Apr 05 2023
Data Analytics

What Is the Role of Data Governance in Healthcare?

Data is omnipresent in healthcare organizations. Governance ensures that the right people can use data at the right time for the right reasons — all within the business systems they use every day.

The World Economic Forum has estimated that the average hospital produces at least 50 petabytes of data each year. It’s widely accepted that roughly 80 percent of healthcare data is unstructured, which means that it must be normalized and standardized in some way to be machine-readable.

These factors make it critical for healthcare organizations to look holistically at how they manage data through a process known as data governance. Getting data governance right ensures that end users across the organization, regardless of their role, can readily access the data they have permission to use, in a format that can be read by the applications they use, so they can make better-informed business decisions.

Implementing data governance — and pairing it with a modern data platform that makes data more readily available to those who benefit from it — can give organizations a solid foundation for gaining insight from their data and help them better prepare for healthcare’s uncertain future.

MDP Sidebar


What Is Data Governance?

To best understand data governance, it helps to bring together definitions from various entities.

  • The American Health Information Management Association focuses on “clearly defined procedures and plans that [assure] the availability, integrity, security, and usability of the structured and unstructured data available to an organization.”
  • Management consulting firm Gartner notes the importance of “[ensuring] the appropriate behavior in the valuation, creation, consumption and control of data and analytics.”
  • The Office of the National Coordinator for Health IT points to the value of data governance in supporting decision-making, “[ensuring] that the organization successfully realizes its desired outcomes and receives business value from data management activities.”

Setting policies for how data is defined, used and consumed may seem like a tall order, but it’s an important process.

“We think of data governance as a relatable framework for setting policies, accessing data and cataloging it, all with appropriate privacy constraints in place,” says Priya Krishnan, head of product management, data and artificial intelligence at IBM. “It’s like brakes on a bicycle: It’s designed to help you go faster, not just to slow you down. It’s a safety net to help you derive value from data.”

Governance also lends itself to transparency, says Kenley Money, chair of the board of directors at the National Association of Health Data Organizations and the director of information architecture at the Arkansas Center for Health Improvement. This benefits not just those who own the data but also anyone who might use it downstream.

“For our stakeholders, we have to be able to trace forward and backward how we use the data,” Money says. “If our analytics result is X, we need to know that the data was right and that we can build it again. We need to be able to tell consumers of the information where we got it and how we got it too.”

Click the banner below to learn how a modern data platform supports decision making.

How Does Data Governance Impact Healthcare Organizations?

Data is omnipresent in healthcare organizations. The more accessible and reliable it is, the more likely you are to develop insights from it, says Jonathan Shannon, associate vice president of healthcare strategy at LexisNexis. On the other hand, he adds, “your entire business suffers in various ways when data governance policies are poor.”

Krishnan describes three telltale signs of poor governance:

  • Duplication of data across clinical systems
  • Localized access and privacy policies
  • Additional costs, stemming from storing and maintaining data in multiple places and from making mistakes

“It’s one thing if a patient gets the same marketing materials twice,” she says. (That can happen if, say, there are records for Sam Smith and Sam S. Smith at the same mailing address.) “But I don’t want my diagnosis to be wrong because physicians don’t have access to all of my records.”

More broadly, poor data governance in healthcare can have significant business implications. Shannon points to the referral process. There are multiple benefits to referring patients to in-network providers: Patients avoid the high cost of seeing physicians not covered by their insurance plan, and organizations keep patient defections to a minimum.

If provider directories are inaccurate — and data from the Centers for Medicare and Medicaid Services indicates that 49 percent of them are — then it’s that much harder to make in-network referrals, he says. “Very important procedures may be sent up the road and out of the network because someone didn’t know.”

Finally, poor governance poses security and regulatory risks. The 21st Century CURES Act requires organizations to make data available to other healthcare stakeholders, including patients. This requires a delicate balance between security and availability, Shannon says.

“Now, there’s more pressure to make data more accessible,” he says, primarily with open application programming interfaces. “Without data governance, you can’t support open access with APIs.”

DISCOVER: How Community Medical Centers powers operations with data.

How Can Healthcare Organizations Improve Their Data Governance?

When it comes to addressing data governance, healthcare organizations tend to fall into one of three buckets, Krishnan says. Some are just getting started and need help putting a general framework in place. Others have a framework but also have many data silos; this is especially tricky when data is on-premises and in both public and private cloud environments.

Still others have made progress but worry about the implications of duplicate records within newly unified data sets, whether it’s difficulty with regulatory compliance or a lag time to get “business-ready data” to the teams that need it. “These organizations want help to scale for new applications,” Krishnan says. “They want to be enterprise-ready and future-proof.”

A common starting point is what Shannon refers to as a “one-time cleanup” of the organization’s data repository.

“By definition, data grows and changes over time. If your organization has been in business for decades, you’ve been accumulating data for decades,” he says.

When technology upgrades are on the horizon, it’s both expensive and counterproductive to move millions of unnecessary records — those that are duplicates, incomplete, from deceased patients, and so on. Through a combination of referential and probabilistic modeling methods, Shannon says a repository with data from 7 million patients could be trimmed to 1 million patient records. As a result, the repository is more accurate, less expensive to maintain, and well suited for use with next-generation applications for decision support, population health management, and predictive analytics and modeling.

For Money, addressing organizational culture is fundamental to improving data governance.

“You can’t buy data governance off the shelf,” she says. “It has to be understood from the highest level of the organization to the bottom. It should be invisible. No matter the organization, what you’re doing is producing quality, usable, effective products, and data governance is a tool to make that happen.”

Kenley Money
You can’t buy data governance off the shelf. It has to be understood from the highest level of the organization to the bottom.”

Kenley Money Chair of the Board of Directors, National Association of Health Data Organizations

What Technology Supports Healthcare Data Management and Governance?

According to Shannon, most organizations that want to modernize data governance begin with a combination of three types of technology: master data management tools that help aggregate and reconcile data; enterprise master patient indices that resolve duplicate identities; and electronic health record systems that serve as a source of truth. The next steps are to layer on an identity and access management system and format data in compliance with the Fast Healthcare Interoperability Resources standard. 

“This multifaceted approach addresses orchestration, availability and security,” he says.

Krishnan suggests that organizations keep in mind four key objectives as they modernize their technology with an eye toward data governance.

  • Scalability: Organizations need data repositories that can handle ever-expanding data sets and still produce a single version of the truth.
  • Integration: The data repository should integrate with traditional clinical systems as well as data science tools, and the same governance rules should apply regardless of where data is used.
  • Quality: Data should be continually monitored as it moves from one system to another so that relevant terminology can be applied, inconsistencies can be flagged and so on.
  • AI Governance: Organizations should apply the same governance principles to the artificial intelligence models they build or license to analyze their data.

“If you can solve these problems, just think of how much more quickly you can innovate,” Krishnan says. “It’s a competitive advantage.”

venimo/Getty Images

Become an Insider

Unlock white papers, personalized recommendations and other premium content for an in-depth look at evolving IT