Clinicians and health care staff use billing codes to document patients’ illnesses and treatments in their medical records. Z codes identify nonmedical but health-influencing social risks such as unstable access to food, employment, or housing. Understanding social factors linked to risks of illness and death can improve our understanding of the overall threats to patients’ health. 

However, a new study by LDI Senior Fellows Paula Chatterjee and Eric T. Roberts, with Statistical Analyst Eliza MacNeal, found that hospitals rarely use Z codes, particularly for the sickest patients who often face multiple social risks. 

This has significant policy implications. “If Z codes are more likely to be used with certain patients and settings,” Chatterjee said, “then policy decisions based on Z code data could be mistargeted or ineffective.”

Biased Z Code Documentation Fails to Describe Social and Health Risks

Chatterjee and colleagues examined inpatient claims for more than 7 million Medicare beneficiaries. The analysis included Z codes recorded during inpatient stays for individuals hospitalized in 2022, 86% of whom were over age 65.

The study confirmed a widespread underuse of Z codes: only 2% of hospitalized patients had at least one recorded Z code. Hospital personnel assigned Z codes less often for clinically complex patients, even though having multiple chronic conditions is associated with higher social risks. Z codes were documented for 1.5% of high-complexity but 2.8% of low-complexity patients.

Patients at the highest risk of dying within 30 days of hospital discharge were least likely to have a Z code documented during hospitalization. The presence of codes correlated with a nearly 1% lower probability of postdischarge death compared to no Z code — counter to evidence linking social risks to increased death risk.  

Patterns of Z Code Use Could Hinder Help For At-Risk Populations

Patients with more health system interactions had more Z codes, suggesting measurement bias against individuals who face barriers to accessing care. Low Z code use may reflect clinician and health system resource constraints. Safety net hospitals and other sites that serve people with low income and complex health conditions may not have the capacity to discuss and document patients’ social risks.

“These results,” Roberts said, “suggest that Z codes may be particularly underused for clinically and socially vulnerable patients.”

These sources of measurement bias can have serious consequences for patients and health systems as Z codes are increasingly used in decisions that affect insurance payments, care quality, and outcomes. Underuse means programs designed for patients with substantial social needs may overlook the very people they intend to help.

Z codes sometimes determine enhanced payments linked to patients’ social risk factors in Medicaid and in a program designed to improve care for traditional Medicare beneficiaries. “Measurement bias in Z code use could distort these payment models,” Chatterjee said, “and penalize providers unfairly or mistarget needed funding and resources.”

Recommendations for Moving Beyond Z Codes

Z code documentation has been rising, but remains low and variable. Instead of using Z codes to understand patients’ social and living conditions, Chatterjee, Roberts, and MacNeal suggest that health systems use existing, validated measures of social risk from their communities, such as local health surveys. Public health infrastructure for these surveys varies nationally, but unlike Z codes, they do not depend on health system interactions for an accurate estimate of community-level social risk. 

At a higher level, the researchers recommend investments to address hazards represented by Z codes. If policymakers continue to use Z code data, they should account for measurement biases against people with clinical complexity and barriers to accessing care. 

The study is part of Chatterjee’s and Roberts’s research to ensure that policymakers use accurate definitions and measures of care. “We hope this work puts this measurement bias on the radar of policymakers and health systems,” Chatterjee said. “We want to ensure that well-intentioned policies to improve the health of socially vulnerable populations actually achieve their goals.” 

---

The study, “Measurement Bias in Documentation of Social Risk Among Medicare Beneficiaries,” was published on July 18, 2025 in JAMA Health Forum. Authors include Paula Chatterjee, Eliza MacNeal, and Eric T. Roberts.

---

Author

---