Researchers compared Area Deprivation Index (ADI) to Healthy Places Index (HPI) in order to assess neighborhood disadvantage as a predictor of in-hospital mortality among sepsis patients.
RT’s Three Key Takeaways:
- Not All Indices Perform Equally: The Area Deprivation Index (ADI), widely used in national policy, misclassified disadvantaged neighborhoods in high-cost cities like San Francisco as “advantaged,” potentially leading to misallocated healthcare resources.
- Healthy Places Index (HPI) Outperforms: Compared to ADI, the Healthy Places Index (HPI) more accurately reflected neighborhood disadvantage and was a stronger predictor of in-hospital mortality among sepsis patients.
- Implications for Resource Allocation: The findings highlight the need to refine socioeconomic tools used in healthcare planning, especially in regions with high housing costs where traditional indices may not capture true neighborhood hardship.
Living in a disadvantaged neighborhood is linked to worse health outcomes, but there are significant differences in how socioeconomic determinants of health are measured. Now new research presented at ATS 2025 finds that a widely used socioeconomic index may not accurately classify patients who live in areas with a high cost of living.
Researchers found the Area Deprivation Index (ADI) incorrectly classified a struggling San Francisco neighborhood as highly advantaged, while the Healthy Places Index (HPI) provided a more accurate assessment of the neighborhood. They also found that the HPI was a better predictor of mortality risk for patients hospitalized with sepsis.
“On a national level and in statewide studies, the ADI is a well validated measure, but its utility is limited in our cohort,” said first author Kathryn Sullivan, MD, a pulmonologist at the School of Medicine at the University of California, San Francisco. “Understanding why this is the case can help us improve our measurement tools.”
The ADI is one of the most widely used area-based socioeconomic measures and is used by the Center for Medicare and Medicaid Services to calculate funding for accountable care organizations. However, a 2023 study in New York suggested that the index may overemphasize home value, raising the concern that in areas with high housing costs, home values could outweigh other components of the index, potentially masking other challenges in the neighborhood.
Sullivan noted that when researchers applied the ADI to San Francisco neighborhoods with a high cost of living where residents struggled with poverty and access to resources, the index classified them as living in one of the most highly advantaged neighborhoods in the country.
For the new study, researchers gathered data from almost 900 critically ill patients with sepsis who were admitted through the emergency departments of two San Francisco hospitals. They used both ADI and HPI to classify each patient’s home address and compared this classification with the patient’s outcomes.
They found a large disparity in how the two indices classified which patients were living in the most disadvantaged neighborhoods. They also found that living in the most disadvantaged neighborhood as measured by the HPI — but not the ADI — was significantly associated with a patient’s risk of dying in the hospital, Dr. Sullivan said.
“Our findings indicate that the ADI is not functioning as intended in our cohort when compared to another measure of neighborhood advantage, which was a better predictor of outcomes in this case,” she said.
The study highlights the need to continue to refine and optimize tools like the ADI and HPI, especially when these indices are used to make decisions about how to allocate health care resources, she said. In addition, while neighborhood disadvantage has previously been associated with negative health outcomes for chronic diseases, the study provides new insights into how these factors can influence health for acute and even critical illnesses, Dr. Sullivan added.
The team hopes to do follow-up research to better understand why the ADI did not perform as expected in this region. They also plan to repeat their analysis in a different region with more rural participants.
