Digitizing patient health data was supposed to make it usable for public and population health research. So far, that hasn’t happened.
Between 2011 and 2015 the government spent upwards of $35 billion subsidizing doctors’ purchases of electronic health records (EHRs) through the Meaningful Use program. It did so mainly in the expectation that EHRs would enable doctors to provide better care for patients and streamline practice and hospital workflows.
But the 2009 Health Information Technology for Clinical Health (HITECH) Act — the law that spawned Meaningful Use — included an additional motive for encouraging EHR use: advancing public and population health. The hope was that digitizing millions of patient health records would produce a treasure-trove of data that researchers could mine for purposes such as spotting disease outbreaks and developing new ways of treating chronic conditions.
Seven years after Meaningful Use, and three years into the COVID-19 public health crisis that has claimed more than a million lives, it’s evident that the hoped-for public/population health benefits of EHRs remain unfulfilled. That raises the questions, why haven’t those hopes been realized? Will they ever be? And were the public/population uses of EHRs oversold?
“We invested $35 billion and just went through a pandemic where we didn’t have enough information to answer the public’s questions about what we should do,” says Shannon West, chief innovation officer at the health information technology company Datavant and former chief technology officer at the Center for Medicare and Medicaid Innovation (CMMI). “And as taxpayers I think we should all be asking, ‘why isn’t this working?’”
And while government and the private sector are taking steps to address the challenges of making EHR data more easily accessible, most observers believe it will be years before those efforts come to fruition.
EHR use now widespread
The challenges of using EHR data for public and population health purposes certainly do not stem from a lack of EHR use. According to government data, by 2019 90% of office-based physicians and more than 95% of hospitals were using EHRs.
Nor is the ability to send and receive data electronically — interoperability — the barrier it once was. And it is likely to become even less so thanks to the growing adoption of technologies such as Fast Healthcare Interoperability Resources, a standard used to access and exchange health care data, TEFCA — a “trusted exchange framework” for exchanging information between health information networks — and applied program interfaces, the architecture that enables development of apps for sending and receiving data.
Why getting EHR data is difficult
The problems lie elsewhere, experts say, largely in incentives and policy choices built into the nation’s health care system. They begin with the challenge of prying patient health data from the institutions that own most of it — the large hospital systems and the practices they own, where most patients now receive care. These institutions are usually reluctant to share their EHR data due to fear it will put them at a disadvantage with competitors and/or payers, without being compensated for it.
“In health care nothing happens unless there’s money behind it,” West says. “At the end of the day if there’s not financial incentive to share medical records among providers or with public health departments the data is going to stay stagnant. It sounds so obvious but that’s the incentive you need for the behavior you want.”
That reluctance is understandable given the highly competitive nature of health care in many communities, says David Blumenthal, M.D., president of The Commonwealth Fund and former National Coordinator for Health Information Technology (ONC). “We don’t require or expect Toyota and BMW to exchange information. But in effect that’s what you’re doing when you ask two hospitals in the same community to exchange health data.”
No business case for data sharing
Blumenthal adds that the lack of incentives for sharing data is part of a bigger problem, which is the priority the health care system gives to treating illness in individual patients versus preventing it among large populations.
“There is no business case for prevention in the U.S. health care system, which is primarily built on fee-for-service," he says. "But there is a very strong case for taking care of sick people. So you can’t expect technologies like EHRs to become solutions to health care problems when they are controlled by humans who will use then for business purposes. Lacking a business case for use of a technology it won’t be used to address that problem.”
The absence of a business case for sharing data was one of the motivations for adopting the 21st Century CURES Act and the Information Blocking Rule resulting from it, says Micky Tripathi, Ph.D., M.P.P., the current head of ONC.
“We recognize the business drivers of the health care system will be to not share information, Tripathi says. “But I would argue that health care has higher needs. That’s why with the Information Blocking Rule we’re telling them [doctors and health care organizations] they are compelled to share that information, or they will be penalized.”
Health care vs. public health
Another barrier to using data EHR data for public and population health research is structural: the separation of health care for individuals--what most people think of as the health care system — from public health functions.
“It’s one of the historical challenges faced by health care providers and policymakers that we’ve kept public health and medical care separate in this country,” says Joshua Vest, Ph.D., M.P.H., director of the Center for Health Policy at Indiana University-Purdue University Indianapolis and a researcher at the Regenstrief Institute. “They are funded and governed separately. And so while integrating public health and health care would be beneficial in many ways there are also a lot of barriers to achieving it.”
Among the consequences of that separation is that public health functions are performed almost entirely at the state and local level. That fragmentation further complicates the process of finding and aggregating the data required for public and population health research.
“There are something like 64 public health jurisdictions around the country, each with its own rules around data collection and sharing,” Tripathi says. “Under those circumstances it’s really hard for the private sector, which controls most EHR data, to figure out all the regulatory issues involved with public health.”
Public health chronically underfunded
Moreover, public health and the health care system face a vast imbalance in financial resources available. According to the American Journal of Public Health, public health accounted for less than 2% of all U.S. health care spending in 2020. Similarly, a 2022 study by the Health Information Management and Systems Society concluded that the nation’s public health infrastructure needs a $36.7 billion investment over the next decade to address current and emerging threats to public health.
“Public health is chronically underfunded,” Vest says. So even if public health departments wanted to partner with health care organizations, they’re generally smaller and less technologically advanced.”
The consequences of that imbalance became painfully clear when COVID-19 struck. Ideally, says West, when a doctor treated someone with the disease that information would have been automatically transmitted to the local and state health department so the outbreak’s progress could be tracked and other doctors would have access to it. “That’s 100% not what happened, and still not happening,” she notes.
“We’d invested all this money in EHRs and created the ability to firehose a tremendous amount of health data,” Tripathi recalls. “But we didn’t make the corresponding investments on the public health side. And we found during the pandemic that when we turned those EHR firehoses on public health agencies they had no ability to receive that volume of data, let alone process it for the kind of response we needed.”
The role of HIEs
The lack of public health funding, and ability to aggregate data for research purposes has been filled somewhat by state health information exchanges (HIEs), networks that enable their members to search for and exchange patient data electronically. While some policymakers view HIEs as a possible way for researchers to access data for public and population health research, West is skeptical.
“One of the problems with HIEs is that some health systems don’t share all their data on individual patients or from every patient visit,” she says. “So you can build the connections (with HIEs) but the connections don’t necessarily all provide the same data. That’s where the idea starts to break down on a national level.”
Blumenthal agrees. “HIEs depend on people and institutions putting information into them to be effective,” he says. “That means their effectiveness depends on the community spirit of local health care organizations, along with the state regulatory framework that facilitates it.”
Grounds for optimism
Despite the obstacles, some observers are optimistic that EHR data will be widely available for research purposes in the not-too-distant future. One is former CMS administrator Seema Verma, who notes that with EHRs now in widespread use and rapid development of standards and technology for data transmission, “it’s just a matter of figuring out how to aggregate and deidentify the data, and making it easy for providers by having their EHR software system send it automatically. I don’t think any of that requires a massive investment.”
Tripathi says that the federal government’s involvement in developing data sharing protocols such as TEFCA will make aggregation and research easier. “It’s only with the federal government stepping in to say, ‘Hey public health agencies, let’s agree on some basic rules of the road and then you can exchange information with each other,’ ” he says.
Blumenthal points out that mining of EHR data is already happening, in the form of partnerships between large health systems and tech companies with the ability to store and analyze large quantities of data. “Every patient encounter in that system end up in the company’s data warehouse where it’s deidentified and the company tries to make money from the resulting information.”
Partnering with the private sector provides the business care for mining EHR data, Blumenthal says. The drawback is the companies doing the research, such as Google and Apple, “are poring over those data looking for what they can make money from, not for ways to help the U.S. improve its health. They might hope that will be a side effect, but it’s not an altruistic activity.”
Regenstrief’s Vest finds grounds for optimism in the development of what he calls the “building blocks” of a comprehensive research network. He points to the many states with all-claims databases that are accessible for research purposes, and the National Patient-Centered Clinical Research Network which, according to its website, providesaccess to data from more than 30 million patient encounters annually for a variety of research purposes, including population health.
The COVID-19 pandemic might also play a role in furthering the EHR-population health connection, Vest says, by driving development of large-scale patient registries among multiple academic institutions for research into the disease. These examples, Vest says, “demonstrate we have the knowledge and experience to do this sort of (public and population health) research using data drawn from multiple sources. Hopefully one day we’ll be in a position to scale them up and build something larger.”
Was the promise of EHR data sharing oversold?
The Health Information Technology for Clinical Health (HITECH) Act of 2009 included about $35 billion for helping doctors and hospitals purchase electronic health records. One of the arguments for including those funds was that digitizing the health data of millions of Americans would enable the data to be used for public and population health-related research.
Fourteen years later, with that goal still far from being achieved, it’s fair to ask whether it was realistic to start with. After all, accessing the vast quantities of data potentially available in EHRs would require overcoming numerous challenges, from staying complaint with HIPAA privacy rules to the technical obstacles of transferring the data.
“Many people don’t realize that working with health data is hard,” says Joshua Vest, Ph.D., M.P.H., director of the Center for Health Policy at Indiana University-Purdue University Indianapolis and a researcher at the Regenstrief Institute. “The data is not always truly representative of the population because it’s a reflection of who can access care and who can’t, who documents the care, and what gets paid for. Collecting it electronically opens the potential for aggregation but doesn’t make it immediately possible or remove all those other difficulties.”
Micky Tripathi, Ph.D., M.P.P., the National Coordinator for Health Information Technology, says the creators of the HITECH Act, may not have anticipated the privacy issues associated with using patient data for research.
“If you’re going to do research using identified data you need patient consent. And consent is complex from a policy and technical perspective and hard to scale,” Tripathi says. And while data can be de-identified, he adds, doing so creates a new set of challenges.
“HIPAA [the Health Information Portability and Accountability Act] requires you take out 18 fields of data,” Tripathi explains. “And once you’ve removed those fields you lose the richness and specificity you need to answer questions like, was a person who just tested positive for COVID vaccinated? If so, which vaccine did they get? Once you’ve de-identified data you’ve lost visibility into those kinds of details.”
Tripathi says creating a digital warehouse for research purposes that contains everyone’s health data was, and remains, unrealistic due to privacy concerns and the sheer quantity of data involved. A more effective approach, he adds, is to continue developing an “open architecture research ecosystem” that will enable researchers to look for and aggregate the information they need from existing data repositories.
Nevertheless, some policymakers think the government delayed the possibility of using EHR data for public health research by not including interoperability among the criteria for qualifying for the Meaningful Use program from the start. Seema Verma, administrator of the Centers for Medicare and Medicaid Services from 2017 to 2021, calls the omission a “lost opportunity.”
“When those (Meaningful Use) regulations were developed they weren’t written around interoperability,” she says. “And once those (EHR) systems were built it was a lot harder to make them interoperable.”
But David Blumenthal, MD, MPP, the National Coordinator for HIT from 2009 to 2011 says there were good reasons for not requiring interoperability in the first phase of Meaningful Use.
“The idea was you had to operate before you could interoperate,” he says. “You had to get people up and running on the systems and get the records in place. Then you could require that they start exchanging it. But what we didn’t anticipate was the extent to which health care institutions would resist information exchange requirements because of the business implications.
“I think that, as is often the case with public health policy, we underestimated the difficulty of achieving the goals we were aiming for,” Blumenthal adds. “But I’m confident that the promise will be realized, if not as quickly as we’d hoped.”