How one CIO built a COVID-19 data collection system from the ground up

Eric Raffin, CIO of the San Francisco Department of Health, talks to Healthcare Finance News about his upcoming session at HIMSS22. 

Photo: Healthcare Finance News

When the number of COVID-19 cases began ramping up in California as they did elsewhere across the country in March 2020, San Francisco Department of Public Health CIO Eric Raffin was tasked with tracking cases, test counts and hospitalizations.

Raffin built a team to collect data from the estimated six acute care hospital systems in the Bay area. The information was input, sometimes manually, onto a spreadsheet.

Hospitals were contacted daily for the information, at a time when staff were doing all they could do to care for patients during pandemic surges. 

"We had daily dashboards, updates (such as) shelter in place, folks experiencing homelessness who needed to get into safe places," Raffin said. "The thing is, a lot of the information in there wasn't accurate."

It was Raffin's job was to validate, by phone, every day, all of the acute care hospital data in San Francisco.

"There was not one day it was error free," Raffin said. "I was taking nurses and other healthcare professionals out of their jobs to validate the information. We realized we had it with that. We had to make a change."

How the San Francisco Department of Health changed to a new COVID-19 Hospitalization Data Repository (CHDR) – pronounced like the cheese – to make the data actionable is the subject of Raffin's HIMSS22 talk, "Transforming Pandemic Hospitalization Data into Action!" Tuesday, March 15, 10:30-11:30 a.m., Room WF3 of the Orange County Convention Center in Orlando.

The first priority was to remove the human from reporting the data, Raffin said. They turned to the EHR and worked with Epic to build out a separate instance of an EHR that contained only the needed information. They needed to know which systems had ventilators and open beds, along with demographic information on where the cases were coming from.

"We were getting none of that data," Raffin said. "We weren't getting meaningful information to start contributing to a predictive model for hospitalizations."

When that information started becoming available, they learned that about 30% of hospital admissions were not from people living in the San Francisco area. They had enough information to start building a predictive model on surge peaks and hospitalization numbers.

A COVID command center was able to work with hospitals on surge and ambulance planning. 

They also were able to corroborate known disparities of care based on socioeconomic status and homelessness and break down that information by neighborhood. For instance, Raffin said, from seeing data showing one case in a housing project rise to four cases the next day, they could send in vaccine and testing help.

Raffin and team built CHDR from the ground up, having access to no other model in the nation to use. "We view the work as pioneering," Raffin said. 

While COVID-19 may never fully go away post-pandemic, Raffin sees CHDR remaining in use for other data collection opportunities.

"There's indispensable utility," he said, "to having a hospital surveillance system that has more detailed information in it."

Twitter: @SusanJMorse
Email the writer: SMorse@himss.org