OR WAIT null SECS
Northwell Health explains how it implemented self-service analytics.
It’s too late to evade chaos when it comes to data and analytics for healthcare. The madness is here, and it’s been here for some time. But there are ways to launch an effective self-service analytics program, according to the executive who did exactly that for Long Island-based Northwell Health.
During the Health IT & Analytics Summit in Baltimore, Maryland, this week, Christopher Hutchins, associate vice president of healthcare analytics for the not-for-profit health system, described how he and his colleagues cut through the confusion and opened an enterprise platform to support self-service analytics, deepening their capabilities and accelerating data-driven learning.
“I felt like I was running behind from right out of the gate,” Hutchins said. “There’s just chaos—you kind of have to acknowledge that it’s there and figure out what you’re going to do to make it better.”
What does he mean by chaos? For one, Northwell is on an acquisition spree, adding something like one new healthcare facility to its ranks every week. Legitimate business priorities also exist, and they’re in competition with enterprise projects. Then there are the limitations of centralized enterprise teams and the availability of subject matter expertise.
Further, healthcare must battle every other industry for skilled data scientists, an expensive labor force. And the more data aggregation that goes on, the more stakeholders want analytics, driving demand in the face of limited bandwidth. All the while, especially in healthcare, organizations must protect the troves of sensitive patient information.
For Northwell, Hutchins enacted a three-part strategy: They opened an enterprise platform for self-service analytics, got tech and clinical staffers to bolster clinical reporting (of which there was a clear deficit), and worked with executives to sketch a road map for enterprise data integration.
“The enterprise platform hopefully makes it less chaotic, but it’s definitely not adding to the chaos,” Hutchins said.
Through this effort, he identified three keys to success for healthcare organizations with similar aspirations.
First, define organizational priorities for data access, using these goals to build subject-focused data marts that enable broad inquisition. For example, any number of areas contribute to hospital readmissions or the opioid crisis. As such, corresponding data marts must comprise each aspect rather than revolving around a single component.
Second, “limit the burden of access provisioning” by using existing models in source systems, Hutchins said.
Third, identify business leaders in the healthcare organization who will support the use of platforms, foster coordination, and connect colleagues who are working on similar projects but don’t realize it. This point person doesn’t need to be an executive; Hutchins’ ace is an associate director of neurology who has a passion for analytics. This person should also advocate accountability and governance, ensuring adherence to proper standards.
High-level guidelines are helpful in ensuring that the data are clean, the analytics are trustworthy, and the users are accountable. “We can make the average Excel jockey look like a rock star with these tools,” Hutchins said. “But that doesn’t mean we should trust the data.”
Healthcare organizations and users should provide enough information in the outputs so that people understand where the data are coming from, the role of regulatory guidelines, and more.
Content review standards, meanwhile, help ensure that a report generated from finance will be accurate before it reaches the board. Prior review from stakeholders and subject matter experts is critical prior to publication, a process that requires collaboration between different teams, like clinical and finance.
“It’s a waste of time to develop anything unless you do this because people won’t trust it,” Hutchins noted.
It’s also important to create style guides to breed consistency and branding to clearly identify the source of the content.
Northwell Health framed its number-cruncher team as an analytics resource center. The reason why is simple: Finance and clinical are likely to be wary of a centralized data team and its ability to “master” their individual informatics. Hutchins found that setting up shop as a support arm—highlighting what they have to offer in terms of collaboration, guidelines, and infrastructure—is an effective means to overcome this challenge.
Hutchins and his colleagues also provide training videos and materials to users. They have opened conversations that led to greater standardization. Broadly, they have built an environment in which education and collaboration rule.
“The business units don’t feel threatened, and they think what we’re doing is helpful,” he said.
The thing is, resources can become tight, especially if adoption is high. When Northwell began advertising its analytical capabilities, it generated plenty of buzz, and people wanted access. To manage this growing pain, it’s smart to steer users toward a central tool and cut the list of approved solutions, Hutchins said. Doing so may encourage more efficient systems administration, licensing management, and performance optimization.
And one final piece of advice from Hutchins: Don’t give everyone, particularly novices, access to everything. Sometimes gating data is the best way to prevent botched analyses.
Get the best insights in healthcare analytics directly to your inbox.