How UCSF created a data platform that leverages 150+ social determinants variables

by Bailey Amber
0 comment

Gaining insights into social factors affecting patients’ lives can be used to augment healthcare delivery in a big way.

That’s according to Dr. Courtney Lyles, an associate professor at the University of California San Francisco, who helped create an expansive social determinants of health-focused data visualization platform. She discussed the platform and its development at the Workgroup for Electronic Data Interchange’s The Quest for Health Equity virtual conference on Feb. 23.

The platform, UCSF Health Atlas, was released last April. Focused on California-based data, it is designed to enable researchers to explore neighborhood-level characteristics and see how they relate to health outcomes. It also includes data on Covid-19.

“It’s pretty clear, and Covid has made it even more clear, that health is really impacted by place,” said Lyles, who is also co-principal investigator of the UCSF Population Health Data Initiative. “Where we live, including our physical and social environment, directly influences health outcomes.”

The interactive platform includes more than 150 social determinates of health variables at different levels of granularity down to the census tract level, which includes between 1,200 and 8,000 people. The platform draws from several sources including the American Community Survey from the Census Bureau, CalEnviroScreen developed by the California Environmental Protection Agency and the California Department of Public Health.

But one of the main challenges of developing a platform like this is deciding what publicly available data to include from the massive trove that is available, said Lyles.

To create Health Atlas, UCSF relied heavily on the Health People 2020 framework created by the Office of Disease Prevention and Health Promotion. This framework helped the UCSF team think through five domains of social factors, said Lyles. These domains were:

  • Demographic characteristics
  • Socioeconomic factors
  • Community characteristics
  • Neighborhood characteristics
  • Health and healthcare indicators

The UCSF team then went in and selected useful variables within each of the five domains. For example, within the community domain, the team selected individual variables like language and foreign-born status as well as wider variables like population density and household composition — that is, people living in households with children versus single-adult households versus seniors living alone.

“Thinking about those variables has really been interesting,” said Lyles. “[It allowed us to] think through what matters for population health and health equity risk.”

Combining all this data on an interactive map enables researchers and clinicians to drill down into granular data on any one variable, and also compare different variables, she said.

UCSF has gone a step further and made it possible to link that social determinants of health data with its EHR data. For example, UCSF combined the two to gain insights into racial disparities in hypertension outcomes. Eliminating this disparity was a pre-Covid goal for the health system, Lyles said.

First, the team extracted every single address that existed within UCSF’s EHR, and then they geo-coded those addresses onto latitude and longitude. They assigned census tract identifiers to those geo-coded addresses so that they could be linked out to the publicly available datasets. Finally, they gathered clinical and demographic data for patients with hypertension receiving care within the system.

By combining all this the system was able to track hypertension patients by race, neighborhood and socioeconomic status in San Francisco on the map and compare these variables with health outcomes, Lyles said. They found that Black patients are concentrated in certain neighborhoods, and they’re struggling with hypertension control.

“This is not showing us something that perhaps we didn’t know already about structural disparities in care or structural disparities in our society,” Lyles said.

But it does show the urgent need for including place-based strategies in health systems’ disparity reduction and quality improvement programs.

“When you put it out there in a visual display, it actually gives you even more impetus to think about neighborhoods you want to target, or [places where] you might think differently about your interventions moving forward,” she said.

Photo: GarryKillian, Getty Images

Related Posts