The Allegheny Health Network is a nine-hospital health system spread across the Pittsburgh area in western Pennsylvania. Of the millions of patients the system sees each year, one in particular stands out to Deborah Donovan, vice president of social determinants of health strategy and operations at Highmark Health, AHN’s parent company.
The patient was diagnosed with stage 4 breast cancer but was not showing up for critical follow-up care. Providers discovered that her hesitancy stemmed from her living situation: She lived in her car with her teenage son, and feared that disclosing her homelessness would lead to his removal from her custody by child services.
To Donovan, the situation exemplifies the relationship between nonmedical issues such as housing, income, transportation and food insecurity – otherwise known as social determinants of health – and a person’s access to care and ability to follow treatment plans.
It underscores the need to “look holistically at what really is impacting overall health,” says Donovan, to ensure that diagnosis occurs and treatment is effective.
Social determinants of health are now understood to account for approximately 60% to 80% of individual health outcomes. Yet making room in clinical care to address problems occurring outside the doctor’s office can be challenging, even if those problems are greatly impacting a person’s health. Assessing the scope of these issues could shed light on how best to resolve them, but health-related data has lagged behind in this area, leaving decision-makers in the dark.
One effort to help bridge this gap came in 2015, when the International Classification of Diseases medical coding system – used by providers to document patient encounters and diagnoses electronically – included an expanded bloc of codes designed specifically to capture social determinants of health. These specific “Z-codes” can be used to inform individual treatment plans and improve patient outcomes, as well as to understand challenges faced by entire populations. They are numerous and far-ranging, encompassing everything from employment problems and family conflict to housing instability and social isolation.
Health care providers’ adoption of Z-codes has been slow, however, and data on social determinants of health has remained sparse and unstandardized as a result, according to a recent study conducted by Dr. Karen Joynt Maddox, an assistant professor at the Washington University School of Medicine in St. Louis.
Joynt Maddox and her fellow researchers found Z-codes tied to social determinants of health were included in less than 2% of admission records for more than 14 million hospitalizations that occurred in 2016 and 2017. The findings echo those of a similar study by the federal Centers for Medicare and Medicaid Services, which found that the health insurance claims of just 1.4% of nearly 34 million Medicare beneficiaries contained Z-codes in 2017.
Joynt Maddox says the failure to use Z-codes in medical documentation represents a missed opportunity to understand and address the role of social and economic issues in the health of individual patients, as well as within entire health systems and communities.
“It’s difficult to know what this means for individual health outcomes or national policy without collecting data,” she says.
If uniform data on social determinants of health were more widely collected, health systems could employ it to direct their often limited resources to address the issues most impacting their patients, and to forge community partnerships with organizations that could further assist, says Heidi Allen, an associate professor of social work at Columbia University.
Policymakers also would gain a better understanding of problems faced by the communities they serve, which could inform legislative efforts and funding allocations. “Having access to information about what is affecting people’s health allows us to shape our policy priorities,” Allen says.
The reasons behind the scarce use of Z-codes are complex. Dr. Julia Skapik, medical director for informatics for the National Association of Community Health Centers, points out that “Z-codes were created for billing purposes” – though social determinant codes are not currently tied to insurance payments – and “not to create action in the field.” Meanwhile, medical providers are often strapped for time – overburdened with full patient schedules, short allotments of time for visits and onerous requirements tied to electronic documentation.
“Expecting physicians to manage one additional domain of activity is probably a failed enterprise,” Skapik says.
In their recent study, Joynt Maddox and colleagues found that among hospitalization records that included at least one Z-code, certain codes tended to be more common – such as one indicating homelessness. This illustrates a tendency on the part of providers to document only “social issues that are immediately germane to the problem,” Joynt Maddox says, as a hospitalized patient’s lack of housing does not bode well for their post-discharge recovery, and could land them right back in the hospital.
Z-codes also were more commonly used when a hospital admission was related to mental health issues or substance use, the researchers found, which Joynt Maddox attributes to the role that social determinants can play in driving these patients to be hospitalized. A job loss or falling-out with a family member, for instance, may heighten the struggles of a person dealing with these issues, leading to hospitalization.
Still, this means less “glaringly obvious” social determinants that nevertheless impact patient outcomes are sometimes overlooked, Joynt Maddox says. Providers may not realize, for example, that a patient hospitalized for diabetic complications can’t afford their medication when discharged, and faces a similar risk of needing emergency care or repeat hospitalization after being released.
Some health care organizations have developed tools for screening patients for social and economic needs, and have built Z-codes into the process of documenting these needs in a patient’s electronic health record. The NACHC helped to develop a comprehensive screening tool for social determinants of health that includes built-in mechanisms for entering Z-codes that correspond to patient responses – yet they’re not always used.
Dave Faldmo is the medical director of Siouxland Community Health Center in Sioux City, Iowa, and an early adopter of the screening tool. He says patients complete an annual questionnaire covering various social determinants of health and their responses are entered in the electronic health record. But Z-codes are typically not added, despite providers having a smooth process to do so.
Patients are typically “dealing with multiple complex medical problems” and Z-codes are “one of a million things we have to do for that visit,” Faldmo says. According to data from the federal Health Resources and Services Administration, more than 80% of Siouxland’s patients have incomes below the federal poverty level, and Faldmo says many are greatly impacted by factors such as toxic stress and a lack of access to healthy foods.
Z-code data would be most useful to share with health insurance companies, says Faldmo – private ones, as well as Medicaid and Medicare managed care organizations.
“If payers knew more about patients’ complexity, it would help when negotiating contracts,” he says. “Managed care organizations are being more proactive about social determinants of health now, and data would help inform them. We could work together to overcome barriers” to addressing social needs.
According to Deborah Donovan at Highmark Health, such data-sharing also could cut costs for health systems and insurance payers, as patients are less likely to need costly medical care if their social needs are met.
Ultimately, a lack of insurance reimbursement may explain why these Z-codes are so little used. Insurance companies pay for services based on diagnosis and procedure codes contained in medical documentation and submitted in claims, but Z-codes for social determinants of health don’t trigger such payments, and this means “there’s not a reason for providers to use them,” Donovan says.
The growing recognition of the profound impact of social and economic conditions on health that has occurred across medicine was driven in part by the transition to value-based payment models embedded in the Affordable Care Act. The law contained various provisions designed to transform care delivery by tying reimbursement to patient outcomes instead of service volume. This forced providers, even those who did not typically “think outside their walls,” to realize that “if we want to make people well, we have to care about this stuff,” Joynt Maddox says.
Some experts have advocated for a model that incorporates risk adjustment for social determinants of health into performance-based payments for quality outcomes such as hospital readmissions. Currently, hospitals are penalized in connection with the readmission of Medicare patients who have certain chronic conditions within 30 days, but these penalties have been criticized for failing to incorporate socioeconomic risk factors and inflicting disproportionate financial pain on so-called safety net hospitals – a problem Congress has sought to alleviate – since they see a larger number of low-income and otherwise vulnerable patients.
Other value-based payment models allow providers to bill more for medically complex patients. And under new coding guidelines released by CMS, providers can account for the impact of social determinants of health on clinical decision-making and bill for a higher level of complexity based on social factors as well, effectively incentivizing them to consider and code for these issues.
Highmark Health is part health care delivery system and part health insurance company, covering publicly and privately insured patients across three Northeastern states. In the future, the company plans to “embed the (Z-code) requirement in a value-based strategy,” says Donovan, which is “how we really would envision and create the encouragement to use them.”
Right now, Highmark pays its providers for care coordination based on a patient’s level of clinical risk, but “I could envision a time where we evolve that care coordination payment to also consider social risk,” she says. Highmark plans to layer patient’s social risk factors into its quality measures as well.
First, though, Highmark is working with the Gravity Project – a national collaborative of health care stakeholders focused on standardizing the collection and exchange of social determinants of health data – to tweak and expand existing Z-codes to improve their applicability. Evelyn Gallego, program manager of the Gravity Project, says Z-codes were developed without sufficient input from the field and contain some significant gaps. One of these is an absence of codes for transportation problems, which is one of the most common social determinants impacting health.
The Gravity Project’s work to adjust Z-codes is “foundational to the social determinant infrastructure that gets built for this country,” says Highmark’s Donovan. If proposed reworked Z-codes are adopted by federal officials this year, they could potentially be implemented in late 2021, at which point Highmark plans to roll out new value-based payment initiatives to incentivize their use. Other payers may well follow suit.
“Z-codes are just critical to our work moving forward. If we don’t have those, we are so much more limited in what we can do and how we can advance programs, value-based strategies, analytic insights and population health management,” Donovan says.