How NIH Uses Health Informatics to Bridge Care and Research

What is the role of health informatics in assisting the integration of clinical research and clinical care?

Health informatics is critical to bringing those worlds closer together and one of the linchpins in doing so. The ultimate vision is for health informatics to help us address researchers’ data and infrastructure needs to enable research that leads us to discovery more quickly, more efficiently. But also, the goal is to facilitate integration of any knowledge that comes from that research or from that discovery into clinical care effectively. The ultimate goal is to make quicker gains in health improvement. Whether that’s helping with diagnosis or treatment, new care delivery models, or understanding how we deliver care differently to be more effective. Informatics has a clear goal in helping us to achieve that vision.

You were part of the group at ONC that created the National Health IT Priorities for Research. In your research, what were the hurdles that you came across preventing wider data interoperability?

Advancing standards and sharing data in a standardized way will go a long way towards addressing interoperability issues, particularly interoperability of data. The opportunities that remain are ensuring that the data can be easily integrated and used across systems.

For example, if you and I are using the Fast Healthcare Interoperability Resources® (FHIR®) specification to exchange the information, and we even agree on some of the underlying data standards so that we can code the data consistently, once I send you a clinical record, where does it land in your EHR? How do you integrate that as a clinician or as a patient? How do you integrate information from across multiple provider EHR systems for use to facilitate consistent interpretation and use of the data?

Because at the end of the day, you want to be able to gain insight and knowledge. Making the data available is one step, but ensuring they can be integrated and used is critical.

"The ultimate vision is for health informatics to help us address researchers’ data and infrastructure needs to enable research that leads us to discovery more quickly, more efficiently."

One of the priorities centered on integration at the point of care. What is the process of making the information useful and contextualized?

There are many colleagues in the informatics community who have been working on clinical decision support for many, many years. There are observations that some of the ways clinical decision support has been designed and implemented could be improved upon so that it is more useful within the context of clinical care. Being able to share the data through standardized application programming interfaces will create opportunities for other developers to think about how the information is presented to clinicians, and to develop new tools to leverage those data. Those tools could sit on top of or alongside existing electronic health record systems. I think we’re moving in the direction of making that a reality.

But doing so will require a very good understanding of what clinicians are trying to do when they treat their patients. In my previous role at AHRQ, we had one funding opportunity that sought to ascertain clinical information needs to help us better inform the design of some of those solutions. I expect that we’ll need some of that research to continue to inform some of these tools moving forward.

What were the considerations for security on the different levels and different types of usage?

What we outlined was the need to ensure privacy and security of the data. We outlined some issues around consent and data sharing in the agenda. Specific to that, there are a couple of different issues to parse out. One is developing and implementing clear privacy policies and practices, and being transparent about those.

Transparency is particularly important, not just with clinicians, but with patients and their caregivers, so they can understand how data are being shared and used. And then implement security practices consistent with those policies.

We talked about the need to innovate around how data are shared, and how we parse which data are shared for what purpose, so that it’s less of a burden on a clinician to make sure that their patient’s data isn’t going where it shouldn’t, but the system has features and functions that allows data to be shared. For example, in the case of COVID-19, to facilitate sharing data in whatever deidentified way is appropriate for public health reporting and tracking, we need additional innovation in that regard to make that a reality.

What needs to happen to ensure that good quality and compliant data is captured and useful for all relevant parties?

We need to move away from thinking about what needs to change at the point of capture and think about how we enable capture of data that are needed, in the level of quality that we hope for through other means. And we need to do that in ways that do not require additional work from patients and caregivers and the clinical team. Unfortunately, sometimes what ends up happening is that the burden falls to patients or their caregivers to capture or report on some of those data. For example, ONC has been exploring the use of automation. There might be opportunities to add instrumentation or leverage data that comes from devices or even from other systems – Human Services, for example. Census tract data can be used to predict certain outcomes.

Identifying what other data sources might be available that we can share in a more automated way might be one way to go. The other is finding ways to address those quality issues that don’t impose an additional burden on clinicians as they’re capturing data. Whether that’s a redesign of the interface, or implementing some quality checks in the backend, and figuring out how to impute or analyze data.

While presenting on a panel last year, a presenter on the same panel argued that data not of quality for research are arguably not of quality for clinical care, either. So, sometimes we shift the conversation towards thinking about quality issues, and how that impacts research. But the reality is that clinicians and care providers need high-quality data too. If we think about addressing some of these needs together, that would also go a long way in both advancing data needs and quality of those data, and reducing burden on clinicians and patients.

"We need to move away from thinking about what needs to change at the point of capture and think about how we enable capture of data that are needed, in the level of quality that we hope for through other means. And we need to do that in ways that do not require additional work from patients and caregivers and the clinical team."

You have spoken about FHIR many times over the years. How have you seen the broader uptake, understanding and usage of that evolve?

The development and adoption of FHIR has actually moved fairly quickly in healthcare compared to the development and adoption of other health data standards. There’s a lot of support and enthusiasm behind that, not just from the health IT developer community and the healthcare provider community. You have insurers, researchers, and the pharmaceutical industry – NIH is encouraging the use of FHIR and issued a notice doing so in 2019. All of that has helped with interest, but also adoption.

There is interest in making sure the specification works as well as possible for as many — as there are different needs from different healthcare stakeholders. Its use is now required under HHS regulation. ONC and the Centers for Medicare & Medicaid Services (CMS) published regulations last year that will require the use of FHIR APIs. ONC, in particular, advanced the use of FHIR APIs because they recognized that FHIR was well- received and adopted within the healthcare community. Use of a standardized API will go a long way towards advancing exchange of health data, not just for clinical care, but for other purposes as appropriate.

The other thing that ONC finalized in its Cures Act Final Rule is the United States Core Data for Interoperability. This is a set of data classes and elements that will be standardized with a set of specified vocabularies. It’s something that will further enable interoperability; we will use FHIR to exchange data in a standardized way, but the data underneath will also be standardized according to specific vocabularies. That’ll help not just for clinical care, but for researchers to be able to bank on those data elements to be in standard form within an EHR.

"The ultimate vision is for health informatics to help us address researchers’ data and infrastructure needs to enable research that leads us to discovery more quickly, more efficiently."

Are there collaborations or partnerships that you think would be good for creating infrastructure?

A more explicit collaboration between healthcare, health IT developers, the informatics community, and the research community to ascertain common needs, and to see if we can carve a path forward across those needs. Even within healthcare, there are different use cases that might have some common needs. Providing care, issues in population health, public health, research, and post-marketing surveillance, and even regulatory decision-making might have some common needs that we can come together as a community to identify and address.

One of the things we mentioned in the ONC National Health IT Priorities for Research agenda was also the need to do more to engage the patient and participant community, and the need to be more inclusive. If we want healthcare to work for everybody, then we need research that is inclusive of everybody.

That will require engaging diverse populations in research and engaging the institutions that serve them as well. We need to either fund partnerships with better- resourced organizations, provide resources or tools that they might be able to use and leverage, and understand what their needs might be. That way the threshold for participating in research for them is not out of reach.

To learn more about Clinical Research as a Care Option, visit CRAACOevent.com.