Momentum Builds for Bolstering Public Health Data Infrastructure
The pandemic exposed several weaknesses in the nation’s public health data infrastructure, but it also has prompted action and investment on the part of the Centers for Disease Control, the Office of the National Coordinator for Health IT, and others to prioritize public health use cases.
One of the clear problems that COVID-19 exposed is just how fragmented and siloed the public health system is from healthcare delivery, says Brian Dixon, Ph.D., director of public health informatics at the Regenstrief Institute and Indiana University’s Richard M. Fairbanks School of Public Health at IUPUI. In order for laboratory results about COVID to get to public health, they had to be faxed; there weren't electronic data feeds. “The other component is that public health departments didn't have a good electronic way to count people who were in the hospital with COVID,” he says. A new, separate system had to be developed in the middle of the pandemic in order for hospital staff members to manually key in how many people were in beds and how many people were in the ICU with COVID. “It demonstrates the fundamental lack of interoperability between medicine and public health,” Dixon says.
The view from Salt Lake City
To give a specific example of the challenges Dixon is describing, Angela Dunn, M.D., M.P.H., executive director and health officer of the Salt Lake County Health Department, recently outlined some strengths and weaknesses Utah experienced during the pandemic. Speaking during a CDC-ONC Industry Day event, Dunn stressed that Salt Lake City’s department has a strong informatics infrastructure, but she noted that the state’s local public health departments all have their own electronic medical records. “We have 13 health districts, and I think seven different electronic medical health records, and they had varying capabilities of handling a surge during a pandemic — from accepting appointments to getting data in— and routinely they crashed during COVID,” she recalled.
“The capacity of our EMRs to link with our immunization system was surprisingly very, very troubling,” she noted. One rural health department, for example, was giving mass vaccination clinics in one building. They had a school building where they did their clinics, and it was across the street from their health department building. A statute required that they had to get all of their vaccine Information in by the next day. They finished their clinic at 10 p.m. All of their records were entered by hand and then they had to go to the health department across the street to import them into their electronic medical record system because they could not access their EMR outside of their health department’s physical location. “That's insane,” Dunn said, “but that still happens even today.”
Dunn has been with Salt Lake County for a year and a half. Previously, she served as the state epidemiologist for the Utah State Health Department. Another problem she recalled at the state level was trying to inventory PPE and ventilators. “At the beginning of the pandemic, I could not get a good count of how many ventilators we had in our state,” Dunn added. “I know that probably seems like decades ago when we cared about ventilators, but it was a big, big deal at the beginning of the pandemic. I didn't know how much PPE we had. We also didn't know our hospital capacities in real time. There was this whole issue of whether or not a bed was truly the capacity of a hospital vs. a staffed bed. That metric alone in the State of Utah determined our intervention. When hospitals were at capacity or 80 percent capacity, we implemented different restrictions. It was a really important data metric to have that we frankly didn't have and we had to get hand-entered daily from all these healthcare systems who are already under so much pressure.”
Dunn can point to several improvements they have made in Salt Lake County, but she knows that there are other health departments that don’t have similar tools. “I have one epidemiologist who runs our COVID dashboard, and it takes her 10 to 15 minutes every week to update it. I have a colleague in another county outside of Utah, who has five to six FTEs who are cleaning data on 800 COVID cases every single week,” she said. “They can't even build a dashboard. The difference between the two is just so striking, and it adds to the complexity for my CDC colleagues to try to figure out how to make this work for all of us.”
Ultimately, she said, public health agencies need flexible systems built with public health in mind. “A lot of technology that has come out is very healthcare system-focused. We need to figure out how to translate that to public health. What we do is different. How we define our populations is very different. I don't worry about my catchment area or my set of patients who have diabetes as population health. I need to be able to define my population by geography, by race/ethnicity, by SOGI, by chronic disease condition or comorbidity. I need that flexibility to define a population in a way that suits whatever pandemic or crisis is going on in the moment.”
She said there is also a need to be able to translate data standards for public health. “We've talked about FHIR and TEFCA, but how do we get that into the hands of public health, when I guarantee you the vast majority of the local health department workforce has no idea what FHIR and TEFCA are. How do we help the public health workforce come up to speed?” Action at the federal level
What can health IT leaders in the federal government do to address some of these issues? Speaking at the CDC-ONC Industry Day on Feb. 27, Suzi Connor, the CDC Chief Information Officer, said that part of the CDC’s ongoing Public Health Data Modernization Initiative involves building a robust, response-ready cloud infrastructure to get data quickly to and from CDC. “We need to reuse common infrastructure. CDC has over 900 systems that support the public health environment. We need to de-silo those and ensure we have centralized systems that are robust to be able to scale for any response and eliminate disease-specific systems over enterprise capabilities. Ultimately, we need to start piloting more advanced innovative solutions in things like artificial intelligence, natural language processing and machine learning so that we can advance our forecasting and analytic capability.”
At the same event, Micky Tripathi, Ph.D., national coordinator for health IT, mentioned several ongoing efforts, including USCDI Plus for Public Health, whose goal is having better integration of public health and healthcare delivery systems, and the creation of a true nationwide public health data model. “With the CDC and other partners, we're working on a USCDI Plus for Public Health to drive toward that goal of a nationwide public health data model that we can all turn to, and it'll help us create more ‘system-ness’ out of our public health system,” he said. The work involves convening public health experts to decide what are the different additional data elements that are needed. “Then how do we create a pipeline of those that can ultimately feed back into standards,” he said.
The country needs to get the public health infrastructure able to do FHIR transactions, Tripathi stressed. “That will allow us to start to think about higher-level functions, which I know is what the CDC team as well as other teams are very interested in. It is not just about being able to get data. It's about being able to do things like subscriptions, like being able to do CDS Hooks or the ability to insert knowledge back into a provider system based on processing of data or analysis of data or the CDC being able to say ‘I'm going to provide information back to the provider now so that they can actually act, with the latest and greatest information from a public health perspective in their day-to day-interactions with patients.’”
Tripathi mentioned that HHS recently had the public announcement of the first approved QHINs under TEFCA, which all committed to going live before the end of this calendar year. “We're working really hard with stakeholders with an eye towards saying let's establish the public health standard operating procedure or use case and then work with early adopters to get those early adopters live on at least one, if not two, public health use cases in 2024.”
Among ONC’s efforts are working with the CDC and HL7 on the Helios project, an HL7 FHIR accelerator focused on public health.
One of the use cases that the Helios project is working on involves immunizations using bulk FHIR, says Viet Nguyen, M.D., chief standards implementation officer at HL7. “There is a great opportunity for us to capture the immunization administration. Imagine that you have a child coming in at six months who should have a certain number of immunizations. We could make sure that that you identify the missing ones, administer them, then send the records to the state registries via FHIR,” he says. There's also an opportunity to improve the quality of information via FHIR, so that school systems can get that data from the state.
Another use case, Nguyen adds, is around surveillance. There are reportable conditions that local public health departments report to the CDC and/or to state public health, and they often involve lab reports. “You could envision that there could be automation of reporting an abnormal lab result using FHIR capabilities,” he says.
Still another opportunity for burden reduction might involve automating case reporting activities. Currently, a lot of that work is very manual. “If we were able to structure those reporting items, we could represent that in FHIR in a Structured Data Capture FHIR questionnaire,” Nguyen explains. “This is a powerful technique for us to be able to collect data via FHIR. If you know what data you want to collect, you structure that data based on FHIR. The data is codified via LOINC codes and other codes. And then you use another FHIR technology called CQL, or Clinical Quality Language, that could be either enabled by the EHR or by a secondary system to go out and actually retrieve that data. Each of those things could be very beneficial, especially for the case manager.”
Nguyen says he encourages regulators to support pilot projects to prove that a particular implementation guide works in a well-controlled area. “You learn the benefits of that, and then scale up.”
A role for health information exchanges
Mimi Hall, M.P.H., has seen the challenges around public health data infrastructure from several perspectives. She recently moved from a career in local public health departments, including Santa Cruz, Yolo, Plumas, and Sierra counties, to leading public health innovation with California’s largest health data network, nonprofit Manifest MedEx.
When she started working as public health director for Plumas County in 1999, they were still submitting progress reports and grant reports by paper to the state with carbon copies. “As data modernization happened in the healthcare world, it did not keep up in public health,” she says. “Even before the pandemic, there was this gradual two-decade-long disinvestment in public health. Somehow the importance of public health fell off of the radar of our decision makers, at the local, state and federal levels.”
Hall says that much of health IT investment has been EHR-centric and hospital-centric. She sees health information exchanges as a big equalizer for public health. If you are an FQHC, a small Medicaid provider in the Inland Empire, you probably do not have the resources to connect directly to your public health department to get data promptly, she explains. But the Admit, Discharge and Transfer (ADT) notices made available by HIEs can help. “If I were a case manager for the homeless and I got an ADT notice that one of my clients was in the ED and then I followed up and found out they tested positive for Hepatitis A, and that test result went immediately to the Public Health Department, we could get two or three weeks ahead of an outbreak, which is a lot of time,” she says.
“When you think about solving a huge public health problem, whether it's the pandemic or syphilis, we don't have a universal system that serves everyone,” Hall says. “If you take the limited resources that you have, and you prioritize them to get the right data at the right time to intervene with the highest-risk populations, that is what I call precision public health,” Hall adds. “We don't have to boil the ocean. We don't have to solve the whole problem. At a macro level or a statewide population level, you need to understand who is affected, who's most at risk for being affected, have the data on those people and target your intervention and your treatment strategies toward those populations. Right now, that isn't happening because there is no digital equity,” she says. “The investment in this basic foundational infrastructure to share data between healthcare and public health and other systems should be a public utility. It is no different than government taking a role in investing in infrastructure like roads or the electric grid, but we're not there yet in our nation.”
Need for continued investment
Indeed, Dixon says that HIEs in Indiana and elsewhere stepped up to help public health in the pandemic by aggregating data from the healthcare system and delivering that to public health, and adding value to the data. “For example, in Indiana for each case of COVID that the state health department or any other jurisdiction found through laboratory testing, they would push that data electronically to the HIE. The HIE could see if that person was hospitalized and provide information back on their past medical history.” Not all states have a sophisticated HIE infrastructure, however, “so our strategy really needs to focus on not just recommending these health data utilities participate in public health, but also making sure that they're available consistently across the U.S. and not just in a few places.”
The CDC has a data modernization initiative under way, but it is facing a funding cliff, Dixon adds. “The fear is that there is not going to be new money coming, because people are kind of over the pandemic and are going to go back to their prior feelings about public health, which is that it's in the background, and not very important,” he says. “But public health organizations have outdated computers and very few of them use the cloud. Few public health departments have an informatics division or people with informatics in their title. They're often dependent on the general IT division of the state or municipality. There are a lot of things that public health is just starting to make some headway on, and additional investment is needed. Otherwise, we might get them from Windows 95 to Windows 11, and then there'll be stuck at that for the next 10 years.”
