Social Drivers of Health (SDOH) Coding
Next Steps for FAHC (2022)
For a more advanced introduction to this topic, see this Back to Basics webinar on coding & data structures for food-based interventions (2022).
Introduction to “Z-Codes”
When we talk about coding for health-related social needs (HRSN), usually that means a conversation about “Z-codes”. (Note - it could mean something much more comprehensive, see for example the Gravity Project work on social informatics, but if you’re getting into that level of detail then you don’t need this explainer page).
“Z-codes” are part of the ICD code schedule, usually ICD-10. These are a set of common diagnosis codes (ICD = International Classification of Disease, 10 = 10th edition). Together with the CPT (Common Procedural Terminology) codes for clinical procedures and services and the HCPCS (Healthcare Common Procedural Coding System) codes, which include and expand on CPT codes, they make up the most basic lexicon for documenting our health care.
There are many more acronyms and taxonomies where those came from. This is the minimalist version, for an expanded introduction to the topic, see this Back to Basics webinar on coding & data structures for food-based interventions (2022).
Within the basic system described here, a few distinctions matter - there is the diagnosis (ICD codes) and the services that go along with that diagnosis (CPT and HCPCS). Within services, there is the distinction of whether the activity is a medical service by a licensed provider working within their scope of practice (CPT) or a different type of service such as care coordination (HCPCS Level II).
The Z-Codes are diagnosis codes. The American Hospital Association provides detailed materials on their use as part of addressing social drivers of health on their website, which offers a good introduction to the topic.
Hospitals can capture data on the social needs of their patient population using the ICD-10-CM codes included in categories Z55-Z65 (“Z codes”), which identify non-medical factors that may influence a patient’s health status. Existing Z codes identify issues related to a patient’s socioeconomic situation, including education and literacy, employment, housing, lack of adequate food or water, or occupational exposure to risk factors like dust, radiation or toxic agents.
For example:
Z59.41 - Food Insecurity
Z59.48 - Other specified lack of adequate food
A health care practice could also document the service that led to this diagnosis. For example, with CPT codes 96160 / 96161.
CPT 96160 = ADMINISTRATION OF PATIENT-FOCUSED HEALTH RISK ASSESSMENT INSTRUMENT (EG, HEALTH HAZARD APPRAISAL) WITH SCORING AND DOCUMENTATION, PER STANDARDIZED INSTRUMENT (96161 is a Caregiver-Focused variation)
You’ll also notice the screening example is a CPT code - that puts it in the clinical / medical realm. The description of its intent is “Health and behavioral assessment is an evaluation of psychological, behavioral, emotional, cognitive and social factors that affect the patient’s response. It is conducted through health focused interviews, observations, and clinical decision making.” (Emphasis added). The standardized food insecurity screening tool does not require clinical decision making, but it can be used in the context of clinical decision making - which this code reflects. Remember that point for the reimbursement section.
While there are codes for determining food insecurity, fewer codes exist for responses to that diagnosis. Sometimes a diagnosis on its own can be helpful to a patient who didn’t know that, for example, their cholesterol levels indicated risk for heart disease. In this case, though, the patient isn’t getting any new information from the diagnosis alone - they’re the ones who told the health care practice they had trouble with food access, so what value did the health care practice provide in exchange for that information?
A recent ICD-10 code update provides an example of building a bridge to recording both diagnoses and relevant services. In 2022, CMS added:
Z91.110 - Patient’s noncompliance with dietary regimen due to financial hardship
This code might go with a number of clinical services, including Medical Nutrition Therapy (establishing the dietary regimen), Chronic Condition Management codes (supporting patients with managing diet-related health conditions), Office Visits with a Primary Care Provider, and so on. You can also see how it would lead to actionable information - a practice can know that a patient is receiving support in understanding and designing a dietary plan for their health goals, but that they need financial assistance to complete the treatment.
Structured Data
Health care practices can accomplish some goals of the ICD-10 / HCPCs / CPT code system without using those particular code sets. Practices can create structured data through their own electronic health record (EHR) system. For example, since 2017 it is becoming increasingly common for EHRs to have built-in fields for food insecurity screening and results, which will prompt the Hunger Vital Sign questions, allow the screener to select answers, and translate to positive or negative. A health care practice could set up filters to interpret this structured data, without adding the ICD-10 code Z59.41. Practices can also build off of the standard EHR structure by adding fields or menus, such as a drop down list of choices for food access programs they might offer as referrals.
At a basic level, this structure can act like an index to a book. If a provider wants to quickly see patterns in a patient’s records, for example asking whether food insecurity concerns coincides with difficulty managing medications or coincides with calendar dates for seasonal employment, then these codes facilitate that review. The structure also allows for analysis across a large group of patients (aka the patient panel). Adding together information in this way allows a practice to answer questions like “how many patients reported a food insecurity concern this month? Has this increased or decreased from last month? How does that compare with other community data sources?” Practices can also create patient cohorts for comparison, for example knowing whether patients experiencing food insecurity who receive a new pre-diabetes diagnosis rely on medication-based treatment more often than patients with the same diagnosis and no food insecurity risk, who might opt to focus on dietary change.
There are drawbacks to relying only on EHR systems to create structure.
It’s difficult to build and maintain an accurate taxonomy - it may be the most expedient route to a simple task (yes / no on food insecurity, for example) but from that starting point are increasing levels of complexity (like tracing the response to a positive screen). You are adding the time taken for developing the data structure to the time needed to train staff, set up data collection systems, review data quality, etc. It may also become expensive, if a health care practice needs to pay for adjustments to their EHR to accommodate the new applications.
A potentially bigger problem is that a bespoke taxonomy only works within the system that built it - it isn’t designed to exchange information with outside organizations, whether they’re other health care providers, social service providers, community organizations, or payers.
Standardized coding systems allow different organizations to talk to each other. For example, the Office of the National Coordinator for Health Information Technology maintains standards that allow different health care information systems to talk to each other - in this case they look at both coding in the sense of computer programming (see, for example, FHIR) and coding in the sense of the data inputted into those programs (see, for example, the SDOH Information Exchange learning forum). And “billable codes” can be used on claims to insurance companies for reimbursement (see next section).
The 211 LA County Taxonomy of Human Services is the classification system used to index information about organizations that provide community services. It includes different types of organizations (hospital, school, library), the services they provide, and the people they serve. Health care SDOH coding work by the Gravity Project, and others, is designed to integrate with this existing classification system used in social services and community resources through matching key terms.
See also the Outreach & Referral section for examples of 211 platforms offering integration between health care and social service referrals.
Codes for Billing
One of the incentives for a health care practice to adopt a system like ICD-10 Z-codes is that they are putting their SDOH interventions into the language of billing.
It’s a complicated language.
Health care reimbursement doesn’t work like a restaurant menu with prices. A health care practice needs to be able to communicate information that lets payers answer questions like what happened in a visit, who performed those services, under what supervision, whether that person was qualified to offer those services both by licensing and by credentialing, how frequently has a patient received the services, for what diagnosis, was a referral or prior authorization required (and, if so, received), and so on.
One example of how SDOH Z-codes can fit into reimbursement is in changes made to Evaluation and Management (E&M) services codes that went into effect in 2021. These are office visit codes and they have different levels of reimbursement. Physicians can calculate the level based on time or medical decision making. In the medical decision making scenario, complexities due to SDOH concerns can increase the payment level. An article from the ICD10 Monitor describes this new system. It provides the following example:
One example from the CDC states that 1 in 4 adults with advanced chronic kidney disease (CKD) are food insecure – food insecure without hunger, food insecure with moderate hunger and food insecure with severe hunger. ICD-10-CM coding options include Z59.4, lack of adequate food and safe drinking water for reporting food insecurities. Raising awareness of food insecurities, asking, documenting, coding and reporting may help to target interventions to improve the health of people with CKD.
This option isn’t a guarantee that there is a financial incentive to incorporate the codes. For example, providers are choosing between time-based and complexity based justifications for their office visit reimbursement level. There is an added burden of documenting the presence of health related social needs and their impact on medical decision making during the office visit. A practice may also observe that incorporating SDOH considerations usually extends the length of a visit. It may be simpler to default to the time-based structure.
This scenario suggests a Catch-22 in introducing new coding to reflect the intersection of SDOH and medical care -- you can’t calculate the strength of the financial incentive until you’ve started to build pilots and experiment with the codes. But practices may not want to start trialing new coding systems until they have a financial incentive to do so. One possible way out of the quandary is to use predictive social analytics to create scenarios and estimates. We did not find any organizations in Vermont performing this analysis.
The question of incentives for SDOH-based coding becomes particularly relevant when looking forward to designing alternative payment models that might cover health care efforts to address health-related social needs. Information tied to claims isn’t reviewed only by the practice and payer, it also supports policy research. Anonymized data sets like all-payer claims databases (VHCURES in Vermont) help regulators and policymakers understand what is happening within a health care system, the impact on patients, and the potential benefits (financial and health-related) of adding new coverage.
For additional discussion, see the following:
Six Levels Down - from the Against the Rules podcast (general interest)
CMS Innovation Center - Part 3 of the Hunger Vital Sign series (studying impact of screening & referral on health care quality and cost)
Documentation of California’s CalAIM initiatives provides examples of putting data collection and common billing into practice for health care funded food services.
Next Steps for FAHC (2022)
In 2020, FAHC reviewed the current status of health care practices’ use of Z-codes in Vermont. We found little-to-no consistent use of these codes. They appeared most frequently connected to specific pilot or research projects.
We also reviewed scenarios in which health care practices might see an operational advantage in SDOH coding. Many of the projects with the initial FAHC focus required only basic data structures, such as recording the results of a food insecurity risk screen. National discussion suggested that practices would not be commonly expected to implement Z-codes (or similar code sets) in this use case. For example, in the FY2023 CMS rule for collecting SDOH data as part of inpatient admissions, their calculation of provider burden assumed a structured EHR form as the common methodology. In another example, FAHC participated in the ONC’s 2022 SDOH Information Exchange learning forum and determined that Vermont does not currently have several of the key factors that led other regions to implement standardized data recording as a step towards participating in these exchanges.
Based on lack of current coding and a perception that codes would be unnecessary for FAHC pilot projects, we did not pursue this topic area.
Three related pieces of information were created in connection with FAHC technical assistance support for pilot projects to use food interventions as part of reducing cardiovascular disease at 3 rural FQHCs:
Mapping a conceptual overview of data structure that would connect actions & outcomes across a range of potential food is medicine interventions. This map is based on common program evaluation structures in the health and agriculture fields.
Creating a structured data dashboard that would allow for comparing the implementation of different food interventions, and their results, across three different practices. This dashboard is based on actual workflows at the practices, offering key terms that identify commonalities across the three examples.
Reviewing whether financial incentives exist for FQHCs to implement SDOH coding as part of detailed data collection and quality improvement for food-based interventions. We found no incentives at this time. See larger cost-savings estimate model linked here. Note: FQHCs do not use the office visit code levels provided in the earlier example, and so this angle was not evaluated.
In 2022, CMS added the ICD-10 code Z91.110 - Patient’s noncompliance with dietary regimen due to financial hardship. This code offers an opportunity to address current limitations in understanding factors that influence the use of clinical nutrition services, as discussed on our Nutrition Services page. If additional funding is received for future projects, one proposed initiative is collaboration with Registered Dietitians and the Vermont Academy of Nutrition and Dietetics to integrate this code into documentation workflows.