Asthma is one of the most significant chronic diseases in pediatric medicine, affecting seven million children in the United States. According to one study, asthma exacerbation accounts for “an estimated 14 million missed school days and more than 1.8 million emergency department visits each year.”1
The most significant non-physiological determinants of multiple emergency department (ED) visits and admissions for childhood asthma include race, ethnicity, and socioeconomic disparities. The links between childhood asthma and social and environmental disparities have been established for decades. Unfortunately, the ability to incorporate those factors into the healthcare journey of an individual or population has been sporadic.
However, key technology advancements and the industry-wide shift to value-based care have given providers the ability to leverage data to look beyond the patient medical record at the socioeconomic circumstances that influence health quality.
These nonmedical elements are grouped under the term social determinants of health (SDOH). “Understanding your patient’s history isn’t just identifying medical history, it’s identifying their vulnerability,” says Michael A. Simon, PhD, principal data scientist at Arcadia.io, a population health management company that specializes in data aggregation, analytics, and workflow software for value-based care. “There’s an opportunity to use data to help infer not just the typical medical causality that we assign to chronic conditions, but also to look at the factors that may be influencing them.”
In other words, medical data alone tells an incomplete story.
“The intention is to have a broad set of data to work off of,” says Rich Parker, MD, chief medical officer, of Arcadia.io. “[Childhood asthma] is a specific example where information technology identifies patients in need of specific care management or a social welfare intervention. This is where we really can get down to the level of identifying the interplay between clinical data and social determinants. Put together a registry of asthmatic patients and cross that with social determinant data. Look at patients who are bouncing into the ED several times a year. Maybe you see that they live in a poorer census block or they live in houses that have more dust and dust mites.”
The addition of SDOH data tells a more complete patient story, Parker argues, and will influence how a provider approaches treatment, education, and care management. From the health information management (HIM) perspective, SDOH presents challenges that extend far beyond coding. HIM professionals need to find creative and innovative ways to acquire, analyze, and apply SDOH data to whole-person healthcare.
“Health information managers understand how to report on claims and how to work with very rigorous data standards,” says Claire Zimmerman, vice president of product Innovation at HealthBI, a technology company focused on care coordination. “But the opportunity and challenge today is to think a little bit differently about the information resources that are available and piece it together with that structured information to tell that person’s story.”
A New Old Idea
Why has SDOH, an idea that has been kicking around since Greek antiquity, suddenly captured the collective imagination of the US healthcare industry? Part of the reason is the decade-long shift toward value-based models of care, according to Zimmerman.
“Providers are held increasingly accountable for both outcomes and costs,” Zimmerman says. “If I’ve taken on risk for the outcomes of a given population, and 60 percent or 80 percent of what drives those outcomes is not actually directly impactable by me, then I need to make sure that I have resources and tools available to help drive the appropriate outcomes.”
A survey from the Deloitte Center for Health Solutions found a high correlation between health systems that were screening for social determinants and those involved in at-risk payment models.2 However, despite pockets of innovation, the gulf between the potential of SDOH and actual results is significant. According to findings published in September 2019 by the Journal of the American Medical Association, only 24 percent of hospitals and 16 percent of physician practices reported screening for SDOH factors.3
“You can find physicians that are doing it and in bits and pieces here and there based upon the type of program they may be in,” says Sita Kapoor, chief information officer of HealthEC, a population health technology company. “But holistically, no one is really gathering the SDOH data elements that we need to help drive health outcomes.”
Unlocking the Data Puzzle
One of the most significant barriers to the mainstream utilization of SDOH is data—how to acquire it, how to analyze it, and how to make it actionable.
“The daily practice of medicine requires compliance with contracts, and those contracts have very specific goals in terms of quality measures and utilization. And there are no specific goals in these contracts around SDOH,” says Parker. “Doctors may see patients who are poor or have been incarcerated or experience food insecurity, but there’s only so much that they can do about it.”
“In order to effectively deploy value-based care and sustain it, I believe that you have to focus on health in addition to healthcare. In order to be able to focus on health, I believe you need to move upstream to extend your reach in the community,” says Steve Miff, PhD, president and CEO of the Parkland Center for Clinical Innovation (PCCI), a nonprofit healthcare analytics research and development organization.
“To be able to do that, I believe we need to better understand our community, our patients, from multiple standpoints, not only healthcare, but their life, their environment,” says Miff. “And to be able to do that, you need much more sophisticated data analytics and ways to digitally share it with entities across the community to coordinate those efforts.”
There’s data available—it’s just not easy to get to. According to a 2019 eHealth Initiative survey, SDOH and behavioral health data are the most difficult types of information to collect and share.4
“The challenge with [SDOH data] is that it resides in much more isolated, less sophisticated systems,” Miff says. “It resides in systems in our local municipalities, in an Excel sheet, the food pantry, or the homeless shelter. We need to figure out how to bring this information together. And to truly start to address social determinants of health, we need to leverage advanced analytics to make sense of this data.”
This leads to the question of what to do with the data once acquired. “Do organizations just grab the data and drop it into a folder on a computer somewhere, and bring it up whenever it’s needed? Or do they have a thoughtful design session or an opportunity to talk through how this information gets incorporated into their grander health IT plans?” Simon asks. “The degree to which folks think ahead about how they want that information used could make a big difference in how well they can act on it, and how much they can get feedback from it and report on it.”
Ideas on Acquisition
A research report by Dell EMC and the research analyst firm IDC predicts that the digital universe will contain 44 trillion gigabytes of data by the end of 2020, a third of which will be collected and stored by the healthcare industry.5
The challenge is that about 80 percent of this healthcare data is unstructured.6 Because those “dark data” elements are difficult to identify and apply to business or clinical challenges, they have little inherent value. For this reason, Joe Nicholson, DO, a Board-certified physician and chief medical officer of CareAllies, a subsidiary of Cigna that partners with providers in the transition to value-based care, argues that provider organizations just stepping into an SDOH initiative may start by peering into the ocean of data they already possess.
Complex organizations entering value-based payment models need connective infrastructure to support network-wide performance. This includes underlying technology to aggregate and analyze data from a range of sources, including electronic health records (EHRs), laboratory results, claims-based payer feeds, and real-time admission, discharge, and transfer (ADT) notifications. Advanced analytics capabilities create new possibilities for whole-person care.
“For HIM professionals, SDOH is a conversation about big data management,” Nicholson says. “I would be looking for all of the touchpoints—pharmaceutical data, EHR data, patient survey data—that can better inform an algorithm and will allow organizations to step into something that feels more like a predictive modeling.”
For example, consider the determinant of housing security. At Arcadia.io, one of Simon’s provider clients hypothesized adding homelessness to one of the organization’s contracts. What sort of story could be told from the organization’s own data?
“We dug through a lot of pseudo- and semi-structured information to try to assign concepts to them. We found direct references to homelessness, but also to living in shelters, temporary housing, living with family—all these sorts of concepts that circle around the idea of housing insecurity,” Simon explains. “There wasn’t a strong workflow or a commitment to coding associated with the data. But, as that information made it into the EHR, we were able to start codifying and formalizing this information to enable better reporting.”
Ultimately, the bulk of SDOH data is going to originate from sources outside the provider organization. A major component of data acquisition is understanding what types of information are critical to an SDOH program’s success. HIM professionals should be educated on social issues and health equity, as well as understand how providers could act on information.
“The connected community component starts with assessment, to understand and create the governance structure,” Miff says. “How do you actually prioritize and start to deploy some of these components? Who are the anchor organizations? They need to be initially part of the governance structure and then start to be able to deploy those work processes.”
Making Data Integrated and Actionable
Acquiring SDOH information doesn’t mean much if it can’t be used to serve patient populations or collaborate with external service entities. Some organizations, such as the Gravity Project, are developing use-cases related to screening SDOH data to identify food insecurity, housing stability, and transportation access, as well as defining and standardizing definitions at the discrete data field level.
However, standardization of codable SDOH data remains nascent. Some provider organizations are attempting standardization within the EHR.
“We need to be sure that SDOH data is always going to be in that exact same spot, it’s always going to be completed in this exact same way,” says Catrena Smith, CCS, CCS-P, CHTS-PW, CPC-I, CPC, president of Access Quality Coding & Consulting. “In the EHR, making sure that the information is housed in the same part of the chart, it’s indexed the same way. So that for one patient, it’s not in the nursing notes, and a different patient it’s mixed in with the physician’s progress note, and a different patient it’s over in some case management note.”
For the HIM professional, the complexity of this interplay means that SDOH is very much a team sport. “When the HIM team tackles this data, it’s going to be a matter of pulling together a team, preferably a disparate team of pharmacists and social workers and doctors to create the sort of data-driven opportunities to identify patients at the highest risk,” says Nicholson.
Connecting patients with the appropriate community resources, then assessing the outcomes of those referrals, requires both interoperability and analytics. Essential to the sustainability of an SDOH initiative is creating a data-driven feedback loop among providers, patients, and community service organizations.
“At the center of this is a governance structure,” Miff says. “To enable this data vision, we need to work with multiple entities across the community, whether it’s other providers and payers, local philanthropic organizations, community-based organizations across the spectrum of food pantries, transportation, daycare, and local municipalities.”
Another piece of the puzzle is integrating the patient into this data collection and analytics infrastructure.
“It’s not enough to just create these connected communities to address the social determinants of health, to address some of the underlying environmental factors that are associated with those conditions,” Miff says. “Engaging individuals themselves is the next level that I strongly believe we need to move toward, whether it’s social determinants of health or whether it’s the broader component of value-based care.”
Putting It Together
Recently, Miff, unveiled the results of an SDOH pilot program launched in 2018 and focused on reducing preterm births.
“One of the key things to population health is being able to identify who’s the high-risk and prioritize your activities and your resources to be able to reach out to those where you can impact the most,” Miff says. “In this case, we wanted to reduce preterm delivery rates and work upstream. One of the key elements was increasing prenatal visit attendance, and then through that, reduce and extend the pregnancy, reduce preterm rates, reduce per member per month costs, and ultimately overall reduce maternal mortality post-delivery.”
The PCCI Preterm Birth Prevention Program was fueled by predictive models combining accurate risk prediction, provider notification, risk-driven and tailored patient education via digital technology, and workflow redesign to improve birth outcomes and reduce the rate of preterm births.
The prediction model incorporated multiple data sources, including claims, eligibility, EHR, and community data as well as demographic, clinical, and socioeconomic data, to predict the risk for preterm delivery at any point during pregnancy.
“We infused that information into the predictive model because we’re risk stratifying 26,000 pregnancies a year,” Miff explains. “If you’re able to map social determinants at the block level, then geo-map individuals to specific blocks and use that as a very strong proxy for the needs that they’re likely experiencing in their day-to-day life, you can incorporate those models very effectively into these predictive algorithms.”
The primary interventions were text messages, including appointment reminders, nutrition tips, and other tailored messages. In the first year of intervention, over 21,000 unique pregnancies were prospectively risk-stratified, with about 7,000 pregnancies risk-stratified every month.
More than 800 at-risk patients received text message interventions, and more than 75 percent of patients reported satisfaction with the program. Compared with matched controls, patients receiving the text messages saw a 24 percent increase in prenatal visit attendance and 27 percent drop in early preterm delivery, Miff says.
The back-end technology infrastructure fueling the initiative was a home-grown creation called Isthmus, a cloud-based platform for acquiring and harmonizing data from disparate sources to create a community data platform for inferring population- and patient-level insights for SDOH.
“Part of our journey was attempting to leverage existing technology infrastructure or license existing platforms, but we ultimately decided to build our own back-end technology infrastructure,” Miff says.
In creating the technology for this initiative, PCCI concluded that it needed to be cloud-based, enable machine learning, and have API-based integration with workflow tools.
“We were very thoughtful about using as much open-source modalities as possible because it facilitates that collaboration and translation of knowledge much more effectively,” Miff says.
In addition to the Preterm Birth Prevention Program, Isthmus has been deployed for other SDOH frameworks. These include a pediatric asthma population health initiative that:
- Reduced emergency room visits by 30 percent
- Reduced asthma-related inpatient admissions by 42 percent
- Realized a 36 percent drop in the cost of asthma care for a savings of $12 million
PCCI also developed a predictive model that in two years has helped prevent more than 2,000 adverse drug events (ADEs) for hospitalized patients, delivering a potential savings of over $17 million by reducing readmissions and ADEs.
During its two years of implementation at Parkland, the program has screened more than 87,000 patients, with 8,731 high-risk patients identified. Of the high-risk patients, 16 percent received timely pharmacy intervention and more than 2,000 ADEs were prevented. For high-risk patients receiving a consult, the 30-day readmission rate was cut by 23.5 percent.
Closing the Loop on SDOH
For many providers, SDOH presents the best chance to build a more effective, efficient, and holistic health system. Data—and how it is applied—will play a central role in the success of every SDOH initiative, which means that HIM professionals are essential stakeholders.
“We know that right now, at least from a coding standpoint, we can’t capture everything that links back to SDOH by way of an ICD-10-CM Z code. We can capture a lot but not everything,” Smith says. “What that means is that we’re not—unless something changes—going to be able to rely on ICD-10 Z codes as a data point for all social factors. We’re going to have to figure out other ways to be able to determine that the patient hit that mark for a particular social determinant of health.”
HIM professionals need to think far beyond coding issues when it comes to SDOH and where they need to be involved. There are many more questions than answers right now:
- How do we ensure the right information is being collected?
- How will the data be analyzed/checked?
- What SDOH data elements are being collected and why?
- Are SDOH data elements being prioritized in terms of being the ones most likely to be influenced for purposes of care delivery and care coordination?
- Are stakeholders (e.g., healthcare institutions, government agencies, clinicians, payers, multi-stakeholder organizations) standardizing the most important SDOH codes to collect?
- Is there consensus about which data elements should be collected?
- Privacy—what is the patient’s role in all of this? Should mimum necessary standards apply? Is patient consent required?
This is certainly not an exhaustive list of questions organizations and the industry must address. It’s very clear that HIM professionals can enhance and inform stakeholder discussions about SDOH.
From their expertise in data integrity and their insights into the person behind the data, HIM can help ensure the best decisions are being made and that the right questions are being asked. Look for ongoing content addressing questions like those listed above—and more—as the Journal delves deeper into SDOH.
- Johnson, Laurie H., Patricia Chambers, and Judith W. Dexheimer. “Asthma-related emergency department use: current perspectives,” Open Access Emergency Medicine 8:2016, pp. 47–55.
- Deloitte. “Addressing social determinants of health in hospitals.” www2.deloitte.com/us/en/pages/life-sciences-and-health-care/articles/addressing-social-determinants-of-health-hospitals-survey.html.
- Fraze, Taressa K. et al. “Prevalence of Screening for Food Insecurity, Housing Instability, Utility Needs, Transportation Needs, and Interpersonal Violence by US Physician Practices and Hospitals.” JAMA Network Open 2, no. 9: 2019. https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2751390.
- eHealth Initiative. “2019 Survey on HIE Technology Priorities.” May 15, 2019. www.ehidc.org/resources/2019-survey-hie-technology-priorities.
- IDC. “The Digital Universe of Opportunities: Rich Data and the Increasing Value of the Internet of Things.” April 2014. www.emc.com/leadership/digital-universe/2014iview/executive-summary.htm.
- Tolson, Bill. “Where Should Healthcare Data Be Stored In 2018 — And Beyond?” Health IT Outcomes. February 20, 2018. www.healthitoutcomes.com/doc/where-should-healthcare-data-be-stored-in-and-beyond-0001.
Matt Schlossberg (firstname.lastname@example.org) is editor at the Journal of AHIMA.
Continuing Education Quiz
Review quiz questions and take the quiz based on this article, available online.
- Quiz ID: Q2019103
- Expiration Date: March 1, 2021
- HIM Domain Area: Clinical Data Management