Social determinants of health (SDOH) are transcending buzzword status and emerging as a key issue in the next wave of reform and improvement in healthcare systems across the country. As we explored last time, it’s still early days for figuring out how best to design, deploy, and gauge the success of new approaches that expand healthcare into health.
The work to date, a mix of grant-funded pilot projects and progressive investments from forward-thinking health systems, has helped to create a foundational evidence base to illustrate key approaches and challenges of building these interventions. Brand-new models like CityBlock are aiming to leapfrog the whole ‘where’s the ROI’ phase, through a spate of partnerships.
The inconsistency of SDOH Data Capture
For the incumbent health systems out there who can’t start from scratch, policymakers are turning their attention to one of the thorniest challenges that’s emerged from this early work: Integrating SDOH data into existing health IT infrastructure. As summarized recently by NYU researchers in Health Affairs, data integration remains tough for myriad reasons:
- A lack of consensus on which screening tools to use to capture SDOH data, exacerbated by a spate of unenforced recommendations from groups like the IOM NACHC, and other acronymed entities, as well as private vendors
- Inconsistent and unenforced approaches to data sharing among community stakeholders, driven by competing population health management platforms from EHR vendors, HIE programs, and homegrown efforts
- Despite the potential of aggregate, community-level data (zip codes, demographic information, etc.) to predict risk, these data have neither been systematically mapped to existing clinical workflows, nor to the litany of needs assessments and screening tools described above
- While ICD-10 introduced new ‘Z-codes’ to document social dimensions of patient’s health back in 2015, these codes can be redundant and conflicting, and they don’t harmonize to the aforementioned screenings. Moreover, research suggests doctors who were diagnosing based on ICD-9 for decades haven’t made much use of the new codes anyway; You can’t teach old Docs too many new tricks
“The result of this situation,” the authors write, “is that existing clinical standards lack explicit translation algorithms to comprehensively integrate results of a screening tool for social determinants of health into an EHR as a relevant clinical finding or problem.” Translation: Too many frameworks and not enough alignment on makes it tough to realize the value of expanding clinical care outwards to address SDOH.
Consensus by Committee and Comity
If you’re wondering, “Isn’t there a national office, dedicated to coordinating these types of health data issues?” you’re exactly right: The Office of the National Coordinator for Health IT (ONC) has made some progress unifying data capture to an existing set of standards (LOINC) across eight domains (e.g. housing, transportation, food security, etc.). Several organizations, such as UCSF’s SIREN team, the Regenstrief Institute, OHSU, and others are also driving research to support policy development on this front.
There’s plenty of this sort of committee-driven, consensus-building work ahead. Ultimately, the ONC will need to hold EHRs and other health IT vendors’ feet to the fire to incorporate these crosswalks and taxonomies into the guts of their informatics engines so that different social needs programs are running on an interoperable (or at least common) data language.
As we saw in the Meaningful Use era, nudging physician practice into new health IT workflow is no cakewalk. Despite the general consensus on SDOH being the right direction for US healthcare, the specific paths forward are already branching out. The American College of Physicians (ACP) recently advocated for “the development of best practices for utilizing EHR systems as a tool to improve individual and population health without adding to the administrative burden on physicians,” as part of a broader set of recommendations for addressing social needs in clinical care.
The ACP’s gentle hedge, about preventing administrative burden, is a wrinkle worth unfolding.
Already, CMS has incorporated different screening tools into their various pilot programs, from CPC+ to the Accountable Health Communities (AHC). Commercial vendors like Healthify, NowPow, and UniteUs are quick to point out that the AHC screening tool sets an almost comically low bar for health systems, so they’re pushing more intensive needs assessment frameworks into the market through their tools.
Meanwhile, various specialties are all leading the charge on social-needs screenings specific to their populations: pediatrics for hunger screening, oncology for specific cancer risk factors, and so on. This helps to drive progress during these early days, but in the future, all of these one-off efforts may require a macro, MACRA-style reset to help streamline how we’re approaching measurement of social health.
All wrinkles considered, the common vision is still fairly straightforward: In the near future, any health system should be able to use a variety of screening tools – standardized, homegrown, or a combination of the two – to gather information on their patients’ social health, into whichever EHR system they happen to have in place. Within that healthIT infrastructure, any new SDOH data should map consistently to appropriate codes for diagnosis and treatment, which will help ensure that the last decade-plus of investments in health IT infrastructure and workflow modification, as well as ongoing and future investments in predictive algorithms and other tools are all informed by the real-world needs of patients and communities.