may give us the data on how many patients have access to affordable housing, it does not tell us the whole story of housing in Baltimore: how Black families were denied housing loans; how entire portions of the city were redlined; how the closing of Bethlehem Steel led to loss of jobs and created white flight out of the city; how a City that once housed a million people now has thousands of vacants and unsafe units waiting for rehab.
SDOH data can back up what many communities already know and experience on a daily basis, but it shouldn’t replace the voices and expertise of those in those communities. How can residents, technologists, and policy officials come together to translate the statistics into collective action on these issues?
3-Technology in the absence of strategy is a minefield.
The integration of SDOH into mainstream healthcare is moving forward at full-force: CVS Health recently announced the launch of a new platform to address social determinants. Epic, the dominant electronic medical record system, is rolling out an SDOH module, and other payors and providers are also moving towards adopting SDOH systems for themselves.
But as with any project that involves the implementation of technology, adopting these tools in the absence of a comprehensive, institutional approach to addressing patients’ social needs leads to unnecessary complexity and duplication. A health system I worked with, for example, shared that their own institution was using five different SDOH technologies—but these systems were not integrated with one another, nor was there a clearly specified role for each one.
Form should follow function. Just like any aspect of healthcare, leveraging technology for maximum impact requires comprehensive strategic planning and attention to the ways in which end-users—every day people like doctors and case workers—will actually use that technology.
As the industry embraces SDOH, we face broader questions about what our new and evolving technology and data infrastructure will add up to. Will we use their builds, and the data that they yield, as an opportunity to consider how we arrived at gaps in our social services to begin with? Will we engage patients themselves in developing creative solutions, both inside and outside of the clinic? And will we ensure that technology plays a supportive, accelerating role in addressing the realities of patients’ lives—rather than dominating the conversation?
Failing to answer these questions could mean perpetuating the same inequities that communities face today. While the answers won’t be found in a slide deck, technologists, healthcare providers and patients must come together to design and implement solutions that —with an assist from technology—could support communities in achieving health.