Data governance: Make analytics 'an amazing asset'

Governance isn't always fun. But it is the most important building block for a healthcare analytics strategy. How the University of Mississippi Medical Center and Dartmouth-Hitchcock are turning the tedious work into an indispensable boon to productivity.
By John Andrews
06:25 AM
Share
Data governance analytics Dartmouth

Elizabeth Stedina, director of the Dartmouth Analytics Institute, said the system is breaking new ground and no single policy can cover all of its needs. 

There is no glory in governance. There are no sparkling data clusters to admire. There are no dazzling 3D images. There is nothing that makes people step back and say: "That is so cool!"

Instead, governance is decidedly unsexy. It is tedium. It is drudgery. It is painstaking diligence. It is conscientious review. It is checking, double checking and checking one more time for good measure. It is an exercise in patience that challenges even the most imperturbable people. It is a seemingly endless chore.

It is also the most important analytics strategy in healthcare.

Given the enormous amount of work involved with governance, it's no wonder that it hasn't progressed farther in the healthcare industry. But its proponents — most notably those who have succeeded with it — insist that the end result is worth the tremendous effort.

"We spent 18 months on governance before we produced any single analysis or report," said John Showalter, MD, chief health information officer at the University of Mississippi Medical Center. "You want unsexy? All that time was spent on procedures and protocols with no analytics. I get this amazed look from people when I say we worked 18 months before generating a report."

It has been 14 months since Showalter's team completed the intensive governance process and in that time he has become thoroughly convinced that it was the right thing to do.

"Now that it is all in place, my team is amazingly productive," he said. "We have produced 40 data visual apps and 1,200 reports with just five report writers. This is possible because they don't have to ask any questions – they know the rules of the road."

Within the fully realized governance structure, UMMC is generating predictive analytics and sending large amounts of data to the cloud, massaged and returned for clinical decision support, Showalter said. The center is also deploying geospatial analytics – the ability to look at environmental and income factors based on a patient's geographic location. For instance, UMMC has an app for pollen counts per block that correlates with asthmatic patients. 

"It is astonishing how much data there is on the block where you live," he said. "There are about 1,500 data points relevant to the concerns about each block even though there are about 100,000 data points out there."

Governance is key to breaking through the stubbornly bureaucratic culture of healthcare organizations, though Showalter concedes that the 18-month governance implementation phase was "an extremely painful" bureaucratic endeavor itself. 

[Related: Why hospitals need governance for information and not just data]

"But the bureaucracy can be turned to your favor and into a trail you can follow," he said. "We turned a detriment into an asset."

UMMC is by no means alone. Dartmouth-Hitchcock Medical Center in New Hampshire and others are also reaping the rewards of hard governance work.

Data as a strategic asset
A recent Harvard Business Review survey revealed that 70 percent of responding organizations across the U.S. economy believe big data is "crucial." Healthcare, specifically, expects that the analytics market will reach $43 billion by 2020. Given those financial stakes, healthcare organizations should be taking a serious look at governance as a means of control, said Tina Foster, vice president of business advisor services at Emeryville, Calif.-based RelayHealth.

Governance, she said, is the vessel that organizations can use to take the necessary steps in transforming data into actionable, enterprise-wide insights. Yet big data and the fastidious process of harnessing it has intimidated many organizations into a state of paralysis regarding its implementation, she said.

"People still haven't embraced the true value of governance," Foster said. "Big data is big, wide and nebulous and population health is also big, wide and nebulous."


Tina Foster of RelayHealth said organizations must understand the data value chain. 


"There is better awareness of governance, but it is still in an infantile state," she added. "People are exhausted from putting so much effort into the data that they haven't progressed beyond that point. Healthcare organizations need to recognize data as a strategic asset instead of a byproduct of the IT system."

Healthcare organizations are recognizing the importance of governance through formation of committees designed to work as think tanks for devising an analytics strategy. But they have been reticent to take the necessary next steps, Foster said. 

"The committees think they are addressing governance, but there are multiple challenges with that because it is not a true governance process – it's a reporting piece that ends up being an IT subcommittee and not an organizational strategy," she said. "They have evolved somewhat, but not a lot. It is triage, basically."

When advising healthcare organizations on governance, Foster said she refrains from using that term and instead refers to the "value chain" that data encompasses.

"Data is a huge part of the equation – it is a critical variable you can't solve the equation without," she said. "Organizations need to wrap their heads around the data value chain – how data is acquired, how it is aggregated and put into a version of truth."

Crux of the tedium
If UMMC has set the template for governance, Showalter isn't sugarcoating it when he describes the 18-month slog that he and his team endured to reach that good place. Neither is anybody else who knows what this process involves.

"The end result sounds really good, but in the meantime it is a lot of work," said David Delaney, MD, chief medical officer for SAP America. "You don't just buy software and a miracle happens. There are a lot of data sources and a lot of cleanings that need to be made. Elements need to be firmly defined. It is a whole lot of blood, sweat and tears and not a lot of (financial) gain from it."

Staying focused on due diligence takes an extraordinary amount of discipline, as Mark Heinemeyer, chief collaboration officer for San Francisco-based Wanda, describes a typical investigative sequence: How to get the data, how many levels of encryption does it have, how is it being contained, who has access to it and why, what documents are included and who might see them, what should be done with it, who owns it and who is impacted.

"You are managing a process of ensuring the integrity of data," he said. "It takes extraordinary discipline and you don't get any glory from any discoveries made."

Does the workload involve logistics? Not as much logistics as planning and design focused on preparing data to transform into meaningful information, said Eric Rock, CEO for Plano, Texas-based Vivify Health.

"That data might be codified … if it's raw textural data, it's impossible to turn into something meaningful," he said. "Be sure you have a platform that understands where data can flow to in the future. The messaging structure for interoperability can change as long as it is properly codified. It is about good design, good practices and the understanding of healthcare through interoperability."

'One source of truth'
Dartmouth-Hitchcock is deeply immersed in data and information governance and as daunting a task as it is, not doing anything about the torrential flood of unharnessed data is worse, said Elizabeth Stedina, director of the Dartmouth Analytics Institute and director of analytics at Dartmouth-Hitchcock.

"On the data governance side, practices are evolving and we're sending a lot more data to organizations and working more closely with researchers, so we're breaking new ground and it feels like no one policy can cover it all, so we're going on a case-by-case basis," she said. "On the information governance side, there is so much data, we need to channel it in a way people don't feel bottlenecked by the centralization of the data. We need to make it accessible and balancing access with the governance of data that is consistent and accurate is the struggle."

A dedicated coalition of stakeholders is essential to provide the oversight, cooperation and commitment to ensuring governance program success. The Dartmouth-Hitchcock team consists of representatives from general counsel, data warehouse, privacy and security, informatics clinicians, research, IT and the analysis and reporting side. About 12 or 13 people compose the data governance committee and the information governance group, known as Measurement Central, has been revived after a period of dormancy while a new IT platform was installed.

"We've gone beyond just creating committees and are building into the organizational and IT platform structure," Stedina said. "Last year we brought in pockets of analysis and data reporting and data management resources together under one umbrella – the Analytics Institute. The rationale was we were working in silos with the same information and different results, so there was mistrust of the data. A centralized function would alleviate that mistrust. We also try to reduce the number of places where people can get data and create a robust self-service environment with one source of truth. It's an uphill battle but we've laid the groundwork."

To date, Stedina believes that "we've made great progress and are making governance a part of our culture. And we are continuing to improve." Launching the initiative did face some resistance at first, she said, but gradually people have started pulling together in the right direction.

"It can be a hard sell to certain groups and we try not to let people bypass the process," she said. "Ultimately our goal is to have a successful centralized place to access data that has been scrubbed, with numerators and denominators appropriately defined and that people who use the information will trust it for clinical and business decisions."

Uncovering unsexy
Governance is an arduous process that requires a lot of heavy lifting in the form of diligence by the healthcare organization's brain trust and workforce. It is deeply introspective, exhaustive and laborious. It is decidedly unglamorous.

"When you're dealing with a topic like this, it is not sexy or glamorous – it is consequences and pain," said Mary Beth Haugen, president and CEO of Denver-based Haugen Consulting Group. "It has been very difficult to get organizations to embrace it."

Making governance sound glamorous might be akin to putting lipstick on a pig, but Haugen believes that better messaging could make a difference in winning over acceptance. Emphasizing the importance of data accuracy across the enterprise and how many systems are prone to distortions could be effective in conveying the reality and urgency of governance, she said.


UMMC's John Showalter, MD, and his analytics team knew when the governance work was done when they drafted rules of the road and everyone signed off on them.  


"Sometimes the data is correct but the information is wrong," she said. "When I'm looking at data fields registering a patient, that data is right; but if I take your medical record number and overlay it with my information, it's all bad if it's wrong. Once a lab value is attached to a record, it becomes the information. So if it is attached to the wrong patient, there is bad information. So we have to govern data fields correctly and also how we use that data from an information standpoint to make sure that is correct as well."

When it comes to governance, Haugen is convinced that it is actually happening in many healthcare organizations, just not on an overarching basis.

"It is being done within organizations, just not at an enterprise level – it is more within departments," she said. "With so many silos still in place, clean data is not system-wide. We are working on tools that enable clean data at intake and within certain departments but if bad data gets through and goes someplace else, it perpetuates the corrupted information."

Establishing a data governance department might help, but any vehicle that facilitates communication between parochial interests can serve the same purpose, she said.

"What has to happen is getting experts in each silo together in the same room – IT talks with clinical, with health information management, with quality personnel about how the data is being used, how it should be collected accurately and properly defined," Haugen said. "When you get the people who are passionate about data around a table, they will find a solution to ensure the information is correct. We have to present that data through portals to patients correctly and in a way that makes sense to them. When you lose trust, it is very hard to gain it back."

And truth be told, "We're not really good at cleaning it up everywhere," Haugen concedes. "With lab data going to multiple systems – how do we ensure that we cleaned up every component? The issue is the lifecycle of the data and getting everyone on the same page. There are multidisciplinary teams for clinical, but not for IT – their experts should get together as well."

Delaney has his own analogy of why governance is unsexy: "At times it seems like the analysis of governance is myopically on visualization, as if it were a beauty contest. But that is like grading a restaurant based on presentation and not the food quality."

The unsexiness also extends to what Delaney asserts is the hardest part of the governance process – "taking ugly definitions of data and mapping them to business-friendly terms. Giving the business user a nice view of the data is critical, but it doesn't obviate the 'garbage-in, garbage-out' scenario."

As the healthcare business model shifts to being paid by value, not volume, hospitals need to alter their thinking about the circuitous route data travels through the enterprise and taking steps to streamline it.

"Until you are able to understand how patient care is delivered at the individual level across the clinical, financial and claims silos, you can't calculate the cost," he said. "By shifting to a data-driven enterprise, you create a virtuous cycle of learning. That enables you to manage things because you can measure them." 

'Implied obligation' to patients
As they typically are able to do with major IT initiatives, large health systems have been able to mobilize for starting governance projects, said Kathy Downing, director of health information management practice excellence at AHIMA, but also what typically happens is that they reach a point where momentum stalls.

"There is still a long way to go across the industry with understanding, acceptance and adoption of information governance," she said. "Organizations know it is needed, but they are under pressure from different initiatives and don't always realize that information governance can assist across the organization in many ways. Record retention and legacy systems are often areas where initiatives can result in cost savings for the organization."

From a business standpoint, the model of protecting data throughout the lifecycle of a patient's record of care is "an implied obligation to that patient," added Adrianna Iorillo, vice president of professional services for Jacksonville, Florida-based CSI Healthcare IT. "The rules in which the organization adheres in maintaining those records not only protect the patient from unwanted data sharing, but can provide meaningful information on how the organization can better serve the needs of their communities."

Ultimately, information governance initiatives "are going to increase patient safety and quality of care through the ability to share trusted, reliable, accurate information with patients, clinicians, and securely to the healthcare community increasing the health of populations," Downing said. "Adopting an information governance program shows an organization's commitment to managing its information as a valued strategic asset."

Don't 'boil the ocean'
Data and information governance should be done piecemeal and that is another reason why it is so unsexy, Rock said. Visualizing the massive task at hand and playing an instrumental role in conquering it has some prestige attached to it. But breaking it down into segments draws out the process, making it hard to see the end result. Which is why many organizations are finding governance such a challenge, he said.

"Most healthcare organizations are trying to boil the ocean, aggregating all data sources, trying to cleanse the data along the way," Rock said. "All this to have the value drop due to the time lag in getting the heavy lifting done before a business team member can gain value from it."

That observation brings up a valid point – how does the governance team know when the process is complete?

In Showalter's experience, he knew the governance work was finished when "we got everyone together, drafted the rules of the road and when everyone signed off on it, we knew we were done."

Putting egos aside, empathizing with others' opinions and keeping the patients' best interests in mind served as the catalyst for the successful campaign, he said.

"As an industry, we have been so focused on EHRs, we haven't spent the time on delivering the promise they bring," Showalter said. "I'd like to say I'm a visionary and saw it, but I knew if I didn't get the tech team, compliance, privacy and research together, we would get no work done. If we are in agreement that we can do high-quality rapid analytics that are good for our patients, we can address any concerns that arise and continue forging ahead."


  Governance will be among the topics at the HIMSS and Healthcare IT News Big Data & Analytics Forum in Boston, Oct. 24-25. What to expect:

⇒ Charlotte hospitals analyze social determinants of health to cut ER visits
⇒ Big Data: Healthcare must move beyond the hype
⇒ Tips for reading Big Data results correctly
⇒ Small hospital makes minor investment in analytics and reaps big rewards 
 MIT professor's quick primer on two types of machine learning for healthcare
⇒ Must-haves for machine learning to thrive in healthcare


Like Healthcare IT News on Facebook and LinkedIn

Topics: 
Analytics